■                                                               ■Northwestern University

Ss maintained on 23-hr. food deprivation cycles were housed in activity wheels (experimental) or standard laboratory cages (control). Experimental Ss were unable to maintain body weight and died; despite high levels of activity, experimental Ss ate significantly less than controls. Interposing a 3-hr. period between running and feeding, adapting Ss to area of feeding, or administering pentobarbital (4 mg/kg) had little influence on this food- intake reduction (“self-starvation”). Chlorpromazine administration (1 mg/kg reduced activity and did have a significant effect on self-starvation. Hypothalamic involvement was discussed in relation to thermo- and glucostatic theories of feeding and a general excitability factor.


The present series of experiments was initiated to clarify the relation between nonadaptive behavior and positive incentive mechanisms. This relation was suggested by a study which showed that rats given the choice between food essential for survival and rewarding posterior hypothalamic brain stimulation chose stimulation, ignored food, and, if allowed to continue (Routtenberg, 1964), perished. This “self-starvation” was seen when electrodes were in the proximity of the medial forebrain bundle at posterior hypothalamic levels, and indeed, the intensity of effect appeared to be related to the closeness of the electrode to this structure (Routtenberg & Lindy, 1965).

To determine the generality of this result, it seemed desirable to explore situations with intact Ss in which a similar effect might obtain. Such situations, if they exist, may suggest certain fundamental processes in the nonadaptive pattern. A brief report that marked weight losses occurring in rats living in activity wheels (Spear & Hill, 1962) suggested such a situation. Earlier reports (Hall & Hanford, 1954; Weasner, Finger, & Reid, 1960) showed, in fact, that rats housed in activity wheels on a food deprivation schedule failed to maintain stable weight; Hall and Hanford (1954) noted that a few rats died, but this result was not investigated further.

Experiments 1-5 determined the extent to which activity-wheel behavior on a food deprivation schedule leads to failure of maintaining body weight, and some of the causes for such a result.

Experiment 1


Subjects. Thirty-six male albino rats, obtained from Holtzman Research, weighed 280-480 gm. Six served as food deprivation controls and lived in laboratory cages; 30 experimental /Ss were housed individually in activity wheels in a temperature-controlled room set at 75° F. Water was available ad lib to all /Ss.

Apparatus. Six standard Wahmann activity wheels were used. A 10 X 6 X 5 in. cage with a sliding door was attached to the wheel; the door could be closed to prevent movement between the cage and the wheel. Water was obtained in the cage.

Procedure. Experimental and control /Ss were fed for 30 min. daily on a feeding stand containing Purina Chow pellets and a dish of water. Because weight loss was appreciable in the first few daj^s of the experiment, both control and experimental Ss were permitted to feed for an additional 15-30 min. on the first 9 days; from Day 10 until the completion of the experiment, Ss were fed for only 30 min. daily. Three control Ss were maintained on this schedule for 26 days and 3 for 41 days.

By use of the sliding door, 6 experimental Ss were restricted to the adjoining cage for 1 hr. before feeding, 6 for 1 hr. immediately after feeding, 9 for 2 hr. before feeding, and 9 for 2 hr. immediately following feeding. Excluding the 30-min. feeding period and the confinement period, all 30 /Ss were permitted free movement between the wheel and the adjoining cage.

As the present experiment was designed to determine whether rats could maintain their weight, a criterion for stable weight was instituted, i.e., the weight on Day 4 of any 4-day period had to be equal to or greater than the weight on Day 1 of that 4-day period.            .


Figure 1 shows the weight of the experimental and control Ss and the activity of experimental Ss on the last 8 days of the experiment. The last day for experimental and control Ss occurred when starvation or the criterion for stable weight was observed. Control Ss were able to maintain their weight, while all experimental Ss lost weight and died. Along with this decreased weight, there was a steady increase in activity, except for a slight decline on the final day, presumably because of inanition. Moskowitz (1959) and Duda and Bolles (1963) have reported similar negative correlations between body weight and wheel activity (—0.99 and —.69, respectively) .

All control Ss maintained weight in a mean of 15.9 days. Experimental Ss died within a mean of 13.5 days. At the extremes, 2 Ss died after 8 days and 1 S died after 25 days of the experiment. There were no significant differences between the before and after groups on measures of activity or weight in the 1-hr., the 2-hr., or the combined groups.

Experiment 2

Experiment 2 was performed to provide an initial assessment of the causes for the marked weight loss shown in Experiment 1. It seemed advisable, at the outset, to measure food intake. In addition, three different schedules were employed to obtain more precise estimates of the feeding durations that would produce the marked weight loss shown in Experiment 1. It seemed likely that a feeding duration that was too short would cause both control and experimental Ss to perish, while too

Fig. 1. Weight and activity of experimental Ss, and weight of control Ss on the last 8 days of the experiment for each S. (The Before group was restricted to a cage adjoining the wheel before feeding; the After group was restricted after feeding.)

long a feeding duration would allow both control and experimental Ss to survive.


Subjects. Thirty adult male albino rats, obtained from Holtzman Research, weighed 288-398 gm. at the start of the experiment.

Apparatus. Five activity wheels similar to those described in Experiment 1 were used. All & were fed Purina Laboratory Chow, and food intake was measured to the nearest 0.1 gm. using a Pelouze balance. Temperature was maintained at 72° ± 3° F. and the relative humidity was 35-60%.

Procedure. The Ss were divided into experimental and control groups. Experimental <Ss lived in activity wheels, controls in standard laboratory cages. Each group of 15 was additionally partitioned into three groups of 5, according to the duration of the feeding period (30, 45, and 60 min.). The 5 experimental /Ss or the 5 control /Ss of a given feeding duration were run at the same time.

Experimental Ss were detained in the cage adjoining the wheel for 1 hr. before feeding. All Ss were weighed before daily feeding. Food was weighed before and after feeding and spillage was weighed in evaluating food intake. Water was available ad lib.

Fig. 2. Food intake of experimental and control Ss on Days 2-7 of Experiment 2.

The weight stabilization criterion instituted in Experiment 1 was used. Since starvation was quite predictable a day or two before S actually perished, in order to avoid unnecessary suffering on the part of the rat, any S that ate less than 1 gm. during the feeding period was considered to have met the starvation criterion and was sacrificed with an overdose of pentobarbital.


Figure 2 shows food intake of experimental and control Ss on Days 2-7. Statistical analysis of these data according to Edwards (1960, p. 233) revealed that the experimental Ss living in activity wheels ate significantly less than control Ss living in standard cages (F = 61.3, df — 1/24,

p < .001).

The significant reduction in food intake was accompanied by an increasing daily activity level, a combination which led to a significantly greater reduction in body weight, in experimental Ss as compared with controls (Treatment X Days interaction: F = 4.0, df = 6/144, p < .001).

Using the criteria for starvation and weight stabilization, all experimental and control Ss in the 30-min. group starved; in the 45-min. group, all experimental Ss and 4 of the 5 controls starved; in the 60-min. group, 4 of the 5 experimental Ss and 1 of the 5 control Ss starved.

It seemed reasonable to conclude that the reduction in food intake in experimental Ss was of importance relative to survival when the feeding period was 1 hr., since eating at or above control levels might have enabled Ss to live. We have described this reduction in food intake as self-starvation, since what S did led to its destruction.

Experiment 3

In Experiment 2 it was shown that, despite high levels of energy expenditure in the activity wheel, experimental Ss consumed less food than controls. While this seemingly paradoxical effect has several alternative explanations, none that we had considered could satisfactorily account for the consistent depression of food intake of experimental Ss on the first day. It should be recalled that in Experiment 2 control Ss were fed in their home cages, while experimental Ss were not. Thus, experimental Ss may have spent the first few days exploring the new environments, and this exploration might have competed with feeding behavior and led to a reduction in food intake as compared with controls. An attempt was made to determine the effect of this exploration by adapting both experimental and control Ss to the feeding cages before deprivation procedures were begun.


Subjects and, apparatus. Ten adult male albino rats, weighing 314-354 gm. at the start of the experiment, were run in the five activity wheels described in Experiment 2.

Procedure. All Ss were fed ad lib in individual home cages for 6 days before the start of the experiment. The procedure used for the 1-hr. group in Experiment 2 was then begun on the seventh day. Starvation and weight stabilization criteria established in Experiments 1 and 2 were used in this experiment. All Ss were mistakenly allowed to feed for an additional 15 min. on the fourth day of the experiment; while this did not represent a significant departure from procedure, statistical analyses of food intake did not use data for that day,


The effects of adaptation were the following. (a) Adapted $s ate significantly more than nonadapted Ss of Experiment 2 (F = 13.8, df = 1/16, p < .01). (b) This effect was most marked on the first 2 days of the experiment, and tended to diminish on subsequent days; such a result is shown in Figure 3 and is supported by a significant Treatment (adaptation/no adaptation ) X Days interaction (F = 4.1, df = 4/64, p < .005). (c) Though adaptation decreased the difference in food intake of experimental Ss relative to controls (F — 5.7, df = 1/16, p < .05), experimental Ss of this experiment still ate significantly less than control Ss {F = 67.7, df = 1/16, p < .001). Thus, while adaptation had a significant effect on the first days of the experiment, it did not preclude the appearance of the food-intake reduction of experimental Ss living in activity wheels. Indeed, as Figure 4 shows, experimental <Ss increased activity, lost weight, and, as in Experiment 2, ate consistently less despite the increased energy requirements. In the present experiment control Ss stabilized weight in a mean of 11.4 days, while all experimental Ss self-starved in a mean of 13.4 days.

Experiment 4

The interpretation of the results of Experiments 1-3 may be related to the thermostatic theory of feeding (Brobeck, 1960). In this theory it is assumed that high temperatures reduce, while low temperatures increase, food intake. While direct temperature measurements from brain and body core are required for a test of the theory, a preliminary approach might be made by looking at the effects of 3-hr. isolation on food intake. According to a thermostatic theory, if temperature augmentation caused by activity should subside in a 3-hr. period, then the depression of food intake caused by the supposed high temperature levels would be reduced, and an S isolated from the wheel for 3 hr. should eat more than an S isolated for 1 hr.


Subjects and apparatus. Ten male albino rats, weighing 271-294 gm, were run in 10 Wahmann activity wheels similar to those described in Experiment 1.

/Ss were allowed to move freely between wheel and adjoining compartment; for 3 hr. before feeding, they were restricted to the adjoining cage by use of the sliding door, and, at feeding time, all (Ss were taken to the feeding cages (standard labora-

Fig. 4. Weight of experimental and control Ss, and activity of experimental & in Experiment 3 on Days 2-14. (On Days 12 and 14, (Ss were removed from the experimental group because they had reached the starvation criterion.)


Fig. 3. Food intake of combined experimental and control adapted (Ss in Experiment 3 vs. food intake of combined experimental and control nonadapted (Ss in the 1-hr. group of Experiment 2.

Procedure. All 10 Ss were housed in activity wheels. The 5 control Ss were restricted to the adjoining cage and were not permitted entry to the wheel during the experiment. Experimental


tory cages) and allowed to feed for 1 hr. Food intake was measured in the manner described in Experiment 2; criteria for weight stabilization and starvation were the same as in Experiments 1 and 2, respectively.


‘ The 3-hr. isolation procedure did not substantially mitigate the result seen in Experiment 2 that Ss living in wheels would eat significantly less than their controls. Using the 1-hr. experimental and control groups in Experiment 2 for comparison, it was found that the difference in food intake between the experimental and control groups was significant (F — 33.69, d/ = 1/16, p < .001) but the effect of duration of isolation and the interactions between isolation duration and days, days and groups, and duration and groups, and the triple interaction were not significant.

A possible effect of the 3-hr. isolation was suggested in the final stages of the experiment. All 5 control Ss maintained weight in a mean of 13.5 days. Three experimental Ss self-starved (12, 14, and 24 days) and 2 maintained weight (17 and 24 days). These results, while perhaps suggesting a reduction of the self-starvation effect, must await confirmation from future experiments in which initial body weight and place of feeding are identical for Ss in both 1-hr. and 3-hr. groups.

Experiment 5

As neither adaptation to the feeding cages nor extended isolation from the wheel prior to feeding completely mitigated the self-starvation effect, it seemed worthwhile to attempt other modifications in an effort to preserve the rat from perishing and so to gain insight concerning the causes for this behavior. One line of reasoning had led to the view that the effect was mediated by overactivity in limbic or reticular structures and that certain drugs might depress such activity.

While the mode of action of chlorproma- zine (CPZ) and pentobarbital (Pent.) has not been precisely delineated, there is some evidence to suggest that Pent, acts primarily on reticular systems, while CPZ acts on limbic structures. Whatever the case, both drugs are depressants and have different electrophysiological and behavioral effects (e.g., Killam & Killam, 1958; Kornetsky & Bain, 1965). It was hoped that one of these drugs might depress the presumed system or systems causing the self-starvation, but would not similarly reduce S’s ability to maintain its weight. This would then lead to weight stabilization and survival.


Subjects. Twenty male albino rats weighed 308390 gm. Ten controls lived in laboratory cages and received either Pent. (N = 5) or CPZ (N = 5); 10 experimental Ss lived in activity wheels and received either Pent. (N = 5) or CPZ (N — 5).

Apparatus. Five activity wheels described in Experiment 2 were used.

Procedure. All >Ss were adapted to the feeding cages 10-13 days before the start of 1-hr. deprivation procedures. Ten Ss received 8 mg/kg of Pent, on the first day of adaptation, and 4 mg/kg on each subsequent day. The other 10 Ss received 3 mg/kg of CPZ on the first day of adaptation, and 1 mg/kg on each subsequent day. These values, according to Kornetsky and Bain (1965)., represented the maximum dosage that had a minor effect on fixed- interval responding for food reward. Other procedures and criteria for starvation and weight stabilization were similar to those outlined in Experiment 3.


The CPZ injection caused a marked depression in activity as compared with Pent, injections. This effect is seen in Figure 5, which shows the activity level of experimental (Ss in the two drug groups.

An analysis of the food-intake data (Figure 6) shows that while Pent, did not materially change the depression of food intake on the part of experimental (Ss, CPZ reduced the food-intake difference between experimental and control (Ss. This impression was supported by a significant Drug X Condition interaction (F = 14.34, df = 1/16, p < .005).

All control (Ss in both groups maintained stable weight levels in a mean of 15.6 days. All 5 Pent, experimental (Ss self- starved, while 3 of the 4 CPZ (Ss maintained weight in 11, 17, and 18 days (the fifth CPZ (S was dropped on Day 26 because it was fed for an additional 25 min.). The weight curves of the first 12 days


Fig. 5. Activity of experimental <Ss in Experiment 5 receiving either pentobarbital or chlorpro- mazine on Days 2-12.

are shown in Figure 7. These curves support the above conclusions since they show clearly that CPZ reduced the difference in weight loss between control and experimental Ss.

Summarizing the results of Experiment 5 relating to doses applied, CPZ depressed activity to a greater extent than Pent. As a concomitant of this depression in activity, there was a reduction in the food-intake depression, normally seen in Ss living in activity wheels. This result may have been of assistance in mitigating the self-starvation effect for the 3 CPZ Ss that survived.


There appear to be two fundamental issues raised by the results of the present



Fig. 6. Food intake of experimental (activity wheel) and control Ss in Experiment 5 receiving chlorpromazine or pentobarbital on Days 2-10.


Fig. 7. Weight of experimental and control Ss in Experiment 5 on Daj^s 1-12.

study: the causes for the rat running in response to deprivation, and the causes of the reduction of food intake. Our present working hypothesis is that deprivation leads to hunger and weight loss, which cause overactivity in certain lateral and posterior hypothalamic nuclear systems. One consequence of this overactivity is an increase in locomotor activity. Concerning the second issue, hypothalamic activity and locomotor activity generate certain products which act as signals to the hypothalamus that are interpreted as food and thus depress food intake.

That deprivation can lead to increased activity in lateral hypothalamus is suggested by the microelectrode work of both Anand, Chhina, and Singh (1962) and Oomura, Kimura, Ooyama, Maeno, Iki, and Kuniyoshi (1964). That overactivity in these structures can lead to increased locomotor activity is suggested by recent brain stimulation work (Scott, 1964). In addition, lesions in these areas tend to depress activity (Gladfelder & Brobeck, 1962), suggesting that the hypothalamus is important in the maintenance of locomotor activity.

That locomotor activity may lead to food-intake reduction is suggested both by the present results and by the work of Cohen and Serrano (1963) and of Stevenson, Box, Feleki, and Beaton (1966), using somewhat different procedures. Our results also show that when activity is reduced by CPZ, food-intake suppression does not occur. This supports the view that there exists an intimate though paradoxical relation between locomotor activity and food intake in the present experiment. It should be recalled in this regard that CPZ typically reduces food intake (e.g., Schmidt & Van Meter, 1958) as well as activity (e.g., Marriott & Spencer, 1965).

What are the consequences of hypothalamic activity that might lead to food- intake depression? Briese (1965) has shown that stimulation in lateral hypothalamus causes a marked increase in body temperature, and Booth, Coons, and Pitt (1965) have shown that electrical or chemical stimulation in similar areas lead to an increase in blood sugar. Increased temperature and blood glucose might be sustained by endocrine consequences of hypothalamic stimulation. Thus stimulation in hypothalamus can lead to an increased release of epinephrine and a consequent rise in body temperature and blood glucose (Bell, Davidson, & Scarborough, 1965). Stimulation in hypothalamus can also increase blood sugar level via the conversion of noncarbohydrates to glucose (gluconeogenesis) presumably by the glucocorticoids (Feldman, Friedman, & Wertheimer, 1966) .2 These temperature and glucose changes could then lead to a depression of food intake according to thermostatic (Brobeck, 1960) and glucostatic (Anand, 1961) theories of feeding.

Hypothalamic overactivity might also lead to certain nonspecific effects that could enhance self-starvation. Stated broadly, it may be possible that the reduction in food intake is brought about, in part, by a general heightened excitability, and that when the rat finally does have the opportunity to feed it is unable to eat efficiently because it is too excited. Some preliminary results serve to illustrate this point. When the rat had to press for food instead of eating pellets placed in its cage, we observed a reduction in bar pressing in a Skinner-box situation that may be related to the food-intake reduction seen in the present experiment. We also observed that several rats ran about the box, and demonstrated hyperactivity when the marked reduction in bar pressing occurred.

2 The gluconeogenesis mechanism was brought to our attention by R. W. Goy.

This was followed by high levels of wheel running suggesting that the reduction in bar pressing was not caused by inanition but by a general rise in excitability level.

The possibility exists, therefore, that the reduction in feeding in the present series of experiments might be brought about in part by a general overexcitability of the rat. It is interesting to consider our CPZ results as having a quieting effect on the rat by reducing hypothalamic overactivity. Of relevance here are the reports that CPZ does depress hypothalamic unitary activity (Zukauskas & Machne, 1964), and that CPZ does increase food intake of patients suffering from anorexia nervosa (Crisp, 1965).

Concerning the initial reasons for pursuing the present study, i.e., self-starvation in the brain reward situation, it would appear promising to view both findings as suggesting that the reduction in food intake is brought about by increased activity in certain posterolateral hypothalamic regions. In the Routtenberg and Lindy (1965) report, this activity was augmented by direct electrical stimulation, while in the present report hypothalamic activity may have been augmented by manipulation of environmental conditions. It would be premature at present to discuss the relation between the two in any detailed fashion.

.                          REFERENCES

Anand, В. K. Nervous regulation of food intake.

Physiol. Rev., 1961, 41, 677-708.

Anand, В. K., Chhina, G. S., & Singh, B. Effect of glucose on the activity of hypothalamic “feeding centers”. Science, 1962, 138, 597-598. Bell, G. H., Davidson, J. N., & Scarborough, H. Textbook of -physiology and biochemistry. Baltimore: Williams & Wilkins, 1965.

Booth, D. A., Coons, E. E., & Pitt, M. E. Blood glucose changes and feeding behavior from stimulating the rat hypothalamus. Biochem. J., 1965, 96, 68.

Briese, E. Hyperthermia in self-stimulating rats.

Acta physiol. Latinoam., 1965, 15, 357-361. Brobeck, J. R. Food and temperature. In G. Pincus (Ed.), Recent progress in hormone research. New York: Academic Press, 1960. Pp. 439-459. Cohen, В. J., & Serrano, L. J. Effects of exercise and confinement on growth, maze learning, and organ weights in rats. Lab. anim. Care, 1963,13, 689-696.

Crisp, A. H. Clinical and therapeutic aspects of anorexia nervosa—a study of 30 cases. J. psy- chosom. Res., 1965,9,67-78.

Duda, J. C., & Bolles, R. C. Effects of prior deprivation, current deprivation, and weight loss on the activity of the hungry rat. J. comp, physiol. Psychol., 1963, 56, 569-571.

Edwards, A. L. Experimental design in psychological research. New York: Holt, 1960.

Feldman, S., Friedman, N., & Wertheimer, H. E. Gluconeogenesis following hypothalamic stimulation. Nature, 1966, 210, 860.

Gladfelder, W. E., & Brobeck, J. R. Decreased spontaneous locomotor activity in the rat induced by hypothalamic lesions. Amer. J. Physiol, 1962, 203, 811-817.

Hall, J. F., & Hanford, P. V. Activity as a function of a restricted feeding schedule. J. comp, physiol. Psychol, 1954, 47, 362-363.

Killam, K. F., & Killam, E. K. Drug action on pathways involving the reticular formation. In H. H. Jasper (Ed.), Reticular formation of the brain. Boston: Little, Brown, 1958. Pp. 111-122.

Kornetsky, C., & Bain, G. The effects of chlor- promazine and pentobarbital on sustained attention in the rat. Psychopharmacologia, 1965, 8, 277-284.

Marriott, A. S., & Spencer, P. S. J. Effects of centrally acting drugs on exploratory behavior in rats. Brit. J. Pharmacol. Chemother., 1965, 25, 432.

MoSkowitz, M. J. Running-wheel activity in the white rat as a function of combined food and water deprivation. J. comp, physiol. Psychol., 1959, 52, 621-625.

Oomota, Y., Kimtjra, K., Ooyama, H., Maeno, T., Iki, M., & Kuniyoshi, M. Reciprocal activities of the ventromedial and lateral hypothalamic areas of the cat. Science, 1964, 143, 484-485.

Routtenberg, A. Self-starvation caused by the availability of feeding center self-stimulation. Amer. Psychologist, 1964, 19, 502. (Abstract)

Routtenberg, A., & Lindy, J. Effects of the availability of rewarding septal and hypothalamic stimulation on bar pressing for food under conditions of deprivation. J. comp, physiol. Psychol, 1965, 60, 158-161.

Schmidt, H., Jr., & Van Meter, W. G. Chlor- promazine depression of food intake in the albino rat. J. comp, physiol. Psychol., 1958, 51, 29-31.

Scott, J. W. Reinforcing stimulation of the brain and elicitation of general activity in the rat. Amer. Psychologist., 1964, 19, 503. (Abstract)

Spear, N. E., & Hill, W. F. Methodological note: Excessive weight loss in rats living in activity wheels. Psychol. Rep., 1962, 11, 437-438.

Stevenson, J. A. F., Box, В. M, Feleki, U., & Beaton, J. R. Bouts of exercise and food intake in the rat. J. appl. Physiol, 1966, 21, 118-122.

Weasner, M. H., Finger, F. W., & Reid, L. S. Activity changes under food deprivation as a function of recording device. J. comp, physiol. Psychol, 1960, 53, 470-474.

Zukauskas, E„ & Machne, X. Effect of Chlorpro- mazine on the response pattern of hypothalamic units, hit. J. Neuropharmacol, 1964, 3, 341— 352.

(Received November 15, 1966)

[1] Experiment 1 was carried out at the Brain Research Laboratory, University of Michigan. The first author is grateful to James Olds for his support and suggestions. Experiments 2-5 were performed at Northwestern University and were supported by the Graduate School and by United States Public Health Service Contract MH 11991.   .