THREE PARIETAL CIRCUITS FOR NUMBER PROCESSING
Stanislas Dehaene, Manuela Piazza, Philippe Pinel, and Laurent Cohen
INSERM-CEA, Service Hospitalier Frédéric Joliot, Orsay, France
Did evolution endow the human brain with a predisposition to represent and acquire knowledge about numbers? Although the parietal lobe has been suggested as a potential substrate for a domain-specific representation of quantities, it is also engaged in verbal, spatial, and attentional functions that may contribute to calculation. To clarify the organisation of number-related processes in the parietal lobe, we examine the three-dimensional intersection of fMRI activations during various numerical tasks, and also review the corresponding neuropsychological evidence. On this basis, we propose a tentative tripartite organisation. The horizontal segment of the intraparietal sulcus (HIPS) appears as a plausible candidate for domain specificity: It is systematically activated whenever numbers are manipulated, independently of number notation, and with increasing activation as the task puts greater emphasis on quantity processing. Depending on task demands, we speculate that this core quantity system, analogous to an internal “number line,” can be supplemented by two other circuits. A left angular gyrus area, in connection with other left-hemispheric perisylvian areas, supports the manipulation of numbers in verbal form. Finally, a bilateral posterior superior parietal system supports attentional orientation on the mental number line, just like on any other spatial dimension.
Did evolution endow the human brain with a predisposition to represent dedicated domains of knowledge? We have previously argued that the number domain provides a good candidate for such a biologically determined semantic domain (Dehaene, 1997; Dehaene, Dehaene-Lambertz, & Cohen, 1998a). Three criteria for domain specificity suggest that number and arithmetic are more than cultural inventions, and may have their ultimate roots in brain evolution. First, a capacity to attend to numerosity, and to manipulate it internally in elementary computations, is present in animals even in the absence of training (Hauser, Carey, & Hauser, 2000). Second, a similar capacity for elementary number processing is found early on in human development, prior to schooling or even to the development of language skills (Spelke & Dehaene, 1999; Xu & Spelke, 2000). This suggests that numerical development follows a distinct developmental trajectory based on mechanisms with a long prior evolutionary history.
Third, it has been suggested that number processing rests on a distinct neural circuitry, which can be reproducibly identified in different subjects with various neuroimaging, neuropsychological, and brain stimulation methods (Dehaene et al., 1998a). The present paper focuses on this last issue, taking into account the considerable progress that has recently been made in neuroimaging methods. The involvement of parietal cortex in number processing was initially discovered on the basis of lesion data (Gerstmann, 1940; Hécaen, Angelergues, & Houillier, 1961; Henschen, 1919). Subsequently, a systematic activation of the parietal lobes during calculation, together with precentral and prefrontal cortices, was discovered (Roland & Friberg, 1985) and extensively replicated using positron emission tomography (PET) (Dehaene et al., 1996; Pesenti, Thioux, Seron, & De Volder, 2000; Zago, Pesenti, Mellet, Crivello, Mazoyer, & Tzourio-Mazoyer, 2001) and later fMRI (Burbaud, Camus, Guehl, Bioulac, Caille, & Allard, 1999; Rueckert et al., 1996). On this basis, some of us proposed that the parietal lobe contributes to the representation of numerical quantity on a mental “number line” (Dehaene & Cohen, 1995). Unfortunately, due to poor spatial resolution and limits on experimental designs, those studies did not permit a finer exploration of the regions involved in different kinds of numerical tasks. This has become critical, however, because recent behavioural studies have made it clear that mental arithmetic relies on a highly composite set of processes, many of which are probably not specific to the number domain. For instance, studies of language interference in normal subjects suggest that language-based processes play an important role in exact but not approximate calculation (Spelke & Tsivkin, 2001). Likewise, concurrent performance of a spatial task interferes with subtraction, but not multiplication, while concurrent performance of a language task interferes with multiplication, but not subtraction (Lee & Kang, 2002). Such behavioural dissociations suggest that the neural bases of calculation must be heterogeneous.
The triple-code model of number processing predicts that, depending on the task, three distinct systems of representation may be recruited: a quantity system (a nonverbal semantic representation of the size and distance relations between numbers, which may be category specific), a verbal system (where numerals are represented lexically, phonologically, and syntactically, much like any other type of word), and a visual system (in which numbers can be encoded as strings of Arabic numerals) (Dehaene, 1992; Dehaene & Cohen, 1995). We initially proposed that the parietal activations during number processing reflected solely the contribution of the quantity system. However, it is now clear that this hypothesis requires further elaboration. First, the left perisylvian language network clearly extends into the inferior parietal lobe. Second, the posterior superior parietal lobes are strongly engaged in visual attention processes that may contribute to the visual processing of numbers. It is thus crucial to distinguish, within the observed parietal lobe activations during number processing, which activation sites, if any, are associated with a semantic representation of numerical quantity and which correspond to nonspecific verbal or visual/ attentional systems.
Fortunately, functional magnetic resonance imaging (fMRI) has recently allowed much finergrained studies of the neuroanatomy of number processing, using paradigms adapted from cognitive psychology. The present review focuses entirely on the parietal lobe activations identified by those recent neuroimaging studies. We use three-dimensional visualisation software to investigate how the parietal activations reported by various studies relate to one another in cortical space. On this basis, we propose that three circuits coexist in the parietal lobe and capture most of the observed differences between arithmetic tasks: a bilateral intraparietal system associated with a core quantity system, a region of the left angular gyrus associated with verbal processing of numbers, and a posterior superior parietal system of spatial and nonspatial attention.
It should be emphasised that our description provides only a tentative model. Although it is based on a synthesis of the existing literature, this model remains speculative and will require further validation by direct experimentation. For each postulated circuit, we first examine the relevant neuroimaging literature, and then consider how those brain-imaging results impinge on our understanding of neuropsychological impairments of number processing. Our account predicts that depending on lesion localisation, three different categories of numerical impairments should be observed: genuine semantic impairments of the numerical domain following intraparietal lesions; impairments of verbal fact retrieval following lesions to the left perisylvian cortices, including the left angular gyrus; and impairments of spatial attention on the number line following lesions to the dorsal parietal attention system.
THE BILATERAL HORIZONTAL SEGMENT OF THE
INTRAPARIETAL SULCUS AND QUANTITY PROCESSING Neuroimaging evidence
The horizontal segment of the intraparietal sulcus (hereafter HIPS) is a major site of activation in neuroimaging studies of number processing. As shown in Figure 1a, this region lies at the intersection of the activations observed in many different number processing tasks (see Table 1). What seems to be common to those tasks is the requirement to access a semantic representation of the quantity that the numbers represent. We propose that a nonverbal representation of numerical quantity, perhaps analogous to a spatial map or “number line,” is present in the HIPS of both hemispheres. This representation would underlie our intuition of what a given numerical size means, and of the proximity relations between numbers. In support of this view, several features of its responsiveness to experimental conditions are worth noting.
Mental arithmetic. The HIPS seems to be active whenever an arithmetic operation calls upon a quantitative representation of numbers. For example, it is more active when subjects calculate than when they merely have to read numerical symbols (Burbaud et al., 1999; Chochon, Cohen, Van de Moortele, & Dehaene, 1999; Pesenti et al., 2000), suggesting that it plays a role in the semantic manipulation of numbers. Its activation increases, at least in the right hemisphere, when subjects have to compute two addition or subtraction operations instead of one (Menon, Rivera, White, Glover, & Reiss, 2000). Furthermore, even within calculation, the HIPS is more active when subjects estimate the approximate result of an addition problem than when they compute its exact solution (Dehaene, Spelke, Stanescu, Pinel, & Tsivkin, 1999). Finally, it shows greater activation for subtraction than for multiplication (Chochon et al., 1999; Lee, 2000). Multiplication tables and small exact addition facts can be stored in rote verbal memory, and hence place minimal requirements on quantity manipulation. Contrariwise, although some subtraction problems may be stored in verbal memory, many are not learned by rote and therefore require genuine quantity manipulations. In another study, relative to five different visuospatial and phonological non-numerical tasks, subtraction was the only task that led to increased activation of the HIPS (Simon, Cohen, Mangin, Bihan, & Dehaene,
Number comparison. The HIPS is also active whenever a comparative operation that needs access to a numerical scale is called for. For instance, it is more active when comparing the magnitudes of two numbers than when simply reading them
(Chochon et al., 1999). The systematic contribution of this region to number comparison processes is replicated in many paradigms using tomographic imaging (Le Clec’H et al., 2000; Pesenti et al., 2000; Pinel, Dehaene, Riviere, & LeBihan, 2001; Thioux, Pesenti, Costes, De Volder, & Seron, 2002) as well as scalp recordings of event-related potentials (Dehaene, 1996). Parietal activation in number comparison is often larger in the right than in the left hemisphere (Chochon et al., 1999; Dehaene, 1996; Pinel et al., 2001). This may point to a possible right-hemispheric advantage in comparison and in other tasks requiring an abstraction of numerical relations (Langdon & Warrington, 1997; Rosselli & Ardila, 1989). However, in comparison, the parietal activation, although it may be asymmetric, is always present in both hemispheres, compatible with the observation that numerical comparison is accessible to both hemispheres in split-brain patients (Cohen & Dehaene, 1996; Seymour, Reuter-Lorenz, & Gazzaniga, 1994).
Specificity for the number domain. Several studies have reported greater HIPS activation when processing numbers than when processing other categories of objects on non-numerical scales (such as comparing the ferocity of animals, the relative positions of body parts, or the orientation of two visually presented characters: Le Clec’H et al., 2000; Pesenti et al., 2000; Thioux et al., 2002). Event-related potentials have also revealed greater parietal activation for numbers than for other categories of words such as action verbs, names of animals, or names of famous persons (Dehaene, 1995). In this study, the first point in time in which category-specific semantic effects emerge during visual word processing was found to be 250–280 ms following stimulus onset.
Figure 1. Regions of overlapping activity for three groups of studies, superimposed on axial and sagittal slices of a normalised single-subject anatomical image. The overlap was calculated by averaging binarised contrast images indicating which voxels were significant for a given contrast (studies and contrasts are listed in Table 1). The colour scale indicates the percentage of studies showing activation in a given voxel. The same colour scale (from 22% to 50% of overlap) is applied to all images. Although no single voxel was shared by 100% of studies in a group, probably due to variability across groups of subjects, laboratories, and imaging methods, Table 1 revealed a high consistency of activations. (A) The horizontal segment of the intraparietal sulcus (HIPS) was activated bilaterally in a variety of contrasts sharing a component of numerical quantity manipulation. The barycentre of the region of maximum overlap (>50%) was at Talairach Coordinates (TC) 41, –42, 49 in the left hemisphere, and –48, –41, 43 in the right hemisphere. Activation overlap is also visible in the precentral gyrus. (B) The angular gyrus (AG) was activated with a strong left lateralisation (TC –48, –59, 30) in 5 studies of arithmetic tasks with a strong verbal component. Posterior cingulate as well as superior frontal regions also show some degrees of overlap. (C) The posterior superior parietal lobule (PSPL) was activated bilaterally in a few numerical tasks (left and right barycentres at TC –26, –69, 61 and 12, –69, 61; and see Table 1). To emphasise the nonspecificity of this region, the image shows the intersection of the overlap between four numerical tasks with an image of posterior parietal activity during a non-numerical visual attention shift task (Simon et al., 2002).
One study directly tested the specificity of the HIPS for the numerical domain in multiple tasks (Thioux et al., 2002). Subjects were presented with number words and names of animals matched for length. The HIPS showed greater activation, bilaterally, to numbers than to animal names. This was true whether subjects were engaged in a comparison task (larger or smaller than 5; more or less ferocious than a dog), a categorisation task (odd or even; mammal or bird), or even a visual judgement of character shape. Thus, the HIPS shows category specificity independently of task context. Further research will be needed, however, to decide whether it is strictly specific for numbers or whether it extends to other categories that have a strong spatial or serial component (e.g., the alphabet, days, months, spatial prepositions, etc.).
Parametric modulation. Parametric studies have revealed that the activation of the HIPS is modulated by semantic parameters such as the absolute magnitude of the numbers and their value relative to a reference point. Thus, intraparietal activity is larger and lasts longer during operations with large numbers than with small numbers (Kiefer & Dehaene, 1997; Stanescu-Cosson, Pinel, Van de Moontele, Le Bihan, Cohen, & Delaene, 2000). It is also modulated by the numerical distance separating the numbers in a comparison task (Dehaene, 1996; Pinel et al., 2001). On the other hand, the activation of the HIPS is independent of the particular modality of input used to convey the numbers. Arabic numerals, spelled-out number words, and even nonsymbolic stimuli like sets of dots or tones can activate this region if subjects attend to the corresponding number (Le Clec’H et al., 2000; Piazza, Mechelli, Butterworth, & Price, 2002a; Piazza, Mechelli, Price, & Butterworth, 2002b; Pinel et al., 2001). In one study, subjects attended either to the numerosity or to the physical characteristics (colour, pitch) of series of auditory and visual events. The right HIPS was active whenever the subjects attended to number, regardless of the modality of the stimuli (Piazza et al., 2002b). In another study, the activation of the bilateral HIPS was found to correlate directly with the numerical distance between two numbers in a comparison task, and this effect was observed whether the numbers were presented as words or as digits (Pinel et al., 2001). Those parametric studies are all consistent with the hypothesis that the HIPS codes the abstract quantity meaning of numbers rather the numerical symbols themselves.
Unconscious quantity processing. Quantity processing and HIPS activation can be demonstrated even when the subject is not aware of having seen a number symbol (Dehaene et al., 1998b; Naccache & Dehaene, 2001). In this experiment, subjects were asked to compare target numbers to a fixed reference of 5. Unbeknownst to them, just prior to the target, another number, the prime, was briefly present in a subliminal manner. FMRI revealed that the left and right intraparietal regions were sensitive to the unconscious repetition of the same number. When the prime and target corresponded to the same quantity (possibly in two different notations, such as ONE and 1), less parietal activation was observed than when the prime and target corresponded to two distinct quantities (e.g., FOUR and 1). This result suggests that this region comprises distinct neural assemblies for different numerical quantities, so that more activation can be observed when two such neural assemblies are activated than when only one is. It also indicates that this region can contribute to number processing in a subliminal fashion.
Taken together, these data suggest that the HIPS is essential for the semantic representation of numbers as quantities. This representation may provide a foundation for our “numerical intuition,” our immediate and often unconscious understanding of where a given quantity falls with respect to others, and whether or not it is appropriate to a given context (Dehaene, 1992, 1997; Dehaene & Marques, 2002).
Neuropsychological evidence Neuropsychological observations confirm the existence of a distinct semantic system for numerical quantities and its relation to the vicinity of the intraparietal sulcus. Several single-case studies indicate that numbers doubly dissociate from other categories of words at the semantic level. On the one hand, spared calculation and number comprehension abilities have been described in patients with grossly deteriorated semantic processing (Thioux, Pillon, Samson, De Partz, Noel, & Seron, 1998) or semantic dementia (Butterworth, Cappelletti, & Kopelman, 2001; Cappelletti, Butterworth, & Kopelman, 2001). In both cases, the lesions broadly affected the left temporo-frontal cortices while sparing the intraparietal regions. On the other hand, Cipolotti, Butterworth, and Denes (1991) reported a striking case of a patient with a small left parietal lesion and an almost complete deficit in all spheres of number processing, sparing only the numbers 1 through 4, in the context of otherwise largely preserved language and semantic functions. Although such a severe and isolated degradation of the number system has never been replicated, other cases confirm that the understanding of numbers and their relations can be specifically impaired in the context of preserved language and semantics (e.g., Dehaene & Cohen, 1997; Delazer & Benke, 1997).
In many cases, the deficit can be extremely incapacitating. Patients may fail to compute operations as simple as 2 + 2, 3 – 1, or 3 × 9. Several characteristics indicate that the deficit arises at an abstract, notation-independent level of processing. First, patients may remain fully able to comprehend and to produce numbers in all formats. Second, they show the same calculation difficulties whether the problem is presented to them visually or auditorily, and whether they have to respond verbally or in writing, or even merely have to decide whether a proposed operation is true or false. Thus, the calculation deficit is not due to an inability to identify the numbers or to produce the operation result. Third, the deficit often extends to tasks outside of calculation per se, such as comparison or bisection. For instance, patient MAR (Dehaene & Cohen, 1997) showed a mild impairment in deciding which of two numbers is the larger (16% errors), and was almost totally unable to decide what number falls in the middle of two others (bisection task: 77% errors). He easily performed analogous comparison and bisection tasks in non-numerical domains such as days of the week, months, or the alphabet (What is between Tuesday and Thursday? February and April? B and D?). This type of deficits seems best described as a category-specific impairment of the semantic representation and manipulation of numerical quantities (Dehaene & Cohen, 1997), rather than with the mere clinical label of
In such patients, calculation impairments often co-occur with other deficits, forming a cluster of deficits called Gerstmann’s syndrome (Benton, 1992; Gerstmann, 1940), which comprises agraphia, finger agnosia, and left–right distinction difficulties (to which one may often add constructive apraxia). The lesions that cause acalculia of the Gerstmann’s type are typically centred in the depth of the left intraparietal sulcus (Mayer, Martory, Pegna, Landis, Delavelle, & Annoni, 1999; Takayama, Sugishita, Akiguchi, & Kimura, 1994). This is compatible with the above brain-imaging results showing intraparietal activation during various numerical manipulation tasks independently of language. Results from a recent brain-imaging study (Simon et al., 2002) shed some light on why the various elements of Gerstmann’s syndrome often co-occur following left intraparietal lesions. In this study, fMRI was used to compare, in the same subjects, the localisation of parietal activations during a number subtraction task with those observed during various tasks that also involve the parietal lobe, such as eye or attention movements, finger pointing, hand grasping, and a language task of phoneme detection. The results revealed a systematic topographical organisation of activations and their intersections. In particular, the intraparietal sulcus appears to contains a “four-corners” region in which four areas of activation are juxtaposed: calculation only, calculation and language, manual tasks only, and an area activated during the four visuospatial tasks (eye and attention movements, pointing, and grasping). The simultaneous lesion of those four areas would predictably result in joint impairments of calculation, word processing (possibly including agraphia), finger knowledge and movement, and high-level spatial reference (possibly including understanding of left– right coordinates). Such a joint lesion might be frequent because this cortical territory is jointly irrigated by a branch of the middle cerebral artery, the angular gyrus artery. Inter-individual variability in the boundaries between cortical territories as well as in the branching patterns of this artery would explain that the different elements of Gerstmann’s syndrome can be dissociated (Benton, 1961, 1992). Note that this interpretation implies that, contrary to a frequent speculation, Gerstmann’s syndrome does not result from a homogeneous impairment to a single representation that would somehow intermingle fingers, numbers, and space (Butterworth, 1999; Gerstmann, 1940; Mayer et al., 1999). Rather, the syndrome may represent a happenstance conjunction of distinct, but dissociable, deficits that frequently co-occur due to a common vascularisation, and that are only loosely connected at the functional level due to the overarching spatial and sensorimotor functions of the parietal lobe.
THE LEFT ANGULAR GYRUS AND
VERBAL NUMBER MANIPULATIONS
Neuroimaging evidence The left angular gyrus (hereafter AG) is also often activated in neuroimaging studies of number processing (see Figure 1b and Table 1). This region is left-lateralised and located posterior and inferior to the HIPS (see Figure 2 for their respective locations). A closer look at the types of numerical tasks that activate this region, detailed below, reveals that its functional properties are very different from the properties of the HIPS. The left AG does not seem to be concerned with quantity processing, but shows increasingly greater activation as the task puts greater requirement on verbal processing. We therefore propose that this region is part of the language system, and contributes to number processing only inasmuch as some arithmetic operations, such as multiplication, make particularly strong demands on a verbal coding of numbers.
Figure 2. Three-dimensional representation of the parietal regions of interest. For better visualisation, the clusters show all parietal voxels activated in at least 40% of studies in a given group.
In support of this hypothesis, the left AG is not merely involved in calculation, but in different types of language-mediated processes such as reading or verbal short-term memory tasks (for reviews, see Fiez & Petersen, 1998; Paulesu, Frith, & Frackowiak, 1993; Price, 1998). In Simon et al.’s (2002) fMRI study of six different tasks, the left angular gyrus was the only parietal site where there was overlapping activity for calculation and phoneme detection, but no activation during the other four visuospatial tasks. This clearly indicates that the left AG is not specific for calculation, but jointly recruited by language and calculation processes.
Even within calculation, several studies indicate a modulation of AG activation in direct proportion to the verbal requirements of the task. First, the AG is more active in exact calculation than in approximation (Dehaene et al., 1999). This fits with behavioural data indicating that exact arithmetic facts are stored in a language-specific format in bilinguals, while approximate knowledge is language-independent and shows the classical numerical distance effect associated with the nonverbal quantity system (Xu & Spelke, 2000). Second, within exact calculation, the left AG shows greater activation for operations that require access to a rote verbal memory of arithmetic facts, such as multiplication, than for operations that are not stored and require some form of quantity manipulation. For instance, the left AG shows increased activation for multiplication relative to both subtraction and number comparison (Chochon et al., 1999; Lee, 2000), for multiplication and division relative to a letter substitution control (Gruber, Indefrey, Steinmetz, & Kleinschmidt, 2001), and for multidigit mulplication relative to a digit-matching control (Fulbright, Molfese, Stevens, Skudlarski, Lacadie, & Gore, 2000).
Even within a given operation, such as single digit addition, the left angular gyrus is more active for small problems with a sum below 10 than for large problems with a sum above 10 (StanescuCosson, Pinel, Van de Moortele, Le Bihan, Cohen, & Dehaene, 2000). This probably reflects the fact that small addition facts, just like multiplication tables, are stored in rote verbal memory, while behavioural evidence indicates that larger addition problems are often solved by resorting to various semantic elaboration strategies (Dehaene & Cohen, 1995; Lefevre, 1996).
In summary, the contribution of the left angular gyrus in number processing may be related to the linguistic basis of arithmetical computations. Its contribution seems essential for the retrieval of facts stored in verbal memory, but not for other numerical tasks (like subtraction, number comparison, or complex calculation) that call for a genuinely quantitative representation of numbers and relate more to the intraparietal sulcus.
Neuropsychological evidence: Dissociations between operations
The finding that the intraparietal sulcus and the angular gyrus exhibit functionally differentiated properties can shed light on the neuropsychology of acalculia. One of the most striking findings is the occurrence of sharp dissociations between arithmetic operations. It is not rare for a patient to be much more severely impaired in multiplication than in subtraction (Cohen & Dehaene, 2000; Dagenbach & McCloskey, 1992; Dehaene & Cohen, 1997; Lampl, Eshel, Gilad, & Sarova-Pinhas, 1994; Lee, 2000; Pesenti, Seron, & Van der Linden, 1994; Van Harskamp & Cipolotti, 2001), while other patients are much more impaired in subtraction than in multiplication (Dehaene & Cohen, 1997; Delazer & Benke, 1997; Van Harskamp &
Cipolotti, 2001). Some have proposed that such dissociations reflect random impairments in a system with distinct stores of arithmetic facts for each operation (Dagenbach & McCloskey, 1992). Here, however, we would like to show that there is much more systematicity behind those observations. Our views suggest that dissociations between operations reflect a single, basic distinction between overlearned arithmetic facts such as the multiplication table, which are stored in rote verbal memory, and the genuine understanding of number meaning that underlies nontable operations such as subtraction (Dehaene & Cohen, 1997; Delazer & Benke, 1997; Hittmair-Delazer, Sailer, & Benke, 1995). According to this interpretation, multiplication requires the integrity of language-based representations of numbers, because multiplication facts are typically learned by rote verbal memorisation. Subtraction, on the other hand, is typically not learned by rote. Although the mechanisms by which simple subtraction problems are resolved are not yet understood, it is likely that some form of internal manipulation of nonverbal quantities on the internal number line is involved, as attested by the fact that very simple subtractions are accessible to preverbal infants (Wynn, 1992) and nonhuman primates (Hauser et al., 2000).
Support for this view comes from several lines of research. First, as noted earlier, imaging studies in normals confirm that distinct sites of activations underlie performance in simple multiplication and subtraction (Chochon et al., 1999; Cohen, Dehaene, Chochon, Lehéricy, & Naccache, 2000; Lee, 2000). Second, all patients in whom subtraction was more impaired than subtraction had left parietal lesions and/or atrophy, most often accompanied by Gerstmann’s syndrome, compatible with an impairment to the left HIPS and to the semantic representation of numerical quantities (Dehaene & Cohen, 1997; Delazer & Benke, 1997; Van Harskamp & Cipolotti, 2001). Conversely, although this is not always thoroughly documented, patients in whom multiplication is more impaired than subtraction typically have associated aphasia (e.g., Cohen et al., 2000; Dehaene & Cohen, 1997). Furthermore, the lesions often spare the intraparietal cortex and can affect multiple regions known to be engaged in language processing, such as the left perisylvian cortices including the inferior parietal lobule (Cohen et al., 2000), the left parieto-temporal carrefour (Lampl et al., 1994), or the left basal ganglia (Dehaene & Cohen, 1997).
Multiplication impairments with spared subtraction have also been reported in two patients with reading deficits in whom the lesion affected access to the language system from visual symbols (Cohen & Dehaene, 2000; McNeil & Warrington, 1994). Amazingly, one of those patients was able to subtract better than she could read the same problems (Cohen & Dehaene, 2000). This confirms the relative independence of subtraction, but not multiplication, from the language system.
Perhaps the best evidence for a dissociation between quantity processing in the HIPS and verbal number processing in the left AG comes from two studies of the temporary calculation impairments caused by electrical brain stimulation. In one patient with strips of subdural electrodes arranged over the left parietal, superior temporal, and posterior frontal regions, a single electrode site was found whose stimulation systematically disrupted multiplication performance much more than addition performance (27% vs. 87% correct; subtraction was not tested; Whalen, McCloskey, Lesser, & Gordon, 1997). Although limited information is available on localisation, this electrode was located in the left inferior parietal region, apparently close to the angular gyrus. Interestingly, multiplication performance was worse when the responses were given orally (27% correct) than when they were typed with a key pad (64% correct), suggesting that stimulation also interfered with the verbal coding of numbers.
A second case presented a double dissociation between subtraction and multiplication (Duffau et al., 2002). Cortical stimulation was performed intra-operatively during the resection of a parietooccipital glioma. Two neighbouring sites were found within the left parietal lobe. The first, located within the angular gyrus proper (approximate Talairach coordinates –50, –60, +30), disrupted multiplication but not subtraction when stimulated. The second, located more superiorily and anteriorily within the intraparietal sulcus (TC –45, –55, +40), disrupted subtraction but not multiplication. An intermediate location was also found where stimulation disrupted both operations. The reported coordinates, although imprecise given the distortions possibly induced by the glioma and the surgery, are completely compatible with the dissociated areas of activation observed in functional brain imaging (Chochon et al., 1999; Lee, 2000).
To close on the issue of dissociations between operations, we briefly consider the case of addition (see also Cohen & Dehaene, 2000). Addition is complex because it can be solved in at least two ways. It is similar to multiplication in that many people have memorised most of the basic addition table (single digit addition facts with a sum below 10). However, addition is also similar to subtraction in that simple addition problems can also be solved by quantity manipulation strategies, something that would be utterly impractical with multiplication. Thus, addition performance is hard to predict. Indeed, in our experience, it varies considerably across patients or even within patients, depending on the strategy that they adopt. The only clear prediction from our model is that addition performance cannot dissociate from both subtraction and multiplication together. That is to say, a patient cannot be impaired in addition, but not in subtraction nor in multiplication (since the latter would imply that both the verbal and the quantity circuits are intact); nor can a patient show preserved addition with impaired subtraction and multiplication (since the latter would imply that both systems are impaired).
If dissociations between operations followed a chance pattern, this prediction should be violated in about one third of cases. In fact, however, it is confirmed by essentially all patients to date (10 out of 11 patients: Cohen & Dehaene, 2000; Dagenbach & McCloskey, 1992; Dehaene & Cohen, 1997; Delazer & Benke, 1997; Lampl et al., 1994; Lee, 2000; Pesenti et al., 1994; Van Harskamp & Cipolotti, 2001). The only exception (patient FS, Van Harskamp & Cipolotti, 2001) is worth discussing. Overall, this patient was 96.3% correct (156/162) in single-digit subtraction and multiplication, but only 61.7% correct (100/162) in singledigit addition, thus superficially qualifying as a straightforward violation of our hypothesis. However, the pattern of errors in this patient was quite different from other cases of acalculia; 87% of his addition errors consisted of selecting the wrong operation (he almost always solved the corresponding multiplication problem, e.g., 3 + 3 = 9). This is very different from the other two patients reported in the same paper: patient DT, who was impaired in subtraction, made only 12.5% operation errors, and patient VP, who was impaired in multiplication, only 3.5%.
In a reanalysis, we excluded patient FS’s operation errors and analysed only the remaining trials, in which he was presumably really attempting to add the operands. In this way, we can estimate patient FS’s conditional success rate in addition, given that he is really trying to add. This success rate is 92.6% correct (100/108), a value which does not differ from the performance observed in the other two operations (96.3% correct). Thus, it can be argued that patient FS experiences little difficulty with arithmetic operations per se, but exhibits a selective deficit in choosing the appropriate operation. Exactly how subjects transform the task instructions and operation signs into the selection of an appropriate information-processing circuit is left largely unspecified in current models. Nevertheless, deficits affecting this task setting level should be kept conceptually distinct from the genuine impairments in arithmetical computation itself.
In summary, a review of neuropsychological dissociations between arithmetic operations indicates that it is not necessary to postulate as many brain circuits as there are arithmetical operations (Dagenbach & McCloskey, 1992). Rather, most if not all cases so far can be accommodated by the postulated dissociation between a quantity circuit (supporting subtraction and other quantity manipulation operations) and a verbal circuit (supporting multiplication and other rote memory based operations).
THE POSTERIOR SUPERIOR PARIETAL SYSTEM AND
ATTENTIONAL PROCESSES Neuroimaging evidence
A third region, observed bilaterally in the posterior superior parietal lobule (hereafter PSPL), with a frequent mesial extension into the precuneus, is also active in several tasks requiring number manipulations. This region is posterior to the HIPS, and occupies a location superior and mesial to the AG in the superior parietal lobule (see Figure 1c and Figure 2). It is active during number comparison (Pesenti et al., 2000; Pinel et al., 2001), approximation (Dehaene et al., 1999), subtraction of two digits (Lee, 2000), and counting (Piazza et al., 2002a). It also appears to increase in activation when subjects carry out two operations instead of one (Menon et al., 2000). However, this region is clearly not specific to the number domain. Rather, it also plays a central role in a variety of visuospatial tasks including hand reaching, grasping, eye and/or attention orienting, mental rotation, and spatial working memory (Corbetta, Kincade, Ollinger, McAvoy, &
Shulman, 2000; Culham & Kanwisher, 2001; Simon et al., 2002). For example, Wojciulik and Kanwisher (1999) have observed overlapping activations in this region in three tasks that all shared a component of attention-orienting. Similarly, Simon et al. (2002) observed that this region was activated during eye movement, attention movements, grasping, and pointing.
The contribution of this region to spatial attention and/or eye orienting probably explains its activation during counting, where subjects are sequentially attending to the enumerated objects. However, spatial attention does not seem to explain its activation during purely numerical operations of comparison, approximation, or subtraction. In all of those tasks, number-related activation in the PSPL was observed relative to a control that used the same spatial distribution of stimuli on screen, as well as a very similar motor response.
Obviously, any reconciliation of those sparse and disparate data set must remain tentative. The hypothesis that we would like to propose is that this region, in addition to being involved in attention orienting in space, can also contribute to attentional selection on other mental dimensions that are analogous to space, such as time (Coull & Nobre, 1998; Wojciulik & Kanwisher, 1999) or number. Psychological experiments indicate that the core semantic representation of numerical quantity can be likened to an internal “number line,” a quasispatial representation on which numbers are organised by their proximity (Dehaene, Bossini, & Giraux, 1993; Moyer & Landauer, 1967). It is then conceivable that the same process of covert attention that operates to select locations in space can also be engaged when attending to specific quantities on the number line. Such number-based attention would be particularly needed in tasks that call for the selection of one amongst several quantities, for instance when deciding which of two quantities is the larger (Pesenti et al., 2000; Pinel et al., 2001), or which of two numbers approximately fits an addition problem (Dehaene et al., 1999).
Neuropsychological evidence: Joint impairments of attention and number processing
Only a few neuropsychological and brain stimulation findings provide some support for our admittedly speculative theory. In a recent study using transcranial magnetic stimulation with normal subjects, Gobel, Walsh, and Rushworth (2001) first located left and right dorsal posterior parietal sites where stimulation interfered with performance in a visual serial search task. The coordinates of those regions correspond to those of the bilateral posterior parietal regions found active in neuroimaging studies of eye and attention orienting (Corbetta et al., 2000; Simon et al., 2002; Wojciulik & Kanwisher, 1999). They then tested the effect of magnetic stimulation at those locations on a twodigit number comparison task. On stimulated trials, comparison performance was significantly slower. Interestingly, the numerical distance effect itself was still present and relatively unchanged (although stimulation on the left tended to interfere more with numbers close to the reference, particularly those that were larger than the reference). This suggests that the stimulation did not directly interfere with a core representation of numerical quantity, but rather with the response decision process itself. At the very least, this experiment confirms that spatial attention orienting and numerical comparison both engage this parietal region, thus confirming previous brain-imaging evidence (Pinel et al., 2001).
Further support for a close interplay between the representations of space and numbers is provided by a study with unilateral neglect patients (Zorzi, Priftis, & Umiltà, 2002). It is a well-known, indeed almost a defining feature of those patients that they perform poorly in spatial bisection tests. When asked to locate the middle of a line segment, neglect patients with right parietal lesions tend to indicate a location further to the right, consistent with their failure to attend to the left side of space. Zorzi et al. tested their performance in a numerical bisection task, where they were asked to find the middle of two orally presented numbers. Strikingly, patients erred systematically, often selecting a number far larger than the correct answer (e.g., Q: What number falls in between 11 and 19? A: 17). This suggests that spatial attention can be oriented on the left-toright oriented number line, and that this attentionorienting process contributes to the resolution of simple arithmetic problems such as the bisection test. Interestingly, these patients were said not to be acalculic and did not show any deficit in other numerical tasks such as simple arithmetic fact retrieval. Indeed, Vuilleumier and Rafal (1999) demonstrated, on a different group of patients with neglect, that a posterior parietal lesion does not impair the mere quantification of small number of items. Neglect patients were able to estimate numerosity with sets of up to four objects even when some of enumerated items fell in the neglected field. Again, this suggests that attentional and numerical systems are dissociable. However, Zorzi et al.’s finding of “representational neglect” on the numerical continuum indicates that spatial attention processes do contribute to some numerical tasks.
AND THE ONTOGENY OF NUMBER
Whether or not our functional characterisation of three parietal subsystems is correct, it is an anatomical fact that those activations sites are strikingly reproducible. It is remarkable that the HIPS, AG, and PSPL are systematically activated in different subjects, often from different countries, with different educational strategies and achievements in mathematics (Stevenson & Stigler, 1992), and with a diversity of linguistic schemes for expressing number (Hurford, 1987). Even the fine dissociation between subtraction and multiplication is reproducible with French vs. Korean subjects (Cohen et al., 2000; Lee, 2000). Such systematicity in the anatomical organisation of parietal numerical processes must be reconciled with the obvious fact that arithmetic is, in part, a recent cultural invention.
Our hypothesis is that the cultural construction of arithmetic is made possible by pre-existing cerebral circuits that are biologically determined and are adequate to support specific subcomponents of number processing (Dehaene, 1997). This hypothesis supposes an initial prespecialisation of the brain circuits that will ultimately support high-level arithmetic in adults. It implies that it should be possible to identify precursors of those circuits in infancy and childhood. Indeed, quantity processing is present at a very young age. Infants in their first year of life can discriminate collections based on their numerosity (Dehaene et al., 1998a; Starkey & Cooper, 1980; Wynn, 1992), even when the numbers are as large as 8 vs. 16 (Xu & Spelke, 2000). Although no brain-imaging evidence is available in infants yet, we speculate that this early numerical ability may be supported by a quantity representation similar to adults’ (Dehaene, 1997; Spelke & Dehaene, 1999). This representation would serve as a foundation for the construction of higher-order arithmetical and mathematical concepts.
The hypothesis of an early emergence of quantity, verbal, and attentional systems leads to several predictions concerning normal and impaired number development:
Brain activation in infancy and childhood. A precursor of the HIPS region should be active in infants and young children during numerosity manipulation tasks. At present, this prediction has only been tested with 5-year-old children in a number comparison task (E. Temple & Posner, 1998). Eventrelated potentials revealed the scalp signature of a numerical distance effect, with a topography similar to adults, common to numbers presented as Arabic numerals or as sets of dots. There is a clear need to extend those data to an earlier age and with a greater anatomical accuracy.
Developmental dyscalculia and the parietal lobe. Deficits of number processing should be observed in case of early left parietal injury or disorganisation. Developmental dyscalculia is relatively frequent, affecting 3–6% of children (Badian, 1983; Kosc, 1974; Lewis, Hitch, & Walker, 1994). We predict that a fraction of those children may suffer from a core conceptual deficit in the numerical domain. Indeed, a “developmental Gerstmann syndrome” has been reported (Benson & Geschwind, 1970; Kinsbourne & Warrington, 1963; Spellacy & Peter, 1978; C. M. Temple, 1989, 1991). In those children, dyscalculia is accompanied by most or all of the following symptoms: dysgraphia, left–right disorientation, and finger agnosia, which suggest a neurological involvement of the parietal lobe. Interestingly, even in a sample of 200 normal children, a test of finger knowledge appears to be a better predictor of later arithmetic abilities than is a test of general intelligence (Fayol, Barrouillet, & Marinthe, 1998).
Two recent reports directly relate developmental dyscalculia to an underlying left parietal disorganisation. Levy, Reis, and Grafman (1999) report the case of an adult with lifelong isolated dyscalculia together with superior intelligence and reading ability, in whom the standard anatomical MRI appeared normal, yet MR spectroscopy techniques revealed a metabolic abnormality in the left inferior parietal area. Similarly, Isaacs, Edmonds, Lucas, and Gadian (2001) used voxel-based morphometry to compare gray matter density in adolescents born at equally severe grades of prematurity, half of whom suffered from dyscalculia. They found a single region of reduced gray matter in the left intraparietal sulcus. The Talairach coordinates of this region (–39, –39, +45) are quite close to the coordinates of the HIPS.
Subtypes of developmental dyscalculia. As in adult acalculia, at least two subtypes of developmental dyscalculia should be observed, and those should be traceable to a differential impairment of quantity vs. language processing circuits. Although several distinctions between subtypes of developmental dyscalculia have been proposed (e.g., Ashcraft, Yamashita, & Aram, 1992; Geary, Hamson, & Hoard, 2000; Rourke & Conway, 1997; C. M. Temple, 1991), most are based on group studies and standardised batteries of tests, which are inappropriate for testing the predicted subtle distinctions between, e.g., subtraction and multiplication. One exception is the single-case study of patient HM (C. M. Temple, 1991), who suffered from developmental phonological dyslexia. His deficit in arithmetic was mostly limited to multiplication facts, while he experienced no difficulty in solving simple addition and subtraction problems with numbers of the same size. Our view predicts that the association of verbal and multiplication impairments observed in this study should be generalisable. Multiplication deficits should be present in cases of dyscalculia accompanied by dysphasia and/or dyslexia, while subtraction and quantity-manipulation deficits should be present in patients with dyscalculia but without any accompanying dyslexia or language retardation. Although this proposal remains largely untested, Geary et al. (2000) do report interesting differences between developmental dyscalculics with or without associated dyslexia. When faced with the same simple addition problems, nondyslexics tend to use fact retrieval much more often than do dyslexics, who rather use finger-counting strategies. This is consistent with the hypothesis that an impairment of rote verbal memory is partially responsible for dyscalculia in children with dyslexia.
Genetics of developmental dyscalculia. If the biological predisposition view is correct, specific combinations of genes should be involved in setting up the internal organisation of the parietal lobe and, in particular, the distinction between quantity and language circuits. Thus, it should be possible to identify dyscalculias of genetic origin. The available data, indeed, indicate that when a child is dyscalculic, other family members are also frequently affected, suggesting that genetic factors may contribute to the disorder (Shalev et al., 2001). Although the search for dyscalculias of genetic origin has only very recently begun, the possibility that Turner syndrome may conform to this typology has recently attracted attention. Turner syndrome is a genetic disorder characterised by partial or complete absence of one X chromosome in a female individual. The disorder occurs in approximately 1 girl in 2000 and is associated with well-documented physical disorders and abnormal oestrogen production and pubertal development. The cognitive profile includes deficits in visual memory, visual-spatial and attentional tasks, and social relations, in the context of a normal verbal IQ (Rovet, 1993). Most interestingly in the present context is the documentation of a mild to severe deficit in mathematics, particularly clear in arithmetic (Mazzocco, 1998; Rovet, Szekely, & Hockenberry, 1994; C. M. Temple & Marriott, 1998).
Anatomically, the data suggest possible bilateral parieto-occipital dysfunction in Turner syndrome. A positron emission tomography study of five adult women demonstrated a glucose hypometabolism in bilateral parietal and occipital regions (Clark, Klonoff, & Hadyen, 1990). Two anatomical MR studies, one with 18 and the other with 30 affected women, demonstrated bilateral reductions in parieto-occipital brain volume, together with other subcortical regions (Murphy et al., 1993; see also Reiss et al., 1993; Reiss, Mazzocco, Greenlaw, Freund, & Ross, 1995). Interestingly, the phenotype of Turner syndrome can differ depending on whether the remaining X chromosome is of paternal or maternal origin (Xm or Xp subtypes; Bishop, Canning, Elgar, Morris, Jacobs, & Skuse, 2000; Skuse, 2000; Skuse et al., 1997). Such a genomic imprinting effect was first demonstrated on tests of social competence (Skuse et al., 1997). It will be interesting to see if a similar effect exists in the arithmetic domain.
We have reviewed the evidence for a subdivision of calculation-related processes in the parietal lobe. A broader discussion of the specificity of the number processing system should also consider the satellite systems that serve as input and outputs to calculation processes. At the visual identification level, pure alexic patients who fail to read words often show a largely preserved ability to read and process digits (Cohen & Dehaene, 1995; Déjerine, 1891, 1892). Conversely, a case of impaired number reading with preserved word reading is on record (Cipolotti, Warrington, & Butterworth, 1995). In the writing domain, severe agraphia and alexia may be accompanied by a fully preserved ability to write and read Arabic numbers (Anderson, Damasio, & Damasio, 1990). Even within the speech production system, patients who suffer from random phoneme substitutions, thus resulting in the production of an incomprehensible jargon, may produce jargon-free number words (Cohen,
Verstichel, & Dehaene, 1997). These dissociations, however, need not imply a distinct semantic system for number. Rather, they can probably be explained by considering that the particular syntax of number words and the peculiarities of the positional notation for Arabic numeral place special demands on visual recognition, speech production, and writing systems.
Even within the parietal lobe, our review of number-related activations suggests that much of the human capacity for number processing relies on representations and processes that are not specific to the number domain. At least two of the parietal circuits that we have described, the posterior superior parietal attention system and the left angular verbal system, are thought to be associated with broader functions than mere calculation. The third circuit, in the bilateral horizontal intraparietal region (HIPS), is a more plausible candidate for domain specificity. As reviewed above, it is systematically activated during mental arithmetic; it is more activated by number words than by other words such as names of animals; and its activation increases with the amount or duration of quantity manipulation required, but is completely independent of the notation used for numbers. Still, we are reluctant to use the term “category-specific” for this brain region, and prefer the terms “core quantity system” or “number-essential” region instead. For a purely empirical point of view, deciding whether a given region is “specific” for numbers seems an extremely difficulty enterprise. Testing for specificity would seem to require a systematic comparison of the target category (e.g., number) against a potentially infinite list of alternatives. It is also complicated by the limited resolution of brainimaging techniques, which cannot yet resolve the fine-grained neuronal and columnar organisation of human cortex. Comparison of group studies, as was done here, may overestimate the amount of overlap between tasks. Studies of multiple tasks within the same subjects will be required to examine whether (1) the very same voxels can be activated by multiple quantity-related paradigms, and (2) those voxels cannot be activated by any other nonnumerical operation. Because such studies are lacking (although see Simon et al., 2002), it is still premature to conclude for or against categoryspecificity in number semantics.
These data show that Dam Salmonella survive in Peyer’s patches of the mouse small intestine for at least 5 days, providing an opportunity for elicitation of a host immune response. Dam– Salmonella, however, were unable to cause disease; they either were unable to invade systemic tissues or were able to invade but could not survive.
DNA adenine methylases are potentially excellent targets for both vaccines and antimicrobials. They are highly conserved in many pathogenic bacteria that cause significant morbidity and mortality, such as Vibrio cholerae (21), Salmonella typhi (22), pathogenic E. coli (23), Yersinia pestis (22), Haemophilus influenzae (24), and Treponema pallidum (25). In addition, because Dam is a global regulator of genes expressed during infection (Fig. 1), Dam– mutants may ectopically express multiple immunogens that are processed and presented to the immune system. Such ectopic expression could elicit a cross-protective immune response between related bacterial strains that share common epitopes. Finally, because the Dam methylase is essential for bacterial virulence, Dam inhibitors are likely to have broad antimicrobial action, hence Dam is a promising target for antimicrobial drug development.
Sources of Mathematical
Thinking: Behavioral and Brain-Imaging Evidence
S. Dehaene,1* E. Spelke,2 P. Pinel,1 R. Stanescu,1 S. Tsivkin2
Does the human capacity for mathematical intuition depend on linguistic competence or on visuo-spatial representations? A series of behavioral and brain-imaging experiments provides evidence for both sources. Exact arithmetic is acquired in a language-specific format, transfers poorly to a different language or to novel facts, and recruits networks involved in word-association processes. In contrast, approximate arithmetic shows language independence, relies on a sense of numerical magnitudes, and recruits bilateral areas of the parietal lobes involved in visuo-spatial processing. Mathematical intuition may emerge from the interplay of these brain systems.
Will it ever happen that mathematicians will know enough about the physiology of the brain, and neurophysiologists enough of mathematical discovery, for efficient cooperation to be possible?
[Jacques Hadamard (1)]
Until recently, the only source of information about the mental representations used in mathematics was the introspection of mathematicians. Eloquent support for the view that mathematics relies on visuo-spatial rather than linguistic processes came from Albert Einstein, who stated: “Words and language, whether written or spoken, do not seem to play any part in my thought processes. The psychological entities that serve as building blocks for my thought are certain signs or images, more or less clear, that I can reproduce and recombine at will” (2). Many mathematicians report similar experiences (1, 3), but some have stressed the crucial role played by language and other formal symbol systems in mathematics (4). Still others have maintained that the critical processes giving rise to new mathematical insights are opaque to consciousness and differ from explicit thought processes (1, 3, 5).
We address the role of language and visuospatial representation in mathematical thinking using empirical methods in cognitive neuroscience. Within the domain of elementary arithmetic, current cognitive models postulate at least two representational formats for number: a language-based format is used to store tables of exact arithmetic knowledge, and a language independent representation of number magnitude, akin to a mental “number line,” is used for quantity manipulation and approximation (6, 7). In agreement with these models, we now demonstrate that exact calculation is language dependent, whereas approximation relies on nonverbal visuo-spatial cerebral networks.
We first used behavioral experiments in bilinguals to examine the role of language-based representations in learning exact and approximate arithmetic. In one experiment, Russian English bilinguals were taught a set of exact or approximate sums of two two-digit numbers in one of their two languages (8). In the exact addition condition, subjects selected the correct sum from two numerically close numbers. In the approximate addition condition, they were asked to estimate the result and select the closest number. After training, subjects’ response times for solving trained problems and novel problems were tested in their two languages. Performance in both tasks improved considerably with training (response times dropped, in approximation, from 4423 to 2368 ms, and in exact calculation from 4285 to 2813 ms; both P , 0.001), regardless of the language in which a problem was trained (response times dropped from 4364 to 2644 ms in Russian and from 4344 to 2534 ms in English). Performance on exact and approximate tasks nevertheless showed different patterns of generalization during the test (Fig. 1). When tested on trained exact addition problems, subjects performed faster in the teaching language than in the untrained language, whether they were trained in Russian or English. This provided evidence that the arithmetic knowledge acquired during training with exact problems was stored in a language-specific format and showed a languages witching cost due to the required internal translation of the arithmetic problem. For approximate addition, in contrast, performance was equivalent in the two languages, providing evidence that the knowledge acquired by exposure to approximate problems was stored in a language-independent form.
Further evidence for contrasting representations underlying exact and approximate arithmetic came from comparisons of performance on trained problems and on novel problems involving similar magnitudes (Fig. 1). For exact addition, subjects performed faster on trained problems, suggesting that each new fact was stored independently of neighboring magnitudes, perhaps as a sequence of words. For approximate addition, performance generalized without cost to novel problems in the same range of magnitudes, providing evidence that new knowledge was stored using a number magnitude format (9). A second experiment extended this phenomenon to more complex arithmetic tasks. A new group of bilinguals was taught two new sets of exact addition facts (two-digit addition with addend 54 or 63), two new exact operations (base 6 and base 8 addition), and two new sets of approximate facts (about cube roots and logarithms in base 2), with one task of each type trained in each of their languages (10). Over training, performance again showed large and comparable improvements for all tasks and for both languages. The exact tasks again exhibited large costs for language-switching and for generalization to novel problems for both languages of training, indicating language-specific learning, whereas the approximate tasks showed language- and item-independence (Fig. 1). These results suggest that the teaching of some advanced mathematical facts such as logarithms and cube roots can give rise to a language-independent conceptualization of their magnitude. Exact arithmetic, however, consistently relies on language-based representations (11).
To examine whether partly distinct cerebral circuits underlie the observed behavioral dissociation, two functional brain imaging techniques were used, one with high spatial resolution and one with high temporal resolution. Functional magnetic resonance images (fMRI) and event-related potentials (ERPs) were acquired while subjects performed tightly matched exact and approximate addition tasks (Fig. 2) (12).
In fMRI, the bilateral parietal lobes showed greater activation for approximation than for exact calculation. The active areas occupied the banks of the left and right intraparietal sulci, extending anteriorily to the depth of the postcentral sulcus and laterally into the inferior parietal lobule (Talaraich coordinates of main peaks: 44, –36, 52, Z 5 6.37; 20, –60, 60, Z 5 6.03; –56, –44, 52, Z 5 5.96; –32, –68, 56, Z 5 5.10) (Fig. 3). Activation was also found during approximation in the right precuneus (4, –60, 52, Z 5 4.99), left and right precentral sulci (–56, 12, 24, Z 5 5.81; 48, 16, 20, Z 5 4.80), left dorsolateral prefrontal cortex (–44, 64, 12, Z 5 4.46), left superior prefrontal gyrus (–32, 8, 64, Z 5 4.75), left cerebellum (–48, –48, –28; Z 5 4.74) and left and right thalami (12, –16, 16; Z 5 4.43; –20, –8, 16, Z 5 4.04).
Most of these areas fall outside of traditional perisylvian language areas (13), and are involved instead in various visuo-spatial and analogical mental transformations (14–16). Cortices in the vicinity of the intraparietal sulcus, in particular, are active during visually guided hand and eye movements (15), mental rotation (16), and attention orienting (17). Previous brain-imaging experiments also reported strong inferior parietal activation during calculation (18), although its functional significance could not be ascertained because of task-difficulty confounds. Here, the parietal activation cannot be attributed to eye movement, hand movement, and attentional or task difficulty artifacts because the approximate and exact tasks were matched in difficulty and in stimulus and response characteristics (19). Rather, it is compatible with the hypothesis that approximate calculation involves a representation of numerical quantities analogous to a spatial number line, which relies on visuo-spatial circuits of the dorsal parietal pathway.
Fig. 1. Generalization of learning new exact or approximate number facts. Mean response times (RTs) to trained problems in the trained language are subtracted from RTs to trained problems in the untrained language (language cost: black bars) and from untrained problems in the trained language (generalization cost: gray bars). In experiment 1 (top two tasks), an analysis of variance on testing RTs indicated significant language-switching [F(1,3) 5 10.53, P , 0.05] and generalization costs [F(1,3) 5 37.64, P , 0.01] for the exact task, but no significant effect for the approximate task (both Fs , 1). The interactions of task (exact or approximate) on each cost measure were also significant [respectively, F(1,6) 5 11.10, P , 0.02 and F(1,6) 5 24.71, P , 0.005]. These effects were observed both with testing in English and with testing in Russian, and performance was similar in the two languages (for trained problems, mean RTs were 3445 ms in Russian and 3272 ms in English). In experiment 2 (bottom three tasks), similar analyses of variance indicated language-switching and generalization costs for base 10 addition, F(1,7) 5 24.23, P, 0.005 and F(1,7) 5 28.61, P, 0.001, and for addition in base 6 or 8, F(1,7) 5 304.06, P, 0.001 and F(1,7) 5 71.10, P, 0.001, but not for logarithm or cube root approximation (both Fs , 1). The interactions of task (exact or approximate) with each cost measure were also significant [respectively, F(2,14) 5 13.06, P, 0.001 and F(2,14) 5 17.31, P, 0.001]. Again, these effects were observed both with Russian and with English testing, and performance was similar in the two languages (for trained problems, mean RTs were 2639 ms in Russian and 2621 ms in English). Error rates were low in both experiments and were not indicative of speed-accuracy trade-offs.
Fig. 2. Design of the tasks used during brain imaging. Subjects fixated continuously on a small central square. On each trial, an addition problem, then two candidate answers were flashed. Subjects selected either the correct answer (exact task) or the most plausible answer (approximate task) by depressing the corresponding hand-held button as quickly as possible. The same addition problems were used in both tasks (12).
The converse fMRI contrast of exact calculation relative to approximation revealed a large and strictly left-lateralized activation in the left inferior frontal lobe (–32, 64, 4, Z 5 7.53) (20). Smaller activation was also found in the left cingulate gyrus (–8, 60, 16, Z 5 6.14), left precuneus (–8, –56, 20, Z 5 5.64), right parieto-occipital sulcus (20, –80, 28, Z 5 5.27), left and right angular gyri (40, –76, 20, Z 5 5.07; –44, –72, 36, Z 5 4.99), and right middle temporal gyrus (48, –16, 8, Z 5 4.68). Previous studies have found left inferior frontal activation during verbal association tasks, including generating a verb associated with a given noun (21). Together with the left angular gyrus and left anterior cingulate, these areas may constitute a network involved in the language-dependent coding of exact addition facts as verbal associations (6).
Because of their low temporal resolution, fMRI data are compatible with an alternative interpretation that does not appeal to dissociable representations underlying exact and approximate calculation. According to this alternative model, in both the exact and approximate tasks, subjects would compute the exact result using the same underlying representation of numbers. Differences in activation would be entirely due to a subsequent decision stage, during which subjects would select either an exact match or a proximity match to the addition result. The higher temporal resolution afforded by ERPs, however, shows that this alternative interpretation is not tenable. Crucially, ERP to exact and approximate trial blocks already differed significantly during the first 400 ms of a trial, when subjects were viewing strictly identical addition problems and had not yet received the choice stimuli (Fig. 3B). At 216 ms after the onset of the addition problem, ERPs first became more negative for exact rather than for approximate calculation over left inferior frontal electrodes, with a topography compatible with the fMRI activation seen in this same area. Previous ERP and intracranial recordings during the verb generation task also reported a latency of about 220 to 240 ms for the left inferior frontal activation (22). Later on in the epoch, starting at 272 ms after addition onset, ERPs became more negative for approximation over bilateral parietal electrodes, with a topography compatible with the bilateral parietal activation seen in fMRI. Thus, the recordings suggest that the two main components of the calculation circuits–the left inferior frontal activation for exact calculation and the bilateral intraparietal activation for approximation–are already active at about 230 and 280 ms poststimulus. This demonstrates that the calculation itself, not just the decision, is performed using distinct circuits depending on whether an exact or an approximate result is required.
This conclusion is also strengthened by previous neuropsychological observations of patients with calculation deficits, in whom the lesion localization fits with the present fMRI results. Several lesion sites can cause acalculia (23). However, on closer examination, at least two distinct patterns of deficit are found (24). Some patients with left parietal lesions exhibit a loss of the sense of numerical quantity (including an inability to decide which number falls between 2 and 4 or whether 9 is closer to 10 or to 5), with a relative preservation of rote language-based arithmetic such as multiplication tables (24, 25). Conversely, aphasia following left-hemispheric brain damage can be associated with a selective impairment of rote arithmetic and a preserved sense of quantity, including proximity and larger-smaller relations between numbers (24). Particularly relevant to the present work is the case of a severely aphasic and alexic patient with a large left-hemispheric lesion who could not decide whether 2 1 2 was 3 or 4, indicating a deficit for exact addition, but consistently preferred 3 over 9, indicating preserved approximation (26). Thus, lesion data confirm that distinct circuits underlie the sense of quantity and knowledge of rote arithmetic facts.
In conclusion, our results provides grounds for reconciling the divergent introspection of mathematicians by showing that even within the small domain of elementary arithmetic, multiple mental representations are used for different tasks. Exact arithmetic puts emphasis on language-specific representations and relies on a left inferior frontal circuit also used for generating associations between words. Symbolic arithmetic is a cultural invention specific to humans, and its development depended on the progressive improvement of number notation systems (27). Many other domains of mathematics, such as the calculus, also may depend critically on the invention of an appropriate mathematical language (28).
Approximate arithmetic, in contrast, shows no dependence on language and relies primarily on a quantity representation implemented in visuo-spatial networks of the left and right parietal lobes. An interesting, though clearly speculative, possibility, is that this language-independent representation of numerical quantity is related to the preverbal numerical abilities that have been independently established in various animals species (29) and in human infants (30). Together, these results may indicate that the nonverbal representation that underlies the human sense of numerical quantities has a long evolutionary history, a distinct developmental trajectory, and a dedicated cerebral substrate (31). In educated humans, it could provide the foundation for an integration with language-based representations of numbers. Much of advanced mathematics may build on this integration.
LATERALIZATION OF THE FRONTAL LOBE
A theme that runs through almost all of this book is that of cerebral lateralization. It has been clear throughout the history of neuropsychology that there are differences between the functional specializations of the two hemispheres. In terms of clinical evidence, this means that some deficits are more frequently associated with right sided lesions and some with left, and some deficits only appear with bilateral lesions (in both hemispheres). It should be emphasized that this is rarely a firm distinction, but is one of relative specialization. The evidence relating to frontal lobe injuries makes this clear.
Benton (1968) constructed a study in which he gave six tests to a group of patients with either left, right, or bilateral frontal lobe damage. Two of the tests (Verbal Fluency and Verbal Learning) were expected to be associated with left sided lesions, and indeed in both tests the patients with left lesions more frequently showed a deficit than those with right lesions. Those with bilateral lesions had difficulty as frequently as those with left lesions for Verbal Fluency, and more often for Verbal Learning. The “right hemisphere tasks” (Block Design and Design Copying) also produced the expected results: those with right lesions or bilateral lesions had a deficit more commonly than those with left lesions. The final two tests (Time Orientation–to see if patients know the time, day, and date and can locate themselves in time–and the Gorham Proverbs Test, in which the patient must explain the meaning of some well-known proverbs) showed that impaired performance was associated much more frequently with bilateral lesions than with unilateral lesions of either side.
These hemisphere specializations can loosely be divided into verbal and symbolic (left hemisphere) versus visuospatial (right hemisphere), although we shall question the validity of this distinction in Part III. In fact, evidence against such a simple dichotomy is already apparent in that failure in the Proverbs task only occurs commonly with bilateral lesions. Also, the test of identification of body parts in egocentric space that has already been mentioned is associated with lesions of the left frontal lobe, and not the right, as might have been expected. This information on lateralization is important, not only because it helps in locating lesions in patients under examination, but also for the construction of neuropsychological models of the organization of the brain.
INTEGRATION OF CORTICAL FUNCTIONS
Now that we have looked at the functions of the four cerebral lobes, although we have yet to discuss the language system, this seems an appropriate point to introduce some comments about the complexity and integration of functional cortical systems. While in order to understand the organization of the cerebral cortex we have to try to dissociate the functions that may be linked with particular areas from other functions in other areas, and this information is in turn useful to the clinician in diagnosing lesions and planning the rehabilitation of patients, this should not lead us to lose sight of the complexity of cognitive functions in everyday life.Itiseasytoadoptamodelofthecortexinwhichsomeareadoesaparticular task, and then another area takes over to do something else, and so on,buttherealityisclearlydifferent.Mostofthebrainisinvolved,mostof the time, in contributing to everyday intelligent behavior.
As I sit at this laptop typing the text you are reading, I am using verbal systems to generate the symbolic form of the semantic content I wish to convey. At the same time I keep reading my notes, looking at the screen to see what has been typed, and planning how I must move my fingers to type the words, introducing little routines for capitals, new paragraphs, and so on. I have to maintain a map of where the keyboard, the screen, my notes, and my limbs are all placed relative to each other, and all the while I am listening for the telephone, trying to ignore that I feel thirsty, and carrying on a separate debate in my head about how much nicer it would be on a beach in the Maldives. Periodically I mutter things to myself, scratch my head, and notice that one of my teeth is aching a little. Not many of the functional elements mentioned in our tour of the lobes are missing from this sample of a few seconds of my behavior.
Take an even simpler sample of behavior: a tennis player about to hit the ball. A motor program is certainly being executed for that stroke, but many other operations are also being simultaneously performed. The body has to be moved to the right place on the court, and balance and control maintained. The ball is being watched, as is the opponent’s position. A map is being maintained of spatial positions within the court, and constantly updated. The sound of the ball striking the racket will be attended to in order to check that the stroke was well executed, and feedback will be obtained through the arm at all stages of the stroke. At the same time future strategy is being planned, previous positions and plays are being remembered, the score is being maintained and calculations being made about points won and lost. Motivational systems are at work, feelings about the present position are being monitored, and an inner verbal dialogue (if not explicit speech) is being conducted.
This is perhaps laboring a point that is obvious. In everyday life most complex intelligent activities involve a great variety of cognitive skills and processes, all of which are interrelated and integrated within the overall operation of the brain. It is unreasonable to expect any of us to be able to grasp the actual complexity of this system, but at least in principle it is important to be aware of it, and not to allow the functional dissection that neuropsychological localization involves to blind us to how elaborate the operation of the brain actually is. No neuropsychologist can specialize in the study of only one area of the brain. The regions of the brain are not like “bolt-on parts,” although the way that much of the research has been conducted might encourage you to think so. The brain, extending beyond only the cerebral cortex, and the body in which it is located must be seen as a whole, and considered as acting within the framework of a single comprehensive system of behavioral control.
The discussion of language functions has been deferred until now because, while it is reasonable to assign most functions to one of the four central lobes (although it inevitably involves some distortion by oversimplification), the language system involves sites spread across a large part of the cortex. We will now examine that system and its disorders, which are known as aphasias.
THE LATERALIZATION OF LANGUAGE
When lateralization of function has been mentioned in previous chapters, it has been qualified as relating to right handed subjects. The reason is that it has long been clear that individuals differ in their cerebral organization, and that one of the variables most clearly associated with this is handedness. This is a complex topic that will be treated in more detail in Chapter 14, but it is of particular relevance to studies of language.
An early piece of evidence for differing speech organization in rightand left-handed people came from the observation that right-side focal lesions rarely produce disorders of speech in the right-handed, but frequently do so in the left-handed. A number of studies have collected data on the relative frequency of aphasia in right- and left-handed patients, and there has been considerable debate over their significance. It became rapidly clear that only a few, if any, left handers have a “reversed” pattern of organization from right handers. Many have left-sided speech representation, which is the typical pattern in right handers, while others seem to have more bilateral representation. Left handers thus have a less clear lateralization of speech, with both cerebral hemispheres contributing to the processing of language. The relative frequency of these different patterns of organization, and how they can be identified, is the subject of debate.
Evidence from a number of techniques of investigation is pertinent to this question (see Part III), but taking here only the data on the frequency of aphasia, a fairly clear conclusion has emerged from the work of Paul Satz. He reviewed the studies between 1935 and 1973 in some detail and analyzed the mathematical fit of the data to some of the models that have been proposed for left-handed speech organization. The results of these analyses (Carter, Hohenegger, & Satz, 1980; Satz, 1979) clearly support a model in which many left handers have speech in both hemispheres. The best fitting model was that in which 76% of left handers have bilateral speech representation, 25% left lateralized, and none right lateralized. For right handers the best model was: 95% left lateralized, none bilateral, and 5% right lateralized.
Another clear piece of evidence that right and left handers differ in the way in which language is organized in their brains comes from the Wada test, in which intracarotid sodium amytal is injected to depress temporarily the function of one of the two hemispheres (see p. 86), a technique developed to provide surgeons with information on speech lateralization. A report of the accumulated data from patients in Montreal who were tested by this method was published by Rasmussen and Milner (1975) and is shown in Table 7.1.
Of the patients without early left-hemisphere brain damage, the vast majority of right handers had left-sided speech representation, as did 70% of the left handers, and the remaining 30% of left handers were divided equally between right and bilateral speech. The proportions were different for those with early left-hemisphere damage, where it is believed that the young brain’s plasticity allows some relocation of language function, and where a higher proportion of patients with right or bilateral speech is therefore to be expected. However, the proportions here differed quite dramatically from those inferred by Satz. This is partly accounted for by differences in the methods used to infer speech representation, partly by differences in criteria for “bilateral” representation, and partly by different ways of classifying handedness, but none of these considerations convincingly accounts for the difference. Resolution of this debate must await further evidence and further development of analytical techniques and theoretical models.
TABLE 7.1. Speech Representation in Patients of Different Handedness on the Wada Test
Speech representation (%)
|Without early left-sided damage
Right handers 140
|Left/mixed handers 122||70||15||15|
|With early left-sided damage
Right handers 31
|Left/mixed handers 78||30||19||51|
Note. Data from Rasmussen and Milner (1975).
There are, of course, practical problems in finding out about the brain lateralization of individual patients. Knowing the handedness of a patient does not allow us to infer the patient’s speech lateralization. If the Wada evidence is correct, then we should expect nearly everyone to have left speech representation, and if we accept Satz’s analysis of the aphasia data, then we should expect most left handers to have bilateral speech representation. The fact is that neither study has allowed the construction of a satisfactory model of the relation of the lateralization of speech to handedness and associated variables. We will, however, return to models of handedness in Chapter 14. It should also be noted here that other methods of determining individual speech lateralization, based upon dichotic listening, seem to hold some promise in clinical applications (see Chapter 12).
A final point about left handers and aphasia is that left handers are considered to suffer more severely from lesions that affect speech in the initial stages of the illness, but then make more rapid and complete recovery from aphasia than do right handers. It is presumed that this is due to the relative bilateralization of their language representation, so that an undamaged hemisphere is more likely to be able to take over the functions previously performed by its damaged partner.
With this qualification about the language lateralization of left handers, we can return to consider the forms of aphasia, and will again assume the typical right handed pattern of left hemisphere speech lateralization.
VARIETIES OF APHASIA
The classification of aphasias has been one of the most hotly contested issues in the history of neuropsychology, and there is as yet no firm agreement. Not only has the terminology differed markedly from scheme to scheme, but the level of complexity of different systems has also varied greatly. Among those who have been willing to subdivide aphasias– and there has not been complete agreement that they differ in anything other than severity–some have considered there to be only two forms, which might be termed as receptive and expressive (Weisenberg & McBride, 1935); fluent and nonfluent (Howes & Geschwind, 1964); or anterior and posterior (Benson, 1967). At the other extreme there are very detailed classifications, often associated with esoteric and unhelpful terminology, that are more commonly found in neurology textbooks. Each scheme of classification has its particular strengths and weaknesses, and useful tabulations of the various common classifications appear in Benson (1979), Kertesz (1979), and Wallesch, JohanssenHorbach, and Blanken (2003).
The scheme used here is the one that seems to be most widely accepted at present, and is known as the Boston classification. It has developed from the work of Geschwind (1970) and Goodglass and Kaplan (1972), and while a number of variants of the classification are used, it forms perhaps the best basis for students of aphasia. This classification divides aphasias into the following six categories (with some roughly equivalent terms given in parentheses):
- Broca’s (motor, nonfluent) aphasia
- Wernicke’s (sensory) aphasia
- Conduction (central) aphasia
- Anomic (amnesic) aphasia
- Transcortical motor aphasia
- Transcortical sensory aphasia (isolation syndrome)
To these forms must be added global aphasia, in which there is massive and severe disturbance of language functions across a number of these categories. Alexia and agraphia, specific disorders of reading and writing respectively, are also included within the classification by some. (Note that while “a-” should imply total loss and “dys-” partial loss, these prefixes are used rather imprecisely in many of these terms.) It should also be recognized that there are other, more peripheral forms of speech pathology, which affect articulation, for example, but are not of primary interest to the neuropsychologist (see Murdoch, 1998).
The background to the Boston classification is the work of Wernicke and Lichtheim at the end of the 19th century, which resulted in a model revived and developed by Geschwind in the 1960s; hence the model is sometimes referred to as the Wernicke–Lichtheim–Geschwind model, or more commonly the Wernicke–Geschwind model. The basic idea is of a series of processes that lead from sensory input in the form of written language or speech through to output in speech or writing. Wernicke’s area is the essential area for obtaining meaning from language, while Broca’s area organizes the output of language. The two are connected by the arcuate fasciculus (see the following section of this chapter for an anatomical description of these structures), although there are also other, less direct routes between the two. The angular gyrus acts to translate information between visual and auditory forms. The actual model is, of course, a little more detailed and complex, but this is the essential basis for the analysis of aphasic problems.
There are a number of reasons for accepting the Boston classification, in particular that there is some agreement that it is clinically valuable and allows sensible distinctions to be made among patients. One aspect of this is that it can be related, at least in broad terms, to lesions at particular sites on the cerebral cortex, and it therefore has some diagnostic validity. It has also proved possible to relate it to the results of various aphasia test batteries, so that the results of these tests can be expressed in terms of the classification. In fact, one of the strongest arguments in its support is that a taxonomic analysis of the results of one of these batteries (the Western Aphasia Battery) on a large group of patients produced a classification that maps very neatly onto the Boston scheme (Kertesz, 1979). This is impressive support, but it should be remembered that the battery was constructed in a milieu in which the Boston scheme was the accepted model of the aphasias, and the result is therefore a little less surprising than at first appears.
One disadvantage of the Boston scheme is that it does not formally recognize the neurolinguistic analyses of aphasias that have become of increasing importance in recent years. Rather than classifying aphasias in terms of performance on various tasks in various modalities, studies have been based upon a linguistic analysis of dysfunction, concentrating on the linguistic structures that have been affected, on distinguishing syntactic and semantic processes, and the parts of speech and characteristics of the language elements that are abnormally processed. Many neuropsychologists are relatively inexpert at linguistic analysis, and neurologists even more so, but the development of aphasiology to include linguistic parameters is logical, and may be of considerable value.
Although aphasias are described here in terms of the Boston classification, there are other valuable ways to subdivide them. One example is the scheme proposed by Luria and Hutton (1977), which is much more in the tradition of the Russian work, and another attractive system has been suggested by Brown (1976). Brown divides aphasias first into anterior and posterior disorders, and then into various levels of linguistic involvement. The anterior series progresses from complete mutism through selective mutism to agrammatism and finally anarthric aphasia (in which the difficulty is making the movements to produce speech), thus moving from global action to more discrete faciovocal activity and finally to speech articulation. The posterior series moves from semantic through nominal to phonemic disorders, being a progression toward increasing specificity in the selection of particular words. Each of these series reflects a structural progression from limbic transitional cortex through generalized neocortex to focal neocortex. Brown’s theory is not easy to grasp when stated so starkly, but the nature of some of the functions mentioned should become clearer as the different forms of aphasia are discussed in more detail. An advantage of his scheme is that it does make explicit reference to linguistic parameters, while still being directly linked to anatomical structures. An added attraction of the scheme is that it treats the cortex as a three-dimensional structure and considers the depth of lesions in subcortical tissue. This aspect of cerebral lesions is too often ignored, and the cortex treated as if it had only two dimensions, relatively divorced from what lies beneath. Brown’s scheme has yet, however, to gain wide acceptance.
Before proceeding to describe the various forms of aphasia, it may be helpful to look again at some of the anatomical locations thought to be involved in the language system (shown in Figure 7.1). Beginning anteriorly, the most important structure is Broca’s area in the inferior posterior frontal cortex (of the left hemisphere, of course). This was the first location to be clearly associated with language function, in Broca’s report of 1861 supported by postmortem findings (although Dax may have some claim to have established it earlier; see Critchley, 1979). Also in the frontal lobe, although separate from Broca’s area, is a region of anterior mesial cortex that, as we shall see, is associated with transcortical motor aphasia.
FIGURE 7.1. The principal structures in the left hemisphere associated with language functions.
Posteriorly, there is a more complex group of structures. Wernicke’s area is in the superior middle and posterior regions of the temporal lobe, not far away from Heschl’s gyrus, which is involved in auditory reception (see p. 73). Wrapped around the posterior end of the Sylvian fissure is an area of association cortex (peri-Sylvian association cortex), and moving posteriorly there are also the areas known as the angular gyrus and the supramarginal gyrus. The region referred to as the angular gyrus may extend rather beyond its strict topographical bounds into the middle posterior temporal region. The only other structure that we must denote is the arcuate fasciculus, which is an important tract of corticocortical (connecting two regions of cortex) fibers running approximately from the region around the posterior end of the Sylvian fissure forward to the posterior regions of the frontal lobe, and thereby serving as a direct link, so it is inferred, between the posterior and anterior language zones.
Before describing how all these different structures contribute to the language system, it should be pointed out that there is some danger of overlooking the considerable variability in the system by abstracting information from it that allows us to make sense of localization. The system should not ideally be described in such neat terms, with apparently accurate cortical localization. Figures presented by Kertesz, Lesk, and McCabe (1977) are derived from isotope scans of lesions producing various forms of aphasia and show that the lesions do no more than center on a particular region. In Figure 7.2, for example, the lesions producing anomic aphasia involve the angular gyrus and middle posterior temporal lobe, and it is clear from the figure that they may commonly involve a variety of other regions.
Not only the isotope scans, but also studies of stimulation of the exposed cortex at surgery have raised considerable doubts about the precise localization of language functions (Ojemann & Whitaker, 1978; Ojemann & Mateer, 1979). Others have more generally questioned the localization of structures involved in language, even to questioning the identification of such well-established landmarks as Wernicke’s area (Bogen & Bogen, 1976; Zangwill, 1975). Care should therefore be taken to recognize both the variation among individuals and the degree of imprecision that may in reality characterize apparently precise accounts of the language system.
FIGURE 7.2. The regions of the brain associated with anomic aphasia. (Redrawn after Kertesz, Lesk, and McCabe, 1977.)
THE FORMS OF APHASIA Broca’s Aphasia
This type of aphasia is associated with lesions of Broca’s area in the inferior posterior frontal lobe. It is the most generally recognized form of aphasia, and is the classical nonfluent form. The principal sign is that the patient’s speech output is severely impaired. Most likely to be affected are articles, adverbs, adjectives, and other function words, so that speech tends to be reduced to nouns and verbs only. In the most extreme cases there is muteness, and in severe cases nouns may be restricted to their singular form, and verbs to the infinitive or a participle. This kind of speech is sometimes described as “telegraphic.” The meaning is often evident, and words tend to be in the correct order, but it is as if speech has been reduced to its essential elements by agrammatic deletions. Short automatic phrases, of the kind common in “small talk,” may well be inserted, and in fact used excessively and inappropriately.
There may also be errors in the actual production of words. Paraphasias, as these are known, are relatively common, but are of the phonemic variety where the target word is usually identifiable. Some error in the selection or production of sound elements seems to occur, but as the word produced usually approximates to the target word, and as the context may also provide some clues, it is generally possible to tell what the patient is intending to say. There is, in addition, some difficulty with repetition and with naming, although prompting by sounds or context can be of significant help with the latter. However, it is probably best to see these particular difficulties as secondary to the problems with speech output, limiting the performance that might otherwise be attained. This is probably not a complete explanation, because written output may well be affected in the same way as speech, presumably by sharing certain output mechanisms, and the difficulty therefore seems to lie in language production generally rather than specifically with speech.
All these problems occur, however, with language comprehension intact. Patients understand perfectly what is said to them, and read as well as they comprehend speech. They often show extreme frustration at their own inability to communicate by speech, while understanding what they read and hear, although not all of them are aware of their speech limitation.
THE REHABILITATION OF APHASICS
Aphasia most commonly results from either traumatic injuries or strokes, the latter being more frequent. The extent of recovery depends on a number of factors, but the most important are probably the cause of the dysfunction and the severity of the handicap. The outcome is generally much better following trauma than following stroke, with more than half of the patients fully recovering their premorbid level of function. This difference is, however, confounded with age, since strokes occur more commonly among the elderly, and this probably accounts for the poorer prognosis. The greatest spontaneous recovery happens within the first three months following the damage, with further significant improvement up to six months. After that, progress tends to slow down, and there are few gains beyond a year after the damage has occurred (Benson & Ardila, 1996; Kertesz, 1993).
Despite there being few well-controlled studies of aphasia therapy, there is some evidence that it is effective in aiding recovery. Most importantly, there is evidence that therapy is more effective if it is begun as early as possible after the injury. A wide variety of approaches to treatment are used by speech therapists with aphasic patients, but most have developed from Wepman’s idea that the patient should be systematically stimulated by appropriate materials and environments (Wepman, 1951). This approach was developed by both Schuell (Schuell, Jenkins, & Jimenez-Pabon, 1964) and Taylor (1964). Typical therapy sessions will involve the acting out of everyday situations and working with classes of objects such as food or clothing, or with similar sounding groups of words. There is a continual attempt to elicit words and stimulate responses, and extensive use is made of repetition and auditory stimulation.
Recently, therapy that aims to reestablish language by specific teaching programs has become more popular. This approach allows a more systematic structure for the therapy program, which may be based on an explicitly linguistic approach, and may also employ behavioral learning techniques such as operant reinforcement. An example of this kind of approach is to be found in Weniger, Huber, Stachowiak, and Poeck (1980). Specific information about the patient’s linguistic deficits is translated into a training program designed to teach these deficient aspects of language, with progress at each stage carefully monitored and the programme adapted accordingly (Marshall, 2003). In some programs, specific rewards will be given for success in producing particular responses, according to a formal and explicit system.
There are a variety of other, less popular, forms of therapy. Among these are Deblocking Therapy (Weigl, 1968), which aims to use the intact aspects of language function to assist in the redevelopment of the damaged areas, and Compensation Therapy (Holland, 1977), which teaches the patient strategies to enable him or her to circumvent the difficulties by using the intact areas of function. Melodic Intonation Therapy (Sparks, Helm, & Albert, 1974) capitalizes on the fact that many aphasics maintain the ability to sing when they cannot speak, and so can learn to “intone” the speech they wish to generate. Finally, Visual Communication Therapy (Gardner, Zurif, Berry, & Baker, 1976) has translated some of the methods used to teach language to chimpanzees, based upon simple figural and geometric symbols, into a method that seems of particular value with patients who have severe global aphasia.
Therapy can be of considerable value to many patients, and aphasics often regain a significant degree of language function. There are, however, a large number of patients whose handicap persists, and who form a major rehabilitational challenge. What are needed are more extensive and better-designed studies of the value of particular therapies so that treatment can be applied more effectively and economically to those with language dysfunction.
Split Brains and Dual Minds
Split-brain patients, or in more formal terms those who have undergone cerebral commissurotomy, provide a natural link between the clinical and experimental studies of the organization of the human brain. These patients are clinical subjects who have undergone neurosurgery, and yet the data they provide are treated as if they were derived from an experimental procedure. The focus of the research is on understanding the brain, not on helping the patient. Historically, split-brain studies also provided a particular stimulus for the development of laboratory techniques for use with intact human subjects.
The commissurotomy operation was introduced by Van Wagenen in 1940 as a treatment for certain severe and intractible forms of epilepsy, characterized by a focus in one hemisphere triggering a major seizure that spreads across the corpus callosum to involve both hemispheres. The idea was, simply, that cutting the corpus callosum and the other forebrain commissures would restrict the seizure to one hemisphere. Ultimately, it was found that even better relief from the epilepsy could be obtained, and that the frequency as well as the extent of seizure activity could be dramatically improved by the operation. A series of patients was operated upon, but interest in the procedure was lost until the early 1960s when Sperry, in collaboration with the neurologist Joseph Bogen and the surgeon Philip Vogel, instituted a new series of operations. (For a full treatment of the theoretical background to this work, see Bogen, 1979.)
Sperry had been working in the 1950s on experimental split-brain preparations in animals, and it was he who saw the significance of the operation for theories about the organization of the brain. A curious fact
about the earlier series was that almost no handicaps in everyday life had been reported, despite the radical nature of the surgery. Relatively sophisticated experimental investigations, for the time, were also carried out without significant deficits being revealed.
However, Sperry, working with Michael Gazzaniga, was able to show by elegant techniques that the patients in the Bogen and Vogel series did suffer from certain subtle deficits that were not apparent in their everyday activity. These deficits are described in the following pages. Following the Bogen and Vogel series, a further series of patients was operated upon by Wilson, and a number of patients also received commissurotomy for the treatment of intraventricular and midline tumors, and for abnormalities of the cerebrovascular system.
THE COMMISSUROTOMY OPERATION AND THE PATIENTS
The operation in its full form involves complete section of the corpus callosum, the anterior and hippocampal commissures, and the massa intermedia (see Figure 10.1). It therefore divides all direct cortical links between the two cerebral hemispheres, isolating the cortex of each hemisphere from its partner. From this division comes the term “split brain,” and also the importance of these patients in helping to construct models of the function of the right and left halves of the human brain. A number of patients have also undergone partial commissurotomy (incomplete section of the corpus callosum), and while data from these patients are of considerable importance, in the interests of clarity only the “complete splits” will be discussed in this chapter. The related operation of hemispherectomy, in which an entire hemisphere is removed, and the naturally occurring cases of absence of the corpus callosum (“callosal agenesis”) are also relevant but will not be treated here. A review may be found in Lassone and Jeeves (1994).
Before we go on to examine the effects of commissurotomy, several points should be made clear. The first is that although split-brain patients are treated as a single group, the operations performed in the three series do differ in significant ways (and this may partly account for the failure to find in the first series some of the deficits that were observed in the later series). In particular, not all the operations involved cutting the anterior commissure, which is now known to be capable of transferring quite considerable amounts of information between the two hemispheres (Risse, LeDoux, Springer, Wilson, & Gazzaniga, 1978). Within the Wilson series, there are two subseries, the anterior commissure being sectioned in the first, but not the later, set of patients (Wilson, Reeves, & Gazzaniga, 1978). The operation performed by Van Wagenen was undertaken in stages, while later series used a single operation. There is insufficient space here to go into the significance of these differences, but a careful study of the effects of commissurotomy should take account of these factors.
FIGURE 10.1. The major interhemispheric cerebral commissures.
Perhaps more important is a word of caution about the patients. The number of patients operated upon is small, and the number reported in the (very extensive) literature is a minority of these. The total number of patients available is probably no more than about 30, and the vast bulk of the literature concerns no more than about half a dozen of these. Particularly as these patients differ greatly in their performance on the experimental tasks set for them, it is dangerous to regard them as a homogeneous group. Beware of any conclusion based on the mean performance of a group of these patients, as it is likely to be misleading. In fact, the data from the patients are increasingly reported as a set of single case studies, which seems more reasonable, but this practice has not been universally adopted.
It is also tempting to treat the data as if split brain patients presented us with neat controlled experimental results. This is not so. These are surgical lesions in abnormal brains. The patients all have long and complex neurological histories, reflected in their pre- and postoperative neurological status. Many have lesions in other parts of the brain; some have had lesions from birth (with the suspicion that the organization of the brain has developed atypically); and some have had the operation at a relatively early age, when a degree of plasticity still remains in the brain. A detailed illustration of how these factors make interpretation of the data from split-brain patients extremely difficult has been provided by Whitaker and Ojemann (1977).
All these factors–the small number of patients, their different preand postoperative histories and response to the operation, the precise nature of the surgery they have received–as well as the effects of their current anticonvulsive medication point to caution in treating the patients as a single group, and in treating the findings as if they came from experimental lesions in a previously normal brain.
THE EFFECTS OF COMMISSUROTOMY
The most striking result of commissurotomy remains the lack of handicap experienced by most patients. There is an interesting film of one female patient doing domestic tasks, like beating eggs, and also swimming, activities that involve integration between the two sides of the body, and doing them quite successfully. Nevertheless, by ensuring that information coming from the external world is restricted to one side of sensory space, and external means of communication between the two sides of the body are prevented (more of this later), the independence of the two cerebral hemispheres, and consequent deficits in integration between the two sides of the body, can be demonstrated.
The clearest example of this is when visual information is presented in the left or right visual half field. Stimuli that appear to the left of the point on which a subject is fixating are projected initially only to the right occipital cortex, while those to the right of fixation (in the “right visual field”) are projected to the left occipital cortex. Therefore, if the subject’s fixation is controlled, this “divided visual field” technique can be used to present visual information to either the left or the right hemisphere alone. While in normal subjects the information is then distributed through the entire brain (see Chapter 11), in split brain patients it is confined to the hemisphere of original reception (see Figure 10.2).
If a visual stimulus is shown in the right visual field, and projects to the left hemisphere, which possesses speech (see Chapter 7), then the stimulus can be verbally identified, and can be picked out from a set of possible responses by the right hand (also controlled by the left hemisphere). It cannot, however, be selected by the left hand (controlled by the right hemisphere). Conversely, a stimulus presented in the left visual field will go to the right hemisphere and can be indicated by the left hand, but not by the right hand, or by speech. The information is restricted to the hemisphere of reception, and can be linked only to response processes that are controlled by that hemisphere. This basic pattern of response to visual stimuli occurs whether the task is to match a visual picture to a real object, to another picture, or to a word, or to try to name it; or whether it is to match a word to another word, to a picture, or an object, or to speak it.
FIGURE 10.2. Connections between the two hemispheres and the hands. In normals, with the direct pathway (a), the right hand responds to the right visual field (solid line) and only one hemisphere, the left, is involved. With the indirect pathway in normals (b) the left hand responds to the right visual field (solid line) and both hemispheres and the corpus callosum are involved. In split-brain subjects the direct pathway still operates (c) but the indirect pathway is unable to function (d). (After Beaumont, 1983.)
Some curious phenomena can be observed as a result of this dissociation between the hemispheres. If a word with two component parts, for example heart (he + art), is presented very briefly so that the subject is fixating between the e and the a, then patients will say that they saw art while with their left hand they will, at the same time, pick out (from among an array of possible responses) he as the stimulus. This behavior can be seen at its clearest in the “chimeric figure” experiments (Levy, Trevarthen, & Sperry, 1972). Chimeric figures are constructed of two half figures arranged so that patients fixate upon the vertical division between the two half stimuli (see Figure 10.3). These stimuli may be constructed from outline drawings or more complex figural stimuli like faces. As we might expect, shown a chimeric figure composed of the left half face of an old man with a beard wearing a hat and the right half face of a young blonde, patients say that they have seen an attractive young woman, as the left speaking hemisphere only knows about the right half of the stimulus. Incidentally, the patients do not know of the chimeric nature of the stimuli and report nothing unusual about what they have seen. At the same time, the left hand will select the old man from among a selection of whole faces that include the complete faces of the old man and the blonde. If the anomaly of the response is pointed out to the patient, he or she may well appear confused, and may make some comment such as that the hairstyle looked rather like a hat, in an attempt to resolve the confusion.
FIGURE 10.3. Chimeric figures. The chimeric figures (below), constructed from whole stimuli (above), are used to test recognition. (After Levy, Trevarthen, and Sperry, 1972.)
Stimuli may also be lateralized by somaesthetic presentation, following the same logic as divided visual field presentation. The patient may be touched at some point on the body and asked to report (in speech) the location of the stimulation or its nature (light touch, pinprick, and so on), or asked to point to the spot with either the right or the left hand. Objects may be placed in one hand (out of sight) and haptically explored for naming, or for matching to another object among an array using either the same or the opposite hand. Hand postures may be set up on one hand (again out of sight) to be reproduced by the opposite hand, or a representation of a gesture presented in one of the visual fields to be reproduced by the ipsilateral or contralateral hand. There is a rich variety of possibilities for the ingenious investigator, and many of the changes have been rung on this kind of experiment. What the results boil down to is, however, that split brain patients have difficulty integrating information across the body midline: pointing with one hand to places on the opposite side of the body, passing information from one hand to the other, or relating visual information on one side of the body to movement and sensation on the other. Further, there is difficulty in expressing the results of right hemisphere processing or leftsided stimulation in speech.
This conclusion concerning somaesthesis has to be qualified a little. The lateralization of somaesthesis is most complete in relation to the more distal parts of the body, especially if finely controlled or skilled movements are required. The effects are seen most clearly, therefore, in the hands and especially the fingertips. Taking more gross body movements, and sensation from the more proximal body areas such as the trunk, the lateral differentiation is less clear. This is almost certainly because of the existence of ipsilateral connections to these areas for both sensation and motor control (see Gazzaniga & LeDoux, 1978).
Audition turns out to be rather less straightforward than vision. This is because the projections from the ears to the brain are not exclusively contralateral, but are to some extent bilateral (see p. 74). Nevertheless, the dominance of the crossed contralateral pathway can be observed in the suppression that follows dichotic (bilateral simultaneous) presentation. Split-brain patients can report stimuli presented to either ear if the stimuli are presented singly, but if two stimuli are presented at the same time to the two ears, there will be a massive bias to report the right ear stimulus (which passes to the left hemisphere by the dominant crossed pathway). It is assumed that the ipsilateral projection from left ear to left hemisphere is suppressed in dichotic presentation, although, while the spoken response may favor the right ear, there may be evidence of reception of the left ear stimulus in response by the left hand (Gordon, 1980; Springer, Sidtis, Wilson, & Gazzaniga, 1978).
While the question of speech function in the two hemispheres is clearly answered by the commissurotomy evidence, the question of the extent of right hemisphere language is not. The left hemisphere clearly has the capacity to generate speech, but there is no evidence to suggest that in the normally developed and lateralized brain this capacity is also present in the right (outside automatic speech and swearing). In early studies there was a tendency to confuse “speech” and “language,” and to assume that language functions were as completely lateralized. There has since been a gradual and continuing revision of this assumption, with the right hemisphere being credited with increasingly sophisticated linguistic abilities, short of speech output.
Early reports tended to suggest that the comprehension of language stimuli presented to the right hemisphere was very limited, and that beyond the recognition of simple and common concrete nouns there was little linguistic ability present. It is now clear that while an asymmetry undoubtedly exists, in that the right hemisphere is simply not as good at performing language functions as the left, the right hemisphere nevertheless possesses a significant degree of linguistic skill. Various studies have now shown that the right hemisphere can understand abstract words and a variety of syntactic structures including verbs, sentence transformations, and long nonredundant and semantically abstract references; is competent at semantic matching; can perform rhyme matches; and can understand mental associations based upon semantically coordinate, contingent and superordinate relationships. In addition, they have shown that the inability to name objects held in the left hand is only an inability to generate the name; it does not imply a failure to comprehend the nature of the object, or to construct semantic associations with it that may be manipulated and expressed in language, although not in speech. A degree of writing is possible for the left hand, although some of the patients show difficulty with this (in contrast with left-hand performance in drawing and spatial manipulation, which is superior to that with the right hand). The precise extent of right hemisphere language is still a matter of debate, despite extensive and elegant studies (notably by Eran Zaidel). Reviews of the work on right hemisphere speech are to be found in Beeman and Chiarello (1997), Code (1987), and Wray (1992). Perhaps the most interesting and telling finding is that of Eran Zaidel (1979), who also compared the performance of the separated hemispheres with the performance in “free vision” (allowing the hemispheres to cooperate), and found that the two hemispheres together produced better performance than either alone.
Relevant to this issue of the abilities of the right hemisphere are the most elegant metacontrol experiments of Levy and Trevarthen (1976). These studies employed chimeric figures that were outline drawings cleverly selected so that the half stimuli could be matched to complete stimuli on the basis of identity, or physical similarity, or by semantic functional association (see Figure 10.4). It was found, as had been expected, that the right hemisphere was relatively proficient at, and generally assumed control of, appearance matches, while the left hemisphere controlled function matches. From these findings come the theoretical arguments for a difference between the two hemispheres that is based upon modes of processing rather than upon the nature of the material or of the response that is demanded. It is suggested that it is the nature of the cognitive operations required that determines the relative proficiency of the two hemispheres. However, we must be careful to note that this result is only one of relative specialization. The data show that while there is an asymmetry between the performance of the two hemispheres, there was a significant number of trials in which the “wrong” hemisphere assumed control of the match. In other words, while the right hemisphere may prefer, or be more competent at, appearance matches (for whatever reason), it is not entirely without the ability to undertake function matches. Dissociations between controlling hemisphere and the strategy employed occurred on a significant number of trials, and therefore the asymmetry is certainly far from absolute.
As reviews of the effects of commissurotomy have tended to emphasize the division and independence of the two hemisphere systems, and to maximize the asymmetries in performance that may be observed, some evidence that points to the limitations of this view should also be noted. I have argued elsewhere (Beaumont, 1981) that the published literature has significantly misrepresented the nature of the effects of commissurotomy by presenting an oversimplified and overgeneralized picture, and I am wary of repeating the error myself here. The problem, historically, was undoubtedly the influence exercised on Sperry’s thinking by the initial one or two patients in the Bogen and Vogel series. These turned out to be not entirely typical for a variety of reasons, and the rather bold and dramatic conclusions that he drew have had to be tempered in a number of ways.
First, there are clear exceptions to the general statements made above about the effects of the operation. For instance, some of the patients are able to indicate with a given hand points stimulated on the opposite side of the body. (The variation in the degree of left hand agraphia has already been noted.) There are also opportunities for the interhemispheric transfer (obviously by routes other than the commissures that have been sectioned) of certain types of information, transfer of the size–weight illusion being just one illustration (Gandevia, 1978). Other evidence has emphasized the continuity of the integrated and articulated operation of the total cerebral system. LeDoux, Risse, Springer, Wilson, and Gazzaniga (1977) examined the performance of one of the patients on a complex, high-level concept attainment and hypothesis testing task, in which there seems no doubt that the cooperation of both hemispheres was required for satisfactory performance. The patient was assessed both pre- and postoperatively, and no postoperative deficit was found, but rather an improvement in performance. They conclude that complex cognitive processes do not depend on the function of the corpus callosum.
A problem in assessing the significance of some of the exceptions to the “expected” findings (although the last-named study seems free of this suspicion) is that many of the patients develop subtle cross-cueing strategies, that is, they develop ways to circumvent their handicap by using external routes for the lateral transfer of information. For example, sound cues may be passed from objects in the left hand to the left hemisphere via the right ear. Also, emotional signals may transfer easily by subcortical routes, so that by feeling a generalized arousal reaction a correct response may be deduced. The left hemisphere may conduct a speech commentary on the responses of the left hand, which it can see in free vision in the right visual field, and so transmit guidance to the right hemisphere. But perhaps the most subtle strategy concerns eye movements, which can transfer subcortically, so that even the extent of eye movement required to inspect a stimulus may give the essential clue to stimulus identity if the range of possibilities is sufficiently small (Gazzaniga, 1970; Gazzaniga & Hillyard, 1971). The degree to which the cerebral hemispheres cannot be considered in isolation from the subcortical centers to which they remain connected (and which are not divided) has been an important theme introduced by Sergent (1986).
These examples alone show how carefully controlled the experiments must be, and how vigilant the experimenter, if the possibility of this kind of external transfer is to be ruled out. Also, it becomes less surprising that the patients suffer little handicap in everyday life, particularly as they can manipulate what appears in which field of vision and employ both hands in order to implement strategies to share information between the hemispheres.
There is one further factor that potentially disturbs the rather neat explanation of the effects of commissurotomy. This is the role of attention, a factor that will feature more prominently in a related context in Chapter 12. The potential intervention of attentional factors is demonstrated by a beautifully elegant experiment by Gazzaniga, Ladavas and colleagues (Gazzaniga & Ladavas, 1987; Ladavas, De Pesce, Mangun, & Gazzaniga, 1994; Luck, Hillyard, Mangun, & Gazzaniga, 1994). They performed the standard experiment with stimuli in the left and right visual fields and obtained the expected effects, as described above. But, they then asked their participants to tilt their head 90 degrees so that one of their ears was touching their shoulder; the rest of the body was kept in the normal position. The experiment was then repeated. What we should expect is that any effects of lateral presentation of the stimuli will disappear. Although the stimuli appear to the left and right of the participant’s body, they are now above and below the fixation point with respect to the orientation of the participant’s head. What actually happened, amazingly, was that the lateral asymmetry was substantially preserved. This cannot be explained adequately with respect to the neural projection of the stimuli into the left and right hemispheres. It suggests the role of some internal representation of extra-bodily space, and processes that direct attention to the “left” or “right” of this space. (For a discussion of the processing of spatial relations in commissurotomy, see Sergent, 1991.) This important finding does not fundamentally undermine our understanding of the effects of commissurotomy, but it does demonstrate that our understanding is incomplete and that there are psychological aspects yet to be determined.
There are numerous reviews of this field in the literature; among the most useful from the period when research was most active are Bogen (1979), Dimond (1972), Gazzaniga and LeDoux (1978), Levy (1980), and Sperry (1973, 1974). References to more recent reviews are given at the end of this chapter. Excellent discussions of the theoretical models employed may also be found in Trevarthen (1975) and E. Zaidel (1978). As has already been made clear, care should be exercised in evaluating the accuracy of some of these reviews. It may nevertheless be concluded that split-brain patients have a deficit in the cross-integration of information, at least under experimental conditions. In a limited sense it is possible to accept the conclusion of Sperry that the patients exhibit “two independent streams of conscious awareness” that are out of contact with each other, each with its own “separate and private sensations; its own perceptions; and its own impulses to act” (Sperry, 1968, p. 724). Whether this really implies that the consciousness of these patients is divided or in some way doubled, and whether this holds implications for the processes that normally underlie consciousness, are questions to which we now turn.
Typically for certain writings in this area, Charles Furst has claimed that “split brain research has opened a new frontier of scientific investigations into the physical basis of consciousness. The vistas which lie beyond this frontier will perhaps revolutionize our traditional ways of understanding the human mind” (1979, p. 161). Is this neuroscience or neurofantasy?
The separation between the cognitive activities of the two hemispheres has led various writers to suggest that dividing the cerebral commissures produces two independent minds. This notion was taken up most energetically by Roger Sperry, who proposed that the split brain patient has two separate minds within the one body, each with its own will as well as its own perceptions and memories (Sperry, 1976). This model, with minor variations, has been the most popular, and it is possible to find several examples of it in the literature (see Beaumont, 1981).
The only major opponent of this theoretical position has been Eccles (1977). He has suggested that, while commissurotomy divides higher-level functions between the two hemispheres, only the left hemisphere contains the seat of consciousness. The right hemisphere is considered as an automatically acting subordinate partner with rather less than normal human attributes. He bases this position largely on the inability of the right hemisphere to produce speech, which he regards as the necessary test of the possession of conscious awareness.
It may have occurred to you to wonder what is meant by “consciousness.” Indeed, much of the debate might be resolved if those who construct theories about this topic were themselves more clear. Some writers, particularly those who espouse double mind theories, include any evidence of intelligent responsiveness, attention, and awareness as indicating consciousness. Eccles, however, employs the very much more restricted requirement that the organism must declare itself conscious by speech. I do not intend to try to resolve this matter here, but it seems to many people that Eccles’s definition is too restrictive. It demands, for instance, that we remain agnostic about whether the congenitally mute are conscious. It also implies that it is not possible to know if a person who has lost his or her voice from a sore throat possesses consciousness. This is patently absurd. On the other hand, “any intelligent cognitive act” seems too loose a definition, for it forces us to attribute consciousness even to lowly animals if they show that they can learn some simple experimental task, even though we are in a wider context unsure that this demands “consciousness.” Possibly a solution can be found by demanding some evidence of self-awareness and purposeful cognitive activity, although this is inadequate as a formal definition. A fuller discussion of this problem may be found in Savage (1976).
By now it may also be clear to you that the debate is embroiled in philosophical mind–body issues (see p. 7). There are many fascinating discussions of these problems by both neuroscientists and philosophers, including the prospect of transplanting separated hemispheres into different bodies and then recombining them at a later date (Bell, 1975; Globus, 1976; Nagel, 1971; Puccetti, 1981, 1985). Such issues raise fundamental questions about our conception of personal identity, but they have to be sidestepped here for want of space. However, if my argument is accepted that it is not sensible to regard the minds of commissurotomy patients as divided, many of the questions become academic. (For excellent discussions of consciousness, with some reference to the split-brain patients, see Rose, 2006, and Zeman, 2001.)
To return to the scientific evidence, there are five central questions that we should ask before deciding on the structure of consciousness in split-brain patients.
- Does the right hemisphere possess language? This question has already been broached, with the conclusion that the right hemisphere may be effectively mute, and is certainly less proficient at most linguistic tasks, but that it does possess a remarkable variety of linguistic skills. As a result, the right half brain of a split-brain patient cannot be considered to lack language. It is, of course, possible that this evidence is in part a consequence of the patient’s early neurological history, and that it does not imply that language is similarly represented in the right hemisphere of a normal brain. The speech developed in association with the right hemisphere by a patient known as PS, who was operated upon at the age of 15 and whose neuropsychological development was almost certainly not normal, provides an illustration of how atypical organization may occur in these patients (Gazzaniga, Volpe, Smylie, Wilson, & LeDoux, 1979). However, the point still stands that split-brain patients have some right hemisphere language capability.
- Are there other abilities that the right hemisphere may lack? While there is much evidence for the relative specialization of the two hemispheres, with some functions better performed by the right and some better performed by the left, no abilities other than speech and language have been proposed as absent from the right hemisphere. In fact, the trend has increasingly been to credit the right hemisphere with sophisticated abilities (Perelman, 1983; Ardila & Ostrosky-Solis, 1984). The most striking example is the demonstration by Sperry, Zaidel, and Zaidel (1979) of self-recognition and social awareness in the right hemisphere of two of the patients. Arrays of pictures and photographs that included some with personal and emotional significance among similar neutral stimuli were presented. The detailed accounts of the patients’ responses leave little doubt that while the patients have great difficulty in describing their thoughts about the stimuli through speech, a characteristic social, political, personal, and self-awareness, such as we should expect for the left, was present in the right hemisphere. For example, a woman was shown (to her right hemisphere) four photographs of men, which included one of her son. She readily picked out her son as the one she recognized, but could not say who it was. She reported good feelings about the image and suggested the names of her daughter and husband, but remained confused. Following further questions to which she indicated further positive feelings (“Yeah, it’s fine, it’s beautiful”), she pointed to the picture of her son and said, “The best-looking one there.”
EXAMINER: Is it you? Your husband? (Patient does not respond.) . . . your son?
PATIENT: (in a very definite decisive tone) Yeah.
Thus helped to overcome the handicap of limited speech output, the right hemisphere showed every sign of being characteristically human.
It has also been suggested that there is a lack of awareness of right hemisphere processes, based on some rather slender evidence, such as the denial of left-hand control, the failure to recognize the anomaly in chimeric figures (although normal subjects may also miss the anomaly under certain circumstances), and the maintenance of independent response probabilities by the two hemispheres in a learning task. This evidence is not sufficiently strong to deny awareness of right hemisphere processes, particularly in view of the kinds of demonstration above.
- Is there any cross-integration between the disconnected hemispheres? It has already been hinted that the separation of the hemispheres is not as complete as is sometimes suggested. There is intermanual transfer of maze learning, accurate target-directed reaching across the midline, and completion across the midline for incomplete geometric figures, pictures, and words. The ambient space around the body, in vision outside the central field, is not divided for the perception of space relations or motion in space (Trevarthen, 1974). Bimanual skills are relatively preserved (Preilowski, 1975).
It is clear that the assumption that cross-integration across the hemispheres is impossible for these patients is incorrect. It is possible to demonstrate deficits in this function in a rigorous experimental situation and on a particular task, but for most cognitive activities (certainly in the unrestricted patient) there are mechanisms that allow substantial cross-integration.
- Are the separated hemispheres ever in conflict? It is more difficult to decide the answer to this question because the relevant evidence is anecdotal. There are various accounts of conflicting activity by the two hands: one hand pulling trousers up, while the other pulled them down; one hand seeking help from the patient’s wife while the other pushed her away. At one time it seemed that most of these accounts came from one patient, and in the early period following surgery, but they have been amplified and extended by a range of personal accounts collected by Dimond (1979). One patient said: “You wouldn’t want to hear some of the things this left hand has done–you wouldn’t believe it. It acts independently a lot of times. I don’t even tell it to–I don’t know it’s going to do anything. . . . It seems to have a mind of its own” (p. 213).
It is still difficult to know what to make of such accounts, particularly as many of the patients are not ignorant of the literature written about them. I feel that these are rare incidents that are not fundamentally different from a variety of the silly errors we all occasionally make (unwrapping a candy, holding the wrapper and candy in opposite hands, and then throwing the candy in the trash and placing the wrapper in your mouth). Experimental attempts to document the conflict have failed, and great significance should not be attached to such anecdotal reports.
- Are there qualitative changes in thinking? This is another question that is difficult to answer. Again, there are intriguing reports, but little firm evidence. Different strategies in performance have been assigned to the two hemispheres on the basis of performance on cognitive tasks, but other reports are essentially anecdotal. It may well be that split-brain patients do suffer changes in the nature of their thinking, but as yet such changes have not been satisfactorily demonstrated.
In view of the answers to our five questions, can we regard the consciousness of commissurotomy patients as abnormal? I do not think so. It is first of all clear that the right hemisphere does not seem in any way basically different from the left, except in speech production. The patients appear to possess high-level abilities, conceptualization, and self-awareness within the right hemisphere and to be aware, although perhaps to a limited extent, of right hemisphere processes. It seems wrong, in the face of this evidence, to deny consciousness to the right hemisphere, and we can therefore reject Eccles’s model.
If consciousness resides in both hemispheres, is it divided? The answers to our questions again suggest that outside very special, highly controlled, and artificial circumstances, it is not. It may be possible, particularly with cognitive tasks in the laboratory, to demonstrate some dissociation between the two hemisphere systems. This is no more remarkable than the fact that I can talk while holding a cup of coffee in one hand and scratching my ear with the other. There is extremely slight evidence for conflicting wills in the two hemisphere systems, and a fair amount of evidence for cross-integration between the two lateral systems. There is none for significant qualitative changes in the thinking, everyday responsiveness, or awareness of these patients. On consideration, this is hardly surprising. Both cortical systems are fundamentally bound onto a subcortical substrate with massive integrated bilateral structures, and they cannot sensibly be considered independently from this substrate. The two hemispheres in split-brain patients are no more disconnected than are my two hands. Under certain circumstances they may be observed to be undertaking different but simultaneous activities, and they may retain certain properties of their own, but they are part of a more general system that maintains the integrity of the mind, just as the hands are joined to the arms and trunk, so maintaining the integrity of the body.
Split-brain patients have prompted a valuable debate about the nature of consciousness and its relation to mental processes. However, it seems a mistake to think that the state of consciousness of these patients has been altered in any radical fashion, or to consider that the studies should lead us to revise our traditional model of a single and unitary mind associated with the entire organ of the brain, forming an integral part of a single physical body.
There has been a decline in the appearance of new literature on the commissurotomy patients over the past decade or so. This is partly because the operation is now rarely performed, except as a treatment for midline or ventricular tumors, and then the “split” is unlikely to be complete. Apart from the increasing success in controlling seizure activity by medication, there has been an increasing reluctance to perform the surgical operation. Ironically, this may in part be a consequence of the fact that certain of the split-brain patients made use of the research literature to sue their surgeons. The very success of the experimental literature has been the cause of its closure as an active area of research.
Considerable significance has been attached to the studies of patients who have undergone cerebral commissurotomy. It has been considered that they provide essential evidence about the independence of the two cerebral hemispheres and their lateral specialization. The evidence suggests that, while there is a degree of disconnection between the two hemispheres in these patients, so that under laboratory conditions it can be shown that information cannot be transferred between the two sides of the brain, we should be cautious about overstating the importance of these findings. Split-brain patients do, however, possess cortical hemispheres each of which can independently perceive, remember, think, and respond, to some degree outside the awareness of the other.
It has also been suggested that these patients show evidence of divided consciousness. From a brief review, it is concluded that this view cannot be supported. Certain unusual phenomena may be demonstrated in these patients, but there is no good reason to believe that they possess two independent minds as a result of the operation, or that they tell us very much about the normal operation of consciousness.
The true significance of split brain studies is the dramatic stimulus that they have given to the development of neuropsychological techniques and theoretical models, which are discussed in later chapters.
Divided Visual Field Studies
The perception of briefly presented lateralized stimuli has been studied by experimental psychologists throughout the last century and into this one. Much work was done in the 1950s to investigate the superior recognition of words presented in the right visual field, but while reference was made to “cerebral dominance,” other factors such as reading habits and postexposural scanning were considered important in explaining the findings. It was the study of split brain patients in the early 1960s by Sperry and Gazzaniga (see Chapter 10) that led to the technique of divided visual field presentation being used to investigate the cerebral organization of normal intact subjects in the laboratory, and laid the foundation for contemporary experimental human neuropsychology. Since the mid-1960s, and especially between 1970 and 1985, an enormous number of studies have used the technique, which now forms the most important method of investigation in experimental neuropsychology. A far from comprehensive review around 1980 contained a bibliography of more than 1,000 references, mostly published within the previous 10 years, and this is some indication of the importance of the technique. If the experimental studies were unanimous in their findings, it would be possible to report dramatic progress in this field, but in fact it is difficult to draw firm conclusions from them. This is partly because of the indifferent scientific quality of many of the studies, and partly because of the complexity of the system under study. Nevertheless, some fairly clear conclusions may be drawn.
The logic of the technique of divided visual field presentation is simple and elegant, and has been referred to in the preceding chapter: if a visual stimulus is presented in the left visual field, then it is projected initially only to the right occipital cortex; if presented in the right visual field, it is projected only to the left occipital cortex; and therefore if the subject’s fixation is controlled, stimuli can be injected into one or the other hemisphere. Because the technique usually involves binocular vision, and the stimulus in a given visual field travels through both eyes and via one crossed and one uncrossed pathway to the contralateral hemisphere, differences between the eyes and the visual pathways are balanced out (see Figure 11.1). The subject’s performance in terms of accuracy of report or speed of response can then be studied as a function of the hemisphere of initial presentation, even though processes following initial reception may result in the stimulus information being distributed to both hemispheres of the brain.
The technique relies on accurate control of fixation, and this is usually achieved by directing the subject’s fixation to a desired point. The stimulus is then presented for a brief duration, too short to allow the subject voluntarily to orient the eyes to the stimulus location. It is difficult to determine just how long an exposure may be before such eye movements can be brought into play, but a conservative estimate suggests 150 msec, and exposures up to 200 msec may be acceptable. A tachistoscope (or its computer simulation) is the most common method for presentation of stimuli, but other methods have been used (see, for example, Dimond and Beaumont, 1974). The use of half-occluding contact lenses to limit vision to one visual field for longer periods, allowing the presentation of a wider variety of forms of stimulation including films and very complex detailed displays, was one exciting development, although the technique may not be entirely reliable and has not been widely taken up (Dimond, Bures, Farrington, & Brouwers, 1975; Zaidel, 1975).
It is obviously wise to maintain a check on the direction of fixation at the time of stimulus presentation, although this has been done less commonly than one might wish. However, both direct video and electrooculographic monitoring of eye movements are possible and have been used with success (Dimond & Beaumont, 1972; Young, Bion, & Ellis, 1980). It has also generally been considered advisable to present stimuli unilaterally, that is, in one field at a time, and in a random sequence so that the subject cannot anticipate where the next stimulus will occur. This guards against the loss of accurate central fixation, and might be especially important where concurrent monitoring of fixation is not undertaken. However, quite a number of studies have used bilateral presentation, or “blocked” unilateral presentation, in which a series of stimuli all appear at the same side. There is remarkably good agreement between the bilateral and random unilateral studies, particularly where these involve some additional control over fixation, and where the bilateral studies report the order in which stimuli have been presented (either from the right, or left, visual field first). This is surprising in view of the general level of disagreement among published results and the heated debate that has surrounded the topic.
FIGURE 11.1. Anatomy of the visual pathways. Note how the left visual field projects to the right visual cortex (solid lines) and the right visual field to the left visual cortex (broken lines).
One extension of the technique, particularly with bilateral studies, has been the presentation of a “neutral” stimulus at central fixation, which is reported before the lateral stimuli as an aid to ensure accurate central fixation. It has been feared that this addition might affect the subsequent perception of the lateral stimuli because of the need to process and report the stimulus in the center, and although such an effect has never been clearly demonstrated, the method has become less popular as a result of this concern.
A final methodological question concerns the degree of overlap down the central meridian of the visual fields, and the eccentricity of the stimuli necessary to ensure lateralized presentation. We noted in Chapter 6 how there was an area of bilateral projection around central fixation that was often spared following unilateral lesions. There is also an area served with direct interhemispheric links through the splenium of the corpus callosum, around central vision. The precise extent of these two areas is not clearly established for humans, and it seems wise to avoid the central three degrees of vision around fixation when placing lateralized stimuli, although studies have reported visual field asymmetries with less eccentric stimuli.
Despite the fact that the method is in principle simple and elegant, some of the methodological problems connected with it are more than trivial (see Beaumont, 1982a; Young, 1982; Bourne, 2006). There has even been doubt cast over the whole field at various times, with suggestions that the findings might be due to something quite other than cerebral organization. However, none of these suggestions has ever satisfactorily explained away the very impressive body of evidence that has been accumulated. While the widespread adoption of the technique has been accompanied by some methodological laxity, the relatively coherent results that have been obtained in scores of different laboratories suggest that the technique is robust and generally reliable. It is now accepted as a valid method of investigating cerebral organization for performance in normal humans.
With such a large body of evidence to review, it is difficult to know how to divide it up sensibly for clear presentation. However, one of the earliest variables thought to underlie the appearance of a left or right visual field advantage was the verbal or nonverbal nature of the stimulus material, and this remains a useful way in which to classify the studies and the one that will be used here. Nevertheless, it should not be taken to indicate that this is the basis on which the observed hemisphere asymmetries rest. It is now clear, as we shall see, that a whole host of variables may influence any lateral advantage that emerges.
A comment is necessary about the terms visual field asymmetry and advantage. An asymmetry may be shown by either a right or a left visual field advantage (some people prefer to speak more accurately about the hemifields, and a hemifield advantage). A right visual field advantage, for example, is when performance shows either more accurate or more rapid response following stimulus presentation to the right of fixation than to the left. It is inferred that reception of the stimulus by the left cerebral hemisphere confers some advantage for the speed or accuracy of subsequent processing. Just how and why this occurs is the subject of the theoretical models that we shall discuss shortly. It should be noted as well that the reported lateral advantage or lateral asymmetry is usually based upon the mean results of a number of subjects whose data have been grouped together for statistical analysis. While an asymmetry may be clear from such data, it does not mean that these average results will be typical of each subject. There are wide individual differences (see Chapter 14), and this variability has been of concern to some writers (Colbourn, 1978).
Poffenberger in 1912 is generally credited as the first to show that there are asymmetries in the detection of simple undifferentiated stimuli such as patches of light. His results have been confirmed by modern experiments, and consist of two main findings. The first is that for a stimulus on a given side of the body, there will be a faster response with the hand on the same side than with the opposite hand. This is interpreted as reflecting the difference between the direct pathway in which detection and response initiation can be performed by the same hemisphere (see Figure 11.2, which already appeared as part of Figure 10.2), and the indirect pathway in which information must be passed across the corpus callosum after stimulus detection for response initiation by the opposite hemisphere. Some investigators have suggested that this model enables an estimate to be made of the time taken to transfer information across the callosum (IHTT: interhemispheric transfer time), with a typical result of about 4 msec, for a simple experiment of this kind. However, it is clear that the model involving a single callosal transfer is too simple, particularly with more complex stimuli and tasks, and that is why firm values for IHTT have not been obtained. Incidentally, the values for IHTT of up to 50 msec, which are often found, are very much greater than would be expected from what is known about physiological rates of neural transmission across the commissures (Cherbuin & Brinkman, 2006).
FIGURE 11.2. (a) Direct pathway: the right hand responds to the right visual field (solid line), and only one hemisphere, the left, is involved. (b) Indirect pathway: the left hand responds to the right visual field (solid line), and both hemispheres and the corpus callosum are involved.
It might well be pointed out that compatibility between stimulus and response is an established phenomenon in human performance, and that to respond with the hand on the same side as the stimulus is a more compatible response. Whether this might explain the finding has been elegantly tested by Berlucchi and co-researchers, by the simple expedient of requiring subjects to undertake trials in which the forearms were crossed and the hands therefore crossed over, thus associating the compatible alternative with the indirect, not the direct, route (Berlucchi, Crea, DiStefano, & Tassinari, 1977). Their results show that while response compatibility is a factor in determining the differences between hands, the anatomical differences between direct and indirect routes are more important.
The second main finding to come out of these experiments is that superimposed upon the advantage of direct over indirect routes is a lateral asymmetry, with the advantage for stimuli in the left visual field. This finding applies not only to flashes of light, but also to other simple stimuli to be detected, such as a solid filled dot among an array of unfilled dots. The finding is not universal (almost none is, using this technique), but there seems general agreement on a right hemisphere advantage for simple stimulus detection.
A similar advantage has been found for sustained attention in a vigilance paradigm by Dimond (1979), who has suggested that the right hemisphere might contain the primary mechanisms for watch keeping (a task such as continuous monitoring of a radar display), although not all investigators have agreed with this interpretation. Similar fundamental differences have been suggested by others to account for asymmetries in elementary perceptual processes. The masking and visual persistence investigations that experimental psychologists use to study the initial stages of perception have been undertaken in this context. During such stages, stimuli have to be selected from a barrage of sensory stimulation, and processed to be ready for categorization and subsequent processing for recognition and storage. There is some evidence that these precategorical mechanisms, whereby stimuli are detected but not yet fully identified, may differ between the hemispheres, favoring the right, but there is also criticism of the experiments that have been performed (Cohen, 1976; Hellige & Webster, 1979).
Hemisphere asymmetries have also been sought for a variety of simple perceptual variables. Judgment of brightness has been shown to result in reports of greater brightness associated with the right hemisphere. Similarly, color perception seems to be better performed by the right hemisphere, although only if a relatively difficult discrimination is demanded. This last aspect is a general feature of these studies. It seems that the task must be perceptually demanding, either because the stimuli are hard to discriminate or because they are relatively complex, if the right hemisphere advantage is to be observed. Another aspect of this is the extent to which stimuli might be named, which brings us directly to the fundamental problem with our scheme: it is very difficult to decide when a stimulus is “verbal” or “nonverbal.” Stimuli that are figural and meaningful are often relatively easy to name or to “verbalize,” which brings verbal characteristics into our “nonverbal” stimulus. Conversely, even words have outlines, length, and so on, and although primarily verbal they have nonverbal attributes. It is sometimes suggested that for a right hemisphere advantage to be clearly observed, stimuli should be perceptually complex and difficult to verbalize, and that this is what is implied by “nonverbal.”
Not only brightness and color, but also depth perception, especially in binocular vision, motion perception, and stimulus enumeration, seems generally to be associated with a right hemisphere advantage. Detection of the length of lines is very difficult to test if a verbal response is to be avoided (so as not to encourage verbalization) and other lateral asymmetries in perception are not to interfere with the result. The findings for stimulus duration are similarly rather unclear because of some difficult procedural and methodological problems. It seems at least possible, however, that some aspects of the perception of duration are linked with a left hemisphere advantage. Two final stimulus variables, localization and orientation, have been associated with mixed results, although once again some sense can be made of the data by considering verbal codability. In general a right hemisphere advantage has been observed, and when a left hemisphere advantage has been found it can be explained by the inclusion of a frame or some other background that might aid coding. Even with stimuli that are (arguably) quite codable, such as a clock face, or with horizontal or vertical stimuli, a left visual field superiority has been found.
Form perception has produced unclear results. White in his review (1972) found no evidence of an asymmetry, and the picture is no less uncertain now. To some extent the outcome depends on the complexity of the stimuli, as studies that have used the classic Vanderplas and Garvin figures in identification or matching tasks (see Figure 11.3) have shown. Even so, the results are not simply that more complex figures (with more corners) produce a right hemisphere advantage, or that the “association value” of the stimuli (which is an index of how verbalizable or codable they are) predicts the asymmetry observed. What is clear is that when complex representations of nameable objects are presented, which are presumably readily associated with a name, then a left hemisphere advantage normally ensues (Andresen & Marsolek, 2005).
Some interesting results have been found with the signs used by the deaf. Adults who cannot “read” them, and for whom they are a meaningless configuration, generally show a left visual field advantage when they are shown, but hearing adults who understand them show, like the deaf, a right visual field advantage (Davidoff, 1982). This fits with the idea that it is semantic or linguistic properties that are linked with a left hemisphere advantage.
Faces have been much investigated as stimuli. In general they yield a right hemisphere advantage, although the results are rather complex. Interestingly, schematic faces produce a much less clear asymmetry than do photographs of real faces. However, the right hemisphere seems better at recognizing and matching faces than the left, and also at recognizing emotions. It was suggested at one time that the hemispheres, right and left, might be linked with negative and positive emotions respectively, but it now seems more likely that the right hemisphere is superior at emotional perception from faces, whatever the emotion (Burton & Levy, 1991; Bourne & Hole, 2006).
FIGURE 11.3. Examples of the Vanderplas and Garvin random shapes, with the number of corners as shown. (After J. M. Vanderplas and E. A. Garvin, Journal of Experimental Psychology, 57 (1959), 147–154.)
If the face is to be named, what should we expect: a right hemisphere advantage because it is a face, or a left hemisphere advantage because it has to be named? In fact, researchers have disagreed, although the balance of evidence seems slightly in favor of a right hemisphere advantage. The duration of presentation may be important here. In some experiments the faces were presented for less time than it normally takes to recognize them in central vision, with the result that basic perceptual aspects of the stimuli may have been emphasized at the expense of the properties that would normally be attended to. Upright and inverted faces have been contrasted in an attempt to control for the perceptual aspects of the stimuli, but just what cognitive processes are invoked by an inverted face seems unclear (Leehey & Cahn, 1979; Young & Bion, 1981).
An intriguing study by Chabris and Hamilton (1992) examined chess masters, who are known to be superior at encoding and recalling the arrangement of a position during a game of chess; they possess a particular semantic structure of the “chess domain.” The study found the right hemisphere of the chess masters to be superior at understanding “sensible” chess positions, but the left hemisphere to be better at grouping pieces in unorthodox positions. The right hemisphere appears critical for chess skill.
It is difficult to summarize the above information on visual field advantages without doing grave injustice to it. Those interested in this topic are advised to consult the extensive review by Davidoff (1982). However, repeating the warning that there are many contrary findings, it seems that if basic low-level perceptual functions are examined, then some, but not all, may yield visual field asymmetries. When an asymmetry is found, it tends to be a right hemisphere advantage. Stimulus complexity and duration (with shorter durations producing perceptually more difficult discriminations) are implicated in determining a right hemisphere superiority. With more complex stimuli, familiar drawings with names produce a left hemisphere advantage, while faces and certain shapes may be associated with a right hemisphere advantage. The verbalizability of the stimuli may be an important variable, as may their symbolic content.
The results of studies using “verbal” stimuli have been less contradictory than the “nonverbal” studies. Certainly if stimuli such as words, strings of letters, and single letters or digits are presented for identification or matching, a strong left hemisphere (right visual field) advantage is normally found (Beaumont, 1982b). In cases where the effect has not been found, there is often reason to suspect that the stimuli have been processed on their perceptual rather than their verbal attributes. When words or letter strings are presented, the right hemisphere superiority (in responding to stimuli in the left visual field) is found whether the presentation is horizontal or vertical, although the horizontal orientation tends to produce the stronger effect. This demonstrates that while left-to-right scanning patterns, perhaps based upon reading habits, may have some influences on the left field superiority, they cannot be a complete explanation (Day, 1979).
What is surprising about these studies in view of the conclusions we drew previously is that linguistic parameters do not seem to relate directly to the left hemisphere advantage. The use of concrete nouns contrasted with abstract nouns, the manipulation of imageability, nouns opposed to verbs, or the effects of word frequency do not seem to relate in any coherent way to the left hemisphere superiority (Day, 1977; Bradshaw & Gates, 1978; Cousin, Peyrin, & Baciu, 2006). Once a lateral asymmetry for the recognition of linguistic stimuli had been found, it was confidently expected that the more “verbal” and less “nonverbal” a stimulus (for example, abstract versus concrete nouns), then the stronger would be the lateral advantage. This has not been clearly demonstrated. Even with strings of letters, or “nonsense” words, it does not seem to be the case that there is a left visual advantage if the string is more pronounceable (“BOV” rather than “ZVH”), or more similar to meaningful English words (Leiber, 1976). This very much undermines the assumption that left hemisphere superiority is determined by the operation of purely linguistic processes.
Although identification of stimuli for simple report or matching has been the task most commonly employed, certain more demanding tasks have also been used. A particularly important one has been the presentation of letter stimuli for physical and nominal matching (Cohen, 1972; Hellige, Cox, & Litvac, 1979; Ledlow, Swanson, & Kinsbourne, 1978). This is a task much studied in cognitive psychology, and requires the subject to determine whether two letters are identical (physical match: “AA but not Aa”), or similar in name (nominal match: “AA or Aa, but not AB”). Specific instructions are given to the subject as to which type of match to perform, and thus determine the cognitive demands of the task. In general, the requirement to make a physical match is associated with a right hemisphere advantage, at least for the “same” responses, although this has not universally been found. Where a nominal match is required, then a left hemisphere superiority is almost always found, although it is again clearer for “same” responses.
It is interesting that the distinction between nominal and physical matching may reflect on tasks where the type of match required is not clearly specified. Although the type of matching may be inherently clear in the form of stimuli presented, subjects may nevertheless adopt different cognitive strategies (Bryden, 1978); that is, they will adopt some preferred or habitual approach to solving the problem of determining a match. If different subjects select different strategies, then this may account for some of the variability found in the research reports. Also, the fact that a left hemisphere advantage is generally found for simple word or letter matching implies that although a physical matching strategy might be sufficient, subjects nevertheless generally adopt a nominal matching strategy. The extent to which very subtle aspects of the experimental instructions and procedure may influence the adoption of cognitive strategies is also of concern to some investigators.
A delay may be introduced between the elements to be matched in the physical/nominal matching paradigm, in order to investigate the role of memory components in determining lateral asymmetry. Some researchers have found that introducing a delay in physical matching produces a shift from right to left hemisphere advantage, implying the necessity of verbally encoding the stimulus, but this has been contested in recent reports (McCarthy, 1980). Nevertheless, what these experiments clearly demonstrate is that purely cognitive factors may influence lateral asymmetry, so that even though the stimuli and presentation conditions may be identical, the task demanded may determine whether a left or right hemisphere advantage is found.
An obvious question that arises is the part played by the response in producing a lateral asymmetry. If a left hemisphere advantage were only to be found when a vocal or right hand response was demanded (both initiated presumably by the left hemisphere), this might well explain the phenomenon. The problem is that it is very difficult to change the response without to some degree changing the task. However, when a manual response is required, most investigators have demanded responses with each of the hands in a balanced order, and some have subsequently analyzed the effect of this variable (Besner, Grimsell, & Davis, 1979; Segalowitz, Bebout, & Lederman, 1979). The results are conflicting, and this may reflect the suspicion that quite subtle aspects of the response, such as how fine a movement is required (with more distinctly contralateral control), may be crucial variables.
It is more difficult to compare manual with vocal responses than to compare responses with the two hands, although in general the evidence is that vocalization in itself is not critical in determining the left hemisphere advantage. This has been shown by contrasting a meaningful response with a vocal, but meaningless, response (such as “pring”) in a go/no-go paradigm where the subject either responds or refrains from responding. Bradshaw and Gates (1978), studying lexical decision (“Is the stimulus a word?”), found a stronger left hemisphere advantage with overt naming than with manual response. Most researchers have been content to assume that response systems are relatively independent of other cognitive operations and, beyond balancing hand of response, have not taken much account of response variables.
Another factor that, it has been suggested, is important in determining lateral asymmetries is the size of the stimulus set (Hardyck, Tzeng, & Wang, 1978). This conclusion was based on a series of experiments in which matching was performed between various combinations of English words and Chinese characters by English and bilingual Chinese subjects. Part of the aim of the experiment was to compare the lateral advantage for perceptually complex characters when they were meaningless (English subjects) or possessed semantic properties (Chinese subjects). It was found, however, that asymmetries appeared only when a very restricted set of possible stimulus items was employed. The suggestion is that the subject must know the range of possible stimuli, or rapidly deduce it, and then adopt a strategy that involves “referencing a table of known values,” if a left hemisphere advantage is to be found with verbal stimuli (see also Chiarello et al., 2006). However, there has been no formal analysis of the literature to support this hypothesis, and many examples can be found where new information is presented on each trial and a right visual field superiority is nevertheless found.
Memory of course plays some role in almost any experiment, but does it play some fundamental role in lateral asymmetry? Studies that have explicitly required a target item or a target set to be held in memory, and to which later stimuli must be matched, have generally yielded a left hemisphere advantage. This might be expected on the basis of the verbal material alone. However, that requiring memory for verbal items does not necessarily produce a left hemisphere advantage is shown by the experiment of Klatzky and Atkinson (1971), who asked subjects to hold verbal items in memory and then presented either single letters or pictures of objects. The initial letter of the depicted object, or the letter presented, were to be matched to items in memory. A left hemisphere advantage emerged for the pictures, but not the letters. All this brings us back to the problem of strategies. We cannot know that subjects will necessarily adopt semantic coding for verbal items, for they might alternatively attend to the spatio-perceptual elements of the stimulus, and attempts to manipulate such strategies directly have not generated clear results (Metzger & Antes, 1976). Memory components are clearly implicated in lateralized cognitive processing, but there is no reason to believe that they are entirely responsible for lateral visual asymmetries.
Practice is another variable that complicates the investigation of hemisphere specialization (Jonides, 1979; Ward & Ross, 1977). Practice clearly does have some effect, but just how is not yet clear. The nature of the stimuli and the task, the difficulty of the task, the familiarity of the material and the experimental paradigm, the compatibility of the response, and the pattern of test trials and rest pauses all interact with the effects of practice. Two factors may be at work, however. One is that with increasing familiarity with the stimuli and task, there is a shift from attention to the spatial and configurational properties of the stimulus material to a processing mode with important verbal and semantic components. This would be reflected in a shift from right hemisphere advantage to left hemisphere advantage. At the same time, increasing familiarity with the task might reduce the task’s difficulty and consequently the strength of lateral asymmetry. These two factors will work to some extent in opposition, and interacting with the host of other variables may result in the very complex results that have been reported.
Finally, we might note that additional tasks have sometimes been added to the primary visual field task in an attempt to elucidate the nature of the lateralized processes (Kinsbourne & Hicks, 1978). In general the addition of a concurrent verbal task to a divided visual field task has been found to reduce or reverse the left hemisphere advantage normally observed. There are difficulties, however, in interpreting such studies (Cohen, 1979; Hellige, Cox, & Litvac, 1979). One simple problem is the ambiguity of activation and interference. Interpretations have suggested that the secondary task may activate one hemisphere and so make its contribution to processing the divided visual field task more prominent. On the other hand, it may interfere with processing, and so cause greater participation of the opposite hemisphere. Such activation and interference will result in opposite effects on lateral asymmetry, but there is no way of knowing which will apply, so that although with increasing secondary task difficulty there may be a shift from activation to interference, there is no independent way of knowing when this will occur. These concepts have tended to result in most unsatisfactory post hoc explanations of experimental results, which have done little to further our understanding of lateralized cognitive processes.
In conclusion, a clear left hemisphere advantage is generally found in studies that require identification of “verbal” stimuli, such as words, letters, and letter strings or digits, although this should not lead us to deny that the right hemisphere is entirely without such functions (Querné, Eustache, & Faure, 2000). The role played by semantic and linguistic parameters remains unclear but does not at present satisfactorily explain the left hemisphere advantage. If physical matching is required, a right hemisphere advantage may be found, depending upon cognitive variables inherent in the task. These results are relatively independent of response mode, although both this and mnemonic processes may play some part in determining lateral asymmetry. Practice is a complex variable that deserves more study, and the whole picture is further complicated by other subject and task variables. Nevertheless, over a very considerable number of experiments, the stability and importance of the left hemisphere advantage in association with verbal stimuli remain clear. (A more detailed and fully referenced review is to be found in Beaumont, 1982b.)
There has been a tendency for divided visual field studies to be reported independently of any explicit theoretical foundation. Visual field differences have thus been reported as if they were little nuggets of fact that had simply been dug up and deposited on a mound of knowledge. Such an approach is limited. At best, data accumulate and can constitute a resource for those who want to understand the processes that underlie performance asymmetries. At worst, the hypotheses and assumptions made in collecting the data are never made explicit and yet play a part in determining the findings that are reported. An associated problem, deriving from the ease with which divided visual field experiments may be carried out, is the large number of small-scale isolated studies that report on a particular task under a single set of conditions. What the theory builder needs are extended studies in which conditions are systematically varied in order to clarify the mechanisms that produce the phenomenon under study.
While several theories accounting for lateral asymmetries have been proposed, there has been little critical conceptual analysis of the various theoretical positions. However, in addition to Bryden’s 1978 review, there is also an invaluable analysis by Cohen (1982). Indeed, the classification of theoretical models of hemisphere specialization suggested below (and see Figure 11.4) owes a great deal to her insightful and creative review.
This has been the most common variety of model adopted in divided visual field researches. If a model is implicit, then it is usually a structural model that is assumed. The general idea behind these models is that psychological functions may show lateral asymmetries, as we know from the behavioral data, and these functions may be localized in cerebral structures that are lateralized to one of the cerebral hemispheres. In this the models reflect current concepts in clinical neuropsychology.
Structural models do not necessarily exclude a variety of other factors that may contribute to lateral asymmetries, such as postexposural scanning or the influence of reading habits. They do, however, assume that when such factors are not themselves a function of cerebral lateralization, they are of minor importance by comparison with the structural cerebral factors.
This raises a whole host of conceptual issues about the allocation of cognitive functions to physiological structures, and accounts for the rather unsatisfactory status of some of the concepts used in building structural models. There is insufficient space to deal with the topic here (see Beaumont, 1982c), but greater attention should be devoted to this philosophical problem.
FIGURE 11.4. Theoretical models of hemisphere specialization. (Modified from Cohen, 1982.)
Structural models may be based upon either absolute or relative specialization. Where the specialization is thought to be absolute, some mechanism must be proposed to effect the transfer of stimulus information if it arrives at the “wrong” hemisphere. This transfer might be automatic, so that all information is transferred and is available to both hemispheres, but only the appropriate hemisphere takes up the processing. Transfer might alternatively be conditional, either following assignment by some initial sorting mechanism, or following the failure of the receiving hemisphere to complete the required processing. A further alternative is that transfer may occur and be followed by both hemispheres processing the material, with the specialized hemisphere reaching a conclusion faster, or some mechanism arbitrating the results. Davis and Schmit (1973) provide an example of such a model. There are problems with all these alternatives. They all seem to rely on mechanisms (for allocation of processing, for deciding when processing has failed, or for arbitrating between solutions) that have an unspecified location and for which there is little evidence. Moreover, in certain cases the systems are wasteful and implausible, such as dual simultaneous processing.
There is also difficulty in squaring absolute specialization models with the split-brain evidence we noted in the last chapter, which suggests (albeit in abnormal patients) that both hemispheres possess some capacity for most functions. Differential error rates may be attributed to “degradation” in callosal transfer, which seems implausible in the large degree often claimed, when other transmission pathways, from primary up to tertiary cortex, are used with much higher fidelity. The failure to establish good stable estimates of IHTT and to account for withinindividual shifts in laterality also argues against the adoption of an absolute specialization model.
On the other hand relative specialization models, in which both hemispheres are believed to be capable of performing most functions but with different levels of efficiency, have the problem that they are too imprecise. They fit very well with most of the data from a variety of sources, but it is very difficult to derive firm predictions from them. Even the inference that increasing difficulty should be associated with greater asymmetry cannot generate firm hypotheses because of the inadequate definition of difficulty, and this imprecision is in turn reflected in confusion among the findings. Thus, although the relative specialization model in many ways seems more acceptable than the absolute model, it only gains its acceptability by a serious loss in explanatory power.
Structural models may be based on either stimulus or processing determinants. We have already seen how inadequate was the division of stimulus type into “verbal” and “nonverbal” as a predictor of lateral asymmetry. It nevertheless figured prominently in early theories and still underlies the assumptions in many procedures designed to exercise left or right hemisphere functioning.
It is now more fashionable to consider processing determinants. These were initially considered in terms of a number of dichotomies that assigned serial, sequential, temporal, or analytic processing to the left hemisphere and parallel, holistic, or gestalt processing to the right. Many have treated these terms as roughly equivalent, which they are not. The description of the processes has frequently been post hoc, and firm criteria for the identification of such processes have almost always been omitted. Although experimental techniques exist for the identification of such processes, they have rarely been employed, and when they have been used, the results have not always been clear (Cohen, 1973; Patterson & Bradshaw, 1975). Of the mode of processing schemes, none accounts well for the experimental data.
The most recent formulations feature stages of processing. These may be in terms of component operations (Cohen, 1977) or informationprocessing models (Moscovitch, 1979).
The component operations model features the contribution of a number of elementary components in cognitive processing, each of which may be characterized by hemisphere specialization. Lateralization will then be determined by the specialization inherent in visual analysis, phonological analysis, or semantic analysis as each is called into operation. This approach is useful in that it allows for some flexibility in the operation of the contributory processes, and it can account for task variability if this is mediated by strategy selection. It suggests a way forward in the attempt to identify and specify the exact nature of the contribution of each element. However, like relative specialization, it is weak at generating firm and testable hypotheses, and perhaps only succeeds in shifting the phenomenon of lateralization to another part of the model without explaining it.
By contrast the information-processing model is more precise. It assumes a model with a clear sequence of processes from perception through categorization to memory and response (from low-level through to high-level). Lateralization is presumed to be characteristic of certain of these stages (mid- and high-level) and is carried through to later stages. The attraction of the model is that it has clear reference to general models in cognitive psychology, although cybernetic models now seem a little old fashioned in this context. The problem is that while the model is precise and relatively well formulated (although Cohen is highly critical of certain aspects), it does not always fit well with the data, and is for this reason not entirely acceptable.
Two more recent developments are worth noting. The first concerns the differentiation of categorical processes, in that a left hemisphere advantage has been demonstrated for “what” (lightness) and “where” (location) systems, while a right hemisphere bias was found for “whatever” (coordinate processing) systems (Parrot, Doyon, Démonet, & Cardebat, 1999). The second is the HERA model: hemispheric encoding/ retrieval asymmetry. In this model the prefrontal cortices of the left and right hemispheres are respectively involved in episodic encoding and retrieval in memory. Developed originally from imaging data (see Chapter 13), the model has been supported by divided visual field data (Blanchet et al., 2001), although how these different processes are related to lateralized modes of processing is not yet entirely clear.
Attentional models are a more recent development and have not gained such wide popularity. The idea underlying them is that the cerebral system has a certain capacity that can be allocated in a flexible way. Arousal, activation, and expectancy will result in attention and associated processing capacity being dedicated (and divided between the hemispheres) in a particular way. The most notable exponent of this form of model has been Kinsbourne (1975).
Such models do not deny the existence of structural differences but claim that these are small in relation to the effects of attentional variables. These effects can be best seen, and are accessible for study, in the mechanism of priming. Priming may be determined by stimulus location, by stimulus type, or by a concurrent task. When the stimulus location is predictable (or when the subject adopts a strategy of expecting stimuli at a given location, even if presentation is random), then the subject may direct attention (although not necessarily his or her gaze) to that area of space, resulting in better perception of stimuli at that location.
The stimulus type may prime the hemispheres either by putting the “appropriate” hemisphere in readiness for a certain type of stimulus, orby directing attention to the side of sensory space opposite the appropriately primed hemisphere. Both these mechanisms would result in better performance in the lateral visual field opposite the more specialized hemisphere. The degree to which such a mechanism is supported by the evidence is uncertain. Some experiments using mixed trials find effects that could be explained in terms of stimulus priming (Berlucchi, Brizzolara, Marzi, Rizzolatti, & Umilta, 1974), but others have failed to find any specific effect of stimulus type expectancy, even when the expectancy can be shown to have an effect on overall performance. Beaumontand Colley(1980),for example, presented subjects with a mixed sequence of shapes or words that appeared randomly in the left or right visual field. In a given set either shapes or words were more common, and the subject was clearly informed about this. While subjects responded more rapidly to the more common type of stimulus in any set, whether a stimulus type was more or less common did not affect the asymmetry for that stimulus type, as attentional models would predict. The effects of perception in the unprimed hemisphere are not, however, always as would be expected from attentional models: presenting pairs of stimuli that could be any pairing from among words, faces, and shapes, presented in balanced order to the two hemispheres,Hines(1975)foundonlylimitedsupportforanattentionalexplanation of the results.
These models have been based upon the hypothesis of attentional bias, but variants of the attentional model have been suggested by Friedman, Polson, Dafoe, and Gaskill (1982) based upon the independent allocation of attentional resources within the two hemispheres, and also through differing allocation of attentional resources, by Nicholls and coinvestigators (Nicholls, 1994; Lindell & Nicholls, 2003).
Reference has already been made to the use of concurrent tasks and the difficulties of interpreting the results of such studies. Nevertheless, some support for attentional models comes from such work. Kinsbourne (1973) required subjects to report the location of a small gap in the outline of a square while performing a concurrent task. When this task was to hold a string of words in memory, a left hemisphere advantage was found, but when remembering a melody, a right hemisphere advantage. Such studies are difficult to explain without recourse to attentional mechanisms, but models that rely on secondary task performance are nevertheless generally inadequate in their explanatory power.
The Dynamic–Structural Model
Despite its rather grand title, this model really does no more than accept that at present there is good evidence for both structural and attentional models. Both have weaknesses and strengths. While the dynamic– structural model has not yet been clearly formulated, the approach is implicit in much current research, and seems the only sensible way to proceed for the moment. Most researchers would regard the structural aspects as more important than the attentional processes, but few wish to deny the real contribution of attentional mechanisms. We are left, therefore, with a model that combines structural features with attentional processes. The structural components confer relative specialization determined by the contribution of particular cognitive processes. The attentional mechanisms include priming in association with cognitive strategy selection. We cannot pretend that this is a satisfactory model of cerebral specialization, but it seems to be the best sense that we can make of the experimental findings that have been accumulated to date.
The divided visual field technique has been the principal method of investigation in experimental human neuropsychology, and has been employed in a large body of research. Somewhat confused findings have resulted but nevertheless, out of this mass of data, a number of conclusions have emerged. These conclusions are now accepted as showing that there are lateral differences in the specialization of the cerebral hemispheres for psychological functions. These differences seem, in general, to reflect the form of cerebral organization that has been inferred from the study of clinical patients.
The mechanisms that produce these lateral asymmetries are still unclear. It is not possible to say why, for example, the recognition of words is associated with superior right visual field performance. However, it is reasonable to believe that certain aspects of cognitive processing, as yet poorly specified but conceptualized in terms, perhaps, of linguistic or phonological or semantic analysis, are more effectively performed in the left cerebral hemisphere. The lateral advantage observed in such processes may be modified by the operation of attentional variables.
This may seem a rather weak conclusion in view of the research effort devoted to the topic. Nevertheless it comprises a significant advance in understanding a very complex dynamic system that, until the introduction of methods such as the divided visual field technique, remained a dark mystery, largely hidden to scientific investigation.
The technique of dichotic listening is the auditory parallel to divided visual field presentation. Like the latter, it was first developed within experimental psychology, and its relevance to neuropsychology was recognized only later. It was Kimura (1961) who first pointed out that the asymmetry that had already been observed could be attributed to cerebral lateralization, thus introducing the technique into human experimental neuropsychology at about the same time as the divided visual field technique was being explored with split brain patients.
Although dichotic listening historically has not attracted as much attention as divided visual field presentation, perhaps because of the added technical difficulty of creating the stimuli, the ability to more easily manipulate stimuli using laboratory computers has resulted in increasing use of the technique in recent years.
The logic that underlies dichotic listening is exactly the same as that of the visual technique. Stimuli presented to the right ear are considered to be directed to the left hemisphere, and via the left ear to the right hemisphere. Subsequent performance measured by accuracy or response latency can be taken to reflect the operation of systems lateralized to the two hemispheres. The only significant difference is that in dichotic listening stimuli are presented to the two ears simultaneously, although associated procedures of lateralized auditory presentation may be also used, in which stimuli are not bilateral and simultaneous. These are not, however, properly termed “dichotic.”
As has already been pointed out in Chapter 4, although the primary projection of the visual system is completely lateralized, that of the auditory system is not (see Figure 4.2). For this reason, the technique is not quite as simple as divided visual field presentation. Dichotic listening relies on the dominance of the crossed contralateral pathways over the ipsilateral uncrossed pathways. The clinical evidence supports this contralateral dominance, at least under conditions of bilateral competition, as does the evidence from split-brain patients (discussed in Chapter 10). There are also physiological differences in the size of the auditory pathways, the crossed tracts being larger.
The extent of the crossed dominance in audition has, however, been questioned. Some workers have shown that the suppression effect seen in split-brain patients (see p. 206) operates only for dichotic speech stimuli and not for pure tones (Efron, Bogen, & Yund, 1977). It has also been suggested that the apparent suppression of left ear material may result from spectral-temporal overlap between stimuli, and so may be a more peripheral perceptual phenomenon than has generally been thought (Springer, Sidtis, Wilson, & Gazzaniga, 1978). These studies indicate that the dominance of the crossed pathway is not so clearly established as was once thought, but for the present we will accept that crossed dominance applies because it enables us to make sense of the experimental findings. Nevertheless, we should remember that if crossed dominance did not apply to a certain auditory stimulus, the logic of the technique used with normal human subjects would be completely undermined.
Careful attention must be paid to the construction of stimulus tapes or computer-generated stimuli for dichotic listening. The characteristics of the stimuli that arrive at the two ears must be carefully balanced, and the onset of the two stimuli carefully aligned. It has become increasingly common to use computers to generate and control the presentation of dichotic stimuli, and this allows very precise control of not only the onset but also the duration of the stimuli.
The way in which subjects are asked to respond is also very important, although it has been little studied. Given that the subject will hear pairs of simultaneous stimuli, whether he or she is asked to report one or both the items, and which is to be given first if both are to be reported, will greatly affect the results. Many workers have simply allowed free report in which subjects must report as many items as they can. Others have directed attention to one or the other ear, or controlled the order of reporting from the two ears. The findings from what little study there has been of the effect of response mode are equivocal. However, while certain forms of response are more appropriate to specific experimental designs, controlled report must in general be superior to free report.
Much study has been made of how to score the results, which is a much less simple problem than it might appear. The asymmetry observed is usually described in terms of a left or right ear advantage, and this description must account for the effects of overall accuracy and the effects of guessing in forced choice response modes. The simplest solution of taking the difference between the scores at the two ears and dividing by the total score ([Right – Left] ÷ [Right + Left]) is not really satisfactory. The difficulty partly hinges on whether we consider that a performance of 20% in the right ear and 10% in the left ear represents the same asymmetry as 80% in the right and 70% in the left; and what if the scores are 100% in the right and 95% in the left: Is that the effect of a “ceiling” on performance? Decisions about how to treat such scores will have a fundamental effect on the interpretation of the results of the experiment (see Berlin & Cullen, 1977). Various solutions have been proposed and the most generally accepted is a slightly complicated index (the e and eg coefficients) developed by Repp (1977). Few studies have, however, employed the more sophisticated scoring methods.
A more detailed discussion of the problems in the dichotic listening technique will be found in Beaumont (1982). For the present, we must recognize that there are some methodological difficulties inherent in it. The production of stimuli may be critical, as may the method of scoring, and there is some niggling uncertainty about the physiological substrate of the technique. Nevertheless, as we shall see, findings using dichotic listening have been remarkably consonant with those from other neuropsychological methods, and this must increase our confidence in its use.
LATERAL EAR ASYMMETRIES
In the classic dichotic listening experiment, three or four pairs of stimuli, one of each pair to each ear, are presented at a rate of one pair every one or two seconds. At the end of the series the subject must report the stimuli. The ear associated with more correct responses (or occasionally the earlier responses) is described as showing an ear advantage. Kimura’s 1961 study had used digits as stimuli and found a right ear advantage. Such stimuli have been the most consistent in producing a clear rightside advantage, but the effect is not limited to numbers. A right ear advantage is also generally observed for words, and not only for meaningful but also nonsense words. Even artificially generated consonant– vowel (CV) syllables yield a stable right ear advantage, indicating that left hemisphere processes are not associated only with linguistically meaningful material. Studies have proceeded to investigate most linguistic aspects of verbal stimuli. These studies are important in revealing, by the degree of lateral asymmetry that may be observed, the processes involved in producing the lateral advantage. They also provide important information for those investigating speech perception and language comprehension. The common factor, however, is that most speechlike and language-related stimuli are reported more accurately from the right ear if presented in dichotic pairs.
By contrast, certain stimuli are associated with a left ear advantage. These include melodies, sonar signals, environmental sounds, and “nonverbal vocal tract sounds.” Environmental sounds might be a running tap, traffic, or teeth being brushed, and vocal tract sounds might be coughs, hums, or grunts. These stimuli seem to have in common that they are patterned and nonverbal. In particular they are not at all speechlike, and this may be an important feature in determining the involvement of the right hemisphere. More recently, both emotional tone (Sim & Martinez, 2005), and the detection of deception (Malcolm & Paul, 2005), have been shown to moderate ear asymmetries and have some association with the right hemisphere.
The simplest model that can be set up for dichotic listening performance looks very much like that for divided visual field performance. Speech and language stimuli are associated with a left hemisphere advantage, while nonverbal stimuli are associated with a right hemisphere advantage.
Before looking at how this model has been developed, it is worth asking whether there is an asymmetry in response to very simple stimuli, reflecting findings in the visual modality. Obviously we cannot perform a parallel dichotic experiment, but we can test the reaction time, shown by the hands, to simple auditory tones presented to the right and left ears. From this test it seems there is some evidence that faster reaction times are associated with the “direct” route from ear to ipsilateral hand, in contrast with the crossed indirect pathway from ear to contralateral hand (Provins & Jeeves, 1975). A study by Elias, Bulman-Fleming, and McManus (2000) has also shown that the Poffenberger effect, illustrating interhemispheric transfer (see p. 224), can be demonstrated in the auditory modality. However, asking subjects to attend only to one ear modifies this effect: if subjects attend to the right ear, the difference in reaction times between hands disappears, if to the left ear, the left hemisphere advantage is retained and there is a faster reaction shown by the left hand (Spellacy & Wilson, 1978). This emphasizes the importance of central, cortically mediated mechanisms in modifying performance through selective attention.
Left Hemisphere Processes
What precisely underlies the left hemisphere advantage? An early idea was that the left hemisphere might be particularly equipped to deal with speech perception. Speech perception is more difficult to study than might at first appear, as is reflected in the slow progress being made with automatic systems for speech recognition. The difficulty is that the sounds which go to make up a word are not strung together simply. Many of the component sounds arrive in parallel, so that the difficulty in speech perception is to unravel parallel components, a process referred to as drastic restructuring (see Liberman, 1974; Springer, 1979). This difficulty is increased because a perceived phonetic segment may result from different acoustic cues in different contexts, that is, the same (perceived) sound may come from differing physical stimuli.
Drastic restructuring has been linked with right ear advantage when hard-to-decode stimuli have been contrasted with easy-to-decode stimuli. Stop consonants (/b/,/d/,/g/,/p/,/t/,/k/) need more restructuring than isolated vowels, and stop consonants with an added /a/ vowel produce a right ear advantage, while synthetic vowels do not. In nonsense words, if the consonants are contrasted (“bip” versus “gif”) then a larger right ear superiority will result than if the vowels are contrasted (“bip” versus “bap”). Sounds that require intermediate levels of restructuring, such as liquid consonants (/r/,/l/), semivowels (/y/,/w/), and fricatives (/s/,/v/,/f/,/z/), have been reported to produce an intermediate right ear advantage.
Darwin (1971) showed the importance of the presence of formant transitions (certain features of speech stimuli) for speech perception, thus supporting the restructuring hypothesis. Other linguistic factors have also been shown to be related to right ear superiority: if grammatical structure is present, even in a sentence of nonsense words (“The wak jud shendily” has grammatical structure, “Bul hudky gu nee” does not; Zurif & Sait, 1970), then it may produce the right ear superiority. Linguistic tone (intonation) may also be involved; if a language like Thai is used (with Thai speakers as subjects) in which the same sound with different intonation can have quite disparate meanings (naa can mean “aunt” or “field”), then a right ear advantage can be seen for stimuli that differ only in tone.
However, that the right ear advantage reflects the involvement of speech perception can only be a partial explanation. For one thing, this advantage may be seen for language stimuli that do not involve the effect of formant transitions. Some very elementary aspects of stimuli, such as intensity, time, and frequency, may also moderate the right ear advantage (Berlin, 1977; Brancucci, Babiloni, Rosini, & Romani, 2005); and using vowels, but masking them by adding white noise, has also been shown to produce a right ear superiority (Weiss & House, 1973). Both these results indicate that purely acoustic factors may also be linked to left hemisphere processes. Some of these processes may still be linked to language perception, but it cannot be language perception alone that is lateralized to the left hemisphere.
Cutting (1974) compared performance using CV stimuli with that using appropriate or inappropriate formant transitions. Both types of stimuli yielded right ear advantages. However, only the stimuli with appropriate formants produced a discrimination curve that would be expected for language-related perception. This seems to point to a purely auditory component in the processing of CV syllables with inappropriate formants, yet they produced a right-sided advantage. Godfrey (1974) has systematically manipulated such features as the signal-to-noise ratio, acoustic–phonetic distinctness, and vowel duration, and his results also suggest that the effects are being created at the auditory rather than the phonetic level of processing.
This literature can be rather technical for those not expert in linguistics and speech processing. However, most workers now accept that there may be two types of processor associated with left hemisphere specialization (Bub & Whitaker, 1980; Springer, 1979). One of these is related to the restructuring and encoding of speech stimuli, and operates at a phonetic level. The other operates at a purely acoustic level, and will function in any difficult acoustic discrimination, even if nonlinguistic in character. It has been suggested that this second processor may be set to detect rapidly changing frequency information, or that it performs temporal order judgments. None of these hypotheses can be regarded as a very satisfactory explanation of the data, and it may be that some better model can yet be found. Nevertheless, it is clear that, while the left hemisphere is implicated in speech and language processing, this is not a sufficient explanation of many of the right ear advantages that have been reported.
Right Hemisphere Processes
That certain non-speechlike sounds can produce a left ear advantage is undisputed; but what processes actually determine right hemisphere lateralization? From what has been said of the left hemisphere, it seems unlikely to be a matter of nonverbal auditory discrimination.
This question has been most commonly tackled through the study of musical stimuli. Certain kinds of musical stimuli produce a clear left ear advantage, and it is possible to manipulate different aspects of such stimuli to discover just what determines the advantage. At least, that has been the strategy behind research in this area. Unfortunately, the results have not yet enabled any clear conclusions to be drawn about the processors resident in the right hemisphere (for reviews, see Craig, 1979; Damasio & Damasio, 1977; Gates & Bradshaw, 1977a; Mitchell & Crow, 2005; Wyke, 1977).
Kimura, in 1964, described a left ear advantage for dichotic melodies, but this result has not always been supported by later similar experiments. For instance, Gordon (1970) found no lateral asymmetry for melodies, although he did find an effect for dichotic chords. Taking even simpler stimuli, the ear advantage for pitch (or frequency) has not been clearly established. There is some suspicion that this may be because pitch is also critically involved in speech perception, and the context of the stimuli may therefore be important in determining whether they are treated as being more or less speechlike. Other factors that have been shown to be relevant are the delay between the stimulus and a subsequent comparison stimulus; the complexity of the stimulus; and the subject’s expectation as to which ear will receive the stimulus. The results with timbre are equally inconsistent, and time and rhythm have received little attention.
It is possible to make some sense of the data, however, by considering just how subjects are likely to treat the stimuli. One difficulty of decomposing complex stimuli into their components is that subjects may not process them in the same manner. As “musical” stimuli become more and more simplified, they are less and less distinct from simplified speech sounds, and it becomes more likely that subjects will process them as if they were speech sounds. This is supported by two pieces of evidence: that, by and large, the right hemisphere advantage is found more readily with structured musical passages, and that characteristics of the subjects themselves can have a profound effect on the lateral advantage observed.
Some of the conflict, between the findings of Kimura and Gordon with respect to melodies, for example, can be resolved by reference to the subjects employed. A number of studies have specifically examined the effects of the musical training and experience of the subjects. For example, in a study by Johnson and colleagues (1977) musicians were asked to recognize conventional melodies or random note sequences. All subjects showed a left ear advantage for the random note sequences. With the melodies, however, the results were especially interesting, for trained musicians who could transcribe music showed a right ear advantage, but the trained musicians who could not transcribe music showed a left ear advantage. The interpretation of these findings is that complex musical stimuli are processed by right hemisphere mechanisms unless the stimuli are both meaningful and can be encoded in a formal symbolic way. If they can be so encoded, as they may be by those who can tran scribe music, then left hemisphere mechanisms may be called into play.
This finding has an interesting parallel in an experiment with morse code (Papçun, Krashen, Terbeck, Remington, & Harshman, 1974). Morse code was presented dichotically to morse code operators and to these unfamiliar with morse code. The operators showed a consistent right ear advantage, but subjects unfamiliar with morse code only showed a right ear advantage if there were seven or fewer pairs of elements. With longer lists, which they were presumably unable to process semantically, or which could not be handled by the sequential analysis of components of the stimulus train, there was a left ear advantage, reflecting a switch to right-hemisphere-based functions.
However, not all studies with trained musicians have found a right ear advantage for complex stimuli: both Gordon (1980) and Zatorre (1979) found a left ear advantage and no effect of the degree of musical training. Also, in an elegant study of the parallel effects of verbal and musical components of the stimulus, which used either numbers superimposed upon piano notes or digits sung in a tonal pattern, a separate right ear superiority for the verbal elements and left ear superiority for the musical elements was maintained. Certain paradoxical effects have also been reported when comparing dichotic tasks with dual task interference studies employing musical stimuli (Lim, Lambert, & Hamm, 2001).
No doubt the solution to this conflicting evidence lies in the precise nature of the experimental task and the expectations of the subjects. Gates and Bradshaw (1977b), following a complex and thorough series of experiments on the detection of pitch, rhythm, and harmony changes, emphasized the importance of the different strategies adopted by subjects, as well as the familiarity of the type of material employed. Musical perception may well rely on contributions by both the cerebral hemispheres, in different forms in different subjects, and this makes it difficult to determine just what is the unique contribution of the right hemisphere.
Both the structural and attentional models can be applied to dichotic listening data, in the same way that they were applied to divided visual field research (see p. 234). We have already seen how the structural model can be used to interpret a lateral ear advantage, but is it sufficient to explain all the results? Various forms of evidence are relevant in answering this question.
The first is whether competition is necessary to establish a lateral asymmetry; that is, can an ear difference be established using monaural instead of dichotic presentation? While the asymmetry is smaller and less stable with monaural stimuli, it seems that it can be observed (Henry, 1979). The study by Kallman (1977) serves as a good example of an experiment that was parallel to the usual dichotic procedure, and yet used monaural presentation. Both speech and nonverbal stimuli of the form used in classic dichotic experiments were used and the procedure was identical with the exception that the stimuli were presented to a single ear. The interactions expected in a dichotic paradigm, between the type of stimulus and performance at the two ears, were found. Belmore (1980) even managed to demonstrate an asymmetry using monaurally presented sentences, although only when the task demanded attention to the meaning of the stimulus material. Results of this kind support the importance of a structural explanation.
There are, however, findings that do not fit so neatly into a structural model. One is that monaural competing stimuli presented to the same ear can yield a right ear advantage (Bradshaw, Farrelly, & Taylor, 1981). This could be interpreted as a simple lateral advantage, but does suggest that the effects of competition do not necessarily arise from interference between different lateral input channels.
That the laterality effect does not depend on physical separation of the stimuli is demonstrated most clearly by the fact that a right ear advantage can still be found even when the stimuli are not presented through headphones but through loudspeakers at the left and right of the subject (Morais & Bertelson, 1973). The effect must in this case be due to the physical location of the stimuli in space, to the left or right, and not to the arrival of the stimuli by different anatomical channels, because the stimuli at each side enter by both ears. Even more remarkable is that the laterality effect can be produced by apparent rather than real physical location to one side of space. We determine the location of sounds by various cues, the most important of which are intensity and time differences. A stimulus to our left, for example, will be louder and will arrive earlier at the left than at the right ear, and this difference in intensity and time of arrival tells us the location of the stimulus. It is therefore possible, by simply manipulating the time and intensity differences between two stimuli, to produce an apparent origin for the sounds that they do not have in reality. That a sound is apparently located to the right side of the body is sufficient for it to be associated with superior performance (Morais & Bertelson, 1975).
Findings of this kind are difficult to explain by a structural model, and support the importance of attentional factors. The contribution of such factors has been assessed in a related phenomenon, the ventriloquism effect (Morais, 1975). Subjects were seated with four loudspeakers visible to them, at 45 degrees and 90 degrees to the left and right of the direction in which they were facing. Behind each of these visible speakers, which were dummies, was a curtain and beyond that an active speaker. The subjects were told to which speakers they should attend for subsequent recall of the stimuli. Simultaneous messages, that might or might not come from the locations expected by the subject, were then presented. Among some rather complex results, the right side advantage was only found when subjects expected the message to be 90 degrees to the right and it was actually located there. This finding clearly shows that neither the structural nor the attentional model is sufficient in itself, both must be involved. If the structural model is a sufficient explanation, the subject’s attention would be irrelevant; there would be a right side advantage whenever the stimulus was at that side. Similarly, if the attentional model were a sufficient explanation, the actual location would be unimportant, and the subject’s expectation would alone determine the advantage.
A similar experiment, but actually using dichotic presentation, examined the effects of ear of entry and subjective location by directing different formants of the stimulus to different ears (Darwin, Howell, & Brady, 1978). Again, the ear of entry was in itself a significant factor, but an additional right ear advantage was attributed to the effects of subjective location (see also Hiscock, Inch, & Ewing, 2005).
A useful discussion of the competing claims of the structural and attentional models was presented by Studdert-Kennedy in introducing a special issue of Brain and Language in 1975 devoted to dichotic listening. He sensibly suggested that an experiment with mixed materials in a random sequence might help to resolve the issue. A structural model would predict that the usual asymmetries for each type of material would be preserved, while an attentional model would predict reduced lateral advantages. An experiment of this type has been reported by Kallman (1978). The speech stimuli produced the expected right ear advantage, while the musical stimuli that were mixed with them tended to give a left ear advantage. The results again indicate that attentional factors, if involved, are not themselves sufficient to explain the lateral asymmetries.
Finally, the fact that the subject’s response strategy can affect the results is pertinent to any model of the processes involved. Both Bryden (1978) and Freides (1977) have shown this to be the case, and the conclusion is supported by Hiscock, Inch, and Kinsbourne (1999). Both of the earlier researchers contrasted a free recall condition with conditions in which attention and report order were more strictly controlled. Both found that the report instructions influenced which ear advantage was found. However, they differed in the significance that they placed upon the result. Freides strongly made the point that if response strategy is such an important determinant of the lateral advantage, then competitional methods such as dichotic listening might be merely measuring the subject’s response strategy rather than something more fundamental about brain organization. Bryden, on the other hand, emphasized how the subject’s strategy reflected the operation of attentional and cognitive factors within the processing system. Whichever view is taken, the effects of cognitive strategy, in the context of our general models of cerebral lateralization, illustrate the contribution of variables other than those related directly to structural mechanisms (Voyer & Flight, 2001). Just as with the visual studies, the evidence clearly points to a dynamic– structural model.
AN INDEX OF LATERALIZATION
Many writers have been keen to point out that the right ear advantage for digits, while associated with cerebral laterality, cannot be used as an index of lateralization (Berlin, 1977; Colbourn, 1978; Teng, 1981). The problem arises from the variability both of different subjects and of any one subject over time. It is easy to be misled by the clear average ear advantage shown by group data, and assume that dichotic performance could be an index of speech lateralization in individual subjects. Studies continue to appear in which dichotic listening data are used in this way.
The evidence is clearly against the use of such an index, with one notable exception: the Dichotic Monitoring Test of Geffen, Traub and Stierman (1978). This test involves the dichotic presentation of monosyllabic word pairs, that include the target word dog among eight “noise words” in each channel that share two phonemes in common with the target (e.g., dig or log), as well as 52 dissimilar words. The subject must detect the target word, and both reaction time and accuracy are recorded in response to occurrences of the target word. The test was given to 4 patients who had undergone the Wada test to establish speech lateralization and to 31 patients whose speech laterality had been assessed by the presence of (temporary) dysphasia following the unilateral administration of electroconvulsive therapy (ECT), and extremely good agreement about language lateralization was shown between the physical tests and the dichotic procedure. An extension of this study by Geffen and Caudrey (1981) reported an agreement of 95%, and a test– retest study for reliability found a shift in assessed laterality in only 3 of 86 subjects. This procedure appears more reliable and more valid than previous attempts to develop an index of speech lateralization that could be used in normal subjects but it has not been widely adopted in a clinical context.
If either dichotic listening or divided visual field methods could yield an index of cerebral lateralization, we should expect a correlation between the asymmetries found using the two techniques in any individual subject. Fennell, Bowers, and Satz (1977) examined just this issue, using a sequence of four tests of dichotic listening to concrete words and divided visual field presentation of letters. They found both the expected right visual field and right ear advantages. The dichotic asymmetry was relatively stable across the four occasions of testing, although the visual asymmetry was significant only after the first occasion. The correlation between the two modalities was also found to be reliable and stable, particularly on the final two occasions of testing. Clinical applications have been discussed by Voyer and Rodgers (2002) and Voyer, Russell, and McKenna (2002).
This study has now been subjected to extremely searching criticism in terms of the general methodological problems inherent in the measures, the composition of the samples, and the statistical treatments used (Berenbaum & Harshman, 1980). The authors of this criticism urge skepticism about the findings of the study by Fennell et al. (1977), and it seems at least prudent to await further investigation of the issue. Other data, collected with a primary interest in the effects of the subject variables of sex and handedness, also suggest that there can be a significant dissociation between the asymmetries observed in different modalities (Searleman, 1980).
While dichotic listening studies have made a major contribution to experimental neuropsychology, they have been less numerous than studies in the visual modality. This is due both to the technical demands of the technique and to the added problems that the bilateral nature of auditory projection brings to interpretation of the results. In consequence, studies in recent years have concentrated more on linguistic variables than on the general principles underlying cerebral lateralization.
The studies have nevertheless provided evidence of lateral specialization in the brain. The left hemisphere seems to be associated with two kinds of processing: one phonetic, the other acoustic. The phonetic processor deals with speech and speech like stimuli in terms of their linguistic composition. The acoustic processor deals more generally with complex auditory stimuli, and may be involved in temporal order perception. It does not deal merely with speech stimuli.
The right hemisphere has been associated with “nonverbal” sounds and with complex musical stimuli. Not all aspects of music are processed preferentially in the right hemisphere, and music perception (like most real-life tasks) involves the contribution of both the hemispheres.
It will by now be apparent that the research undertaken in a purely experimental context is generally of a much higher standard than that performed with clinical patients (discussed in Part II). This mainly reflects the difficulty of doing scientifically sound research in a clinical context, but the general level of both methodological sophistication and theoretical discussion has been superior in experimental neuropsychology.
Direct comparison between experimental and clinical findings is difficult, partly because clinical work has concentrated on localization, which has been ignored in experimental research in favor of lateralization. Also, many of the tests and tasks employed in clinical research cannot be presented, for procedural reasons, in experimental paradigms. Integration of the two approaches has not been attempted in any systematic fashion, although there seem to be few fundamental disagreements between the two areas of research. The evidence from dichotic listening agrees remarkably well with both clinical findings and data from divided visual field studies.
The only model that at present accounts well for the whole body of dichotic research is the dynamic–structural model, which is also the model supported by visual studies. Attentional and cognitive factors play a significant role in auditory as in visual lateralization.
Dichotic listening has proved neither sufficiently reliable nor sufficiently stable to allow the determination of speech lateralization in individual subjects. Recent research may, however, show that an index of this kind can be developed.
OTHER METHODS IN EXPERIMENTAL NEUROPSYCHOLOGY
While divided visual field and dichotic listening techniques far outweigh all other methods in importance, some of the alternative methods should be briefly mentioned. A review of all these methods is to be found in Beaumont (1982).
Lateralized stimulus presentation has also been used in tactile perception. Stimuli can be presented to the right and left hands and subsequent performance recorded. It has been possible to demonstrate lateral asymmetries in this way, but only when fairly fine manipulation or tactile exploration is required. The projection of tactile information to the cortex (see p. 96) involves both ipsilateral and contralateral pathways, and this means that it is difficult to be sure that stimulus presentation has been appropriately lateralized. However, a “dichhaptic” technique originally developed by Witelson (1974) and involving bilateral simultaneous stimulation has been used in a number of studies and produced interesting results.
Lateral eye movements have also attracted some interest. Kinsbourne (1972) sparked off this work by the observation that solving “verbal” problems was often accompanied by conjugate gaze deviation to the right, and solving “spatial” problems by gaze deviation to the left. He explained this phenomenon in terms of his attentional model, hemisphere activation being considered to produce eye movement toward the opposite side of space. Following much debate about the reliability and validity of such observations, recent reviews have suggested that lateral eye movements can be reliably elicited and that they do reflect hemisphere lateralization (Ehrlichman & Weinberger, 1978). Whether the evidence justifies their use as an index of an individual’s cerebral organization, as is sometimes done, seems doubtful.
Some lateral asymmetries are also to be observed in free vision; for example, there are lateral asymmetries in aesthetic composition that seem to relate to hemisphere specialization, although the mechanism behind such asymmetries is under dispute. It has also been reported that the left side of the face has been more commonly presented in formal portraiture, and this side of the face is also judged as more emotionally intense. Asymmetries in spatial orientation and judgment, as seen in map reading and direction finding, have also been demonstrated.
Finally, there are the lateral performance asymmetries, of which the most important is handedness. Differences between left and right handers are discussed at length in Chapter 14, but even in right handers the difference between the preferred and nonpreferred hands can be studied as one aspect of cerebral organization. Various aspects of manual skill have been measured, and there has been a renewal of interest in tapping performance. As with tactile presentation, inferences about cerebral lateralization are complicated by the bilateral nature of certain aspects of neural organization. The Torque Test, which measures the direction of drawing tendencies as clockwise or anticlockwise, has been much discussed but has also been subjected to criticism.
All of these methods can contribute to our overall model of neuropsychological organization but because of methodological or theoretical complications none is as satisfactory or important as divided visual field and dichotic listening research. All experimental data have nevertheless some part to play, and the development of techniques in neuropsychology may well spring from refinements of these alternative methods.
Electrophysiology and Imaging
At the time of the first edition of this book, the introduction of electrophysiological techniques into neuropsychology was one of the most exciting developments of recent years. For the first time it was, in principle, possible to observe in “real time”–that is, as they happen– cognitive processes and the physiological events that are believed to be associated with them. A technique was available that might make it possible to construct a bridge between mental and physiological events. If so, then there was a solution to the problem that has dogged so much of neuropsychology, that of directly investigating mind–body relationships.
Twenty-five years later, electrophysiology has failed to fulfill its promise, partly because of the methodological complexity and technical demands of the technique (discussed below). One has to feel a little sorry for electrophysiology. Its clinical importance in identifying the location of lesions was supplanted in the 1970s by the introduction of CT (computerized tomography) scans, and its anticipated role in neuropsychological research, which seemed to offer a new area of application, was never fully established, and was in turn supplanted by more advanced imaging methods (such as fMRI) of the 1990s. Nevertheless, this should not lead us to discount electrophysiology. It remains a clinically important technique, particularly in the analysis of seizure disorders, and it continues to have a place among the resources available to experimental neuropsychologists. In light of its recent history, it is difficult to predict with confidence how it may next be employed. However, its potential in addressing mind–body relationships, and at considerably less cost than advanced imaging systems, remains.
Before being carried away by the exciting possibilities of electrophysiology, one must recognize that there are a number of difficult technical problems yet to be solved with these methods of investigation. This area of research is probably the most technically demanding in human neuropsychology, and there continue to be myriad experimental problems, as I hope to show. Nevertheless, despite the difficulties and some disappointments, we should not lose sight of what the techniques offer: the opportunity to observe subjects engaged in some intelligent task, and at the same time to observe the processes within the brain that are the physiological aspect of that mental activity. The degree to which other imaging techniques also offer this opportunity is discussed below.
Before introducing the techniques themselves, the distinction between clinical and research electroencephalography should be noted. Clinical electroencephalograms, or EEGs, are recordings of electrical activity in the brain used to investigate suspected pathological processes within the head. They are performed in a relatively standard way, using a variety of appropriate techniques, and play an important role in the physical investigation of neurological and psychiatric patients, particularly if a seizure disorder (epilepsy) is suspected. Research electroencephalograms, however, are used to investigate brain activity in normal subjects, and involve a much less well defined set of techniques that have been derived from, and are associated with, the clinical techniques.
Electroencephalographical techniques fall into two principal groups: recording “ongoing” activity, and recording evoked potentials. The procedures involved in both groups are described first, followed by some typical experiments and findings.
The recording of the continuous electrical activity of the brain, or ongoing EEG, has been practiced since the 1930s. However, advances during the 1970s in amplifier design, and the involvement of laboratory computers in analyzing recordings, have enabled the technique to be used more widely as a research tool.
If two electrodes, traditionally small silver cups coated with silver chloride and filled with a conducting gel, are glued at two positions on the scalp, with a third electrode attached elsewhere on the head to ground them, it is possible to detect a fluctuating difference in electrical potential between the two electrodes. This constantly changing potential difference can be amplified (about 20,000 times) and written out on a moving paper chart to form the EEG trace. It is usual to record from more than a single pair of electrodes (a single channel) and a typical recording might have four, eight, or many more channels simultaneously recorded (up to 24 in clinical EEGs). The chart forms a convenient way of inspecting the recording, although the data written out on the chart will be fed directly into a laboratory computer for analysis. As the computer will sample the trace for each channel at least 128 times a second, yielding 61,440 numbers for each minute of activity across eight channels, a great deal of data is generated by even a relatively short period of recording. An example of a normal human EEG is shown in Figure 13.1.
The first problems–of what exactly is being recorded and how to interpret the trace–arise when the sites for placing the electrodes are selected. Since the electrical activity of the brain is being recorded through the skull, scalp, and other tissues, the electrode is picking up activity from a relatively large area of underlying cortex, and presenting a rather distorted aggregate image of what is going on in an area of about 1 square centimeter. However, it is known from comparing scalp recordings with recordings made directly on exposed brain tissue that there is some validity in the method. Even if it is imperfect, it is all we have. Matters are sometimes further confused in that the electrode may pick up more activity from distant sites than from the directly underlying tissue, but these problems, caused by the directional nature of the propagation of the electrical activity, need not worry us for the present.
FIGURE 13.1. Normal human EEG: three channels recorded from the right (above) and left (below) hemispheres; common extracerebral reference. Bursts of alpha activity may be seen, especially to the right of center.
The electrodes are usually placed directly over the area of the brain that is of interest, and usually at points located with reference to a system of labeling known as the 10–20 system (because it divides distances on the head in terms of 10% and 20% of the distance between fixed landmarks), which is shown in Figure 13.2. Modern systems commonly employ a “cap” that fits snugly over the head and is intended to automatically locate the electrodes, although its success in accurately achieving this is somewhat in doubt. A set of electrode positions is referred to as a montage.
Although, as mentioned previously, we cannot directly record the activity from one electrode, but only the difference between two electrodes, we can accept that we are recording the difference between two “active” electrodes (both recording electrical activity from the brain) and so make a bipolar recording. Such recordings are quite common in clinical work, but they present difficulties in research. When the potential difference between electrodes changes, which electrode has changed? Almost certainly both will have changed, and it is impossible to unravel the contribution of each to the final observed change in the EEG.
In order to get around this problem we may choose to make a monopolar recording, where one of the two electrodes is relatively “inactive.” However, it is almost impossible to find sites on the head at which brain electrical activity will not be recorded, including the tip of the nose or the tongue, and we must also avoid recording eye movements, muscle potentials, and other artifacts. We might move our inactive electrode off the head (extracerebral reference), but then we find the activity of the heart and other bodily processes contaminating our recording, although there are techniques that allow these to be filtered out. The usual solution is therefore to choose a site that will pick up some EEG activity but will be relatively inactive with respect to the activity in which we are interested. Such sites are often on the midline, and frequently the vertex (Cz) or midpoint on top of the head is chosen. The mastoid processes (behind the ears) and other sites are also used, sometimes in linked pairs to cut out any lateral bias (although linking is probably ineffective in doing this).
FIGURE 13.2. Principal points on the 10–20 system of electrode placement.
Further, it is common to employ a single reference for all the active electrodes being used, in a common reference recording. The reference may not be truly inactive, but at least when a comparison is made across channels, its contribution to each channel will be the same, and differences between the channels associated with active electrodes will reflect differences between the activity at those sites.
There are other more technical problems associated with a selection of electrode site and recording montage, especially if lateral asymmetries are to be investigated, which has been the aim of much of the research in this field (see Beaumont, 1982; Donchin, Kutas, & McCarthy, 1977). Many of these problems can be solved by careful experimental design but one fundamental problem, as yet unsolved, is that of underlying asymmetries of the brain. We know that the brain is asymmetrical, particularly in certain regions, and there is now fairly good evidence that these asymmetries have functional significance, but electrodes are applied to symmetrical points on the scalp. If we detect asymmetries in the EEG at homolateral points, do these represent lateral asymmetries in the activity of homolateral regions or merely indicate that we are not recording from anatomically homolateral points on the brain? This is an extremely vexing problem, which, until we can determine individual differences in anatomical asymmetries, is insoluble.
Once we have battled with all these problems, have selected the task to set the subject (also fraught with difficulties that may produce artifactual results in the EEG), and have made our recording, the problems are not over. A form of data reduction and analysis must be selected. The difficulties here spring, quite simply, from ignorance: if we really understood what we see in the EEG trace, decisions about how to analyze it would be less arbitrary. While much is known about the origins of the EEG (see Thatcher & John, 1977, for a good introduction), we still do not know how a complex pattern of changes relates to mental events. It is usual however to analyze the frequencies present in the recording, and to divide these up into several frequency bands:
|Delta:||0 to 3.5 cycles/second (Hz)|
|Theta:||4 to 7.5 cycles/second (Hz)|
|Alpha:||8 to 12.5 cycles/second (Hz)|
|Beta 1:||13 to 19.5 cycles/second (Hz)|
|Beta 2:||20 to 29.5 cycles/second (Hz)|
These bands are not, however, precisely fixed, and there is some variation in the limits applied to them. Research initially concentrated almost exclusively on alpha activity, although all the bands are potentially of interest and there is now a recognition that equally interesting results may be found in other bands, especially theta and beta 1.
The power of a frequency band, or of a specific frequency, is usually calculated, and represents an overall measure of the strength of that component in the EEG during a specified period. Mental activity usually results in a decrease in the power of alpha activity, and an increase in the power of other frequency bands, especially beta 1. The former effect is referred to as alpha attenuation or alpha abundance, and the latter as enhancement. Power, however, confounds the amplitude with the duration of the component, that is, a very powerful component present for part of the time may result in the same overall power as a weaker component present for all of the period. These power analyses, however, are a useful if very crude way of quantifying the activity.
Online computers have enabled a variety of more complex analyses to be performed, in particular certain forms of frequency analysis. One of the more promising of these is coherence analysis. This analysis takes a pair of channels and describes the amount of shared activity at the two channels at each frequency, but independently of the power of that frequency in the two channels (Shaw, 1981). The particular value of this form of analysis, apart from its independence from power, is the fact that it can be calculated from relatively short periods of EEG recording, down to about 0.5 seconds, although a number of such short epochs have to be combined to provide a reliable estimate of the coherence. It is possible to identify some period when a particular cognitive process might be active, and then to examine the coherence between sites during that period. We can therefore begin to build up a dynamic map of the brain changes associated with relatively specific cognitive events, and to do this separately for each frequency across the range of the whole band of the EEG. Often associated with the coherence spectrum, as the plot of the coherence across the whole band of frequencies is called, is a phase spectrum, which can indicate which channel leads the other when the two share a significant amount of activity at some given frequency.
There are of course difficulties with coherence, technically and in interpretation. It is, for example, still uncertain whether we should expect coherence to increase or decrease with task-related activity in some region. It seems that there is a local increase, but a decrease with reference to more distant sites, and much more fundamental work is required before we can appreciate what this and other recently developed techniques can tell us.
In the EEG, we do have a fairly direct record of the activity of the brain, and while we may have difficult decisions to make about how to analyze the complexity of what is recorded, we are at least dealing with brain processes in a relatively immediate way. We may still expect advances in understanding the EEG, and the brain processes that it reflects, in the future.
Recording Evoked Potentials
Evoked potentials (EPs), also known as averaged evoked potentials (AEPs), averaged evoked responses (AERs), and event-related potentials (ERPs), are an alternative way of studying psychological processes in the electrical activity of the brain. The common use of these abbreviations, further complicated by the addition of the modality of stimulation (hence visual, auditory, and somatosensory evoked response: VER, AER, SER), can be somewhat confusing, but they all indicate measures based on the same basic technique.
This technique relies upon computer averaging, and has only been possible since the advent of relatively inexpensive laboratory computers. The idea behind it is that there is a relatively invariant response by the brain to a given sensory event. It is not possible to see this response in the EEG because it is masked by the large amount of background noise in which the response is embedded. However, if the noise is considered to be random, and if a number of examples of the response to a given event are collected and added together, then the noise at any time will tend to cancel out, allowing the event to emerge out of the background noise. Calculating the average, rather than simply adding to find the total, allows the number of examples collected to be taken into account. An illustration of the principle behind the technique, using an artificial waveform, is shown in Figure 13.3.
In practical terms, a simple sensory stimulus that can be accurately reproduced is selected, and this is presented a large number of times (commonly between 64 and 512) to the subject. Time-locked to the stimulus presentation, a short epoch of EEG is collected (between 500 and1,000 msec, around the stimulus), and stored in the computer. These epochs or samples are averaged as they are collected, and usually displayed during data collection. Subsequent analyses can then be performed. Examples of typical evoked potentials from different sensory modalities are shown in Figure 13.4. (A note of warning here: physiologists prefer their graphs negative up, while engineers and psychologists generally adopt positive up, so take care about which way up an EP has been plotted.)
FIGURE 13.3. Illustration of the principle of waveform averaging: (a) the underlying waveform; (b) with added random noise; (c) after averaging 32 samples of the noisy signal; (d) after averaging 512 samples of the noisy signal, showing that the waveform has been accurately extracted.
The recording electrodes and initial stages of the recording are just as for an ongoing EEG and many of the problems surrounding electrode placement apply equally to EP recording. A rather different set of problems arises, however, when we come to analyze the evoked potential records. Even though the absolute amount of data will be less than in an EEG record, there is still a formidable problem of data reduction. Again, our difficulties in deciding how to extract the most important features from the EP stem from our relative ignorance of what it is that we are looking for.
FIGURE 13.4. Typical average evoked potentials (positive up) in: (a) visual, (b) auditory, and (c) somatosensory modalities.
Most EPs contain a series of peaks and troughs, known as components, which presumably relate to significant brain events. It is usual to identify these components and to measure their location in time with respect to the stimulus event (latency), together with the amplitude of the response (commonly in the range of ± 15 microvolts). We can then compare the amplitude and latency of given components across different recording sites and in different experimental conditions.
Things are rarely so simple, however. Although most subjects show a similar response pattern to a fairly clearly defined stimulus, a given component may simply not be present in the record from a certain subject. There may be two peaks of equal amplitude very close together: which is the component we are measuring, and should we average the latencies? One component may be superimposed upon another, and it may be difficult to extract the “true” amplitude and latency information about both from the record that has been obtained. Also, the amplitude of one component may be influenced by the preceding component, and should this be taken into account? The use of peak-to-peak measurements of amplitude is one attempt to account for these sequential effects. What should be our baseline for measurement: is it zero volts, the average voltage of the whole response, or some prestimulus baseline?
None of these problems is readily soluble, although there are conventions that have been adopted to deal with most, and a number of computer algorithms that may help with the description of the components present in a set of evoked responses. There is, however, a further problem: that of latency jitter. One of the assumptions of the averaging technique is that the underlying response is invariant. If, however, there is some basic variability in the latency of a component, this variability will only be seen in a reduction of the amplitude of the averaged component (see Figure 13.5). Care must therefore be taken to ensure that amplitude differences between sites or conditions represent true amplitude differences and are not the effect of variable latency in the component. Inspection of the variability of the samples that make up the average, together with filtering techniques that correct for the latency of components, can help to deal with this problem. (Important discussions of EP methodology are to be found in Callaway, Tueting, & Koslow, 1978; Desmedt, 1977; and Swick, Kutas, & Neville, 1994.)
FIGURE 13.5. Latency jitter: averaging samples of differing amplitude but constant latency (a) yields the average (b); averaging samples of the same constant amplitude as (b) but differing latency (c) yields the average (d), which is of lower amplitude than (b).
There are some conventions about the labeling of commonly occurring components seen in EPs but none of them is universally accepted. One approach is to number the components sequentially, so that P1 is the first main positive peak and N1 the first negative inflexion, followed by P2, N2, P3, and so on. This, however, tends to be rather confusing as the components tend to have different latencies with stimuli in different modalities. An alternative is to label the components as N100, P300, and so on, to indicate their polarity and approximate mean latency. An extension of this is that P300 comes to mean the theoretical component associated with P300, even if its latency is perhaps as long as 450 msec from stimulus onset.
Unfortunately but inevitably this is confusing. However, it is worth noting that the two components that are perhaps most discussed are N1, which occurs in the visual modality at about 90–120 msec from stimulus onset, and P300, which has a latency of 300–450 msec. Both these components have been clearly linked to cognitive processes. In general, the early components, up to about 100 msec from stimulus onset, are thought of as being exogenous components, determined by the sensory reception and perception of the stimulus. The middle-range components that follow (100–300 msec) and the subsequent late components (up to 1,000 msec) are thought of as endogenous and related to cognitive manipulation, stimulus evaluation, and decision making. In one special case, the brain stem evoked potential, it has been possible to identify components associated with each stage of the sensory transmission up to the cortex. With later endogenous components, things are much less well understood, although there are a number of theories about the cognitive significance of various components.
Mention should finally be made of the so-called slow potentials, generally seen as slow negative changes in the EEG preceding an expected event. These responses, or CNVs as they are sometimes called (contingent negative variation), can be recorded by similar averaging methods, and are thought to represent preparatory processes within the brain.
Evoked potential recording has been perhaps more successful than direct EEG studies in eliciting phenomena of psychological significance. Particularly where the stimulus event and task are relatively simple and clearly defined, the technique is extremely valuable. With complex tasks and stimuli, it becomes more difficult to use the technique, and the need for a relatively large number of repetitious trials enforces a certain artificiality upon the experimental design. Despite this, and the numerous methodological problems and pitfalls, evoked potential recording has become an important specialty within neuropsychology.
Lateral EEG Asymmetries
As the bulk of research with ongoing EEG measures has been directed to establishing lateral asymmetries in the EEG, a review of this topic will serve to illustrate the neuropsychological research using such measures.
All the substantial reviews (Beaumont, 1982; Donchin et al., 1977; Kertesz, 1994; Marsh, 1978) have commented upon the methodological shortcomings of the studies that have been carried out. In fact it would be fair to say that all studies (including those in which I have been involved) can be regarded as having methodological limitations. The view that one takes of the literature depends upon a judgment of the seriousness of the methodological shortcomings.
The approach in most research designs has been to set the subject some task or tasks to “engage” the left or right hemispheres, and then to look for asymmetries in the accompanying EEG. These asymmetries might be seen by comparing the two hemispheres directly (between hemisphere effects) or by comparing intrahemispheric (within hemisphere) effects occurring on the left and on the right. The better studies have also recorded behavioral task performance alongside the EEG. This not only enables an examination of the relationship between EEG and performance parameters but, more mundanely, helps to ensure that the subject actually carries out the task that he has been set. It can also help to provide an independent measure of whether tasks selected to engage the left and right hemispheres have been satisfactorily matched in difficulty.
We might ask first whether there are relatively “good” studies that do find an effect of cognitive task. Davidson, Taylor, Saron, and Snyder (1980), for example, gave subjects tasks such as an embedded figures task, the Kohs blocks, and verbal analogies and reading comprehension tests, and found an asymmetry in right-handed subjects, although only for those with sinistrals (left-handed individuals) in the family. Ehrlichman and Wiener (1979, 1980) also used various verbal and spatial tasks and asymmetries in integrated alpha power, finding a significant reduction in power (which is what would be expected) over the engaged hemisphere. They repeated their experiment using covert as well as overt responses, because of the suspicion that overt responding may introduce artifactual asymmetry into the EEG, and found their earlier results supported. It should be noted that their tasks, as with those in all similar studies, are only expected to engage one of the hemispheres preferentially; there is no independent evidence to demonstrate that they will do so (apart from clinical investigations).
Tasks such as counting the words the or a in an auditory passage, or listening to music, were used by Moore (1979), and similar tasks but without counting were used by Moore and Haynes (1980), resulting in different asymmetries for the two tasks, as was predicted. Parallel results have come from Ornstein, Johnstone, Herron, and Swencionis (1980) and Rebert and Low (1978). The latter study also involved listening to verbal material, but had as the spatial component a block construction task performed entirely in imagination, which seems rather odd. Beaumont and Rugg (1979) used two memory tasks, carefully matched for difficulty, for which there was clinical evidence of association with left and right anterior lobes; the tasks involved judging which of a pair of either words or abstract paintings had been previously presented in a sequence of similar items. A task asymmetry in intrahemispheric coherence (see p. 268), at the expected cortical sites, was found. There are a number of other studies that would also support the existence of taskrelated asymmetries on various EEG variables.
We must also ask, however, whether there are “good” studies that have failed to find the expected asymmetry. The study by Beaumont and Rugg just mentioned found the effects on coherence that were predicted, but found quite anomalous enhancement of alpha power with the cognitive task; further data on coherence are provided by Weiss and Mueller (2003). In the study by Ornstein et al. (1980), although task effects were found at centrally sited electrodes, they were not found at parietal placements. Rebert (1977) has also found alpha enhancement, rather than the expected attenuation, in the right hemisphere with a dot detection task. Rebert and Mahoney (1978) found no task effects at all in raw power measures taken during verbal and nonverbal target detection tasks. Visuospatial tasks designed to be analytic or synthetic in their cognitive demands failed to produce clear task-related effects (Tucker, 1976), and although Dolce with others (Dolce & Waldeier, 1974; Dolce & Decker, 1975) found complex and interesting effects in the beta, theta, and delta ranges, no task effects upon alpha activity were seen. Asymmetries in relation to mood and cognition have been reported by Gotlib, Ranganath, and Rosenfeld (1998), and Bell and Fox (2003). McCarthy and Donchin (1978) used a task derived from the “chimeric metacontrol” studies with split-brain patients, in a laudable attempt to produce well-matched “left and right hemisphere tasks,” but found no task-dependent asymmetries in alpha power.
Finally, we must ask whether there are “good” studies that find effects but suggest that the effect may be artifactual. The studies of Gevins and co-investigators (Gevins, Zeitlin, Yingling, et al., 1979; Gevins, Zeitlin, Doyle, Schafler, & Callway, 1979; Gevins, 1981) are particularly relevant here. They conducted a series of experiments in which particular efforts were made to control for stimulus characteristics, the effects of eye and limb movements, and other performancerelated factors, such as the subject’s ability and task engagement. Their conclusion was quite clear: when such factors were effectively controlled, task-related asymmetries, which had been otherwise observed, disappeared.
Other artifactual sources of asymmetry are suggested by the work of Amochaev and Salamy (1979) and Haynes (1980). In the first study, different sites for the reference electrode were compared, and they found the stability of the task effect, using a variety of tasks, to be related to the site chosen. The study by Haynes, in which subjects listened to sentences that were either to be imitated or followed by the construction of a new sentence, found left hemisphere activation only when imitation was demanded. This was interpreted in terms of the preparatory phases of motor programming for ensuing speech, and suggests that even when EEG is recorded during a period of inactivity, subvocal preparation, or preparation for other response processes, might be contaminating the EEG.
What are we to conclude? If we adopt a fairly relaxed criterion of methodological rigor, then the bulk of studies point toward task-related asymmetries in the EEG. The asymmetries are those that would be predicted from the divided visual field and dichotic listening literature. That is the general trend of the results. However, the failure of a number of reputable studies to find effects, and the clear demonstration of a number of possible sources of artifact, must mean we have to conclude that clear and reliable task-related effects have not as yet been demonstrated. It does not mean that the search for them should be abandoned, and improvements in methodology should lead to a clearer answer to whether asymmetries are to be observed in ongoing EEG.
Evoked Potential Studies
When employing the evoked potential methodology to answer neuropsychological questions, there are two major strategies available. The stimuli that are used to elicit the EP may be task relevant (and may or may not be lateralized) or else may be incidental to the task. If task-relevant stimuli are used, there are problems about the control of such necessarily complex stimuli and also about obtaining strong potentials. On the other hand, it is not clear that if EPs are evoked incidentally during performance of some cognitive task by simple “irrelevant” stimuli, the response to these “probe” stimuli will reflect the concurrent task engagement. Studies have been reported using both approaches, and have tended to be in either the visual or auditory modalities.
Taking visual EPs first, a good example of a study using taskrelevant lateralized stimuli is that of Ledlow, Swanson, and Kinsbourne (1978). They used the nominal/physical letter-matching task (see p. 230) and collected performance data together with the EPs in order to establish independent evidence of lateral asymmetry in cognitive processing during the task. Their results are complex, but they found an overall difference in the amplitude of left and right hemisphere potentials, and for some components the direct pathway (when stimuli were presented to the hemisphere contralateral to response: see Figure 11.2) was associated with responses of smaller amplitude but reduced latency. At P300 there were differences between direct and indirect stimulation that also varied with type of match (nominal or physical), and also an effect of same or different cases at P130 and N170.
A similar experimental design was used by Rugg and Beaumont (1978). Letters were employed as stimuli throughout, but subjects were asked either to respond to letters containing an “ee” sound (verbal task) or to letters containing a right angle (spatial task). Variations in the amplitude of middle latency components with field of presentation were found that differed according to the cognitive task. A lateral asymmetry was found for late components, but only with the spatial task.
By contrast, Friedman, Simson, Ritter, and Rapin (1975) collected responses to sequentially flashed words that made up one of three sentences: “The wheel is on the axle,” “The heel is on the shoe,” “The peel is on the orange.” On some trials, the first part of the second word was not presented, so that from “-eel” the subject could not tell which sentence was being shown until the final word arrived. The second word therefore might deliver information or be indeterminate. The latency of P300 was longer for words delivering information, and “syntactic closure” was also shown to affect P300 amplitude. However, no lateral asymmetries were found between the hemispheres.
McCarthy and Donchin (1978), as already indicated, have used the structural–functional matching task first used with split-brain patients and have collected not only EEG data but also EP recordings. There was not only a visual EP to the stimulus figures, but also a warning tone that in some conditions served to indicate the type of match required and that evoked an auditory response. Although there was no effect of mode of matching on the EEG, during the preparatory period following the warning tone a component was seen that was sensitive to the mode required, and that had an asymmetrical distribution across the electrode sites employed.
The alternative of incidentally collecting EPs to flash stimuli during an ongoing left or right hemisphere task was developed by Galin and Ellis (1975). They found changes in the asymmetry of the potentials recorded at left and right temporal and parietal areas that parallel taskdependent asymmetries in both EEG alpha power and in performance measures. These effects have, however, not always been confirmed by other researchers (Beaumont & Mayes, 1977; Mayes & Beaumont,
In the auditory modality, the “identical stimuli but variable task” design has also been used, for example by Wood (1977). Reaction times and EPs were collected during tasks that focused upon various phonetic characteristics of the stimuli, with the result that differences were found in the left, but not the right, hemisphere between 60 and 140 msec after stimulus onset. This was taken to support the independence of auditory and phonetic processes. Tanguay, Taub, Doubleday, and Clarkson (1977) presented monaural voiced stop consonants (e.g., /ba/, /da/, /ga/) and found increased amplitude of response contralateral to the ear of presentation, but no hemisphere differences in latency or amplitude. Nevertheless, using musical chord stimuli, Taub, Tanguay, Doubleday, Clarkson, and Remington (1976) found lateral asymmetries as a function of hemisphere stimulated, in middle-range components and at sites close to Wernicke’s area.
The auditory parallel to the visual ambiguity task, described above, has been extensively investigated by Brown and Marsh with others (see Brown, Marsh, & Smith, 1979). They used sentences that contained an element with a given sound but with different meanings according to the context, for example, the words led and lead in “The horse was led” or “The metal was lead,” and fire in “Ready, aim, fire” or “Sit by the fire.” The results from these studies are complex, but effects have been found that are sensitive to parts of speech, as well as lateral asymmetries in certain components, some of which suggest the involvement of left anterior cerebral processes.
Task irrelevant probes with an ongoing auditory task have been used by Shucard and Shucard (1979) and found to be associated with engagement in either musical or verbal tasks. Musical performance has also been examined by Harris and Silberstein (1999) and by Aaron and Tobias (2004). Both auditory and visual stimuli have been examined by Davis and Wada (1978), who performed spectral analysis on their EPs and found greater coherence in the left hemisphere with click stimuli, while flash stimuli produced this effect in the right. Not only were these hemisphere effects observed, but they were also found to vary with the speech lateralization of the subjects, which had been independently assessed.
Attentional variables have been assessed through evoked potential recording, and in both auditory and visual modalities. There is considerable debate about the effects of attention on EPs, especially upon the N1 component (see Näätänen & Michie, 1979), with various groups of researchers contributing elegant experiments. Most of these experiments rely upon the presentation of lateralized auditory material, with attention being directed to one of the ears. Lateralized recording sites have been included as well as central sites, and some but not all of these have revealed asymmetrical components. Notably, Buchsbaum and Drago (1977) have reported clear asymmetries in an N140 component with visual presentation, which could be manipulated by attentional variables. The importance of these experiments lies not in the precise results, but in the demonstration that manipulation of the cognitive variable of attention may have effects that can be seen in electrophysiological recordings.
Research has not been limited entirely to the visual and auditory modalities, and somatosensory potentials (Barrett, Halliday, & Halliday, 1978; Saron, Foxe, Schroeder, & Vaughan, 2003) as well as movementrelated effects (Kutas & Donchin, 1978; McCallum & Curry, 1979) have been shown to exhibit lateralized components.
The experiments mentioned above serve as examples of the kind of research that has been conducted using evoked potentials. The research literature is quite extensive, the pattern of results reported is extremely complex, and, as already indicated, methodological difficulties raise considerable problems in considering many of the studies. Happily there are several good reviews of the literature (Hillyard & Woods, 1979; Marsh, 1978; Rugg, 1982a, 1982b), and there seems no reason not to accept what is the general opinion of these reviews: that EPs appear sensitive to both the locus of cognitive processes and to factors that reflect the mode of stimulus presentation, and the resulting effects upon EPs are the product of a complex interaction of such factors.
MINOR PHYSIOLOGICAL TECHNIQUES
This seems an appropriate point at which to note that, while the major electrophysiological techniques employed in experimental neuropsychology have now been considered, there are a number of less widely used physiological techniques that have contributed to our present knowledge.
Psychophysiological measures such as skin conductance, muscle tension, and finger pulse volume have also been used, but the results, while interesting, are not entirely clear. The interpretation of results is complicated by the ipsilateral efferent projection of the autonomic nervous system, and it is difficult to predict the direction of change in autonomic variables.
Anatomical investigations are also of interest to neuropsychologists. It is now clearly established that there are certain marked anatomical asymmetries in the topography of the cerebral cortex, that these are present from birth, and that there is some reason to associate them with functional specialization (Galaburda, LeMay, Kemper, & Geschwind, 1978).
If there is a revolution occurring in neuropsychology, it is in the contribution of medical imaging. Indeed, things are changing so fast that, whatever I write may be substantially out of date before I have even finished writing this chapter, let alone before the book appears in print. All that I can do is provide a rather rough guide to the techniques and their uses. Beyond that, students are strongly encouraged to take a trip to the library or, more likely, sit down at their electronic literature search facility to ascertain the current “state of the art.”
Although computerized tomography (CT; or in the earlier literature, computer-assisted tomography, CAT) now seems rather mundane, its introduction in the 1970s was quite remarkable and initiated the modern revolution in medical imaging, earning its developers, Godfrey Hounsfield and Allan Cormack, a Nobel Prize in 1979.
The technique employs X-rays directed at the body from a large number of orientations, with the X-ray source and detector continuously positioned opposite each other, and the body in between. The information from each of these images, which can be obtained using very low doses of radiation, is combined by computational methods to reconstruct “slices” through the body. A CT scan of the brain produces a set of images through consecutive planes of the brain in which, as in conventional or “plain” X-ray images, bone and other dense material appears white, air and water appear dark, and the brain substance is represented as varying shades of gray. The image is a representation of the structural state of the brain.
CT scans (see Plate III for an example) now appear very crude by comparison with more recently developed techniques, but they are relatively inexpensive and widely available, and are an important clinical tool for the neurologist and the trauma specialist. CT scans of the head are now routinely performed following head injury, or to investigate suspected neurological abnormalities, and are an important clinical tool. However, CT is not capable of imaging all abnormalities, has relatively low definition, and is no longer of significance in neuroscience research.
Magnetic Resonance Imaging
A far more detailed image of the anatomical structure of the brain can be obtained by employing magnetic resonance imaging (MRI; also less commonly known as nuclear magnetic resonance imaging, NMR). Essentially, the body is placed in a very powerful magnetic field so that the atomic particles of the tissues being scanned are aligned in the same direction. Radio waves directed at the body then give off radio signals that differ according to the type of tissue; as in CT, “slices,” or 3-D reconstructions of the brain can be reconstructed as images. By comparison with CT scans, MRI images possess extraordinary detail and clarity (see Plate IV), and this form of imaging has become the gold standard for the visualization of normal and abnormal anatomical structures, in the head or in any other part of the body. More expensive than CT scanning, it is nevertheless widely available and routinely applied, but it can provide no more than a picture of structures in the body; it cannot tell us what these structures do (Perani & Cappa, 1999).
Since the early 1990s the MRI technique has been developed in order to provide images, not only of structure, but also of function. Functional magnetic resonance imaging (fMRI) achieves this by detecting the flow of glucose and oxygen in the blood to areas where the neurons are most active and therefore consuming these substances. It is possible to acquire four or more images every second and therefore to obtain a dynamic, if rather crude, picture of the brain in action as it deals with cognitive processes.
An important aspect of fMRI technology has been the use of the subtraction technique. This is based on the idea of comparing the activity of the brain engaged in a particular task with the state of the brain “at rest,” or performing a neutral task. The differences between the two sets of images are calculated by simple subtraction, revealing the areas of activity that are particularly associated with task performance; it is an image of these differences that is displayed (for an example, see Plate V).
Functional magnetic resonance imaging has replaced the earlier technique of recording regional cerebral blood flow (rCBF) in which short-lived radioactive isotopes were attached to molecules in the blood through an injection, and their progress through the head, and areas of local concentration, detected. Although employed within limits of safety, rCBF was limited by the number of scans that could be obtained and by how often it could be repeated. In contrast, fMRI is without such safety concerns and there are no problems in performing repeated or multiple scans. Although expensive, and only available in a small number of specialized centers, fMRI has become the most important of the functional imaging techniques in experimental neuroscience.
An example of the way that fMRI is currently used within neuropsychology is the study of Boronat and colleagues (2005). They were interested in the idea that conceptual knowledge is distributed through the various sensory, motor, and verbal-declarative attribute domains in a way that reflects the activity of these systems at the time the knowledge was acquired. Answering questions about this knowledge should then reactivate these systems. In their study, answering questions about object manipulation was predicted to activate the sensory–motor system employed in object use, and questions about object function to reactivate the verbal-declarative system. Participants viewed picture or word pairs identifying manipulable objects and were asked to decide either if the objects were manipulated similarly, or if they had a similar function; fMRI demonstrated that decisions about manipulation were associated with greater activation of the left inferior parietal lobe, more especially with pictures than with words. This is the region that is thought to support skilled object use, according to more traditional neuropsychological studies. No brain regions were found to be particularly associated with decisions about object function. Their results therefore partly supported the hypothesis concerning distributed representations.
There is now a large research literature employing fMRI, although the contribution it has made to neuropsychological understanding, to date, has not been proportionate to the research effort involved. Attempts continue to be made to develop increasingly powerful machines, involving magnets weighing many tons, in order to improve the clarity and specificity of the images, but the technique has an inherent limit of its temporal resolution, and also requires participants to be tested in a rather unusual environment. Although the scanning environment with the most modern machines is less frightening than the claustrophic tubes into which participants were previously inserted, there is evidence that the assessment environment in itself may change the nature of cognitive functioning and, for example, impair memory performance in some individuals (Gutchess & Park, 2006).
Positron Emission Tomography
A precursor to fMRI, which is still applied, is positron emission tomography (PET). PET, which is a direct development of rCBF, is an alternative method of dynamically imaging the function of the brain. Although the images are exceptionally clear (and often quite beautiful), the technique is expensive, participants may only be subject to about 10 scans within a 12-month period, and it takes about 30 seconds to acquire each scan. There is therefore little hope of identifying specific cognitive processes from PET scans.
Nevertheless, PET scans continue to be of some importance, and an interesting example is provided by a previous student of mine, Eleanor Maguire (Maguire, Frackowiak, & Frith, 1997). She studied London taxi drivers, who are required to pass a stringent test of their knowledge of London streets (known as “the Knowledge”) before gaining their license, and naturally access this information in their memory on a daily basis. Experienced London taxi drivers showed an activation of the right hippocampus together with a network of other structures when recalling complex routes around the city, in comparison to baseline and nontopographical memory tasks. Nontopographical memory retrieval involved the left inferior frontal gyrus, with no activation of temporal regions (associated with the hippocampus). Incidentally, for those who might dismiss standard MRI as a neuropsychological technique, this group has also shown that the hippocampus is structurally more developed in these taxi drivers than in comparable controls without “the Knowledge.” Of course, whether this is a development consequent upon acquiring these special topographical skills, or whether individuals with greater potential based upon brain structure self-select to enter this form of employment, is an open question.
A further illustration of the way in which PET has been employed to investigate neuropsychological models is to be found in Peigneux et al. (2004), who investigated the components of upper limb apraxia by requiring participants to undertake tasks of pantomiming familiar gestures to verbal command, imitating familiar gestures, imitating novel gestures, and matching objects for functional use. They were able, on the basis of their results, to refine and extend an existing model of apraxia based upon the clinical neuropsychological literature.
The most exciting, and latest, development in imaging is magnetoencephalography (MEG). MEG at last offers the prospect of both high spatial and temporal resolution in functional mapping, is entirely noninvasive, and promises to make a considerable impact in neuroscience (Preissl, 2005; and see Plate VI).
This technique can be thought of as similar to EEG, but rather than detecting the electrical fields associated with groups of neurons, MEG detects the magnetic signals that accompany brain activity. These magnetic fields are extremely weak, and recording them has been compared to trying to detect the sound of the footsteps of an ant in the middle of a rock concert, but they can be detected by an array of superconducting coils. The efficiency of the device is increased by operating it in a special room shielded from magnetic and radio frequency signals and bathing the coils in liquid helium at –269 degrees Celsius, within an insulated cryogenic vessel known as a dewar. This is therefore a demanding and expensive technique and MEG machines are not yet widely available.
The remarkable potential of MEG is that it is capable of identifying specific areas of activity as small as 2 mm and with a temporal resolution of 1 ms. In this it dramatically outperforms any of the available imaging and electrophysiological techniques–at least, this is what is claimed by the proponents of MEG, although the degree to which the technique can precisely identify particular locations is still rather controversial. An example of the use of MEG can be found in the study of Kaehkönen, Rossi, and Yamashita (2005) in which MEG was used alongside traditional electrophysiological recording of auditory evoked potentials (AEP). (The use of different imaging techniques in combination has been a general strategy to improve the information that can be obtained, and to aid interpretation.) They examined the effects of alcohol on the auditory processing of frequency changes, presenting infrequent deviant frequency tones and novel sounds in a double-blind, placebo-controlled, design. Alcohol was shown to decrease the amplitude of AEP potentials, but not their latency, and to have a differential effect for tones, frequency change, and novel sounds. The MEG results confirmed these changes but provided additional and more detailed information not revealed by the AEPs.
A final recent innovation deserves a brief mention. This is an alternative development of MRI into diffusion-tensor imaging (DTI) and fiber tractography (FT) (Mori & van Zijl, 2005). These are anatomical, rather than functional, techniques and are likely to be of greater importance clinically than as a research tool for investigating cerebral function. The technique relies upon the fact that water diffusion occurs differently in the gray matter of the cortex and the white matter of the “wiring.” Essentially, water diffusion occurs more rapidly along a white matter tract than across the boundary between the tract and surrounding tissues. Magnetic resonance imaging has been adapted to detect these patterns of diffusion and so create 3-D images of the fiber tracts within the brain. The images are often superimposed upon “standard” MRI images to give context and orientation. The images are striking, and reveal a complexity to these pathways that belies our simplistic notion of regions of activity rather directly connected by simple bundles of fibers; but then the brain is undoubtedly more complex than the ideas that we can hold about it. An example of the application of tractography to an understanding of blindsight (see p. 127) can be found in Leh, Johansen, and Ptito (2006).
Having briefly reviewed the techniques employed in electrophysiological investigations and studies employing brain imaging, and the kinds of experiments that have been performed, let us consider what general result has emerged. On the one hand it would seem that there is remarkably good evidence, given the technical demands of the techniques, for the identification of cerebral processes that mirror those inferred from human performance studies. The results appear to fit remarkably well with what might be expected from divided visual field and dichotic listening research.
On the other hand, it is worth noting that there are considerable differences in the kinds of activity studied in clinical and experimental research. The experimental techniques, and especially the electrophysiological methods, demand a more passive involvement on the part of participants, and it has to be recognized that rather different domains of behavior are being studied by each of the different approaches and techniques.
It must also be remembered that there are considerable methodological difficulties in both the EEG and EP literature. Close inspection of the reported results also reveals little unanimity about the precise effects to be observed, and there is little replication of experimental findings. Much of the interpretation of results is uncomfortably post hoc, and our ignorance about just what to look for in either the EEG or in EP components means that it is almost impossible to construct sound experimental tests of precise neuropsychological hypotheses.
Brain imaging is currently the most fashionable area of experimental neuropsychology. I have a tendency to feel, as a personal view, that it has yet to fulfill its promise or, at least, to deliver the outcomes that were promised for it. The inherent limitations of the various techniques have not always been fully acknowledged or accepted, and a consequence has been that less glamorous, but more soundly established, techniques have been relatively neglected. Almost 20 years ago, a very level-headed friend who works in imaging told me that he thought it would take 20 years before imaging made a real contribution to neuropsychology. It starts to look as though he was entirely right. After a period in which too many grand claims–that could not really be supported–were made for imaging, a more recent wave of advances employing more sophisticated research designs is beginning to reveal novel information about the neural bases of psychological processes. Examples have already been given in the areas of memory (see p. 89) and altered states of consciousness (see p. 171). Substantial advances in the future are likely to lie in combining image data with data from clinical lesion studies (currently neglected) and allowing each approach mutually to inform the other.
All these techniques described still hold considerable promise, and it may well be that the bridge will yet be established enabling us to link, precisely and accurately, the processes that we can observe in task performance with the associated cerebral events that we can observe in electrophysiological recordings and in functional imaging.
Gender and Handedness
Happily, we are not all alike, and one of the major historical themes of experimental psychology has been the investigation of just how and why we differ. This interest has inevitably been reflected in human neuropsychology.
There is a whole range of variables upon which individuals differ and which may be relevant to neuropsychological function. They include sensory function and perceptual abilities; metabolic and endocrine factors; motor skills and manipulative ability; intellectual performance and cognitive style. Even emotional traits, personality differences, and anxiety bear some relation to cerebral function. However, two particular characteristics of the individual subject have come under extensive scrutiny, and they are gender and handedness or sinistrality.
GENDER DIFFERENCES IN CEREBRAL ORGANIZATION
The evidence in general psychology about gender differences in cognitive ability has pointed to a superiority among males for spatial and mechanical skills, and a superiority among females for verbal skills (Hutt, 1972; Maccoby & Jacklin, 1974). While this dichotomy has not been accepted uncritically, it is generally recognized that there are some differences between the genders along these lines. Whether such differences can be attributed to biological factors, or result from social and cultural influences, is still hotly debated. It seems reasonable that both are involved, although the extent of the contribution of each is unclear.
It was almost inevitable that these ideas would be carried across into neuropsychology, particularly considering the battle to explain the very considerable degree of variability shown by subjects in performing experimental tasks. The kind of dichotomy proposed for gender differences also appeared to reflect, appealingly, contemporary ideas about the dichotomy between the specializations of the two hemispheres.
The work of Buffery and Gray (1972) gave a significant impetus to research by suggesting a model in which males were considered to possess less cerebral lateralization than females. The male brain was seen to have language represented more bilaterally, which in turn implied that spatial abilities must also be represented more bilaterally to occupy the remaining capacity. The result was, for males, a relative deficit in verbal ability, which suffers as a result of bilateral representation, and a relative advantage in spatial ability, which might benefit from ambilaterality.
Subsequent research and theorizing have almost unanimously failed to support the Buffery and Gray hypothesis. Nevertheless, it illustrates the kind of argument that has led to current theories. These, when they have accepted the presence of gender differences, have proposed that the female brain is more bilaterally organized than that of males. The current debate is whether the evidence justifies such a conclusion. Before considering the evidence, some methodological issues must be raised.
There is a particular difficulty about the manner in which much of the gender difference data have been collected. Particularly because of the general feeling that gender differences might be important, an experimenter will typically include equal numbers of males and females in the participant sample, and will probably include the factor of gender in the statistical analysis of the results. If there is a significant effect of gender, it will be reported; if there is no significant effect, it may well not be reported at all. Reference may not even be made to gender in the report of the analysis and discussion of the results. Even if such a bias were not present in the reporting of results, the incidental way in which information has accumulated, from studies not specifically designed to investigate gender differences, must be regarded as unsatisfactory.
The question as to whether there are differences between the genders in the effects of clinical lesions raises another problem, which is the preponderance of males in clinical samples. Where comparable groups of male and female brain-injured patients are studied, there are almost inevitably differences between the two gender groups in terms of the cause of the lesion, the age of the patient, the severity of the lesion, and so on (see p. 47). McGlone (1980), from the evidence of her 1978 study, and from her review of similar studies, reports that while males show effects of the laterality of the lesion on the Verbal–Performance IQ discrepancy (left lesions affecting Verbal, and right lesions Performance, IQ on the Wechsler Adult Intelligence Scale, or WAIS), these effects are not to be found in females. It is also claimed that the aphasic effects of left hemisphere lesions are less severe in females, who also recover more rapidly, and this hypothesis has been more recently supported by Frith and Vargha-Khadem (2001) with respect to literacy skills. The commentaries that accompany McGlone’s 1980 review together with the additional data presented there, and the data set out in Fairweather (1982), show that while some studies support the conclusion of female bilaterality on the basis of clinical evidence, there are also a number that do not.
Experimental neuropsychology has yielded a very confused set of results on gender differences. To illustrate this confusion, in the early studies using the divided visual field technique and verbal stimuli, some researchers found the predicted greater right visual field advantage among male subjects (Bradshaw & Gates, 1978; Kail & Siegel, 1978). However, there are also well-conducted studies in which no evidence of a genderrelated effect was apparent, for example, that of Hannay and Boyer (1978). Other researchers found differences, and then failed to replicate them (McKeever & Van Deventer, 1977). A similar picture emerged from the studies employing nonverbal stimuli, with both positive (Davidoff, 1977; Rizzolatti & Buchtel, 1977; Sasanuma & Kobayashi, 1978) and negative (Bryden, 1976; Kail & Siegel, 1978) findings.
Merely tallying up the number of studies that find a certain result, irrespective of their methodological adequacy, may be a dangerous procedure, but it is interesting that Fairweather (1982) calculated that among 49 studies with verbal stimuli, 42 found no gender effect, 5 found more lateralization among males, and 2 more lateralization among females. With nonverbal stimuli, of 62 studies, 13 showed greater lateralization for males and 4 for females. It is particularly striking that of the 111 studies reporting on gender differences, 87 found no effect of that variable. Fairweather concluded that there was no evidence for gender differences on verbal divided visual field tasks, and slender evidence for it on nonverbal tasks. The only hint of a consistent finding was in facial recognition, stemming from the work of Italian laboratories (Umilta, Brizzolara, Tabossi, & Fairweather, 1978). McGlone’s (1980) review of the same material was much more selective, and reached the conclusion that the most parsimonious explanation for the reported findings was that there is less functional brain asymmetry in females.
Matters are complicated by the presence of differences in sensory and perceptual performance between the genders, and by the suggestion that there may be differences between males and females in cognitive strategy. The idea here is that females prefer to adopt “verbal strategies,” and males “nonverbal” ones (Bryden, 1978). Incidentally, Metzger and Antes in 1976 showed that instructing subjects explicitly to adopt a certain strategy had a general effect on performance, but it did not affect the basic pattern of cerebral asymmetry inferred to underlie task performance. However, the concept of differential strategy utilization is still often discussed as either a cause or an effect of fundamental differences in cerebral lateralization. Strategy adoption, as a gender-linked factor, clearly could have the observed effects on task performance (and criticisms of the mechanism have often confused within-individual effects with the effects that might be seen when the results of a group of subjects are collated). However, the operation of this mechanism has yet to be experimentally demonstrated, rather than proposed as a post hoc explanation.
Fortunately, there have been a series of more recent meta-analyses that have helpfully reviewed the accumulating evidence. Zaidel, Aboitiz, Clarke, Kasier, and Matteson (1995) reported an analysis of 31 divided visual field experiments, and Voyer (1996) also reported on an extensive series of studies, finding results that essentially supported McGlone’s (1980) view. However, a considerable service has been provided by Hiscock and colleagues who, between 1994 and 2001, published four major papers reviewing the literature on, in turn, dichotic listening, divided visual field studies, tactile asymmetries, and dual task experiments (Hiscock, Inch, Jacek, Hiscock-Kalil, & Kalil, 1994; Hiscock, Israelian, Inch, Jacek, & Hiscock-Kalil, 1995; Hiscock, Inch, Hawryluk, Lyon, & Perachio, 1999; Hiscock, Perachio, & Inch, 2001). The first two papers reviewed 140 and 217 studies respectively, in which gender differences were analyzed, and the studies were further classified in terms of the stringency of their design and methodological controls. For dichotic listening, 9 of the 11 “stringent” studies found increased hemisphere specialization in males by comparison with females, although only 21 of the 140 found a weaker but consistent effect. A similar pattern was found for divided visual field studies: 17 of 23 studies meeting the stringent criteria demonstrated the increased specialization for males, but of the less stringent only 27 studies found a consistent parallel effect, and 6 studies reported contrary findings. The similarity of overall findings in these two meta-analyses suggests that this is a real effect, but the conclusion can be no more than that there is a weak population-level difference between the genders in the degree of lateralization, with greater lateralization seen in males. This is certainly not of a magnitude from which predictions might be made about individual cases. The overall contribution of gender differences to the variance of lateralization is no more than 1 to 2%, and this cannot be regarded as of real significance.
The two later meta-analyses of tactile and dual-task studies were performed with smaller numbers of studies (30 and 51 respectively) and the number demonstrating unambiguous gender differences was small. No clear conclusions could be drawn from either of these studies alone, but the findings do not contradict the earlier, larger studies and so lend some support to the hypothesis of increased bilateralization of functions in females.
Two more recent studies are worth a mention. Sappington and Topolski (2005) analyzed SAT scores for mathematics in a very large number of American students. They found that being male and nonright-handed and having an earlier onset of puberty were all associated with higher math scores, while being female, being non-right-handed, and having a later onset of puberty were linked to very poor scores. On the assumption that greater lateralization confers an advantage for formal intellectual tasks, such as mathematics, this finding is consistent with the general view now being adopted. However, employing fMRI, Clements et al. (2006) found increased left lateralization for males on a phonological task with more bilateral activity on a visuospatial task (as might be expected), but for females bilateralization on the phonological task (also expected) with increased right lateralization on the visuospatial task.
An intriguing sideline to the investigation of gender differences has been the study of sexual orientation and laterality. Again, there is a fairly recent meta-analysis to assist us (Lalumière, Blanchard, & Zucker, 2000). Analysis of 20 studies revealed that individuals with a homosexual orientation had 39% greater odds of being non-right-handed and, by inference, having a less lateralized cerebral organization. The effect was much greater for women, where the increase in odds was 91%, while for men it was only 34%. It is reasonable to suspect that the basis for this effect lies in early neurodevelopment, through prenatal exposure to sex hormones, to a maternal immunological reaction, or simply in developmental instability. However, the precise mechanism that creates this effect remains unclear.
What view are we therefore to take? Is it that there really are gender differences in brain organization, but that they are subtle and often masked by other sources of variability in task performance? The withingender variability is certainly large in comparison with the mean between-gender difference, and the more substantial effects of individual differences in memory function may also swamp any other effects that might be present. That so few studies are designed with the intent of studying gender differences may also explain why so few find these effects on subsequent analysis. We may adopt the position of McGlone and Hiscock and colleagues and believe that the effects are elusive, although real, and that when they can be demonstrated they show that females have a less lateralized brain. Alternatively, we may follow Fairweather and consider that the evidence is too thin and inconsistent, or the contribution of the effect too small, to allow us to include gender in any theory of cerebral laterality (or laterality in any theory of gender differences). My personal inclination, formerly for the latter position, is to accept that the differences are real but that they are not of any (realworld) significance. Students should read the literature and arrive at their own opinion. However, significant gender differences have not to date been shown convincingly as present either in the effects of clinical lesions or in the neuropsychological performance of normal subjects. We have to await the accumulation of further evidence to prove this view too cautious.
Handedness is one of the most obvious ways in which the performance of some people differs from that of the majority. Although the precise figure depends on how left-handedness is defined, most surveys find that between about 8% and 12% of the population is left-handed. Although there is some geographical variability (Raymond & Pontier, 2004), this figure seems typical of most contemporary cultures and even traditional societies (Faurie, Schiefenhövel, Le Bomin, Billiard & Raymond, 2005). The Eipo, who live in the forests of Irian Jaya, an area of Papua New Guinea, wear a plug in their earlobe on the opposite side to the hand they use for shooting with a bow; with this knowledge it has been possible to establish handedness rates from photographs where the ears are visible. The rate also seems to have been invariant through recorded history. There is even evidence that in prehistoric humans roughly 10% were sinistral, or left-handed.
The study of handedness is a major topic in its own right, with sociological, cultural, and even religious aspects. There have been many theories about the origins of left-handedness, but ever since asymmetries in the nervous system were first recognized, the major theories have all linked handedness in some way with cerebral laterality (see Herron, 1980; Coren, 1990).
Against this background, the contribution of genetic factors to the determination of handedness has never been denied. If you have one lefthanded parent, you are 2–3 times more likely to be left-handed than if both your parents are right-handed; if both your parents are left-handed, the probability is 3–4 times (Bryden, Roy, McManus, & BulmanFleming, 1997). Left-handed children are therefore more commonly, although not exclusively, the children of left-handed parents, and it can be shown that this is not merely due to cultural transmission. In fact, it has been found that left-handed parents may more strongly discourage sinistrality in their children because of their awareness of its disadvantages. In recent years there has been a debate between two opposing theories: that of Annett (2002) and that of Levy and Nagylaki (1972). Much of this debate is relatively technical, and the Levy–Nagylaki model is well specified. However, although a more specific model, it seems to account less well for all the available data than Annett’s model. The latter is known as the “right shift” theory, because it involves the inheritance of a single factor that shifts handedness, which would otherwise be determined by chance, toward right preference. Of individuals who do not inherit the factor, half will be of each handedness by random chance variation; among those who do inherit the factor, most, although not all, will be right-handed. One of the simplest predictions of this model is that of the children of two left-handed parents (who in most cases cannot inherit the right shift factor), almost half will be left-handed. This, and more complex, predictions seem to be borne out fairly well by the data. Nevertheless, there continue to be some who argue that this model is too simple (McManus & Bryden, 1993; Ecuyer-Dab, Tremblay, Joanette, & Passini, 2005), and that handedness may be an X-linked characteristic (McKeever, 2000) which would be incompatible with Annett’s theory.
Much of the research on handedness assumes that the measurement of handedness is a simple matter. It is not. There is no true dichotomy between right and left handers. Although the majority of people perform all skilled actions preferentially with the right hand, the remainder of the population forms a continuum out to “pure” left-handedness. Much of the discussion about handedness fails to take this adequately into account. It is, however, recognized in the questionnaires generally employed to assess handedness (Annett, 1970; Bryden, 1977; McFarland & Anderson, 1980).
As a final general point, it should be remembered that handedness occurs in a general setting of biological asymmetries. Although the link with some of these may not appear very direct, for example the hairwhorl on the crown of the head, fingerprints, nostril size, and the arrangement of the male testicles, the association with others, such as sighting dominance (which eye you use for a telescope) or footedness (with which foot you step out, or kick a ball), may be more clear. However, while all these phenomena show a statistical relationship with handedness due to an underlying general biological factor, none shows a very direct relation. For example, ocular dominance has less connection with handedness than is often assumed. The association, if any, is a very complex one (Birkett, 1977; Porac & Coren, 1976).
Handedness and Cerebral Dominance
It is now unfashionable to speak of “cerebral dominance” as a general factor determining laterality, although contemporary theories continue to use the idea quite freely under a number of guises. They may speak of “dominance for speech” or “dominance for phoneme discrimination” or whatever specific function is being considered, but in fact all theories involving handedness imply a general factor determining the overall pattern of laterality. There are four ways in which this factor might be linked with handedness:
This idea has a very long history and is simply that stress or trauma in early life, especially around birth, may result in left-handedness. The observation of an unusually high rate of left-handedness among many groups of abnormal children and adults, especially those with some dysfunction of the central nervous system, has always been used as an argument to support this idea. If we accept that intelligence is normally distributed in the population (and psychologists arrange that measured IQ follows this theoretical statistical distribution), there are more lefthanded people at the lower end of the range than would be expected by statistical variation. This can be explained by the depression of intelligence as a result of accident or pathology. In the same way, handedness is thought to be affected by early damage or injury. This argument reveals the cultural bias that has often linked sinistrality (“sinister”) to weakness and even evil, and we might well question whether the depression of intelligence can be considered a parallel to a shift from right- to left-handedness. However, pathology is still an important factor in some explanations of left-handedness, and there is no doubt that severe abnormal states can be associated with disorders of the nervous system and may result in a variety of atypical patterns of handedness. There is much more doubt whether less severe birth stress and trauma, which do not produce any overt abnormality, can be considered to lead to a shift in hand preference. The evidence about this is discussed below.
The Contralateral Rule
This concept is much more simply stated and underlies many popular accounts and discussions of handedness. The concept is that speech dominance is always located in the hemisphere contralateral to the preferred hand. Stated another way, if almost all right handers have left hemisphere speech dominance, then left handers must have right speech dominance. Curiously, although it has never received much support from scientific research, the idea often crops up in discussions by anthropologists, palaeontologists, and biologists of the development of language, tool use, and manual skill, when speech lateralization in the evolution of language is linked to asymmetries of skilled performance. To be fair, it is not necessarily implied that the relation between speech lateralization and skilled manual performance must also hold for left handers, but such a relation is often assumed, so that the mirror image nature of left preferrent performance is extended to a mirror image model of the nervous system.
This is probably the most popular position adopted, but is also the most vague. It states that left handers, including “mixed” handers either implicitly or explicitly, have a lesser degree of lateralization than the right-handed, just as we have seen argued for females. Occasionally the small group of “pure” left handers is excluded (and then follows the contralateral rule), but this is not generally the case. The reasons for this relative bilateralization are rarely clearly stated, and it is more common for this position to emerge from a review of the available evidence on lateralization of function and handedness.
The Right Shift Model
Annett (2002) has argued that the right shift model of genetic determination of handedness (see p. 298) predicts a pattern of cerebral asymmetry for the left-handed that is distinct from the bilateralization model. Annett very clearly demonstrates how the strictness of the criterion for what we would consider left-handedness has an important effect on the pattern of cerebral asymmetry to be expected for that group. If the criterion is strict, so that only the more extreme sinistrals are included, then very few will have the right shift factor, and just over half will have left hemisphere speech. As the criterion becomes more lax, more individuals will be included who have the right shift factor, and the proportion with left hemisphere speech will rise as a result since the right shift factor is, of course, considered not only to increase the probability of righthandedness, but, at the same time, of left hemisphere speech. The direct association between speech lateralization and hand preference is still evident, if less directly determined.
All these models make predictions about the pattern of asymmetry to be observed in clinical phenomona and in laboratory performance. The predictions to be made from the pathological model are least clear, but are generally taken to imply some reversal, perhaps with a degree of disorganization, of observed laterality. The contralateral rule predicts reversed patterns of asymmetry, and the bilateralization model that no asymmetry will be observed. Annett’s model predicts a reduced degree of the normal pattern to be found with right handers. It would seem, therefore, easy to test between these different predictions. In practice it is not. The pathological model admits results indicating a weak degree of asymmetry, especially if they are in the “reversed” direction. The bilateralization model admits all results that show any weak degree of laterality in either direction, allowing for sampling bias, errors of measurement, and so on. The right shift model is compatible, on the same grounds, with no asymmetry and even weak reversal, and with a strict criterion of sinistrality it becomes practically indistinguishable from the bilateralization model. Nevertheless, we must try to evaluate these models in the light of the available evidence.
Three final factors must be mentioned before we turn to the evidence. One is familial sinistrality, that is, whether the individual has any relatives who are left-handed. This factor is often included as an added dimension in studies of laterality, although its theoretical significance is not always clear; it should, of course, distinguish between “inherited” and “pathological” handedness, if such a distinction were to prove valid. It is sometimes linked to the second factor, the strength of handedness. This variable can be extracted from questionnaire responses or from measures of lateral dexterity or performance speed. Although Searleman, Tweedy, and Springer (1979), for example, have shown strength of handedness and familial sinistrality to be unrelated, it is still often assumed that left handers with sinistral relatives are purely or extremely left-handed, and hence right handers with sinistral relatives must be less clearly right-handed than those without. It is expected that these variables will be found in the laterality observed on experimental tasks.
The third and last factor is that there is a potential interaction between gender and handedness when considering bilateralization. As studies of gender in handedness almost unanimously show a greater frequency of left-handedness among males (see Annett, 2002), a more bilateral brain might be associated with both male gender and lefthandedness. Or is hand preference a confounding factor in studies of gender, and vice versa? These are questions without, as yet, clear answers in the scientific literature.
The Evidence about Left-Handed Brains
The first question to be answered is whether the concept of pathological left-handedness receives any significant support. Bakan (1977) has been one of the foremost proponents of this idea, and has defended it against a number of criticisms. He has also shown that there is an excess of left handers among first and late (fourth or later) positions in birth order, which he claims links sinistrality with birth stress. He goes further and specifically proposes that the left-handedness results from damage to the left hemisphere pyramidal motor pathways following perinatal anoxia. Nevertheless, other investigators have failed to support Bakan’s findings. For example, Hicks, Elliott, Garbesi, and Martin (1979) constructed a study that they considered would give the greatest chance of replicating the Bakan results. They found, however, that the relationship between handedness and a combination of factors known to be associated with high-risk birth was “trivial.” Annett and Ockwell (1980) found only the weakest relationship for daughters but none for sons, while Leiber and Axelrod (1981) found the slightest of links for sons but none for daughters. The current situation is that the gravest doubt is cast upon the validity of a contribution of birth stress to the determination of handedness.
Turning to the clinical evidence, two methods have already been introduced (see p. 86). The Wada technique by which sodium amytal is introduced into the carotid artery, so interrupting function in the hemisphere on the same side as the injection for a brief period, has given us some data on the speech lateralization of left handers. Rasmussen and Milner’s 1975 data, which have already appeared in Table 7.1, show that of the left handers studied, about 70% had left-hemisphere speech, 15% right hemisphere speech, and 15% had speech bilaterally represented. Isaacs, Barr, Nelson, and Devinsky (2006) have recently reported similar findings; 9% of strong right handers had atypical language lateralization, while in the ambidextrous it was 46%, and 69% in strong left handers. Further support comes from a recent fMRI study (Szaflarski et al., 2002).
The second method mentioned in Chapter 7 was the study of the frequency and severity of aphasia following unilateral lesions of the left or right hemisphere. This is a complicated issue when studied in detail, but Satz’s analysis of the studies reported between 1935 and 1973 is probably the clearest summary for our purposes (but also see Coren, 1990). In its most recent form (Carter, Hohenegger, & Satz, 1980), it suggests a model in which 24% of left handers have left-sided speech, none right-sided speech, and 76% bilateral speech.
A third, although less important, method might also be mentioned: the study of unilateral ECT (Warrington & Pratt, 1973, 1981). ECT (electroconvulsive therapy) for depression is now often administered unilaterally to avoid some of the undesired aftereffects of bilateral treatment, which may include interference with verbal memory. Such interference is minimized if the shock is given only on the side that is not dominant for speech. The studies that established this aspect of the unilateral method, and the early treatments that may be given to either side in order to establish speech laterality, also yield information that is of relevance for our question. What the studies suggest is that left-sided speech is to be found in about 70% of left handers, in line with the Wada technique results. Incidentally, Warrington and Pratt found that neither dichotic listening indices nor the hand used for writing could be used to determine speech laterality with any confidence.
What are we to make of this clinically derived evidence? There are problems associated with each of the methods discussed. All are concerned with abnormal subjects, and the general problems of comparing the effects of clinical lesions in different groups apply to the aphasia evidence. The Wada technique is also highly stressful for the patient and imposes severe practical limitations upon testing. The inferences drawn from the study of ECT are particularly indirect, and depressed patients may well have an abnormal pattern of cerebral lateralization (see Chapter 16). Criteria for left-handedness also differ between studies. Finally, it is probable that the nature of the testing and the way that responses are scored make a definite finding of right or left lateralization more likely for the Wada and ECT methods than for the aphasia assessments. So, bearing in mind this point, we might conclude that very few, if any, left handers have right hemisphere speech, at least a third have clear left hemisphere speech, and the remainder, between a half and two-thirds, have relatively bilateral speech in which the greater contribution comes from the left hemisphere.
What about the evidence from experimental neuropsychology? The work of Geffen and colleagues on a dichotic listening test that would predict speech laterality, and their encouraging results, were discussed in Chapter 12 (see p. 255). The studies of this group have included left handers, and have shown that a left ear advantage or no advantage is more common among left handers than right handers, although handedness did not affect the relative magnitude of the asymmetry. Familial sinistrality in male left handers increased the probability of finding a right ear advantage, and therefore left hemisphere speech (Geffen & Traub, 1979, 1980).
There are a very considerable number of studies using the dichotic listening and divided visual field techniques that have included handedness as a subject variable. Rather, as with gender, it has been easy to add handedness as a variable in a study never designed to investigate this factor and the result has been equally unsatisfactory. Happily, there are some valuable reviews of this rather baffling literature: Annett (1982, 2002), Hardyck and Petrinovich (1977), Levy (1980), and Coren (1990). Both Annett and Hardyck include a tabulation of the most prominent studies and their results. Unhappily, however, the reviews do not entirely agree in their conclusions.
Annett (1982) shows that the usual kind of divided visual field study, using verbal stimuli and obtaining a right visual field advantage for right handers, generally produces no asymmetry in left handers. Of the 15 studies quoted, only one found a left field superiority, while two found the right field to be superior. The picture was much the same for the 13 nonverbal studies cited. Most showed a left field advantage for right handers, but no asymmetry for left handers. For this group, two studies found a left, and three a right, visual field advantage. Familial sinistrality was reviewed independently in 19 studies and the variable was considered to add “nothing substantial” to the results. It was occasionally found to reduce the lateral advantage of right handers, and to be linked with a trend to reversed asymmetry in left handers, but these effects were neither strong nor clear. Annett concludes that on visual tasks, left handers show no overall asymmetry, especially if lefthandedness is strictly defined.
Hardyck, by contrast, considers familial sinistrality of significance, and his review leads to a model in which the right-handed without sinistral relatives are the most strongly lateralized. Left handers with sinistral relatives are the most bilateral. Right handers with left-handed relatives and left handers without left-handed relatives are intermediate and show weak (but not reversed) lateralization.
Levy reaches yet a different conclusion. She considers that a few sinistrals have strong right hemisphere language (although this is based principally upon the clinical evidence), but that the majority show weaker left hemisphere language in comparison with right-handed subjects. (This is having discounted the small number of sinistrals whose left-handedness can be attributed to major early disruption of the left hemisphere, and whose speech functions are then controlled from the right hemisphere.)
In so far as all these studies can be seen to present a common view, they suggest that very few, if any, left handers have right hemisphere speech. When the latter is observed in individual subjects, it is more commonly in a clinical rather than in an experimental context. It may well be that the small number of cases of reversed dominance results from abnormal processes, and that in normal subjects right-hemisphere speech is a rare phenomenon. All the reports agree that the majority of left handers will show weak or relatively bilateral left hemisphere speech lateralization. (The proportion of these individuals whose tendency to left-sided speech will be sufficiently pronounced to consider them to have clear left speech dominance will depend on the criteria for dominance, and will probably also be influenced by the criteria adopted for left-handedness.) There seems a clear agreement, therefore, that lefthandedness is in general associated with a reduced degree of cerebral lateralization.
Having reached this conclusion it is not surprising that there are reports, as we have already noted, of a weak relationship between handedness and speech laterality as assessed by dichotic listening (Searleman, 1980); that left handers show greater variability across occasions of testing (Hines, Fennell, Bowers, & Satz, 1980); and that for left handers there is no clear correlation between the results of visual and auditory tests (Hines & Satz, 1974).
The conclusion concerning familial sinistrality is less clear. While it seems to have been agreed that the existence of sinistral relatives is likely to be associated with a shift away from strong left hemisphere speech, it has not been agreed how important is the contribution of this variable.
In terms of the four models with which we started, pathological processes appear to provide a significant explanation in only a very small number of cases, and they cannot be considered relevant to any general explanation of left-handedness. There is also no evidence to support the operation of the contralateral rule. However, the evidence is broadly in agreement with both the bilateralization and the right shift models. As we noted, it is very difficult to devise a critical test between these two models given present techniques. The clinical evidence would marginally favor the bilateralization model because of the large proportion reported to have clear left sided speech. However, it could be argued that this is not entirely incompatible, given certain methodological considerations, with the right shift model. Annett has shown that the experimental evidence can fit very well with the right shift model, and if this model is accepted to explain the inheritance of handedness, then it also seems to serve fairly well as a way of interpreting observations of cerebral laterality in different handedness groups. It certainly seems that this, or some other version of the general bilateralization model, provides a good description of how left-handed brains differ from right-handed ones.
Differences in Cerebral Mechanisms
While most researchers have been content to present the empirical evidence of performance differences between left and right handers, a few have asked how these differences might be expressed in cognitive operations. Semmes (1968) developed an influential model of hemispheric function, suggesting that organization was “focal” in the left hemisphere and “diffuse” in the right hemisphere, with consequent advantages for verbal and spatial processes respectively. She admitted, however, that the model could not account for the cerebral organization found in many left handers.
Levy (1974) developed a model based upon a different dichotomy, between “analytic” and “global” processing. Within this model, left handers are specifically taken into account, and are considered to have bilateral representation of the functions based upon analytic processes normally located in the left hemisphere in right handers. However, while speech, language, writing, and calculation are represented in both hemispheres, there is no corresponding bilateralization of “normal” right hemisphere functions. These suffer in left handers by being crowded out by the verbal functions. Levy has expanded this model to include gender differences in cerebral organization and orientation of writing hand (see p. 308; Levy, 1980; Levy & Gur, 1980). This theory has become quite complex. Nevertheless, it still contains the idea that left-handedness generally indicates a bilateralization of certain functions, which is in turn seen in a superior level of these abilities. The functions for which no fully specialized hemisphere exists will show a deficiency in performance. Handedness is, however, only one factor among others, including birth stress, genetic factors, and fetal gender hormones, which may be linked with a depression in the level of ability.
A different model was proposed by Beaumont (1974). This was that left handers showed a general diffuseness of cerebral organization in all functions in comparison with the relative focal organization of the righthanded. This was considered to lead to an advantage on certain cognitive functions, but a deficiency on others. The more complex the integration required by the task, the greater the advantage for the left hander, irrespective of whether the task was “verbal” or “spatial.” This model was supported by evidence from divided visual field studies.
These last two models have been reviewed by Hardyck (1977). He finds more support for Beaumont’s than for Levy’s model, but rightly criticizes the lack of an independent criterion of task complexity. His second criticism, that a more diffuse system must imply longer processing times for complex tasks, does not really follow from the model. However, he himself develops a new form of the model, which does have certain advantages over its predecessor. Hardyck’s idea is that the most bilateral individuals, left handers with sinistral relatives, will employ more interhemispheric processing. This idea can be extended to generate specific predictions about certain tasks, for instance, that left handers will have an advantage for identity judgments in divided visual field performance, and will make more errors with mirror image stimuli. This model fits well with much of the data, and deserves more attention than it has attracted. In particular, tests of some of the specific predictions would be welcome.
Are Left Handers Less Able?
An issue related to that of the cognitive mechanisms that may distinguish left and right handers is whether left handers are less proficient in certain abilities. Levy (1969) had reported that sinistrals were inferior on nonverbal or “performance” intelligence subtests, such as are found in the Wechsler Adult Intelligence Scale (WAIS), and this finding was confirmed by Miller (1971), using different tests. That left handers could have such a fundamental handicap was greeted with some skepticism, and it was argued that the populations studied were in some way unusual. In both studies they were students, who are generally selected for their verbal abilities. This might have biased the results in that the left handers examined may have been in some respects atypical of the majority of left handers in the pattern of their abilities.
The doubt cast on these studies was strengthened by the findings of three large-scale studies of the general population: the Oxfordshire Villages Survey in Britain (Newcombe & Ratcliff, 1973); a study of 7,119 schoolchildren in the United States (Roberts & Engle, 1974); and a survey of 7,688 children tested by Hardyck, Petrinovich, and Goldman (1976). These studies found no evidence that handedness was in any way related to the pattern of verbal and performance abilities. In view of the serious implications that such a handicap would have for left handers, this has been thought to be the more reasonable result.
Nevertheless, some evidence continues to appear of such differences between handedness groups. Bradshaw, Nettleton, and Taylor (1981) again found the WAIS Performance deficit among sinistral students, particularly among those with left-handed relatives. Hicks and Beveridge (1978) confirmed some of their earlier work by finding that left handers were inferior to right handers on a general factor of intelligence, and although Heim and Watts (1976) did not find differences in the level of ability on three intellectual factors, they did find a specific deficit in numerical ability. This last study was of a large sample of 2,165 nineto twenty-year-olds. Lastly, a specific deficit in reading abilities was reported by Cohen and Freeman (1978). Left handers were found to be poorer at silent reading, although they had adequate comprehension. They also suffered more, while reading out loud, from visual distortion introduced into the text, although they were less affected by linguistic distortion. This suggests a different strategy in the left-handed for reading, based more on visual analysis, which in some situations may prove less efficient.
One aspect of Annett’s recent research has been to show that there may be a disadvantage associated with inheriting the right shift gene from both parents; the heterozygote (a single right shift gene) confers an advantage. The disadvantages of a more “extreme” right shift can have an impact upon spatial and mathematical reasoning, and have implications for abilities in art, music, surgery, and sport (Annett, 2002).
It is a common observation that there is an excess of left handers (above the 10 to 12% that we might expect) among the unusually gifted, among high-achieving sportsmen (Grouios, Tsorbatzoudis, Alexandris, & Barkoukis, 2000), and in certain professions. A recent study in a tradition of studies examining different occupations found an excess of left handers among architects and lawyers and a dearth among orthopedic surgeons, mathematicians, and librarians; psychiatrists tended to be ambilateral (Schachter & Ransil, 1996).
It therefore appears that while there is no good evidence for a general and serious deficit in the intellectual abilities of left handers, we cannot discount that in certain groups of sinistrals there may be a limited range of functions that are impaired in comparison with right handers; however, there are also situations where left-handedness may be an advantage. How widespread and significant such handicaps and benefits may be has not yet been satisfactorily determined.
A final topic that, for a time, attracted a great deal of interest is that of writing posture. Levy and Reid (1978) reported that speech lateralization, as shown by a divided visual field task, could be predicted by the hand and posture used for writing. They divided hand postures into the “normal” and the “inverted.” Although the criteria for this division have never been quite clear, in the normal position the pen points to the top of the page while writing. In the inverted posture it points toward the bottom. Inverted postures are rare in right handers, but are common among the left handed. In terms of speech lateralization, the normal posture was associated with speech control contralateral to the hand of writing, while the inverted posture indicated ipsilateral speech localization. Right handers with normal posture and left handers with inverted posture therefore must have left-sided speech, while left handers with normal posture must have right-sided speech. If this model were supported, then it would be of tremendous value in providing a simple means of assessing speech laterality. Levy and Reid’s data were most encouraging.
Unhappily, as seems the normal course in this area, subsequent studies did not unanimously support the model. Two clearly supported it, and two failed to provide any support, and indirect evidence has been found to support the model in only two out of three relevant studies. Finding partial support for the model, McKeever and Hoff (1979) attempted to develop it by introducing the idea of intrahemispheric disconnection of left hemisphere visual and manual motor areas. An additional problem is that the model does not seem to work when lateralization is assessed in the auditory modality by dichotic listening, rather than in vision by a divided visual field task (Beaumont & McCarthy, 1981; McKeever & Van Deventer, 1980). This is at best an embarrassment for the model.
While there may well be something of significance about writing posture that can provide information about handedness and cerebral organization, the main findings have not as yet been clearly established. The difficulty of measuring writing posture has not helped to generate clear replicable results. What is clear is that on current evidence writing posture does not constitute a valid index of cerebral lateralization in individual subjects. Its initial exciting promise has yet to be fulfilled.
The study of individual differences in cerebral organization is an important area of research, particularly in view of the practical necessity in clinical applications of being able to establish the pattern of laterality in an individual patient. Of the two principal variables investigated, the findings with respect to the first, gender, are too inconsistent to allow for definite conclusions and the cautious reader would accept that clear gender differences have yet to be demonstrated. Nevertheless, abandoning caution a little, there seems to be some body of evidence that supports a greater degree of cerebral lateralization in females.
The second variable, handedness, shows more consistent findings in that left-handedness appears to be associated with an abnormal pattern of cerebral lateralization, with most left handers having a relative bilateralization of function. This may be mediated through a relatively diffuse functional system that requires greater interhemispheric intercommunication in the left-handed. Although there is still some suspicion that left handers may suffer distinct cognitive handicaps, large studies have generally failed to support this conclusion. Writing posture seems to reflect certain aspects of cerebral function, but cannot be used as an index of speech lateralization at the present time.
The neuropsychologist who applies a knowledge of the functions of the brain to clinical practice is likely to find that a major, although not the only, role is assessment of patients. The assessment may be to assist in the diagnosis and localization of any cerebral pathology, but it is more likely to be undertaken to establish a patient’s current functional status, described in psychological terms, to act as a basis for the design of remedial or rehabilitative therapy, and to monitor the progress of treatment and recovery.
Neuropsychologists are now centrally involved in the design and execution of therapy programs, contributing their specialized knowledge of the structure of human abilities, of the psychological aspects of training, and of specific behavioral methods to the efforts of the remedial team. This role for neuropsychologists within rehabilitation may seem obvious, but until about 30 years ago neuropsychology was principally concerned with diagnosis, rather than treatment. It was the introduction of scanning technologies, initially the computerized tomography (CT) scans (see Chapter 13) that liberated neuropsychology from its then primary role of assisting diagnosis in neurology and neurosurgery. Nevertheless, although considerably more clinical time is being devoted to rehabilitation and management, the major concern of most clinical neuropsychologists continues to be the assessment of cognitive abilities. In any clinic, neuropsychologists will typically be expected to provide an opinion on one of three questions: (1) whether the patient is suffering from any cerebral pathology and, if so, its nature, and whether focal or diffuse; (2) if focal cerebral pathology is present, which behavioral functions have been affected; and (3) how the patient can best be helped to return toward normal functioning.
The way in which answers might be sought to these three questions is sketched out below, but a general point should first be made. In trying to introduce as clearly as possible the relation between lesions and deficits, and the procedures of assessment and treatment, there is a danger that neuropsychological practice may seem to be simply a matter of following a well-charted routine. Nothing could be further from the truth. Neuropsychological practice involves sensitive and emphatic insight into the patient’s problems (and not just the cognitive problems), and a creative and questioning intellectual approach to the dysfunction, no less than in any other area of clinical psychology. As a naive graduate student I expected neurological patients to be much easier to deal with than psychiatric patients–after all they were only medically ill, like people with broken legs or liver complaints; they would, I thought, be cooperative, well motivated, and grateful for my help. My view of all types of patient has since matured. Many neurological patients are indeed a pleasure to meet, but, largely as a result of their neurological state, some may be poorly motivated, distractible, and irritable. They often tire easily and suffer from headaches, causing bad temper, and consultation requires patience and empathy. The most important personal quality in any neuropsychologist is patience; an assessment may have to be spread, in brief sessions, over many days, and rehabilitation might take months of daily regular work. This makes clinical practice no less challenging and rewarding, but it is as well to remember the difficult conditions under which much assessment and therapy is carried out when reading the following sections.
TESTING FOR BRAIN DAMAGE
This is one of the classic demands directed to clinical psychologists or specialized neuropsychologists. It is, however, the demand that is least easy to satisfy. When psychologists are asked to give opinions as to whether a particular patient shows any form of brain damage or not, they are likely to refuse to answer so general a question, or at least to feel little confidence in their ability to answer it.
The issue may seem a little outdated, and in part it is since the introduction of modern scanning, but the question still arises. A patient has suffered a mild or moderate head injury in a road traffic accident some years previously but continues to complain of cognitive and behavioral problems. The scans reveal no abnormality. Does the patient have enduring neuropsychological consequences of the head injury, despite the fact that we might expect them to have cleared up within the first year or so after the accident? Or are these problems resulting from a psychological reaction to the accident and its aftermath, and not directly attributable to the head injury at all? This is a common issue in the neuropsychological clinic, and it is important to find the correct answer. If a direct result of brain injury, at over 2 years from the accident, the problems are likely to endure for the remainder of the patient’s life. If the problem is a secondary psychological reaction, then there is a good prospect that if this is treated, most commonly by cognitive behavior therapy (CBT), the patient may recover.
The difficulty in identifying the presence of damage to the brain arises with the assumption that there is some validity in the notion of a generalized phenomenon that can be called “brain damage.” It should by now be clear that there are a variety of pathological processes that can result in cerebral lesions: trauma from blows that may or may not pierce the skull (open or closed head injuries); tumors; cerebral diseases; poisoning by toxic substances; strokes; brain hemorrhages; and the side effects of disorders of other bodily systems. These pathological processes may result in a diffuse pattern of lesions throughout the brain, or in more or less localized or focal lesions. The lesions may be in continuous progressive development, slow or rapid; or they may be static, the result of a single traumatic event. They may be chronic, of long standing, or acute and recently acquired. Lesions, of course, affect different hemispheres and different lobes of the brain, and the evidence presented in Part II showed how these variables could be related to different resulting deficits. Also take into account the patient’s age, the premorbid level of achievements and abilities, and the current level of motivation, and it is easy to see what a dramatic range of behavioral disturbances, in both variety and severity, may follow damage to the brain.
Much time and effort were devoted during the decade following World War II to trying to establish some single factor that would characterize “brain damage”; it was hoped that by measuring such a factor it would be possible readily to determine the presence or absence of brain damage. The result of this research was, in essence, that no single factor could be identified. The only single aspect of psychological function that has any claim to be generally affected by any cerebral pathology is psychomotor speed. This is the speed with which simple tasks, usually requiring some active motor participation in response, can be carried out. It involves both speed of thinking and speed of motor execution. This might be useful were it not that this factor is equally affected by most functional psychiatric states, ruling out its application to the detection of brain damage.
The problem of screening for the presence of brain damage is nonetheless a practical clinical issue, and a number of single tests are employed to test for the presence of brain damage. Heaton, Baade, and Johnson (1978) have usefully reviewed studies of the performance of these tests in making what is often the most difficult discrimination: between organic cerebral dysfunction and psychiatric states. They found that if chronic and process schizophrenics (those with long-lasting and severe psychotic illnesses) were included, then the tests performed very poorly. If, however, these patients were excluded, then the median success rate for determining brain damage was about 75%. The tests, in descending rank of success, included the Bender Gestalt Test (success enhanced by use of the Background Interference Procedure), the Benton Visual Retention Test, the Trail Making Task, and the Memory for Designs Test. The ability of these tests to discriminate between braindamaged and psychiatric patients, excluding the chronic and process schizophrenics, was about equivalent to their ability to distinguish between brain-damaged and normal patients. It is a matter of opinion whether you regard these tests as successful and justified for use in clinical practice, but it is obvious that discrimination rates in the region of 75% leave very considerable room for improvement.
An additional factor must also be taken into account in assessing the utility of these and other assessment tests: the effect of base rates. There is an interesting statistical phenomenon (most lucidly explained in Vernon, 1964) that results in the ability of a test result to identify a certain characteristic being dependent on the base rate of that characteristic, the base rate being the proportion of individuals in the tested population who possess the characteristic. Discrimination will be best when the characteristic occurs in 50% of the population, and will decline dramatically as the base rate falls. In research studies that, as part of their design, employ equal-sized groups of those with and without brain damage, the success of a test in picking out those with brain damage will be considerably greater than when trying to screen for brain damage in a population where such damage might be expected to occur in, perhaps, only 20% of cases. Care should be taken that the base rates of the research studies match those of the target clinical populations.
It is also the case that when older people are to be tested, and determination of cerebral dementia in older adults is a difficult and yet clinically important problem, the tests that are often employed may perform even less well. The Newcastle-upon-Tyne study (Savage, Britton, Bolton, & Hall, 1973) found that indices derived from the Wechsler Adult Intelligence Scale (WAIS) tended to identify only a third to a half of the patients with organic conditions, at the same time misclassifying up to a third of normal subjects or psychiatric patients as organic. The Allen Index, which takes the difference between the summed scaled scores for the Information and Comprehension subtests and for the Digit Span and
Digit Symbol subtests, correctly identified 52% of the patients with organic problems but went on to declare 48% of the psychiatric patients and 62% of the normal subjects as brain damaged! Two more specialized tests, a paired associate learning test and the Modified New Word Learning Test, performed rather better, but not at a satisfactory level. It is sometimes pointed out that in screening, false positives are less serious than false negatives because they will result in further investigation, following which pathology will be ruled out, rather than resulting in pathology being missed. This is to some extent true, but it depends somewhat on the context. In the case of the aged, where organic and functional illnesses are most difficult to distinguish, it might well mean that a patient failed to receive treatment for a remediable functional psychiatric illness on the assumption that the problem was a progressive untreatable cerebral dementia.
Battery tests, which we shall shortly discuss, have also been used to detect generalized brain damage. They have had rather better success than the single tests, the Halstead-Reitan Neuropsychological Test Battery showing a general level of discrimination of about 80% (Boll, 1981), and the Luria-Nebraska Neuropsychological Battery a level of about 86%, for the presence of brain damage. This is to some extent because they sample a broad range of psychological functions, and therefore have a greater opportunity to pick up whatever signs of dysfunction are present. The improved success comes at the expense of very lengthy testing, which makes these procedures impractical as a general screening device for the presence of brain damage.
Before moving on to discuss the examination of specific functions, some of the difficulties that underlie the assessment of behavioral deficits should be pointed out. Two logical approaches can be taken to testing: direct and indirect. In direct testing, the patient’s performance is measured premorbidly and then again following the onset of the lesion. This is the ideal state of affairs, but it is rarely possible. Occasionally, particularly in the case of servicemen, there will be recent formal assessments of intelligence and cognitive abilities that form an accurate baseline against which to measure subsequent decline. However, when this information is not available, the only direct approach is by repeated testing, perhaps weekly or monthly. Leaving aside the problems of teasing out the effects of familiarity with the test instrument, repeated testing may enable the development of the functional state to be charted, but it will not give any information by which to assess the extent of decline from premorbid abilities.
As the direct approach is often not feasible, an indirect approach must be adopted. One version of this is to find some index by which the patient’s premorbid level of functioning can be inferred. At one time it was thought that vocabulary ability was not affected by cerebral damage. This idea was based upon a further notion that vocabulary ability remains unaffected when other abilities decline during normal aging (which is, incidentally, not true when stated this simply). It was argued that if brain dysfunction was like normal aging, then vocabulary would remain stable and could be used as a pointer to premorbid intelligence. It is not surprising that this method has not been found to be valid. Vocabulary is, of course, affected by some cerebral lesions, and the way in which it is affected depends on the previous levels of ability and education; even hospitalization, in itself, is known to produce a decline in vocabulary ability. This approach has therefore fallen out of favor. However, one more recent development, the New Adult Reading Test (NART; Nelson & O’Connell, 1978), which is a simple test of the patient’s ability to read a list of words, has been shown to predict with reasonable accuracy the premorbid intelligence level of adults with dementia, and is also widely used to infer premorbid ability following all kinds of trauma. The list employed is composed of irregular words–those that cannot be read by the standard grapheme to phoneme conversion rules– and so the pronunciation has had to be learned for these words, also sometimes called “exception” words. An example would be salmon; the rules predict that this should be pronounced “sal-mon” but, since it is an irregular word, this is not the correct way to pronounce it. You just have to know how to say it. The NART is now in a second edition and has been joined by the Wechsler Test of Adult Reading (WTAR), which is constructed on exactly the same principle.
The only other approach that remains to the neuropsychologist, and another indirect one, is the use of inferences from the patient’s history. It is possible to piece together from the patient’s educational attainments, occupational choice and success, and pastimes and interests, a fairly accurate picture of the level and pattern of the patient’s cognitive abilities. It will be easier in some cases than others, particularly if examinations have recently been passed, or particular occupations successfully taken up. The prudent clinician will generally check an estimate derived from one of the tests of adult word reading against this kind of estimate; it is not uncommon to see individuals who simply were more interested in sports, or motorcycles, than in their academic subjects, and these individuals may attain relatively poorly on the formal test while having achieved more successfully in their subsequent occupational life in an activity for which they are motivated, often based on practical rather than academic skills. However, these data are all that the clinician may have to work with; this is yet another challenge to the neuropsychologist’s professional skill in creating an accurate and valid assessment from the information that is available or may be collected, both by direct observation and by indirect inference.
THE ASSESSMENT OF SPECIFIC FUNCTIONS
It was noted in Chapter 1 that there are three distinct traditions in the approach to neuropsychological assessment of specific functions. These traditions are behavioral neurology; the use of neuropsychological batteries; and the individual-centered normative approach. They can loosely be said to typify the historical traditions of clinical neuropsychology in the Soviet Union, the United States, and Britain respectively.
This approach derives from the influential work of the Russian neuropsychologist A. R. Luria and is individual centered and clinical in nature. The goal of neuropsychological assessment is not a quantitative measurement of patients’ difficulties, but a qualitative analysis and description of their problems. Rather than employing psychometric procedures to identify abnormal performance by statistical means, with reference to a normal population, the emphasis is on behaviors that any normal individual of the age, background, and general ability of the patient should be able to perform. When such behaviors cannot be generated, then a deficit has been demonstrated. Particular attention is paid to the qualitative aspects of how a task is performed, instead of merely to the absolute level of performance that is observed.
The neuropsychological examination has four essential aspects. First, the psychologist begins from his or her knowledge of the different types of dysfunction that follow cerebral lesions, that is, from a model of the organization of the brain. Second, in order to locate the areas of dysfunction to be investigated in depth, the initial stages of examination explore in a preliminary fashion the optic, auditory, and kinesthetic processes together with motor behavior. Third, in analyzing the behavioral deficits observed, a distinction is maintained between deficits that follow from a primary failure in the system under investigation and those that are a secondary result of some more general fundamental failure. Last, the examination must include tests of complex integrated activity, such as speech, reading, writing, comprehension, and problem solving. The aim is a careful qualitative analysis of the entire range of patients’ activities, of the difficulties that they experience, and the mistakes they make.
The investigation therefore begins with a general evaluation of the basic “individual analyzers” (optic, auditory, and so on) of Luria’s model, considered in terms of the levels of direct sensory reaction, mnestic organization, and complex mediated operations (Luria’s terms). The tests available are short and are selected for their appropriateness to the patient. There is no rigid pattern of administration or scoring. The investigation then moves into a second phase designed to investigate in detail the problems of the individual patient, a period of individualized qualitative exploration.
While maintaining a high regard for the insightful and persuasive nature of much of Luria’s model, and an admiration for Luria’s clinical skills, most Western psychologists have found this approach difficult to apply. However, considerable assistance to those attempting to employ the approach was given by Christensen’s publication of a systematized version of the tests used in Luria’s clinic (Christensen, 1974). While covering only the initial stages of the investigation, it provides a source of materials and procedures for use in assessment. The text is divided into main sections on topics including “Acousticomotor Organization,” “Higher Cutaneous and Kinesthetic Functions,” “Impressive Speech,” and so on. Each of these is divided into subsections for which a number of simple test procedures are supplied. The Investigation of Acousticomotor Organization is broken down into the investigation of perception and reproduction of pitch relationships, and the perception and reproduction of rhythmic structures. The first of these is further subdivided into the perception of pitch and the reproduction of pitch relationships and musical melodies, the second into perception and evaluation of acoustic signals and motor performances of rhythmic groups. For each of about three or four dysfunctions that might be observed under each of these subdivisions, a lesion site is suggested. For example, in motor performance of rhythmic groups, using the four little tests of rhythmic reproduction suggested, if the patient does not fully apprehend the incorrectness of his or her performance, a right temporal lesion is indicated, but a lesion of the frontotemporal division of the cortex is indicated if the patient exhibits marked incoordination in all tests. This version does not reflect fully the philosophy behind Luria’s work but it conveys something of the character and flexibility of the approach. For a good description of the application of Luria’s approach, the paper by Luria and Majovski (1977) is a useful introduction to the fuller account in Luria (1980).
The advantages of Luria’s method are first that it is based upon an explicit theoretical foundation, his model of cerebral organization, although this model (as noted on p. 59, for example) has not always been supported by empirical evidence. Second, it emphasizes the qualitative aspects of performance and is flexible in approaching the diagnosis of functional deficits. Particularly in view of our imperfect knowledge of cerebral organization and function, this latter aspect might be expected to result in greater accuracy and finer resolution in the final description of a patient’s difficulties.
The disadvantage is that the system depends almost entirely upon the clinical acumen and skill of the neuropsychologist. In the hands of Luria, the results were impressive, as the published case reports demonstrate. However, the approach demands a comprehensive grasp of the details of Luria’s complex theoretical model. There are no rigorous investigations of the validity of the procedures in routine application, and the result, in the West, has been some suspicion of the “clinicalanalytical” approach, and a reluctance to apply it in regular practice.
The strict psychometric approach, in complete contrast with behavioral neurology, has been more popular in the United States than elsewhere. While it would be false to give the impression that this is the only approach adopted there, it is the only region where the employment of neuropsychological batteries has been taken up with any enthusiasm and, if anything, the trend is toward increasing popularity.
Historically, there have been a number of batteries that have been widely used in clinical practice. However, two are currently of major importance: the Halstead–Reitan and the Luria–Nebraska.
The Halstead–Reitan battery has been gradually developed by Reitan and co-investigators over a number of years from tests originally selected, on a largely pragmatic basis, by Halstead in the 1940s. The exact tests that make up the battery vary a little from clinic to clinic, but the tests seek to cover as comprehensive a range of behavioral functions as is possible. The Halstead Category Test (see Figure 3.1) is a concept attainment task involving learning, memory, mental efficiency, and adaptability. The Tactual Performance Test is a formboard test with repeated presentations in which improvement in speed can be observed, culminating in a blindfold trial in which the patient must remember the location of the holes that take each shape. The Rhythm Test is drawn from the Seashore Tests of Musical Ability (see p. 76), and the Speech Sounds Perception Test involves multiple-choice identification of auditorily presented nonsense words. The Finger Oscillation Test examines simple motor speed. Most versions of the battery include all the above tests.
Also employed are the “allied procedures,” comprising the Trail Making Test, in which numbered or lettered circles are to be joined in sequence, or by alternating letters and numbers, by a pencil line; the Strength of Grip Test; and the Sensory–Perceptual Examination, a test of tactile, visual, and auditory modalities. To these are added three tests of tactile perception and a modified form of the Halstead–Wepman Aphasia Screening Test. The full WAIS is often also given.
It will be appreciated that this battery takes a long time to administer. Even with uninterrupted testing a period of 6 to 8 hours is required, and with a distractible or easily fatigued patient testing may have to be undertaken over many days. Nevertheless, the battery has been widely used, and a great deal is known about its reliability, validity, and psychometric structure. The battery, in common with most procedures, has some difficulty in identifying chronic schizophrenics as psychiatric patients, but otherwise the results are quite impressive. Reitan, working from the test protocols alone, without any additional information about the patient, is able to identify the presence of brain damage, distinguish diffuse from focal lesions, and lateralize and localize focal lesions, with an accuracy of around 80%. This may to some extent reflect Reitan’s clinical skills, but even relatively untrained personnel can show remarkable diagnostic success using the battery. There has been extensive investigation of the performance of the battery when applied to criterion groups of brain-damaged and psychiatric patients and normal subjects in a long series of studies for validation and cross-validation (in which diagnostic criteria established in one research sample are tested out upon an independent sample).
One particularly interesting development has been the creation of a computer-supported automated diagnostic process that operates by a number of classificatory keys from the set of 41 scores yielded by the full battery (Russell, Neuringer, & Goldstein, 1970). This program had an accuracy rate of 88% for detecting brain damage, 80% for lateralizing left and right hemisphere lesions, and 62% for identifying diffuse lesions. This compares well with traditional methods, and the computer agreed with clinical neurologists and neuropsychologists as often as they agreed with each other.
In favor of this battery, then, is the extensive evaluation of its validity that has been published. It covers a broad range of functions and has the ability to identify complex test patterns and relationships across a number of tests, which may typify certain specific conditions. Nevertheless, the test carries a heavy penalty in the time taken in administration, and it does not cover all possible neurological deficits. Its heavy reliance on motor performance also rules out its use with certain patients. It has, however, proved a useful instrument, and may form the basis of further improved versions.
The second, and much more recently developed, battery is that generally known as the Luria–Nebraska Neuropsychological Battery, more formally referred to as the Standardized Version of Luria’s Neuropsychological Techniques. This is just what it is, an attempt to take Luria’s procedures and assemble a battery of normatively based tests from them.
The battery is formed of 269 items that cover the main components of the Luria Investigation as described, for example, by Christensen (see p. 324). These components comprise motor function; rhythmic and pitch skills; tactile functions; visual (spatial) functions; receptive speech; expressive speech; writing; reading; arithmetic; memory; and intellectual processes. Each item is scored according to formal protocols. Summary Pathognomic, Left Hemisphere, and Right Hemisphere scores are calculated, and a profile across the individual scales may be drawn.
There is much less research available on the Luria–Nebraska battery than the Halstead–Reitan battery. Nevertheless, reports to date are extremely encouraging and suggest that it is at least as powerful in identifying both the presence and the type of brain damage. A recently published cross-validation study (Golden, Osmon, Moses, & Berg, 1981) found that the presence of brain damage was correctly diagnosed in 86% of cases, the lateralization was correct in 78%, localization in 92%, and the quadrant of the brain in which the lesion was situated was correctly identified in 84% of cases.
This battery does have the advantage that it takes considerably less time to administer than the Halstead–Reitan, at around 2½ hours, and it relates more directly to the patient’s problems on a clinical level. However, it requires that the user be aware of Luria’s formal model, and the greatest clinical value will be derived from it only when qualitative as well as quantitative aspects of performance are taken into account. The authors of the test (principally Charles Golden) specifically state that the test depends upon both the standardized and qualitative interpretive systems that the battery can provide. Nevertheless, this battery has been taken up with some enthusiasm, particularly in North America. (Several books that clinical case examples of the use of both these batteries are listed at the end of the chapter.)
The Individual-Centered Normative Approach
This approach, typical of clinical practice in Britain, stands between the two approaches already described. It relies to some extent on a formal psychometric approach, but it emphasizes the need to tailor the assessment to the nature of a particular patient’s difficulties. British neuropsychologists have generally considered batteries to be inefficient, wasteful of time and resources, and unlikely to provide a full and accurate description of the dysfunction under investigation. The aim of formulating such a description, which goes beyond a simple diagnostic classification to an understanding of the behavioral deficits in psychological terms, has always been kept firmly in view.
The strategy of assessment has therefore been, as in the Russian approach, to conduct a broad general survey of the patient’s functions, identifying areas of difficulty and pursuing an analysis of these particular areas. Each investigation thus becomes an individual examination of the possibly unique state that the patient exhibits. Unlike the Russian approach, however, the particular tests employed are rooted in the empirical tradition. Performance is carefully analyzed and scored, and reference made, whenever possible, to established norms against which to judge the patient’s performance. Where individual experiments are set up to test some hypothesis about the patient’s cognitive state, attention is paid to the scientific design of the procedures, and there will be a formal statistical evaluation of the results.
The procedure of neuropsychological consultation, after collection of informal data about the patient’s complaint, and a preliminary interview with the patient, might often begin with the WAIS (Kaufman & Lichtenberger, 1999; Tulsky et al., 2003). Although never designed with such an application in mind, the quality of this test and the range of functions that it examines make it the most popular instrument for detailed general evaluation of intellectual functions. The degree to which a neuropsychologist may be willing to infer deficits from the pattern of results on the WAIS varies widely, and few are likely to consider it a sufficient assessment, although a great deal can be inferred by a skilled clinician from this test alone.
Having identified the general areas of difficulty, and undertaken tests of primary sensory functions if appropriate, the neuropsychologist will have begun to formulate hypotheses about the nature of the deficit, and perhaps the lesion site. The next stage is to employ specific individual testing procedures to evaluate these hypotheses. The tests are of the type described in Chapters 3 to 6 when discussing the effects of specific lesions, and include the assessment of language function, which was discussed on pp 150–151. Because these tests have been shown to be affected by lesions of certain areas, then as a corollary the tests can be used to locate the sites of unidentified lesions. There is an enormous number of such tests available. Lezak (Lezak, Howieson, Loring, & Hannay, 2004) in her valuable and comprehensive guide to neuropsychological assessment lists about 300, but most clinicians will routinely use perhaps a couple of dozen, calling on less popular tests when an unusual or baffling case is encountered.
For example, if a frontal lesion is suspected, there are three areas of function that are likely to be examined. Concept attainment and abstraction may be tested with the Wisconsin Card Sorting Test (Figure 3.3), or the Halstead Category Test (Figure 3.1). Planning and the integrated execution of complex actions may be investigated with the Porteus mazes (Figure 3.5), the Trail Making Test; or the tests contained within the Behavioural Assessment of the Dysexecutive Syndrome (BADS). Finally, the Verbal Fluency Test might be used to check on “frontal” aspects of verbal function. There will, however, be wide variation in the practice of individual neuropsychologists (Rabbitt, 1997).
Because these tests are insufficient to provide a comprehensive coverage of behavioral functions, the gaps will be filled by the development of special experiments to investigate individual hypotheses about the deficits under examination. If a patient’s poor motor performance is thought to be due to poor kinesthetic feedback, then artificial feedback could be introduced under experimental conditions to see if performance improves. If the problem is thought to be due to a deficit of verbal memory, then additional cues could be used to determine the level of the patient’s performance. As another example, a patient’s difficulty in performing mental arithmetic could be examined by presenting problems in different forms, verbally encoded or formally abstract, in written or mental form, in order to tease out whether the difficulty is one of abstracting the problem, of employing an appropriate strategy for solution, or of undertaking the mechanics of computation. The possibilities are limited only by the ingenuity of the neuropsychologist and the sophistication of his or her model of cognitive processes.
It must be remembered that the investigation aims not only to classify the lesion, but also to understand the patient’s difficulty in psychological terms. This form of cognitive analysis (Shallice, 1988) attempts to specify the dysfunction in terms of the rate and accuracy of information processing, and the operation of the normal repertoire of cognitive strategies. The component processes are identified and measures of the efficiency of each employed. This psychological analysis can in turn be related to areas of localized cerebral function, but it can also form a basis for the management, treatment, and rehabilitation of the patient. This would be lacking from a simple diagnosis of the lesion and its site. An excellent demonstration of how the cognitive psychologist’s knowledge of reading processes can help in assessing patients, and of how data from the clinic can be fed back to develop our understanding of normal function, has been provided by Patterson (1981).
This approach relies heavily on the skills and insight of the individual clinician. There is the risk that areas of function will be neglected, or that complex patterns of functional interaction will be missed. Nevertheless it seems the most intelligent approach to neuropsychological assessment, if practiced by a capable and sensitive clinician.
ASSESSMENT IN PRACTICE
All this may seem rather bewildering; there are various approaches and a myriad of tests that might be employed. While the choice of approach and specific procedures is generally dictated by the clinical question of what is to be discovered about the patient’s state, and for what purpose, it may assist in clarifying matters if I describe my own practice in con ducting examinations for medicolegal purposes. Here the mission is fairly broad, to perform a psychological analysis of any deficit from which the individual is suffering, normally as a consequence of a road traffic accident or an occupational accident, and to analyze the impact of any such deficit in everyday life, now and in the future. I conduct these examinations about twice each week, and the reasonable period for which it is possible to assess an individual is limited to about 3 hours. I should very clearly emphasize that this is just an example and that while many of my colleagues perform slightly different examinations, I do not claim that mine is superior or that there are not other equally valid methods for examining clients. Nevertheless, this basic assessment has evolved over many years, and some considerable thought has gone into it, partly because every so often I have to stand up in court as an expert witness and defend it!
I should also make clear that while this is the basic form of the assessment it is regularly varied to take account of specific issues and circumstances. The client may be so severely affected cognitively, or by visual or motor handicaps, that these tests are inappropriate. The client’s first language may not be English, which creates specific problems for assessment. It is rare that this set of tests is administered as described without some adaptation; but I do have a notional standard form of the assessment that illustrates the principles adopted in conducting it.
Before seeing a client it will have been usual to review the client’s full medical history, and to have read accounts of the traumatic incident and seen medical records relating to hospitalization following the accident. There may also be school records and occupational records, and expert opinions already prepared by other neuropsychologists and experts in related fields such as neurology, neurosurgery, orthopedics, psychiatry and rehabilitation medicine. So, one is reasonably well informed about the history before the client arrives. Even so, after trying to put the client at ease, it is important to spend at least a further half hour obtaining the client’s own view of his or her problems and clarifying the history. Actually observing clients while they describe their difficulties and their experience provides important data, as does observation of their performance on the formal tests. That is why in Britain, unlike in North America, it is considered important that the neuropsychologist actually administer all the tests, rather than have a technician conduct the testing and provide the test scores. While the formal psychometric data are the core of the assessment, interpretation and analysis are also informed by the qualitative clinical observations made of the client.
The formal assessment generally begins with one of the adult word reading tests, and I happen to prefer the WTAR, despite certain limitations. The results of the WTAR will be combined with information about the client’s educational and occupational history to infer an estimate of his or her likely general cognitive ability before the accident occurred. This will then be compared (formally and statistically) with the assessment of the current level of general cognitive ability obtained from the third edition of the Wechsler Adult Intelligence Scale (WAISIII). This battery provides not only overall IQ indices for the major functional areas, but also factor-based index scores that may indicate a significant slowing in mental speed, or a deficit in working memory reflecting problems with attention and concentration, for example, even in the presence of intact “core” verbal and visuospatial abilities. Highly specific deficits may be apparent on particular subtests.
After this battery, which may take up to an hour in a slightly modified form, the client usually needs a break. A cup of tea or coffee helps to dissipate fatigue, and I often enjoy the opportunity simply to chat with clients in an informal way. It is also surprising and relevant to note how once clients are engaged in conversation on a topic that is of interest to them, their mood may lighten and their psychological function improve.
Next comes a battery of memory tests and my preference is for the Wechsler Memory Scale, third edition (WMS-III), partly for its inherent psychometric quality, and partly because formal statistical comparisons can be made with the results of the WTAR and the WAIS-III to determine whether current memory function is impaired with respect to estimated premorbid ability, or with respect to the current level of general cognitive ability. However, certain other batteries are in widespread use, and I commonly use the Camden Memory Tests or the Doors and People Test as additional or alternative tests of recognition memory. A scale of verbal learning may also be specifically employed, or the Rivermead Behavioural Memory Test if the standard tests prove too demanding.
This survey of general cognitive abilities and a more detailed exploration of memory provide a broad assessment of cognitive functions but with one important omission: tests of frontal executive functions. It was noted in Chapter 3 how severe and handicapping problems of executive function can occur in the absence of any abnormality apparent on other tests. Some specific tests of this area of function are therefore next included. My favorites are the Cognitive Estimates Test; a test of verbal fluency; the Trail Making Test; and the Brixton Test (which may be considered a briefer and more palatable alternative to the Wisconsin Card Sorting Test). Elements of the BADS might also be administered. Specific tests of other areas of function such as word naming or visual object and space perception might also be included here.
My 3 hours are starting to run out by now, but time remains to conduct a brief survey of the client’s psychological state through the Beck Anxiety and Depression Scales and, in cases where the interview has suggested that it might be important, a scale to assess symptoms of posttraumatic stress disorder. I have always considered it appropriate and courteous to provide some information to clients about the outcome of the assessment and potential treatment and rehabilitation options, and to answer any questions the client may have. The client leaves, and it just remains to score the tests, think hard about the implications of the findings, and prepare a lengthy and detailed report!
It is difficult to overemphasize the dramatic change that has occurred in neuropsychology with respect to rehabilitation. Only 25 years ago when preparing the first edition of this book, I wrote: “I wish that it were possible to devote a full chapter to treatment and rehabilitation, but unfortunately this would not be justified. While there are encouraging developments, neuropsychologists have not traditionally been heavily involved in therapy for brain-damaged patients.” How things have changed, and for the better.
Clinical neuropsychologists now work in a variety of environments, and it is probably a minority who work in an acute setting in direct contact with neurologists and neurosurgeons, commonly in Britain in a regional neurosciences unit. A majority of specialist clinical neuropsychologists now work in a rehabilitation context, either within a brain injury rehabilitation unit that is likely to be, at least in part, residential or, increasingly, attached to a community rehabilitation team providing services in the client’s own home. Many of the managers or clinical leaders of these units and teams are neuropsychologists. Nevertheless, rehabilitation teams are strongly multidisciplinary and depend upon active collaboration among specialists in neurorehabilitation aspects of neurology, neuropsychiatry, nursing, occupational therapy, physiotherapy, rehabilitation medicine, social work, and speech and language therapy.
One of the difficulties in developing neurological rehabilitation has been to formulate a model of how function is reinstated, when that does occur. As there is no significant regrowth of damaged tissue in the central nervous system (although neuroscience may yet provide a solution to this critical problem), what process results in lost functions being regained? There are at least four ways in which this might happen. First, recovery may result from the reinstatement of function at its original site, in tissue that has not been irreversibly damaged but only temporarily upset by processes associated with the acute effects of the lesion. Second, recovery may result from the adoption of the function of the damaged area of cortex by an area that did not originally serve it. Some have considered that this process of neural relocation might be fundamentally programmed into the system. Third, there may simply be adaptation to the deficit. As most high level performance can be achieved by the mobilization of a variety of component processes, it may be possible to learn to perform the old skills by new processes that avoid the site of dysfunction in the system. Finally, some have thought that recovery reflects the original tissue regaining control of functions that have been temporarily taken away by other sites during the acute phase of the illness (Andrewes, 2001; see also Chapter 10). Without an accepted model of how recovery is achieved, it is difficult to plan a rational program of therapy that will be both effective and efficient.
However, in practice, there are two approaches to reestablishing function: relearning an ability through training and practice or adapting to the deficit by learning alternative methods or strategies to achieve the desired function. The latter can be thought of as similar to the behavior of traffic when a road in a city is closed. Vehicles simply find a diversion around the obstacle (the lesion in the road network) and are nevertheless able to reach their goal (achieve functional performance).
The progress of recovery, unaided or with the assistance of rehabilitation, can last over a very long period, and improvement continuing over at least 6 years has been recorded in some studies. Nevertheless, it is generally accepted that the most important period of recovery, if it is going to occur, is during the first year following injury, and especially during the first 6 months, with a gradual plateau developing during the second year, at the end of which natural recovery is essentially complete. There have, however, been few formal studies of the course of recovery. The exceptions have been long-term studies of the war injured over periods of up to 30 years (Newcombe, 1969; Teuber, 1975), which show certain deficits persisting throughout that period, and recent studies of the psychosocial effects of closed head injury (Ponsford, 1995; Richardson, 2000). These studies have shown great variation in the social functioning regained by the patients, related in part to the severity of the injury, but also to the patient’s premorbid personality and intelligence.
However, there are a number of forms of active therapeutic intervention by neuropsychologists. Aphasia therapy, shared with speech and language therapists, has already been discussed (see p. 151) and forms an important aspect of this work. In more general terms, training has been the dominant strategy employed. This has sometimes been within the context of a belief that practice of a skill might actually play some part in the direct physiological reinstatement of that function, and it is sometimes known as “brain function therapy” (Buffery, 1977; Powell, 1981). More often, the approach has been simply pragmatic, in that it is accepted that the patient cannot perform a certain activity, and psychological knowledge of learning is applied to teach the patient to perform it.
The kind of contribution that the psychologist can bring to the design of training programs is in performing task analyses and constructing appropriate training routines (Golden, 1981). This is where the cognitive analysis carried out during assessment can be of great value. The components of the dysfunctional skill must be recognized if the training is to be accurately directed at the dysfunctional processes. Once the components have been identified, they can be practiced within tasks in which the level of difficulty can be varied, and the errors committed by the patient controlled. Error-free learning has been shown to be particularly effective in neurorehabilitation (Wilson, Baddeley, Evans, & Shiel, 1994) and the principle is now widely employed. Emphasis should, in any event, be placed upon careful monitoring of the patient’s progress, and on providing immediate, direct, and clear feedback about actual performance. Biofeedback (for example, of neuromuscular activity) and other forms of concurrent enhanced monitoring of performance have sometimes been used to aid patients in maintaining an accurate internal representation of their performance and its effects.
Where direct learning approaches are inappropriate or ineffective, it may be productive to introduce new cognitive strategies to the patient. For example, the patient with a verbal memory deficit may be taught to use explicit visual-imagery associations and nonverbal mnemonics to improve memory performance. Alternatively, patients may be encouraged to deepen their encoding of stimuli by thinking directly about associations with the material, constructing appropriate cues, and practicing regular rehearsal, so that words are less likely to be forgotten and may be retrieved when recall is attempted. If there is currently a dominant approach to neuropsychological rehabilitation, it is cognitive rehabilitation, in which attempts to intervene in the operation of cognitive processes are seen as facilitating the reacquisition of function (Riddoch & Humphreys, 1994; Stuss, Winocur, & Robertson, 1999).
Specific behavioral methods have also been explored, using operant, or occasionally classical, conditioning techniques, although they are currently less widely employed than cognitive approaches. Existing behavior may be shaped or new behavior established by the use of explicit reinforcement strategies, the patient being rewarded for the appropriate performance of some target behavior. Inappropriate behavior can similarly be reduced by the withdrawal of rewards, or by punishment regimes, although punishment is now regarded as singularly ineffective and in most cases ethically unacceptable.
An example of an elegant and purely behavioral approach is provided by Wood (1987). A patient regularly sat with head bowed, despite having the muscular ability and motor control to maintain a more normal posture. In this position he was understimulated by events occurring around him and unlikely to engage in social interaction. The patient was provided with a personal music player, and his favorite music, but the player was also equipped with a mercury tilt-switch so that the music only played when the patient lifted his head, stopping when his head dropped. This simple intervention resulted in a clear increase in the time the patient held his head in a more appropriate position and, after a period, not only when the player was on his head.
Recent developments in behavioral approaches, which may be particularly important when the patient is so impaired cognitively that benefit cannot be obtained from cognitive approaches, have centered around positive programming. Rather than being concerned with undesirable behaviors and the need to direct rehabilitation at reducing them, the emphasis is on increasing the frequency of desirable behaviors. It is generally possible to be more successful in increasing desirable behaviors than in removing undesirable ones, and an increase in positive adaptive behaviors, especially if they are incompatible with those that are undesired, simply results in their replacing the problem behaviors (Rothwell, LaVigna, & Willis, 1999).
Reality orientation therapy should also be mentioned as it is so widely employed with confused patients, particularly confused older people (Miller, 1977). This form of milieu therapy has been successfully used with elderly patients with dementia and is designed to provide artificially the basic information that the patient requires in order to conduct sensible interaction and generate appropriate behavior. A simple example is to greet Mrs. Brown, wandering about the ward at 2 a.m., with “Hello Mrs. Brown, you are in Barchester Hospital, and it’s two o’clock in the morning. You ought to be in bed, asleep. Can I help you, dear?” rather than simply “What’s the matter, dear?”, which does not assist the patient to maintain appropriate orientation for time and place, and behave accordingly. The rationale for this form of therapy is clear.
In general, there has been a fruitful and increasing emphasis on holistic approaches to rehabilitation (Ben-Yishay, 1996; Prigatano, 1999; Sohlberg & Mateer, 2001). Here, the idea is not to concentrate upon specific handicaps, but to consider the whole individual in his or her personal, social, and possibly occupational, context. Rehabilitation may not be an attempt to restore the individuals to their previous condition, as this may be inappropriate and unrealistic for a variety of reasons, but to work toward the best adaptation that can be achieved for patients in their new personal and social context. There is an interaction between the condition of the patient and the environment in which he or she lives, and rehabilitation will be about adapting the environment appropriately, as much as it is about trying the “fix” the patient’s problems.
The efforts of the neuropsychologist are of course only one element in the operation of a clinical team. In an acute context, besides the neuropsychologist, and the neurologist or neurosurgeon who heads a team of medical staff, there are radiographers, electroencephalographers, pathologists and biomedical laboratory staff, medical physicists, speech and occupational therapists, medical social workers, and of course the nursing staff. A neuropsychiatrist may well also bring special skills into play. The diagnostic information and the recommendations for management and treatment that the neuropsychologist makes will therefore be evaluated in the light of findings and opinions contributed by other members of the team. The level of accuracy of neuropsychological assessment may seem disappointingly low, and there is undoubtedly considerable room for improvement in the validity of assessment procedures, but taken in the context of other medical and physical investigations, it can contribute crucial information to the understanding of the patient’s condition.
A variety of general approaches and specific procedures have been used in determining the presence of brain damage and in assessing its nature and location. Many of these methods show a moderate degree of success, and are of undoubted clinical value, but at present none seems outstanding in accuracy or efficiency. However, with a shift in emphasis to the psychological analysis and functional description of a patient’s neuropsychological problems, neuropsychological assessment has found an important new role in the understanding of handicap, the planning of rehabilitation and management, and the future adaptation of those with injury or disease. In this role neuropsychological assessment has proved more successful.
In a dramatic change over the past 25 years, the majority of clinical neuropsychologists now work in a context in which rehabilitation is the primary goal. In a team, not uncommonly led by a neuropsychologist, the neurorehabilitation disciplines together provide intensive rehabilitation to assist those with neurological injury or disease to minimize their handicaps, and to optimize their adaptation to the physical and psychological problems of everyday living in a context of good psychological health.
Neuropsychiatry and Neuropsychology
Early in my career, I submitted an academic paper to a highly respected psychiatric journal on the subject of hemisphere laterality and schizophrenia. The anonymous, but presumably eminent, reviewer rejected the paper not on the merit of its contents but because it was considered that to look for abnormal brain organization in schizophrenics was an outdated and futile exercise, of no interest to current psychiatry.
In A Pelican at Blandings, published shortly before in 1969, P. G. Wodehouse has Galahad Threepwood say: “That was the night you were so disturbed because she hummed and giggled, giving you the impression that something had gone wrong with the two hemispheres of her brain and the broad band of transversely running fibres known as the corpus callosum and that she was, in your crisp phrase, potty.”
Wodehouse seems to have been the more prophetic of the two in view of the dramatic increase in interest in neuropsychological variables in psychiatry over the past 30 years. Neurologists and psychiatrists have always been interested in mental states that result from cerebral pathology, the area known as neuropsychiatry or organic psychiatry, but this interest has now been extended to disorders that would not traditionally fall within this field, such as schizophrenia and affective psychosis. Neuropsychiatry has now grown into an independent medical specialty, standing astride psychiatry and clinical neuropsychology.
It has been clear for a very long time that abnormal mental states can follow from damage to the brain. Some of the changes associated with focal lesions that were discussed in Part II are obviously of a psychiatric nature, and there are in addition changes that follow from diffuse brain diseases, and from generalized trauma associated with blows to the head.
It should, however, be remembered that not all neurological conditions involving the brain produce psychiatric symptoms. In a large study of more than 3,500 war veterans with brain injuries, about a third of whom had penetrating wounds, lasting cognitive impairment was present in only 2%, character changes in 18%, psychosis in 8%, and severe neurosis in 11% (Hillbom, 1960). The incidence of mental changes following the development of tumors is rather higher, in the range of approximately 50% to 80% among patients in various studies, but on the other hand the incidence of cognitive and psychiatric sequelae after closed head injuries producing concussion and temporary amnesia is remarkably low. Such cases do nevertheless occur.
The variety of psychiatric conditions associated with organic states is a special area of study in its own right, particular patterns of psychiatric impairment being associated with, for example, the various cerebral diseases (see David, Fleminger, Kopelman, Lovestone, & Mellers, 2008). However, there are a number of general symptoms that are often seen to accompany organic nervous disease. Patients frequently show some disturbance of consciousness, varying from mild inattention to disorientation, delirium, unconsciousness, and coma. Sleep may also be disturbed, either in quantity, in pattern, or in the electrical rhythms recorded through the night. Impairment of memory is also common, and patients tire easily and tend to perseverate in their thoughts or their activity. They are often irritable and exhibit frequent changes in emotion, and their behavior may also be relatively disinhibited.
Some of these symptoms have been found to be associated in the postconcussional syndrome (sometimes called posttraumatic neurosis), which is characterized by severe headache, giddiness and blackouts, fatigue and difficulty in concentrating, loss of confidence, depression, anxiety, irritability, and intolerance of noise. Patients may also complain of insomnia and show a reasonable but particular concern for, and awareness of, their head.
Epilepsy might also be expected to be associated with psychiatric disorders. However, although epilepsy is a common aftereffect of cerebral trauma or disease, epileptics in general show relatively low rates of psychiatric disorder. Studies of the incidence of mental abnormality in epileptics commonly find that only about 5% are ever diagnosed as suffering from a serious psychiatric condition. This compares with a rate of about 1% for the general population. There does, however, seem to be a much clearer relationship between temporal lobe epilepsy and psychiatric disorder. This form is found disproportionately among epileptics admitted to mental health facilities.
The abnormal behavior can take the form of psychiatric symptoms following seizures, or less commonly of a chronic psychotic state rather similar to paranoid schizophrenia. A particular form of personality, typified by stubbornness, concreteness, lack of spontaneity, egocentricity, and unctuosity has also been described as typical of epileptics. However, these descriptions were based upon institutionalized epileptics, not those living in the general community (as almost all are), and it seems probable that any personality changes that do occur can be regarded as secondary reactions to being handicapped by epilepsy, rather than as primary effects mediated by changes in cerebral systems.
Cerebral dementia is, of course, invariably accompanied by psychiatric changes. Dementia, seen most commonly in older people but also in certain presenile degenerative diseases, involves the generalized and progressive loss of cortical tissue from the brain. Most dementias are, at present, irreversible. As the cortex is lost, mental functions decline progressively. There is a global deterioration of memory, thinking, motor performance, emotional responsiveness, and social behavior. As the illness develops, speech may be affected, behavior becomes increasingly unreasonable and often disruptive, until finally the control of basic and vital processes becomes disorganized. It is a distressing and pathetic progression familiar to many with aging relatives.
Finally, of course, it should be remembered that there are a range of specific psychiatric effects that can follow damage to the brain, and these will depend upon the site of the lesion, its speed of development, the nature of the pathology, and particular characteristics of the individual patient.
There are also certain rather bizarre neuropsychiatric conditions that may follow head injury. These conditions are generally described as rare, and they are certainly uncommon, but their apparent rarity is, in my experience, partly attributable to the reluctance of individuals with head injuries to report these problems; they are afraid of being regarded as “mad” and it often takes time to gain the trust of the patient before the symptoms are described. As these syndromes are striking, and have recently been of particular interest to neuropsychologists, four typical conditions will briefly be described: Capgras, Cotard, and Frégoli syndromes, and obsessive–compulsive disorder. A fuller account, with case studies, can be found in Halligan and Marshall (1996).
In Capgras syndrome the patient believes that imposters have replaced family members and friends. These imposters behave exactly like the people who they have replaced, and are essentially indistinguishable from them. While this is a rare feature of a rare condition, there are cases in which the patient has threatened, harmed, or even killed the person they believe to be an imposter, as a way of trying to unmask the perceived deception. While not all cases of Capgras syndrome are linked to neurological disease, up to half the cases are, and in particular abnormalities of the right hemisphere have been reported in these patients. Ellis and Young (1990) have provided a cognitive interpretation of this syndrome, and a more recent review can be found in Ellis and Lewis (2001). The syndrome can be surprisingly disabling. A teacher whom I saw after a head injury sustained in a road traffic accident experienced these delusions not only with respect to members of her family but also her colleagues at school, and was quite unable to sustain a return to work because of the distress which it caused to her.
Cotard syndrome appears rather similar to Capgras syndrome and, in my experience, the two can occur together. Cotard syndrome is often described as a syndrome of nihilism. Patients describe the feeling that they have died, and may also have the delusion that their bodies have been replaced by corpses. A complaint may be made of having lost possessions, status, or strength, or that certain internal organs are missing. Right hemisphere lesions are also associated with Cotard syndrome and, as with Capgras syndrome, there may be impairments of face processing. Some patients with Cotard syndrome may be clinically depressed, and a distinction is sometimes drawn between two forms of the disorder: one form linked to severe depressive illness, and one form based upon a delusional state.
Another form of delusional misperception occurs in Frégoli syndrome. Commonly, there is a paranoid element to this condition and patients identify an individual whom they believe to be persecuting them. This paranoid delusion is, however, extended to the idea that this person can disguise him- or herself to appear like other people with whom the patient comes into contact, usually for a sinister purpose. Despite the lack of any similarity between the (believed) persecutor and the person believed to be the persecutor-in-disguise, the patient believes them to be the same person. This syndrome can occur in association with degenerative conditions and also, transiently, in toxic states including cannabis use. A cognitive interpretation for Frégoli syndrome is also offered in Ellis and Young (1990).
Obsessive–compulsive disorder (OCD) is less rare following brain injury, and obsessional features that fall short of a full psychiatric diagnosis are not at all uncommon, especially after frontal brain injury. Often, the patient has a previous tendency to mild obsessionality that is exacerbated by the head injury. Excessive orderliness, a concern with cleanliness, a fixed routine, and excessive checking behavior are the common features. However, severe compulsions can sometimes occur, and I have seen patients who cannot sleep because they feel compelled to continually check that the door is locked or the gas stove turned off. Very rarely, the OCD may take a form that includes violent compulsive thoughts. Such thoughts are of too grotesque, macabre, and disturbing a nature to be described here, but one of mildest images described to me by a patient was that as we sat and talked he could not stop himself from imagining that he was removing all the skin from my arm with a cheese grater. These thoughts that intrude into everyday consciousness occur in individuals without previous violent tendencies or interests and are typically so distressing that they avoid all social interaction, either staying alone within their home, or moving to a remote and isolated location where they are unlikely to meet other people. For a neuropsychological account, see Tallis, Pratt, and Jamani (1999); for an overview of these and other similar disorders, see O’Carroll (2003). The neurological basis of delusions, more generally, is discussed by Gilleen and David (2005), and the role of the frontal lobes in Salloway, Malloy, and Duffy (2001).
This, very briefly, is the subject matter of organic psychiatry, and is composed of organic states or reaction types. In traditional psychiatry these have always been distinguished from functional states. Although the distinction is not entirely clear-cut, functional states are those where no clear physical pathology can be established. Whether this is because they do not follow from physical pathology, or because we are ignorant of the physical basis of each condition, is a matter for debate, but the distinction has nevertheless been traditionally maintained. Functional disorders have generally been accepted to include the major psychotic states of schizophrenia and bipolar affective illness. However, there has been a need to reexamine this distinction, particularly in the light of new evidence suggesting that abnormal neuropsychological organization might be shown to be typical of certain of these disorders.
A word should be said here about the psychiatric terms referred to in this chapter. A distinction is usually made between neuroses and psychoses. In the neuroses, which are less severe and usually require no more than outpatient treatment, the patient maintains contact with reality and has some insight into his or her problems. The symptoms are likely to be anxiety, depression, or avoidance behavior, but without hallucinations or extreme deviations in thought, mood, or action. Psychoses are more serious and generally require admission to a hospital. The symptoms may be primarily in mood (manic-depressive or affective psychosis, mania being intense excitement and overactivity, depression a state of mute, expressionless, and perhaps agitated withdrawal) or in changes in thought or action (hallucinations, delusions, bizarre behavior) in states usually termed schizophrenic. These psychotic patients generally have little contact with reality and limited awareness of their own condition. (If you feel uncertain about the distinctions being made here, it would be worth reading the section on abnormal psychology in an introductory psychology text.)
FUNCTIONAL STATES AND LATERALITY
Some general points should be made before considering in detail the studies of lateral cerebral organization in psychiatric patients. The first is that a rather simple model of psychiatric disorder is often assumed, in which mental illness is seen as exactly like a physical disease. It has long been recognized that there are problems with this view, particularly in accounting for the effects that psychological and social variables have upon abnormal behavior. Secondly, although considerable advances have been made in recent years, there is still considerable unreliability in the use of psychiatric diagnostic labels. Some terms, particularly schizophrenia, are rather loosely used and encompass a variety of rather different disorders. Thirdly, scientific rigor is lacking in many of the studies. Clinical studies are difficult to conduct, but in comparison with the experimental studies discussed in Part III, research with psychiatric patients has been methodologically of poor quality. Little care has been taken over the specification of patient groups, over the measurement of cerebral asymmetries, and over the interpretation of data collected. A variety of experimental findings have been squeezed uncomfortably into a mold formed by the researchers’ expectations. The findings, and the strong claims made for them, must be treated with some caution, although the research questions remain of considerable interest and importance.
Studies of lateral cerebral organization in psychiatric patients seem to have been undertaken in a number of laboratories at about the same time, but one important influence was the work of Flor-Henry. Early on he observed that different psychiatric symptoms appeared to be associated with left- and right-side temporal lobe epilepsy. A great variety of forms of evidence are now quoted in support of laterality differences in functional psychiatric states, but many of the early observations were also of this kind, concerned with lateralized organic states or lateralized symptoms and signs.
Among lateralized signs in psychiatric patients are conversion symptoms, in which disturbance is expressed by some overt physical manifestation, such as hysterical paralysis or loss of sensation. Such “conversions” have been reported to be found more commonly on the left side of the body than on the right, especially in females (Galin, Diamond, & Braff, 1977), although in another study, of 759 cases with conversion symptoms, only 29 showed lateralization, and among these, 12 were left- and 17 right-sided (Bishop, Mobley, & Farr, 1978). Other studies have found at best only marginal significance for a left-side preponderance of psychogenic symptoms. Fleminger, McClure, and Dalton (1980) used an interesting suggestion test to show that suggestion would operate more powerfully upon left-sided responses. They did indeed find such an effect, not only for the psychiatric patients they studied, however, but also for the psychiatric nurses they used as controls.
Pain has also been reported to be more frequent or intense on the left side of the body, especially when it can be considered psychogenic in origin. The inference is that the increased pain is associated with a more “emotional” right hemisphere. However, some of the most careful studies have failed to find any asymmetry, in particular a study of a large sample of patients attending a pain clinic, using data that were not collected with any intention of later analysis for lateral asymmetry (Hall, 1981).
An alternative approach has been to examine the performance of psychiatric patients on standard neuropsychological test batteries to see what kind of organic dysfunction such performance might suggest. FlorHenry, Fromm-Auch, Tapper, and Schopflocher (1981), used a Reitantype battery with hysterical and schizophrenic patients. The results were complex, but it was argued that they supported a model of bilateral frontal pathology being associated with hysteria, but with greater importance being placed upon a dominant (left) hemisphere dysfunction. Hysteria, it was argued, can be viewed as a special form of schizophrenia in this context. Similar results, of schizophrenia being linked with dominant hemisphere dysfunction, particularly of the temporal and temporalparietal regions, were also found by Taylor and co-investigators using an aphasia test and a neuropsychological test battery (Taylor, Greenspan, & Abrams, 1979; Taylor, Redfield, & Abrams, 1981). Not only were Kronfol, Hamsher, Digre, and Waziri (1978) able to show, using standard tests, a right hemisphere abnormality in depressed patients before treatment, but they were also able to show that unilateral ECT (see p. [old 258]), administered to either side of the head, improved “right hemisphere” performance on the tests.
In this context, a study by Hommes and Panhuysen (1971) using the Wada technique with depressed patients has been much quoted. They found that there was a negative relationship between the depth of depression and the degree of left hemisphere speech dominance. They also found that the injection of sodium amytal resulted in improved mood, whether the injection was to the left or right, but especially when it was on the left. They concluded that normal mood regulation depends on the integrity of left hemisphere functional dominance, although their results have also been taken by others to implicate a right hemisphere dysfunction in depressed mood.
While research into lateralized symptoms in neuropsychiatric disorders has declined in recent years, a more recent review can be found in Green, Sergi, and Kern (2003).
The study of ECT has already been mentioned above and in the context of laterality research, and the findings can be turned around to provide evidence for lateralized brain dysfunction in depressive conditions.
A second physiological approach has been to look at electrophysiological parameters. It has long been known that psychiatric patients may show mild abnormalities in their EEG. Abrams and Taylor (1980), for instance, found 44 of 159 psychiatric patients with functional psychoses to have clinical EEG abnormalities, and these were more commonly observed in the left temporal region, and in those patients who had “formal thought disorder.”
Some studies have taken a more systematic approach to the study of ongoing EEG, and Flor-Henry, Yeudall, Koles, and Howarth (1979) found atypical brain function in the left frontal region in obsessive– compulsive patients. Similar results, of greater left hemisphere variability of alpha in paranoid schizophrenia and greater right hemisphere variability in depression, have also been reported (Rochford, Weinapple, & Goldstein, 1981). Coherence studies (see p. 268) have yielded rather complex results, but have if anything tended to suggest a deficit in interhemispheric integration, rather than a lateralized dysfunction in one of the hemispheres (Shaw, Brooks, Colter, & O’Connor, 1979; Weller & Montagu, 1979). The theme of functional disconnection as demonstrated by EEG coherence has been extended by Higashima and colleagues (Higashima et al., 2007), who specifically identified a disconnection between the frontal and temporal lobes of the left hemisphere as being responsible for acute psychotic symptoms in schizophrenia.
As might be expected, evoked potentials (see p. 269) have been studied as well as on-going EEG. Some studies have supported a left hemisphere dysfunction in schizophrenia (Roemer, Shagass, Straumanis, & Amadeo, 1978; Shagass, Roemer, Straumanis, & Amadeo, 1980), although this last study also found evidence of a similar left hemisphere dysfunction for depression, rather than the right-sided dysfunction that was expected. The hypothesis of a failure in interhemispheric communication was supported by the study of Jones and Miller (1981) examining somatosensory evoked potentials, which found no difference in the latency of components following ipsilateral or contralateral stimulation. This is an area of research from which it is often difficult to draw conclusions given the complexity of the findings, but a useful review may be found in Gruzelier, Galderisi, and Strik (2002).
The modern imaging techniques (see Chapter 13) have also begun to make a contribution to an understanding of neuropsychiatric conditions. A PET study of medicated schizophrenics identified an abnormal laterality of the supplementary motor area as responsible for auditory verbal hallucinations (“hearing voices”), interpreted as a failure to correctly attribute speech generated by the brain to the self (Stephane et al., 2006). More generally, left hemisphere abnormalities in brain function in schizophrenia, determined by a variety of imaging methods, have been described by Gur and Chin (1999), and links with neurotransmitter systems proposed by Doty (2003).
MRI studies have also established that there is a greater asymmetry of the Sylvian fissure in schizophrenia characterized by hallucinations and thought disorder (Shapleske, Rossell, Simmons, David, & Woodruff, 2001). Brambilla and colleagues (Brambilla et al., 2007), also using MRI, found an abnormally reduced and inverse hemisphere cerebral blood volume in a large group of patients with schizophrenia, and they argued that abnormal distribution of the blood within the brain may not only be associated with schizophrenia, but account as well for the brain atrophy and cognitive deficits also seen in this group.
Finally, it seems relevant to note two drug studies. Carr (1978) has reported evidence that piracetam, which is known to facilitate interhemispheric neural transmission, had the effect of improving certain aspects of schizophrenic performance, including scores on certain memory and learning tasks. Chlorpromazine, which is one of the major tranquilizers used in the control of schizophrenic symptoms, has also been shown to decrease lateral asymmetry in auditory temporal discrimination (Hammond & Gruzelier, 1978).
None of this evidence seems of great significance taken alone, but it all contributes to a general picture of disturbed cerebral organization that might characterize some psychiatric patients.
Beaumont and Dimond (1973) were the first to use the divided visual field technique with psychiatric patients. They examined within- and between-hemisphere matching using a variety of stimuli and found evidence for some degree of lateralized dysfunction, but also for a more significant deficit in interhemispheric communication. Some later researchers have found a similar effect (Pic’l, Magaro, & Wade, 1979), but others have found clearer evidence for a specific lateralized deficit (Gur, 1978; Tucker, Antes, Stenslie, & Earnhardt, 1978). Colbourn (1982) has thoroughly reviewed the divided visual field studies with psychiatric patients and concludes that it is possible to interpret the findings in terms of a left hemisphere dysfunction being associated with schizophrenia, although the findings for other conditions are much less clear. However, he goes on to point out that other interpretations of the evidence should also be examined.
Dichotic listening has been employed rather more often than the visual techniques. Here again the evidence has been rather confused. However, it again falls roughly into two areas. There are those who argue for a lateralized deficit from the ear differences that have been obtained. Bruder and Yozawitz (1979) found evidence for a right hemisphere abnormality in patients with affective disorders, but no evidence of an abnormality in schizophrenia. With a similar group of patients, an abnormal pattern of lateralization, which returned toward normal with ECT treatment, was also found by Moscovitch, Strauss, and Olds (1981). Schizophrenics were found to have an enhanced right ear advantage in the study of Nachshon (1980), and Tucker et al. (1978) also found a right ear attentional bias in patients with anxiety syndrome. Findings from dichotic listening were also considered alongside other sources of data by Mitchell and Crow (2005). Their thinking focused on the role of language functions in the right hemisphere for successful social communication, and the idea that left and right hemisphere functions were inadequately segregated in psychosis. Their hypothesis is that language functions supported by the right hemisphere are essential to accurately interpreting the intent of an interaction, and that a deficit in these functions in schizophrenics may be the basis of their problems with social interaction.
Other researchers have interpreted their findings differently. Lishman, Toone, Colbourn, McKeekan, and Mance (1978), studying recently recovered psychotic patients, found evidence of abnormal auditory laterality but concluded that this was more typical of an interhemispheric integration deficit than of a left hemisphere dysfunction. Against the general trend, Wexler and Heninger (1979) found increasing laterality to be associated with a decreasing probability of psychotic thought and behavior, so that they formulated their view of the schizophrenic deficit in terms of a failure of interhemispheric inhibition. Lastly, Walker, Hoppes, and Emory (1981) have reanalyzed the data of Nachshon’s study, which they argue rather support a hypothesis of faulty interhemispheric transfer.
Any conclusion with respect to auditory laterality has been considerably complicated by the finding of Gruzelier and Hammond (1979) of asymmetries in hearing thresholds, with diurnal variation, in schizophrenic patients.
There are also studies of lateral performance asymmetry in the tactile modality, but they all point toward a deficit in intermanual transfer and integration, rather than specific lateralized deficits (Carr, 1980; Dimond, Scammell, Pryce, Huws, & Gray, 1980).
A relatively high level of agreement has also been shown by studies of lateral eye movements. These have, in general, found a tendency for an increase in rightward conjugate lateral eye movements (or a reduction in left movements) in schizophrenia (Gur, 1978; Schweitzer, 1979; Tomer, Mintz, Levi, & Myslobodsky, 1979; Tucker et al., 1978). The interpretation placed upon these findings has generally been that this demonstrates the presence of some abnormal process originating from the dominant left hemisphere.
It should also be noted that there have been reports of abnormal distributions of hand preference within psychiatric groups. The overall results have been, however, somewhat unclear. Some investigators (Gur, 1977; Lishman & McMeekan, 1976) have reported an increase in left handedness among schizophrenic patients. Alternately, some have found an excess of right handers among these patients (Taylor, Dalton, & Fleminger, 1980). There are undoubtedly difficulties in assessing the handedness of psychiatric patients by questionnaire measures, and this may contribute to some of the variability in the findings.
By now you may feel thoroughly confused by the variety of studies and the diversity of the findings. A whole range of different methods of investigation has been used with all kinds of different patients, and the result has been a very complex set of research data. This rather selective survey of the studies, with rather more references than usual, has been presented in order to illustrate just how difficult it can be to make sense of some particular field of investigation. However, we must try and make what sense we can out of the research, and it seems useful to do this in the light of the two major theoretical hypotheses that have been adopted.
Three types of model have been proposed to link lateral asymmetry with functional psychiatric disorder. The first was proposed by Galin (1974, 1977), and is the only one to be based upon psychodynamic concepts. Galin derived his ideas from split-brain research and argued that psychiatric states might be likened to the activity of the disconnected right hemisphere. Specifically, the right hemisphere could be associated with the Freudian “primary process” and by its relative disconnection from the conscious, speaking left hemisphere could be the source of unconscious processes and repression. This is an intriguing theory, but partly because a psychodynamic approach is not fashionable in contemporary psychiatry, certainly among those psychiatrists likely to be interested in biological correlates of mental illness, relatively little attention has been paid to Galin’s ideas.
The other two theoretical models will probably already be apparent from the sketch of the research evidence given above. These are of specific lateralized cerebral dysfunction and of impaired interhemispheric integration. The concept of a specific dysfunction has been more actively promoted, especially by Flor-Henry (Gruzelier & Flor-Henry, 1979; Flor-Henry, 1978, 1979). Flor-Henry based his theories on extremely diverse evidence, of the kind that has been cited already, and while the theory is fairly complex, it boils down to the idea that functional disturbances can be linked with dysfunction of the frontal and temporal regions of the brain. If the abnormality is on the dominant side, then the illness will be of a schizophrenic nature. If the abnormality is on the nondominant side, then the disorder will be an affective one with symptoms of depression more likely than symptoms of mania. Flor-Henry, supported by others, argues quite directly that the research evidence unequivocally supports his theoretical model.
The alternative model, which stems from the work of Dimond (Beaumont & Dimond, 1973; Dimond, 1980; Dimond et al., 1980), proposes that certain functional states, principally schizophrenia, can be viewed as a disconnection syndrome. The psychiatric symptoms are associated with a failure of interhemispheric transfer across the corpus callosum. In this it reflects Galin’s model, but is formulated in terms of information-handling processes rather than psychodynamic mechanisms. It has not been claimed that such a model can explain all the evidence that is sometimes cited for neuropsychological effects in psychiatric illness, but it is proposed as a more satisfactory explanation of many of the findings obtained by using well-controlled experimental tasks with schizophrenic patients. As we have seen, the results of a number of studies provide clear support for such a model.
A variant on the disconnection model is the valence model proposed by Davidson and colleagues (Davidson & Irwin, 1999). Their idea is that the anterior parts of the brain, together with the amygdala, on the left side support positive emotions and approach behavior, while the same structures on the right support negative emotions and withdrawal behavior. Damage to one or the other frontal lobe will disturb the balance between the two systems through a loss of inhibition by one hemisphere over the other. The evidence for this hypothesis has been reviewed by Andrewes (2001), and is derived not only from experimental tasks but also from stroke patients. Some of the evidence is particularly persuasive; the study of patients administered unilateral sodium amytal (the Wada technique; see p. 86) found depressed mood to be associated with left sided suppression, while elated mood was found after right sided suppression. However, drawing conclusions from studies of patients with strokes is far from straightforward, and the issue is complicated by the possibility of either denial or alexithymia, the faulty perception of one’s own emotions.
Before assessing the relative validity of these models, one or two methodological and analytical points should be noted. In this book it has often been necessary to stand back and consider methodological issues before assessing the value of any conclusions that might be drawn, and it often seems that the methodological difficulties overshadow the significance of the outcome of the research. This is simply part of the nature of neuropsychology at present, and whether it stems from the immaturity of this field of research or the complexity of the processes under investigation is difficult to say.
There are a whole host of procedural problems that I do not intend to spell out at length. These include the difficulties of undertaking laterality research and of obtaining laterality measurements, which have been discussed in preceding chapters. Added to these are the problems of undertaking clinical studies, and particular problems associated with psychiatric research. It is difficult to standardize the diagnostic procedures employed to classify groups of psychiatric patients; in addition many have been institutionalized for varying periods, and almost all are being treated with a variety of powerful drugs acting upon the central nervous system. These problems have been extensively discussed in a number of reviews (Colbourn, 1982; Marin & Tucker, 1981; Merrin, 1981).
One problem that cannot go undiscussed, however, is the difficulty of interpreting abnormal lateral asymmetries. This is particularly relevant when considering a model of specific lateralized dysfunction. The problem is that once the concepts of interhemispheric inhibition or overactivation are introduced into the discussion of research findings, then almost any model of lateralized dysfunction can be supported by post hoc interpretation of the results. Let me take two studies as examples. This is not to spotlight these studies as examples of bad science–in fact both seem well conducted and yield valuable data–but they are useful to illustrate the problems that are frequently encountered in interpretation of the findings.
Gruzelier and Hammond (1980) examined schizophrenic performance in a dichotic listening task. Among other findings they found that in the serial position effect with digit pairs of equal loudness, there was a tendency for the left ear to lack primacy in report. They discussed this as evidence in support of the expected left hemisphere processing impairment. Their argument was, roughly, that as the reporting processes rely upon verbal encoding, which can be considered a left hemisphere activity, then even though the effect was at the left ear, it demonstrated a left hemisphere impairment. They may be correct, and I have taken their result out of the context of their other findings, but it still seems that the result could as easily be taken as direct evidence of an alternative right hemisphere deficit.
In a second study, Gur (1979) asked subjects with schizophrenia to judge pairs of altered pictures that were presented simultaneously or successively. Control subjects did better overall, and were equally fast in the simultaneous and successive conditions. The patients with schizophrenia were faster in the successive than the simultaneous condition. Gur argues that this pattern of performance is typical of subjects with right brain damage, which must therefore be evidence for left hemisphere overactivation in schizophrenia. Even assuming that the parallel with brain damage allows a valid inference to be drawn, the conclusion might obviously just as well be that people with schizophrenia have a defective right hemisphere.
So what conclusions may be drawn about laterality and functional psychiatric states? Of the scholarly reviews, one concludes that despite the methodological difficulties, there is consistent support for a model of left hemisphere overactivation or dysfunction in schizophrenia (Newlin, Carpenter, & Golden, 1981). Three other reviews, however, are driven to the view that there is no consistent support for any of the models proposed, and each makes certain recommendations that are designed to improve the quality of research in this field (Marin & Tucker, 1981; Merrin, 1981; Wexler, 1980).
My own view continues to be that this last position is the only possible conclusion at the present. Many of the theoretical discussions have been far too selective in the evidence that they have considered, and most have indulged quite shamelessly in post hoc interpretation that has been in line only with the hypotheses being promoted. The basic research evidence is still too sketchy to allow any clear conclusions to be drawn, and it is certainly much too early to proclaim that functional psychiatric states can be explained in terms of (a particular model of) abnormal neuropsychological processes. There are a variety of other psychological models of psychotic behavior, for instance in terms of an attentional disturbance associated with heightened arousal, and many of these models are at least as powerful in explaining the data as are the neuropsychological theories. Whether neuropsychological models will come to reflect models cast in other conceptual terms, and derived from different levels of behavioral analysis, remains to be seen, but the thought undoubtedly provides at least one avenue for future research. What is now established without reasonable doubt is that there is a neuropsychological dimension to many psychiatric disorders, even if the precise mechanisms that underlie the dysfunction have yet to be clarified.
Neuropsychology has been able to make an increasingly valuable contribution to the study of the mental sequelae of organic pathological processes in the area known as neuropsychiatry or organic psychiatry. The psychiatric effects of gross damage to the brain have long been recognized, and neuropsychology is able to clarify and illuminate the processes involved by developing its own models of the processes that underlie brain–behavior relationships.
Almost no clear conclusions have emerged from the upsurge of interest in the neuropsychological concomitants of functional psychiatric states. There has been a refreshing reevaluation of the processes that might contribute to schizophrenic and affective psychotic disorders. Many exciting theoretical models have been formulated, but despite the growing research effort, it is still too early to say whether a particular neuropsychological dysfunction can be identified as accompanying a given functional state. This area is nevertheless one of the most exciting and challenging of any in contemporary neuropsychology.
The problems and difficulties that arise when we attempt practical applications of neuropsychology illustrate not only the fundamental deficiencies of our knowledge but also the importance of understanding how physical systems generate the highest levels of thinking, feeling, and consciousness. We have made considerable progress in unraveling the nature of brain–behavior processes, and yet there is much still to be discovered, not only about the details of functional relationships, but about the fundamental principles that govern the operation of those functions. To discover the answers involves an engagement with essential philosophical issues, a rigorous and creative approach to experimental design, and a questioning and critical appraisal in the construction of theories. The answers–and they will be found–will be of fundamental importance for our understanding, not only of disordered behavior, but also of everyday human action, and the essence of the humanity of humankind.