Animals without water do not eat as much food as usual, and hungry animals do not drink much water (4, 8, 13, 16, 17, 20). Animals drink more after meals than at other times. The dog and hen (13) and the rat (20) show a drop in food intake during water de privation. Dogs (5, 11) and rats (20) similarly drop in water intake during food deprivation. Rats drink more after a period of water deprivation during which food is available than after a similar period with no food available (17). The corresponding cas e for food intake apparently has not been investigated. After protracted periods of food or water deprivation, rats exceed in both drinking and eating their average value before deprivation (4).
No systematic sets of data are available on these phenomena as they are encountered in studies of learning, although their signifigance for theoretical formulations of "motivation" and learning has not escaped some investigators (9, p.234; 22). Recent studies on "drive interaction," "drive discrimination," and on "cognitions" have involved the control of the behavior of food-deprived and of water-deprived rats by food and water placed in goal boxes and alleys (21). These have not had uniform results, so that it is pertinent to examine the matter more closely.
Today's learning theorists are in fair agreement on a definition of "drive." This concept is an intervening variable, explicitly involving two sets of operations and implicitly a third.2 The first operations establi sh drives; e.g., for hunger and thirst, the animal is deprived of food and water, respectively, for a stated number of hours. The second class of operations is the measurement of classes of behavior (running, bar pressing, eating) that vary with the durat ion of the preceding deprivation. The third, implicit, operation is that of satiation, usually giving the animal access to food and water long enough so that it neither eats nor drinks for a specified period. "Satiation" operations vary consider ably.
In this experiment, the operations are depriving the animal of food or water, or both, through stated intervals of time following free feeding and free drinking. The measure of behavior chosen is the total weight, in grams, of food and water ingested by t he animal in the first hour following the period of deprivation. The adequacy of these measures has been established by others (1, p.128; 2, 7, 15, 18, 19).
The general plan of the experiment and the values of the variables investigated
have been chosen to provide data useful for the interpretation of experimental
data in the field of learning.
Subjects. Twelve male albino rats (Hisaw strain) ranging in age from 97 to 100 days at the beginning of the first experiment were the Ss. Before the experiment, they were caged in groups of six, with continuous access to food and water.
Experimental room and apparatus. The rats lived in a small internal room, with one door (closed) and no window. An automatic timer turned on two fluorescent ceiling fixtures at 1900 daily; they were turned off at approximately 0700. A heater was tu rned on by a thermostat when the room temperature fell to 65°F; no attempt was made to control humidity. A continuous, 24-hr. record showed that the mean temperature throughout the experiment was 70.8°F with a range from 63.3°F to 75.5° ;F; the mean relative humidity was 50.0 per cent with a range from 39.5 per cent to 62.8 per cent.
Each animal lived separately in an 11-in. by 8-in. by 8-in. wire-mesh open-bottom cage resting on a wire-mesh grating mounted on a metal rack. A paper-lined tray under each cage caught feces and food particles that fell through the mesh. No attempt was ma de to control coprophagia.
Measurements. Weighed rations of food, greater than the amount the animal could eat, were placed in each cage at the beginning of the period. At the end of the period, the food remaining was taken from the cages. Food particles were carefully colle cted from the trays, dried, and added to the food taken from the cages, and the whole then weighed. The difference between the initial amount and the amount recovered was taken as the measure of food eaten. The error of measurement in this operation was a pproximately .05 gm.
An inverted, rubber-stoppered, 250-cc. chemical graduate was mounted on the side of each cage. A short length of copper tubing led from the stopper through the side of the cage. The amount of water, to the nearest .5 cc., removed by the animal was determi ned by reading changes of water level from the calibrated graduate. A water bottle fitted to an empty cage yielded corrections for losses due to dripping and evaporation. This measure was converted into grams.
The experimental diet consisted of dry Purina Laboratory Chow biscuits and tap water. A small piece of fresh lettuce was given daily during the free-feeding period as a dietary supplement.
Preliminary handling. For 17 days prior to the beginning of the experiment, the animals were kept on a 24-hr. feeding schedule. At 2200 daily, they were given food and water. At 2000 on the following day, food and water were withdrawn.
Daily Schedules. The experiment was performed in 12 experimental periods of six days each. The first five days of each period were experimental ones; the sixth was a recovery day, during which the animals had unrestricted access to food and water. Each experimental day began at 2200. From 2200 to 1900 of the following day the animals underwent a 21-hr. deprivation period. At 1900, food and water ingestion through this period were measured. Then followed a measurement period extending from 1930 to 2030. At 2030, food and water ingestion were again measured. At 2045 all animals were given free access to food (including the lettuce) and water, and the trays were changed. Shortly before 2200, food and water were removed and dean trays pla ced under the cages. The deprivation period of the new experimental day began at 2200, when the experimental rations were given. Each day, then, incorporated a 21-hr. deprivation period, a second deprivation period of 1/2 hr., and a 1-hr. measurement per iod. The remaining time was devoted to experimental routine and to free feeding and drinking.
On the fifth experimental day of each of the 12 mental periods, from 2200 to 2130 of the next calendar day, all animals had free access to both food and water. Food and water were then removed, and at 2200 the new experimental period began.
Experimental treatments. Through all five deprivation periods of each experimental period, any of four different treatments could obtain. The rat might be given: (a) food and water (FW), (b) food but no water (F), (c) water but no food (W), (d) neither food nor water (0). These, together with the immediately following 1/2 hr. period with neither food nor water, establish four deprivation conditions: (a) FW-1/2 hr. food deprivation, 1/2 hr. water deprivation; (b ) F-1/2-hr. food deprivation, 21 1/2-hr. water deprivation; (c) W-21 1/2-hr. food deprivation, 1/2-hr. water deprivation; (d) 0-21 1/2-hr. food deprivation, 21 1/2-hr. water deprivation.
During the 1-hr. measurement period the animals could
be given (a) both food and water (fw), (b) food but no water
(f), (c) water but no food (w).3
Measurements of both food and water ingestion were made at the end of both deprivation and measurement periods.
If we take all combinations of the two periods, there are 12 different mental treatments of deprivation and measurement. Each rat underwent all 12. The order in which any one animal went through them was determined by picking cards from a hat. The experim ent was performed through the period February 8 to April 21, 1951.
Means of both food and water ingestion were computed for each treatment for the first two and for the last three days. Despite a slight tendency for intakes to be lower on the last three days, no significant differences appeared. For this reason, the data presented are the arithmetic means,4 for each rat, of daily food and water intake over all five days of each experimental period.
Deprivation period. Table 1 presents the number of grams of food (la) and water (lb) ingested by the animals during each of the three kinds of deprivation periods, subdivided according to the associated measurement periods. In Table 1 two-tailed matched t tests of the differences show that the differences between columns are significant at better than the .001 level in all cases, and that the differences within columns are not significant except for the greatest difference in t he right-hand column of Table la, which falls between the .02 and .05 levels of probability; this may be expected by chance when 24 comparisons are made. Ingestion during the deprivation period did not depend on the measurement period of the preced ing day. The data in each column were therefore lumped. In the 21-hr. period, the animals ingested approximately 51 gm. of substance when both food and water were available, 39 per cent of which was dry Purina Chow. This may be compared with Strominger's 33 per cent (20).
When food-deprived, the rats drank .41 as much as when food was available, and when water deprived, they ate .57 of their normal intake. These numbers are the W/FW and F/FWratios, respectively. Under these conditions the foodless animal limi ts its intake of water, and the waterless animal reduces the amount of food it eats. Water consumption is reduced more in this way than is food consumption.
Measurement period. Tables 2a and 2b present, respectively, the amounts of food and water taken during the 1-hr. measurement periods, together with their standard deviations, and the f/fw, and w/fw ratios. Matched t I> tests have been made of the difference between each mean and every other mean of each table.
In Table 2a all differences are significant at better than the .01
level, except three: (a) FWf and FWfw differ between the .02 and .05 levels
of significance; this f/fw ratio, however, is in conformity with ratios
obtai ned over the 22 hr. of the ingestion period. (b) No significant difference
appears between Wfw and Ofw. Whether or not water is available during food
deprivation makes little difference in the amount eaten at the end of the
period provided it is available when the food is given. (c) Ffw and Of
do not differ significantly. Rats that have been deprived only of water
eat just as much food when water is given to them as do rats first deprived
of both food and w ater and then given food without water.
During First 21
Hr. of Preceding
21 1/2 Hr. (Deprivation Period)
|Food and Water (FW)||2.52||1.05||3.60||1.22||.70|
During First 21
Hr. of Preceding
21 1/2 Hr. (Deprivation Period)
|Food and Water (FW)||3.20||1.37||4.77||1.35||.67|
In Table 2b all differences are significant at better than the .01 level except as follows: (a)Between Fw and Ffw; waterdeprived rats drink large quantities of water irrespective of the presence of food when water is given them . The w/fw ratio deviates from 1.00 in the direction expected from the deprivation period. (b) Between Ow and FWfw, and Ow and FWw. The rat first deprived of both food and water, and then given water, drinks no mo re than an effectively satiated animal. A strong "hunger drive" seems to suppress drinking and might be said to "inhibit thirst."
The "self-imposed" deprivation reveals itself in the w/fw and f/fw ratios. The ratio reaches its maximum when, for 22 1/2 hr., the animals have been deprived of the substance measured, and its minimum when the animals have been dep rived of the other substance.
Food and water intakes are thus a function not solely of food and water deprivation, respectively, but of both these and of the substances available at the time of measurement. Speaking loosely, thirsty rats are hungry rats and vice versa, and the simple failure of S to take either food or water offered alone after the animal has been satiated with it, will predict not at all what S will do if presented with food and water together.
During the deprivation period, a food-deprived animal cuts down its water intake and effectively becomes water-deprived as well. Similarly, a water-deprived animal cuts down its food intake and becomes food-deprived. The present data show that effects of these self-imposed deprivations appear when measurements are made of food and water intake together in the hour following termination of deprivation.
We have performed two further experiments. In the first, we deprived the animals of water for 21 1/2 hr., then for 1 hr. made different combinations of food and water available to various groups of animals. In the hour following this, we measured food int ake. In the second study, a parallel procedure has been followed with food deprivation.
The second experiment was run through five days. The 12 Ss, the apparatus, and basic method were the same as those used in the first -ment. At the time this experiment began, Ss ranged from 201 to 204 days in age. The daily schedule was as follows: At 2200 all animals were given unllmited food but no water (f). At 1900 of the following day, 3 randomly assigned animals were given food and water (Group fw'), 3 were given food but no water (Group f'), 3 were given water but no food (Group w') , and 3 were deprived of both food and water (Group 0,). All food and water were removed at 2000, and at 2010 all annnals were given a measured amount of food but no water (f). At 2110 all food was removed; the amount eaten was measured. At 2127 all anima ls were given access to food and water with supplementary lettuce. Water was removed at 2200, the beginning of the new experimental period.
Table 3a presents for the four groups of animals the means and standard deviations of the quantities of food eaten in the 1-hr. measurement period. Significance of differences was determined by t tests. Since the preceding experiment led us to expect particular differences, these t tests were one-tailed. The amount of food eaten by the first group (w') is significantly different, at the .01 level, from that eaten by any of the other groups. Group fw' differs, as well, from Group f' at the .02 level. Other differences, although not statistically significant, fall in the a propriate direction and are of the appropnate magnitudes. No effects of age, or ot the sampling procedure, are evident.
The eating behavior of these animals is in remarkable quantitative agreement
with the behavior under analogous conditions summarized in Table 2. The
data conform with the predictions based on the results of the preceding
The data of Experiment III were collected by Mr. Ogden R. Lindsley in the course of another experiment performed with the senior author.
following period of 1 hr. measured the amount of water drunk in the absence of food. These procedures again define a deprivation period, an ingestion period, and a measurement period.
Subjects:. The Ss were 48 male albino rats, run in three squads. The first squad contained 16 animals of the Hisaw strain, 119 days old at the beginning of
||Available in 1-hr.
||food and water (fw')||
||food and water (fw')||
Experimental room and apparatus. Throughout the experiment the rats were housed in a small internal room with one door and no windows. Two fluorescent ceiling fixtures came on at 0530 daily and turned off at 1730; A coil-type heater fan, blowing ov er an open vessel of water, was turned on by a thermostat. Temperatures measured at the beginning of the daily experimental sessions averaged 71.7° F, ±3.0°, and the relative humidity averaged 67.6 per cent, ±5.0 per cent The cages, f ood, water graduates, and tubes were the same as those used in Experiment I.
Measurements. Drinking was measured with the same apparatus and methods as in Experiment I. In this experiment, however, reading accuracy was to the nearest 1 cc.
Preliminary handling. For 8 days the animals were housed in the room and stabilized on the temperature and light-dark cycle with free access to food and water. For the next 15 days the rats had free access to water, but access to food for only 1 hr . per day.
Daily schedules. The experiment proper ran for 25 days. Each day included the following operations: At 0530 the lights went on. At 1300 temperature and humidity were measured and the rats weighed. Water was next removed from pairs of cages at 6 min . intervals, and the experimental diet of the 1-hr. ingestion period was correspondingly given to the animals, according to their group memberships. General procedures otherwise conformed with those followed in the preceding experiment. The staggering was necessary because these measurements were made in the course of another experiment.
Summarizing, each animal was submitted to a deprivation period, 20 hr. and 50 min. in length, with access to water but no food (W). During the next hour, the ingestion period, the experimental groups received differential treatment. Followin g this, after a short delay, came a 1-hr. measurement period, when water alone was available. At the end of this time, each animal was given access to food for 1 hr., in addition to the water, and the experimental day was over.
Groups. Four of the 16 animals of each squad were
randomly assigned to one or another of four groups, which differed in the
diet given during the ingestion period. Group fw' received food and water,
Group f' food without water, Group w' only water, and Group 0' received
The experiment was run through the period November, 1951 to April, 1952.
No significant differences appeared in the drinking of rats of the Hisaw and of the Wistar strains. The data of all the animals under each treatment were therefore lumped and treated together.
Table 3b presents the data obtained during the first 5 days and through the whole 25 days of the experiment. One-tailed t tests of significance were made. Over the first five days, Group f' differed significantly, at better than the .001 lev el, from Group fw'. .Group fw' differed at the .06 level from Group 0'. Over the full 25 days, all three groups differed from one another at better than the .001 level. Through neither period did Groups 0' and w' differ significantly from one another.
The results are in substantial quantitative agreement with those reported
in Table 2. Group f' is effectively rendered thirsty (as evidenced by drinking
behavior) by the termination of the 22-hr. period of food deprivation.
These results confirm and extend findings already available in the literature we have cited. More important, they clarify the quantitative relationships involved.
There is reason to expect that noningestive habits acquired under food and water deprivation will behave much as do eating and drinking (7, 18, 19). More pertinently, Kendler (10) has observed a depression of food- conditioned responses when the animal is deprived of water as well as of food for 22 1/2 hr., and he has found, too, that food-and water-deprived animals learn a simple T maze more slowly than animals that have been deprived of food alone. Webb (22) has also presented data showing that water d eprivation controls habits that have been rewarded by food.
On the other hand, Miller, Bailey, and Stevenson (12) found that animals that have been subjected to operative techniques involving the hypothalamus may show a considerable dissociation between amounts of food eaten and the strength of behavior acquired u nder food deprivation with reinforcement by food.
The large size of the interaction effects we have found and the lack of strict reciprocity between food and water intakes after long periods of deprivation suggest reasons for the anomalous results in experiments on "latent learning" (6). The lo gical controversy over the behavior of "thirsty" and "hungry" animals loses much of its force when the assumptions that have been made, implicitly or explicitly, about the behavior of "hungry" rats toward water and of "t hirsty" rats toward food prove unjustified. That "low motivation" may be a condition necessary for the exhibition of certain classes of behavior may follow from our results on simple eating and drinking.
Simple formulations of "drive" and "drive summation" such as those that have been made on the basis of Hull's 1943 theoretical treatment (9) seem to require revision. Perhaps new classes of "drives"-"latent drives" or "concomitant drives"-need to be introduced. Perhaps it will be more fruitful if these endeavors are held in abeyance until further experimentation clarifies the phenomena associated with deprivation.
Several physiological mechanisms may account for these results. Perhaps a thirsty rat cannot very efficiently eat dry food such as we used. Probably more important is the role of water in the digestive and metabolic processes (3). It is not clear, however , that the physiologists are yet in a position to give us a reasonably coherent description of the processes involved.
Further experimentation will probably have two merits: it should enable
us to specify the conditions of deprivation and reinforcement associated
with efficient learning, and it should provide the data which seem to us
indispensable for further production of theories of drive or of latent
1. The amount of food eaten by water-deprived rats is some 60 per cent of the amount eaten by those which are not water-deprived.
2. The amount of water drunk by food-deprived rats is some 40 per cent of the amount drunk by those which are not food-deprived.
3. If a 21-hr. period of water deprivation is terminated, the amount of food eaten rises to some 6 gm., which approaches the amount of food eaten after a like period of food deprivation.
4. If a 21-hr. period of food deprivation is terminated, the amount of water drunk rises to some 7/10 cc., which approaches the amount of water drunk after a like period of water deprivation.
5. These increases in water and food intake are of the same order of magnitude, whether the measurement is made at the time the experimental deprivation periods are being terminated by eating or drinking, or in the period following such termination.
6. Considerable questions are raised by these results for some current
treatments of "drive" and for the interpretation, in terms of so-called
"irrelevant drive" and "high" and "low" motivation, of the results of la
tent-learning experiments involving the behavior of food- and water-deprived