\Hl‘lllll LL \ H TL-LE QUANTLFLCATLON 0F 003% L LNCENWE VALUES 0F F000 , D AND WNTFLL 7 Tho-é; fer F910 Mose? 'PthL _ MLCHLGAN STATE causes , Robe'iff H Davis I ‘ 1953 This is to certifg that the thesis entitled The Quantification of Drive, I: Incentive Values of Food and Water presented by Mr. Robert E. Davis has been accepted towards fulfillment of the requirements for m.— degree in M08? Date Febm! 2!. 1952. EH E8153. THE QUANTIFICATICU (F DRIVE I. INCENTIVE VALUES OF FOOD AND WATER by N 93’“ Robert H.” Davis A THESIS Submitted to the School of Graduate Studies of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR 0F PHIL$OPHY Department of Psychology 1953 Robert E. Davis The present study was conducted in order to establish incentive values for food and water, and in an effort to determine the feasibility of quantifying the drive construct once such values were known. A push-panel apparatus was constructed in which activity levels could be measured simultaneously with response amplitude and latency. Thirty-six male, albino rats were divided into two major groups, both of which were subdivided again into three groups. 1. High Drive:‘ is) Large Food Reward b) Small Food Reward (c) Medium Water Reward 2. Low Drive: (a) Large Food Reward (b) Small Food Reward (c) Medium water Reward Each of the 36 animals was habituated to the box, assigned to one of the subdivisions, trained to open the push-panel for either food or water, and then tested for a total of 40 trials, 20 trials under a high drive and 20 trials under a low drive. Half of the animals began their test series under a high drive and half began their test series under a low drive in order to counterbalance the trials. Activity level for six minutes before the exposure of the pushrpanel, and the latency and amplitude of each response was recorded. At the close of the test series, all animals were extinguished under either high drive or low drive. The results were as follows: 1. Latency: Amount or type of reinforcement was not a significant variable with respect to latency. The incentive value of small food Robert H. Davis reward, however, more nearly matched the incentive value of the amount of water employed. Such differences as do exist are largely confined to the first half of the test series. Early in the test series, latency appears definitely to be a function of the drive level under which it is measured, but not late in the series. 2. Activitv Level: Activity level offered some promise as on inde- pendent measure of drive. Activity shows a consistent upward trend through~ out the test series. This cannot be accounted for in terms of some general~ ized increasing drive but seems to he a consequence of learned anticipation. A significant negative correlation was obtained between activity level and latency. There was a significant difference between activity levels taken following long deprivation and those taken following short deprivation. Type of reinforcement was unrelated to activity. 3. Amolitude: The amplitude of the response as measured in the present study did not prove to be related to either the amount or type of reinforcement, or to the enount of deprivation. 4. Extinction: No difference we“ found between animals extinguished under high drive and those extinguished under low drive in number of respOnees to extinction. 5. One of ,he significant findinns of the study was the discovery ‘3 r that differences often appear to be a consequence of the point in the test series at which measurements are taken, rather than a simple func- tion of some variable such as drive or reinforcement. ACKNOWLEDGEMENT Grateful acknowledgement is made to Dr. M; Ray Denny for his advice and assistance throughout the course of this research. 111 TABLE OF CONTENTS Page I. Introduction ‘ 1 A. Drive as a Construct l B. Activity Level and Deprivation 2 C. Activity Level and Learning 3 D. Theoretical Interrelation Between Drive, Learning and PerfOrmance 5 E. Establishing Drive as a Construct 6 1. Problems Associated with Task to be Learned 7 2. Variation of Deprivation after Learning 10 3. The Independent Measure of Drive 11 4. Control of Variables Contributing to sEr 11 F. Purpose of Study 13 II. Subjects 14 III. Apparatus 15 IV. Procedure 22 A. Handling 22 B. Habituation 22 C. Training 23 D. Test Period 24 E. Extinction 27 V. Results 30 A. The Effect of Differential Reward on Performance 30 l. Latency 30 2. Amplitude 36 B. The Effect of Differential Deprivation on Performance 1. Latency 2. Amplitude C. Activity Level and Performance 1. Increase of Activity per Experimental Day 2. Activity and the Type of Deprivation 3. Activity and the Period of Deprivation 4. Activity and Latency D. Extinction 1. Number of Responses to Extinction 2. First Six Trials: Latency 3. First Six Trials: Amplitude VI. Interpretation A. sEr and Size of Reward B. sEr and the Type of Reward C. Deprivation and sEr l. Latency 2. Amplitude D. Extinction l. n as a Significant Indicator of sEr 2. The Increase of D Following Removal of Reinforcement E. Activity as an Independent measure VII. Summary Bibliography iv 43 43 43 46 A6 49 49 51 51 56 56 59 59 59 6O 61 61 62 63 65 69 Table l. 2. 3. 4. 5. 9. 10. LIST OF TABLES Deprivation Schedule for Hungry Animals in a L-H Group Deprivation Schedule for Thirsty Animals in a H-L Group Design of the Experiment Sums of Individual Latency Measures for all Groups Arrangement of Experimental Design to Comply with the Assumption of Independence of Latency Scores Analysis of Variance of Latency Scores of Three Groups of Subjects Tested Under Different Quantities and Types of Reward Arrangement of Experimental Design to Comply with the Assumption of Independence of Amplitude Scores Analysis of'Variance of Amplitude Scores of Three Groups of Subjects Tested Under Different Quantities and Types of Reward Activity Levels for Hungry and Thirsty Animals Compared in Terms of Hours of Deprivation Analysis of Variance Design to Test the Difference in Number of Responses to Extinction Between Animals Extinguished Under High Drive and Those Extinguished Under Law Drive Page 28 29 32 33 34 35 40 45 50 TABLE OF FIGURES Figure l. 2. 3. 4. 5. 9. 10. Cross-Sectional Drawing of Apparatus Discussed in Text Sketch Illustrating the Wiring of Timer to Guillotine Door and Push-Panel Side and Rear Exterior Views of Apparatus Described in Text, Showing Particularly Push-Panel, Metal Rod, Plastic Wheel, and.Feeding Tray Sketch of Watering Device Mounted Behind PushePanel Illustrates the Steady Rise of’Activity Level for all Animals over the Four Test Days Illustrates the Decline of Latency with Increased.Activity Illustrates the Decline of the Mean Latency on the Initial Extinction Trials Illustrates the Decline of the Median Latency on the Initial Extinction Trials Illustrates the Rise of Mean Amplitude on the Initial Extinction Trials in the Case of High Drive Animals Illustrates the Median Rise in Amplitude on the Initial Extinction Trials in the Case of Both High Drive and Low Drive Animals vi Page 17 18 20 21 39 47 52 53 54 55 INTRODUCTION A. DRIVE AS A CONSTRUCT Two fundamental issues in contemporary psychology concern the nature of drive and the nature of reinforcement. There can be little doubt but that the adequacy of any future learning theory will depend upon the extent to which these two concepts are understood and their conditions described. Although drive is frequently treated as a generalized drive state, its empirical definition in terms of antecedent conditions and resultant behavior is inadequate. To consider drive as a unitary, empirical construct implies that each measured value of drive will bear a constant relationship to certain behavioral consequences. In other words, regardless of the principal conditions under which drive is produced (privation of food, or water, or sex, etc.), the relationship of drive to the reaponse variable in question should be the same. Since the current literature sheds little light on this matter, it is possible that drive may not be a genuine scion“ tific construct. The behavioral consequences of deprivation may be distinguished as to: (l) the general activity level and (2} learned performance. Often, activity and performance are observed to vary concomitantly, but perform- ance shifts are also related to learning fatigue, etc.. It is, therefore, often held that general activity is a better indicator of drive then is performance (31).1 1It is important to note that activity amplitude does not necessarily reflect the "need" of an animal, however. B. ACTIVITY LEVEL AHD DEPRIVATICN A series of ectivity level experiments have been carried out over the pest several decades , end they have demonstrated that there is a. definite relationship between genersli zed activity level and the experimental meni- puletion of many features in the environment. mm. many reviews have been published in this area (18) (1’7) (36), the relevance of activity level studies to the present experiment make it imperative that we inspect a number of these studies and investigate the possibility of assessing drive level, not by the hours of deprivation of some relevant need, but by the general activity level which an animal manifests. Specific ectivities in animals tend to be rhythmic. Such specific activities have been extensively investigsted by Richter (21) (22). Where food is present at all times, for example, eating activity is still periodic, taking plece every 3-4 hours. Richter hes demonstrated thet drinking, uri- nation, defecetion, end muting, all are cherecterised by e certain rather specific periodicity. P. 2. Young (37) hes reported studies conducted in his laboratory demonstrating a. diurnal drinking pattern in the rat with periods of maximum drinking occuring in the late efternoon and at night. Actually, it might be ssid that activity level always shifts es a result of scene change in the physiological state of the animal, but fre- quently this shift appears to take pleee es the result of some specific change in the external world. For example, ectivity level is e function of temperature (13) (2), previous activity (26), end illumination (11) (3). In addition to these differences in general activity level which appear as a consequence of those physiological and environ-ental condi- tions sighted above, there are aany other conditions which have been correlated with activity. Along these are, for exanple: inheritance, endocrine condition, drug, cerebral inJury, opportunities for sexual outlet, etc. Finally, perhaps the most inportant and interesting cor- relations for our purposes have been established between activity levels and the deprivation of some material substance which the organist requires for its existence, particularly food and water. Whenever an animal is deprived of a needed substance its activity level rises. With reference to food it has been repeatedly demonstrated that Just prior to their regular feeding periods, animals become very active, even though this period,- if it were not regularly followed by food, would normally be an inactive one (21) (28). It has also been demonstrated that activity continues to rise up to about 96 hours of deprivation when an aniaal is deprived of food alone, but falls off at 72 hours if an aninal is deprived of both food and water(22) (32). warden (33), using an obstruction box, found that animals would cross a charged grid a greater number of tines at 24 hours of deprivation of food than at any other number of hours of deprivation up to 6 days. Held and Jackson (32) found that rats when deprived of many substances were in- creasingly active until satiatsd. C. ACTIVITY LEVEL AND LEARNING A nunber of years ago, several studies were conducted which bear directly on the problen of the relationship between activity level and performance. In two of the earliest of these studies no appreciable relationship was demonstrated (24) (l). A later study, conducted by Tuttle and Dykshorn (30), indicated a definite correlation between activity level and learning. The Na were small : 7, 6, 5, 7, 7 with correlations of .57, .60, .30, .61., and .82 respectively, and the data was derived from a study primarily concerned with the influence of gonadectomy. Rundquist & Heron (23) reanalyzed the data of Tuttle and Dykshorn. When groups were combined to achieve larger Na and controls were introduced for sex and gonadectomy, a correlation of .30 was found for males and an r of -.15 for females between activity and learning. Using a different measure of learning, larger groups, and a t-test for significance rather than a Pearson r, Rundquist and Heron demonstrated marked differences in the learning and performance of active as com- pared with inactive rats. in interesting aspect of the Rundquist and Heron study in terns of the present investigation is the fact that these investigators shifted drive levels at the 23 and 32 trials in the series of 1.0. On the 23 trial the daily diet was cut to one half the normal'idiet and on the 32 trial it was raised to one and one-half times the normal diet. While this resulted in no significant shift in errors for the inactive animals there was a significant drOp in errors for the active animals on the block of trials 21. through 31, as well as the block 32 through 1.0. This was particularly true of the block of trial 24-31, on whichthedailydietwascutinhalf. There would appear to be evidence here indicating that (1) active animals learn better than inactive ones, (2) active animals perform better than inactive ones, and (3) that shifting the deprivation levels Ir ,1 of active animals causes a marked decrease in errors and a significant increase in performance levels. D. ‘IflEORETICAL INTERBELA‘IIQI BETWEEN DRIVE, LEARNING, AND PERFWCE It has been pointed out then that deprivation has an influence on activity level, and that activity level, in turn, apparently is func- tionally related to performance. Accordingly, it might now be asked, "How are these three complex factors interrelated?" It is possible that the systematic investigation of deprivation, activity, and performance might ultimately lead to the developaent of a drive-construct, which would resemble these constructs found in the more exact sciences. If it could be demonstrated, for example, that regard- less of the technique which was utilised to produce drive, a measure could be established which would predict its measurable consequence in perfonance, then "drive“ would assume the status of a genuinely valid scientific construct. Performance (sEr) is said by Bull to manifest itself in four ways, and was originally said to have been a function of drive (D) times habit strength (sEr). Accordingly, in any situation in which sEr was manifested, Hull's original postulates asserted that this sEr resulted from the inter- action of at least those two complex factors. In order to determine which of these two factors influence an observed behavioral change, it would be necessary to control the other. he control of er involves the manipulation of environmental variables which have not thus far been discussed. ‘Within Hull‘s original system, er is assumed to be a function of several variables, particularly reinforcement. Hull (9) stated in his Bringiplgs that both the kind of reinforcement and its amount were learning (er) variables, influencing performance through learning. The empirical verification of this assumption has not been accomplished. Hull's reformulated postulates, therefore, now express the strength of er as a function of the number of reinfOrcements exclusively, and have recognised the influence of variation in amount or type of reinforcement by giving it the status of a "motivational" variable. In so far as learning is concerned, reinforcement is said to be an "all or none" affair. Performance, however, is apparently a definite function of the amount and type of.reinforcement. Effective reaction potential, at the termination of learning, is now believed to be a multiplicative function of a negatively accelerated increasing monotonic incentive function (K), drive (D), stimulus intensity (V), delay in reinforcement (J), and habit strength (sEr), i.e., 5m: vaxx'XJXaHr. E. ESTABLISHING DRIVE AS A COHSTRDCT [A critical question which arises in connection with this formula and the status of drive as a scientific construct is this: if performance is a function of all of these things, then how are we to detect the influr ence of drive alone on sEr? And, in this connection, the most difficult problems arise relative to the incentive function, X. If the aajor problem consists of defining the conditions said to influence drive and of'measuring the effect of these on perfOrmance (‘D /‘I so as to establish drive as a construct, then what.mnst be done to accomplish this? Briefly: 1. Animals must learn some prescribed task. 2. Once animals have learned this task, drive leve1-as produced by different types and degrees of deprivation-must be varied. 3. Ideally, drive should be measurable in units which are inde- pendent of those involved in its production. 4. Finally, throughout the experiment, the factors of‘V, J, K, &:er must be controlled. Let us consider these requirements individually and in greater detail. 1. mmmmmmmm.umg(am or habit strength must first be established and then held at a constant level in all groups. To accomplish this, animals must be trained for an equal number of trials in the performance of some specific task. ‘What levels of drive and reinforcement should be used to accomplish this? 6 There is a growing body of evidence to indicate that drive level, as well as amount of reinforcement, is not a critical learning variable. ‘With.reference to the drive level under which a task is learned, Finan (6) in 1940 published evidence to show that rats learned better under 12 hours of deprivation than at any other level. Animals were trained to a criterion of 30 reinforced trials under different levels of depri- vation and then extinguished with all animals under the same level of deprivation. ‘Nhile animals trained under 12 hours of food deprivation required more responses to extinction than those trained under 1, 24, /§ or 1.8 hours, the differences were not statistically significant except in those comparisons involving the 1 hour group. Actually, the evidence here for any real differences between groups is thus extremely limited. In 191.5 , Kendler (12) published a study in which a relevant drive (hunger) was held constant at 22 hours and an irrelevant drive (thirst) was maintained at two different degrees, 12-22 hours, during learning. No evidence was obtained to show that degree of deprivation during learning effected the number of bar pressing responses to extinction. In view of the contradictory implications of the studies of Kendler and Pinon, Strassburger (29) undertook to do a genuinely definitive study of the problem. Hhile the study which Strassburgor conducted was essentially a replication of Finan's, it was expanded, and he attempted to assess the strength of sEr after different numbers of reinforcements, as well as under different drive levels. The general conclusion of the study was that, although response was definitely a function of the number of reinforced trials during learning, no consistent relation existed be- tween degree of hunger in conditioning and resistance to extinction. From these studies, it can now be inferred that the problem of . , what drive should be used during learning, offers no particular diffi- culties. Deprivation level can, of course, be held constant through- all groups during this phase of the experiment. It is interesting to note in this connection that it follows from the above studies that the results obtained on a stuw such as the one proposed here would be applicable to studies in which animals were trained under different levels of drive. ,1. - The same is largely true of reinforcement during learning. Our second problem in connection with the strength of er as it partiéipated in the formula for sEr concerned the quantity of reinforcement which should be utilized during learning. while the evidence in this case does not lend itself to a straightforward interpretation, it now appears that learning is not influenced by the amount of reinforcement available to the animal per learning trial. Performance, however, definitely appears to be a function of this variable, and Hull's revised formula for sEr is the explicit recognition of this fact. The new forumla holds that er depends only on the amber of reinforced trials, whereas sEr is equal to the habit structure times certain other factors, one of which is K or quantity of reinforcement. Grindley (7) demonstrated as early as 1929 that speed of running was related to the amount of reinforcement given an animal per trial, and in the years which followed Grindley' s original study, several additional experiments were published in this area. Cowles and Nissen (4) correlated delay interval in chimpanzees with size of reinforcement. A later abstract, published by Fitts in 191.0 (38) indicated that animals given 1 trial per‘day and rewarded with 10 grams per response to a bar pressing apparatus required more responses to extinction than those rewarded with .2 grams. hem these studies, it appears that amount of reinforcement has a universal influence on the rate at which learning takes place. Volfe and Kaplon's stud? (35), published one year after the abstract reported by Pitts , casts considerable doubt on the universal application of this 10 assumption. With certain procedural modifications and using pop corn rather than rice, Wolfe and Kaplon repeated Grindley's original study. They demonstrated that one large piece of pep corn and a piece one- quarter its size were equally effective in producing lower running times, but that four one-quarter pieces were more effective than one one-quarter piece alone, and that four one-quarter pieces were even more effective than one large piece. The problem of the amount of reinforcement and its role in learning was finally systemtically attacked in 194.2 by Orespi (5) . Using a long runway and large differences in the amount of reinforcement, Crespi found significantly smaller running times in animals with the larger incentives. These differences in running times characterized both the learning and post—learning periods. Furthermore, it was demonstrated that a shift in incentive caused a corresponding shift in running time, so that, for example , large-incentive-fasts-running animals, when shifted to smaller incentives , were observed to reduce the speed of locomotion, and vice versa. Recently, it has become increasingly evident that while speed of response is definitely a function of the amount of reinforcement, it is doubtful that learning, no: 3,9,- is correlated with this factor (19). These studies make it reasonably evident that the amount of rein- forcement, as well as the level of drive, is not a critical learning variable. 2. m m 93 W19! winning. Because our second requirement involves the manipulation of an independent variable , no control difficulty is offered by the degree of deprivation. ‘ . l - = . - \ _ - . z - - I - . ‘ II I n. : I - II; { l '- ' .-I- ‘ i ’ l' c - I I I ~ . c ' ' t . .. ._ __ - l g - t _' . 1 ' . ' .- a -. "- I t . .. 5-K! .-: -'-= t r1 '. '1 . r3 . ' ) ;:— I uni. : I. F i . |lj c ' U 0 . . - _ . ' - .u: - -. ' — " “" "- t ' " 3. n: independent m of 31:11:. It is desirable that drive should be measured in response units which are both serum to various types of deprivation and independent of the conditions involved in its production. These units should be so correlated with behavior that a knowledge of them will permit a trained observer to predict behavioral consequences. The studies reported in the area ' of general activity offer some evidence that these units might be supplied by an activity measure obtained either immediately before or during the learning and performance situations. 1.. m m1 at W W 1.9 m. 'me control of habit strength or er Ins been considered in (1) above. Attention was directed to the drive and reinforcement levels under which learning should take place (which should be used during learning). It is assumed that this factor can be adequately controlled by observing the considerations previously set forth. Stimulus intensity dynamism (V) is controlled by the fact that all animals learn and respond in essentially the same environment. Delay of reinforcement (J) Operates to influence sEr as a consequence of the time intervening between the occurrence of the response and the reinforcement received. Since all animals would be trained under identical drive and reinfercement levels, there is little reason for assuming that J would differ from one group to another. The importance of the incentive function (K) has been considered with reference to the influence of different m of reinforcement on learning and performance. With reference to amount, the difficulties are not particularly formidable. But, it will be recalled that ideally drive . I \- should be produced not only with different amounts of deprivation, but also with different m of deprivation, e.g. food and water. This brings forth the last and most difficult obstacle blocking the way to a thorough and systematic attack on the problem of drive as it influences animal behavior. It can readily be seen from the formula for sEr that in so far as quantity is concerned, this variable could be held constant from group to group and thereby controlled. But, the problem is to investigate the conditions said to influence drive and these conditions include more than just one type of deprivation. Although many other conditions are believed to influence drive, it would be both impractical and possibly even impossible to manipulate all of these, but the deprivation of water, as well as food, would certainly seen to be both feasible and desirable. This being the case, the important question which now arises is: what quantity of water shall be used so as to be equal in reinforcing value to what quantity of food? The influences exerted on behavior by different deprivation levels of these substances will never be comparable unless we are certain that the quantity of reinforcement utilised to investigate learning and performance under different types of deprivation are of equal reinforcing value. In other words, everything must be con- trolled in the situation with the exception of the independent variable, D. This can only be done by substituting values of K which are constant from one type of deprivation to the next. There is no evidence in the literature bearing directly on this problem. n. . s l\ ._. . ea I I. 1 I .- l l . l m n . . . s . n. I It ..... . l l- I l u . J . . . .. . n u . . s . . . . t Accordingly, we here are presented with a question which must be answered before the larger problem of validating the drive construct can even be attacked. F. PURPOSE The purpose of the present study is twofold: (a) To determine the effects, which different quantities of food reward have on the learning and performance of hungry animals in a panel pushing apparatus as cempared with the effect of a specific quantity of water reward on the learning and performance of thirsty animals in the same situation. (b) To assess the extent to which the general activity level of an animal~~as measured in the experimental situation--can be used to predict the strength or amplitude of a learned response, the speed of a learned response, and the number of such responses to extinction. II. W The subjects in this experiment were 36 male albino rats from the colony maintained by the Department of Psychology of Michigan State College. Ages ranged from 85 to 125 days, with a mean of approximately 90 days. 15 III. APPARATUS In order to investigate the present problem.an apparatus was con? structed in which it was possible to measure activity level, response latency, response amplitude or force, and the number of responses to extinction. It consisted of a 1/2" plywood box with overall dimensions of 20" x 16" x 11". Figure 1 presents a cross section of the apparatus. At the bottom of the activity chamber there was a false floor which was supported by 3 springs and a rubber ball at its exact center. it the fear corners of this false floor, small, attached, rubber balls served as stops, preventing the floor from tipping any more than l/A”. A guillotine door at one end of the activity chamber, when raised, gave access to a hinged, 4" x.2' panel. This panel was constructed of a thin rectangular piece of wood, l/l6" diameter; at the upper end of the panel was a small piece of l/2" plywood 2-l/2" long, which farmed a base for the hinge and brass strips. (See figures 1, 2, and 4.) The flat grey interior was illuminated by a 7-watt bulb which was situated at the end of the box Opposite to the guillotine door and was covered by a piece of opal-flashed glass. Entrance to the box from.the top was gained through a 10 3/4" hinged door, in the center of which was placed a large clear-glass, observation window. 16 Activity level was measured by a device consisting of a GE 2-3651 mercury switch, suspended vertically beneath the false floor and con- nected in series to a Gorrell and Gorrell 115 volt electric counter. The mercury switch was situated beueath the floor in such a manner that movement by the animal caused the liquid in the tube to move, momentarily making and breaking the circuit in accordance with the strength and number of movements. (See figure 1.) A thin metal rod, hinged at the top of the panel was twisted so as to extend to the back of the panel in one direction and to the tap of the box in the other. The rod was so designed that the lower half of it “rode" back on the panel as it was pushed Open, and the upper half came forward toward the activity chamber. By means of this rod’, the force applied to the door was transmitted to a slender stick of wood which was attached to a light, plastic wheel, mounted on a plastic axle. The force of the response, which was applied to the door, was thus trans- mitted into the movement of the wheel. The distance which this wheel moved was measured in degrees. Because the wheel offered very little resistance to the metal red, the initial movement of the push-panel invariably caused it to turn out of range of further movements of the metal rod. That is, the degrees which the wheel was displaced depended upon the 29229 with which the door was m struck, and not merely upon the distance through which the door was moved. Response latency was measured by a Standard Electric Timer, con- nected in series through two switches. The first switch consisted of two brass strips, one being placed along the tap of the push-panel, and 17 HE E membome mafiamdmmim .8 ozbfidm fiéoHHQMmlmmomu 2m MMDCHM sopanm hence: $90th mo uooam omaom make wcfipoem Hosea sovasm Henna nonsmco haabwvo¢ mamas Hoe£3 moo: mum on com Hope: a H as Hoes: owpmmam noon ocesoflflesu mason noon oseeoaaeso Loafizm Ioooar \ Guillotine Door Frame l Microfiwitch \ 1 \\ Jamb \ I \ Panelfiwitch Standard Electric Panel Timer M Z Plug FIGURE 2. "‘-‘~<’t-*;’_i.‘"_"--i If-LUSTRATING 'I'HE xiiifll‘} OF TIMER TC .4 .3...“ ‘ GUST“: ”ZINE DOOR AND PUSH-Pgmy, 19 the other being attached to the upper pushrpanel jamb, directly above the strip on the pushrpanel. A second switch was attached to the top of the guillotine door. When the guillotine door was raised, this switch was closed. Thus, when the pushepanel was closed and the guil- lotine door open, the circuit was closed. when the panel was opened by approximately 1/32", the switch was opened. This is illustrated in figure 2. Thus the timer started when the guillotine door was raised and stopped as the panel was being pushed Open by an animal. ~Single reward pellets were placed on a tray which was located approximately one quarter of an inch below the lower panel Jamb. Metal walls were built up on either side of this tray to discourage exploratory behavior. The corners of these walls were bent towards the door, ferming steps to prevent the animals from.forcing the door and breaking it. This also reduced exploratory behavior. 0n the tray itself a small wall of solder was constructed to hold the food pellets in place. These cons struction details are illustrated in figures 3 and 4. Water reward was administered through a curved tube supplied by a standard water bottle which was retractable. This is illustrated in figure 4. The water bottle was mounted on a rectangular piece of wood which was attached to a length of l/Z" dowelling. The dowel was inserted in slots, which were cut in both sides of the box behind the pushepanel. When the dowel was rotated, the attached bottle also turned. ‘water reward could thus be presented or retracted by the experimenter. 'e e w u . - loi . - a , _. _ e . "' s u . 9. .1. e ‘- 9 . . . 4 a I a l - - - a I n. on ei- 0 w p a . _ , e . ‘l t I . . I .. . .; . i. - .l. :- l ..' .‘_ Lie-e *3 u e . - e ‘- '..--w ,1... ' . :- . -'-' I . I ' I: .- x I- .- . V. ."a - .. _ _ . .e. I..- , u . .I. u . . - ‘. : i- -.- I. = u. . I I... e. . [-- . u ' ‘ . I . . - .- .I M .. . e .- _. a I. . I I .L_'_-- ..‘_. e _ . "t- ' ' .l -. 0+}..- :.f ' bird“? . -.--.-‘ "-m ‘ div-53:"! "be" Just '71 :..L1' "tr-nit :. ! .L H D '-'-= --:'1'-"- 20 :5 czmmalmm 53 Same; Damian» eqom 3.9m: .smmznmmcmmooe waraiemomafiem 0238mm .HE E QmmHmommm mbagiamd n6 rim?“ monmafim rémm 3?. am . I A (r \ (1‘: E? L": 0‘ ‘34 V A b .H =[ [I ll is _ _ FLLI \ \§ ‘\ \ 3.2.2 f \. »( -.,A./ — f , // l " FIGURE 4. SKETCH OF madame 331311-22: MOUNTED BEHIND PUSH=PANEL .21. 22 IV. .EBQQEDHBE A. HANDLING All animals were handled for three minutes a day for three days prior to the pretraining phase of the experiment. After handling, animals were placed on a large flat surfaced table and allowed to explore for a period of 30 minutes. During this entire three-day period, food and water were available at all times in the home cage. B. HABITUATIW Animals in the present study may be conveniently divided into two major groups: (1) a thirst group (Nu-12) and (2) a hunger group (DI-21.). Each day for five days, animals from both groups were placed indivi- dually in the activity chamber of the drive level box for a six-minute period and their activity was recorded. Following this six-minute activity period, animals were placed in individual feeding cages where the relevant reward was made available after a delay of six minutes. In the case of hungry animals, this reward consisted of pellets of the same composition as these later used as reward pellets. For the first three habituation days, animals remainéd in the indi- vidual cages for 30 minutes following the introduction of the food and water and were then returned to the home cage where 11. grams of food per animals and water were available. On the fourth and fifth habitu- ation days, hungry animals were placed in the individual feeding cages -'.t'. . "1 - fl .' - can . . l iinu-lm .. 1.5-? 141 23 where-after a sixeminute delayb-their standard reward pellets were made available for a period of 15 minutes. These were followed by a wet'mash, composed of 60% water and 40% ground Purina dog chow by weight. This mash was made available for an additional 15 minutes. Once or twice during this period, the mesh was removed from the individual feeding cage and stirred. Following this, these animals were returned to their home cage where water was available for the next 24 hours but not feed. In the case of hungry animals, this procedure insured against building up a large residual drive. 0n the fourth and fifth habituation days, thirsty animals were placed in the individual feeding cages with Purina pellets immediately available, and water was made available after six minutes. These animals remained in the individual feeding cages for a period of.45 minutes and were then returned to their home cage where feed was available for the next 24 hours, but no water. Both hungry and thirsty animals were thus under a 24 hours relevant drive when introduced into the activity chamber on the fifth day of habituation. C. TRAINING For a three-day period, animals were trained to open the pushrpanel. Once per day, each of the anomals was placed individually in the activity chamber for a period of six minutes. At the close of this sixeminute period, the activity level was recorded, the guillotine door was raised, and the pushepanel was presented. This was done on each of the three training days in accordance with the following schedule: w L; 'il' .,..-1- -- r. Day 1: Door fully open for eight trials; door open l/i” for two @1818 e Day 2: Door fully apen for two trials; door open l/L" for eight trials. Day 3: Door open l/L" for two trials; door fully closed for eight trials. Hungry animals were presented on each of these trials with reward pellets of a.medium.size (.20 gnu).2 Thirsty animals were given access to the drinking tube (See figure 4) and allowed to drink for five seconds per trial. Deprivation conditions for both groups were set up in the same way as on days four and five of the habituation period. That is, all animals were trained under a deprivation level of 24.hours. At the close of the third day's training, animals were placed in the individual feeding cages and treated as outlined above. Hungry animals were fed pellets for 15 ‘minutes and mash for an additional 15 minutes. Thirsty animals were given access to food and water for 45 minutes. D. TEST PERIOD It was desirable that all animals be tested under high and low drive levels. Furthermore, these levels were to be roughly comparable for hungry and thirsty animals. It was also desirable to arrange the study so that low drive animals would be sufficiently motivated to respond to the door fer the full number of test trials each day. The number of test trials was set at 10 per day for fear test days. Each 2Large pellets infthe present study weighed .32 gm., and small pellets weighed .08 gm. n . a . . .) , , r w . ) ~. .1 . T x -. ~ ~.‘. . .. H... 25 animal was given a total of 40 test'responses, 30 successive trials at.a low drive level, and 20 successive trials at a high drive level. Animals ‘were counterbalanced in terms of the position of the blocks of twenty ‘high and low drive trials. Half of the animals were tested under high drive first and then shifted to a low drive, and half of the animals ‘were tested under a low drive first and then shifted to a high drive. In order to insure that animals Operating under a low drive would respond for the full ten trials, it was necessary to empirically deter- mine the minimum.length of time which must elapse following satiation before animals would make the necessary 10 test responses consistently. This value was determined to be five hours in the case of water depri- vation and eight hours in the case of food deprivation. While many studies have apparently demonstrated that animals will respond under fewer hours of deprivation than those used in the present study, none of these studies have employed as strict a criterion of performance as was required in this experiment. Furthermore, the animals in the present study were under no residual drive and all measures are from.ggnnlgfig satiation. It is interesting to note that, though some animals responded once or twice to the pushrpanel at drive levels below those employed, the problem consisted in finding a level which, while of a relativoiy low value, was still one at which all animals would consistently respond. Even at the relatively high number of hours of deprivation for "low" drive employed in the present study (8 hours hunger; and 5 hours thirst), two hungry animals, which were begun on the low drive, failed to complete all 10 responses on one of their two low drive days. W") 26 A further difficulty in the present study was the inevitable resis- tance offered by the push-panel. This resistance resulted from the attached device for measuring amplitude and may have tended to inhibit responses when the drive level was at a low value. The 21. hungry animals were divided in two maJcr groups, which were further subdivided for the purpose of counterbalancing. (l) Lg (H-L): (2) Lg (L-H): (3) S!!! (3.1-) 3 (4) Sn (II-H): Animals in this group received one large pellet (.32 gms.) on each of the 1.0 test trials. 0n the first two test days (20 trials) these animals were tested under a high drive and on the last two test days these animals were tested under a low drives is in the case of the above group, this group received one large pellet for all forty trials, but drive conditions were reversed, and animals were initially run under a low drive. This group was treated exactly like the Lg (Ii-L) ‘ group except that on all 1.0 test trials, these animals received a small pellet of food (.08 gls.). his group was treated exactly like the Lg (Ii-L) group except that on all 1.0 test trials these animals received a small pellet of food. The 12 thirsty animals were given only one level of reinforcement but were split into two sub-groups for the purpose of counterbalancing. (5) Mi (H‘L) : (6) Md (Ir-H) : This youp received a reward of five seconds drinking time per trial on each of the 40 test trials. 0n the first two test days (20 trials) these animals were tested under a high drive and then switched to a low drives This group was treated exactly like the lid (H—L) except that deprivation levels were reversed. All animals were run only once a day for 10 trials between the hours O‘BPO Me 811(1le be 27 Deprivation scheduling for hungry animals in a LPH group is illustrated in table 1, and deprivation scheduling for thirsty animals in a H-L group is illustrated in table 2. E. EXTINCTION For the purpose of extinction, all major groups were subdivided equally into two sub-groups, and the animals were extinguished under either or high or a low drive. For example, three of the animals in the Lg (HéL) group were extinguished to a three minute no response criterion under 48 hours of deprivation and three were extinguished to a three minute no response criterion under 8 hours of deprivation. Six minute activity levels were recorded before the extinction series was begun. The latency and amplitude of every trial was recorded. In some cases the activity was taken at two minute intervals throughout extinc- tion. An error in assignment caused four of six animals in the Lg (H-L) group to be extinguished at 48 hours rather than three of six. In order to equalize the number of animals extinguished at 8 as compared with 48 hours, it was necessary to extinguish feur of the six animals in the Sm.(LeH) group at 8 hours. During the entire process of running animals, it was the practice to assign animals randomly to groups. Systematic error was avoided to an even greater extent by running animals from the various groups simul- taneously. . .- n I. I. - e-_. u 28 TABLE 1 DEPRIVATICB SCHEDULE son HUNGRY ANIMALS IN A L-H GROUP Number Condition Day of Study of Feeding Regimen Trials Training: 2/. hrs. First 10 Fed to satiation 6 Drive Second 10 minutes after being Third 10 trained. Test: Low Drive (8 hrs.) Fourth 10 Fed to satiation every Fifth 10 running day 8 hrs. be- fore testing and 6 min- utes after testing. High Drive (4.8 hrs,)Seventh 10 Fed to satiation every Ninth 10 running day 6 minutes “tar “at trims htinction 8 or Tenth or Fed to satiation 8 hrs. Variable before running. [.8 hrs. drive Eleventh Fed 6 minutes after last run and extinguished 1.8 hrs. later. W TABLEZ IEPRIVATION SCHEDULE FOR THIRSTY ANIMALS IN A H-L GROUP Number Condition Day of Study of Feeding Regimen Trials Training: 24 hrs. First 10 Fed and watered to Drive Second 10 satiation 6 minutes Third 10 after every day's training trial. Test: High Drive (36 hrs.) Fifth 10 Fed and watered to Seventh 10 satiation 6 minutes after every r day and fed and watered to satiation at noon on days four and six and run 36 hrs. later. Low Drive (5 hrs.) Eighth 10 Fed and watered to Ninth 10 satiation six minutes after test on days 8 and 9 and 5 hrs. before on these some days. Extinction High (36) or Eleventh or Variable As above. Low (5) Tenth .- .. . . .-. . .... . . .. .a...... —--_-- -.a-..—.-. «n... u... . ‘...— - 'iul-fl-‘F ..,....... .__ . . .. .9 a..." -- n. -. .— quanta-o .. a .. -u.n.-.a—~-—--.—-.--.- . . . _._.a ._._- - . u- -:.-.l. - -.-..-.... . ...........a.... -... ..-. .-....‘.._— . a.-- no. --¢_-... . . .. .— . a «l -~- -. . I; 05.. ii. l'e- ' - ago-,5 . ' -‘ .-'. . .- s . .-.‘ ----"e . a . '. I . . a . _s «I... . I . g. . . . l ' - l- ‘ I"- r- . ' ' ‘ , .' 1' . . . ‘... .) - . . . . . - . .- p . _ e. I a . . a. _I.'. I _ - _ ) .. I . .. __) .. . . . .5 . a... - —.. .. . . ,..- ... .... .. n—.---.._.-.... . .- -. .... .........-... .... . .... ... .—........, ......—--- 30 V: BhaflLTS A. THE me! or DIFFERENTIAL mm on PERFORMANCE l. m. Table I. sumarises the latencies for all 36 animals. Each block represents the performance of six different animals over a series of twenty trials. It can be observed that the values range from a total of 196.1 seconds for thirsty animals, which were begun under a high drive and shifted to a low drive, to 1068.2 seconds fer thirsty animals which were begun under a low drive and then shifted to a high drive. The lar- gest total difference is between the 12 animals which were run to a large food.reward and the 12 animals which were run to a water reward. While the general trend of the data fer animals rewarded with food is in accord- ance with our theoretical anticipations, that is, small reward animals exhibited considerably longer latencies than large reward animals, it was necessary to test the significance of this difference. Likewise, one of the major aims in undertaking the present study was to discover a quantity food reward which very nearly equalled a water reward of a fixed value. The differences between these total latencies were tested to determine which-if either-of our food rewards is‘ngt statistically dif- ferent from.the water reward. The most obvious statistical tool to apply in.the present problem was an analysis of variance. One of the basic assumptions in the use 31 of analysis of variance, however, is that the measures are independent.3 An inspection of table .3 reveals that the individual measures in this experimental design did not comply with the assumption of independence. The design, however, may be rearranged in such a way as to compare the m for any given subject or group of subjects, and this results in a simple factorial design of the form illustrated in table 5. Since the subjects are randomly assigned to groups and measured independently of one another, their sums may be compared without violating the assump- tion of independence. Table 5 illustrates the comparisons which may be made legitimately when the influence of trials and drives is neglected. The results of the analysis of variance applied to this design are sumari zed in table 6. 3A second assumption involved in such an analysis is that the variance is homogenous. Bartlett's test of homogeniety of variance was applied and an uncorrected 12 of 7~..829 was obtained. TABLEB DESIGN CF THE EXPERIMDIT 32 large Food Small Food Hater L-HBB 34 35 High Drive 1.0.0.0000000000 210.00.00.0000000 10.00.000.000... 210.00.00.0000000 1.0.0.0000000000 210.00.00.0000000 Low Drive Trials 20 21000000000000... 40 40 leeeeeeeeeeeeeee 20 20 ZlOOOOOOOOOOCOOOO 40 40 leeeeeeeeeeeeeee 20 20 21....00000000000 40 40 leeeeeeeeeeeeeee 20 .‘L IsleonCIIIOIOO .eee-leiveltln elmmmleeeeeease recessions-lose sene-OQs-eeetll law-VOIIOIOIIII ueqlleovecespee Illnesses-pleas assets-eeeeeela IQOCDOCOIQIQIIC TABLE4 SIDE OF INDIVIDUAL LATENCI MEASURES FOR ALL GROUPS 33 High Drive Low Drive 0 H-L 318.8 313.6 "2 8w 3 3 g L-H 301.4 411.1 3 .4 H-L 306.5w 203-3 ‘3 §§ an 38 ‘5 m a. 209.9 7.2 :3 H H‘L 196e1 298e6 “2 3 {‘2‘ g L-H 232.6 1068.2 .-. 0 1565.3 3182.5 3 O H 34 TABIES ARRANGWT (F EXPERIMWTAL IESIGN TO CG‘IPLY WITH m ASSUWTIQI W INDEPDIDENCE CF LATENCY SCORES High to Low Low to High g : 632.4 : 712.5 a. 39 : 6 : 6 g : 510.3 1: 1097.1 m. a 2 6 z 6 no :3 : 6 : 6 at .u .9. a." . mp ." m. e s q 6... . VI .- . u. n. .. u.. .n : . .. . 35 TABLE 6 ANALISIS OF VARIANCE OF LATENCY SCORES OF THREE GROUPS OF SUBJECTS TESTED UNDER DINNERENT QUANTITIES AND TYPES OF REWARD SS df MS F P Total Between Subjects 4978.99 35 142.26 1.59 Groups 2298.10 5 459.62 5.14. .01 Procedure 1506.75 1 1506.75 16.86 .01 Size 213.41 2 106.70 1.19 P X S 577.94 2 288.97 3.23 Same Groups (Residual) 2680.89 30 89.36 __ u ‘- .‘qI .. . ‘ . .. K. a ".r- . . ' O O c ' 1 | I‘d“ . '_ I I "’5 5h E I 4 . ~ n 1 '9 v... .- ‘. ' I y. 36 Two of these variances are significant at the .01 level of confidence: Procedure and.Groups. This may be interpreted to mean that, regardless of the size or type of reward, animals begun under the high drive and than shifted to the law were significantly faster than those begun under the low drive and then shifted to the high. Despite the relatively large difference existing between animals rewarded with water and animals re- warded with a large food pellet, this difference did not prove to be statistically significant. With.reference to the difference between the individual groups, it is of interest to ask between which of any two of the groups in table 5 is the difference great enough to warrant the conclusion that they are significantly different from.one another. Applying the formula diff (M1 - M2) 8 2N (Error) the to our data, it can be shown that a difference of 534 seconds between any two groups is necessary to reject the null hypothesis at the .01 level of confidence. Such a difference can be feund only between animals rewarded with a large pellet in the low to high drive group and those rewarded with water in the low to high drive group. 2. Alnlitnde. Corresponding to each latency measure, there is a measure of amplitude. Thus we have two parallel problems, one concerned with amplitude and the other with latency. Everything relative to experimental design, which has been said thus far in relation to latency, applies to amplitude. Table 7 summarizes the results, and the analysis of variance for this parallel problem is given in table 8. 37 It will be noted that none of the observed differences is significant beyond the . 05 level of confidence. The large difference between animals rewarded with water and those rewarded with food is, however, noteworthy. Water reward animals appear to apply a stronger pressure to the push-panel than do the ether animals. This may be the result of an arti- fact in the design of the apparatus or it may represent a true difference between the subjects which were tested. Two considerations may help the reader to arrive at a conclusion relative to this problem. Figure 1. illustrates the method whereby water reward was administered. From this drawing, it can be seen that the tip of the fired and of the watering tube was, in the case of water-rewarded animals, immediately over the point at which hungry animals found their food reward pellets. The push-panel, in other words, had to be Opened in either case approxi- mately the same distance, and it would therefore seem that the initial pressure applied was not controlled by a difference in the spatial point at which the twa types of reward were found. A second way, which might be suggested to account for the difference between animals is in terms of the manner in which reward was administered. While hungry animals obtained their pellets and retreated to eat them, thirsty animals had to hold the push-panel open while drinking. During this time in which thirsty animals held the push-panel open, the push- panel obviously did not remain perfectly stationary. It might be claimed justifiably that thirsty animals in holding the push-panel open caused the amplitude indicator to move forward. The peculiar arrangement of . . § 1 t. a. .. e. . . .. _ . .. II. - . . A I! u -\ i I .r . . . - the anplitude measuring device makes this seem unlikely, for once an animal had struck the door, the projecting thin wooden spoke on the plastic wheel invariably swung out of range of further movement of the metal rod attached to the push-panel. 38 L TOTAL ACTIVITY FOR 36 ANIMALS FIGURE 5. ILLUSTRAT‘E‘S THE STEADV RISE OF ACT... ITY LEVEL FOR ALL 22000 2I000 20000 I 9000 l8000 I 2 3 4 TEST DAYS ANIMALS OVER THE FOUR TEST DAYS 39 TABLE 7 ARRANGEMENT'GF EXPERIMENTAL DESIGN T0 COMPLY WITH THE ASSUMPTION OF INDEPENDENCE OF AMPLITUDE SCORES High to Low . Low to High. Total Large Food 6848 7448 14,296 Small Food ; 8031 9476 17,507 Water 16051 9382 25,433 30,930 26,346 57,236 - z . . - . . . - - . ., . . .= . I. -. . . . . . . .. . . . .- -... .1“... ----..._,. ....-.... —.... . . .. -.. . . ... . . . .1. ._ _.. ., ._.,. ..... n-i- an... “unravel-I, - .. - . I - I. u. . u - -. i v I I 4 ll . I -- I! l-I‘ e - I" - -A mast—- ’- a m — y A II O-Dfllm wiflfl I. - ‘ . g . . ' .'. . w w . ‘ _ an!“ . .- .r . .‘a- - g g s- i 1 ' 1 D 1'!- 't a . ‘ _ . n - m... .. nn-p‘ . a . .. 4-H-.. .: .2" ."...:“':4. 23-.-. I'm ms ANALYSIS W VARIANCE W AMPLITUDE SCORE OF THREE ' GROUPS W SUBJECTS WED UNDER DIFFERENT QUANTITIES AND TYPES W REWARD SS df IS F P Total Between Subdects 636,147.05 35 18,175.6 1.36 Groups 234,675.5 5 46,935.1 3.51 .05 Procedure 17,773.4 l 17,773.4 1.33 Size 136,919.98 2 68,459.9 5.12 .05 P'X S 79,982.2 2 39.991.1 3.00 Same Groups (Residual) 401,471.6 30 13,382.4 .-.... B. THE EFFECT OF DIFFERENTIAL DEPRIVATION ON PERFORMANCE l. Lamgngz. According to the analysis set forth above, procedure con- tributed significantly to the variance observed in the present study. This may be interpreted to mean that a difference existed between those animals which began the test series under the high drive level and those which began the test series under the low drive level. Consequently, to test the effect of differential deprivation on response latency, it was necessary to compute two separate t's. One of these compared high drive and low drive animals which had begun their test series with the indicated drive, and the other t compared high and low drive animals which had com- pleted their test series with the indicated drive. Thus, the difference between those animals begun under high and those begun under law were tested separately from those which completed their test series under high or under low. I The mean latency for the first 20 trials of 18 animals begun under the low drive level was 131.47. The mean latency for the first 20 trials of the 18 animals begun under the high drive was 45.63. ‘A t calculated from this data was found to be equal to 6.77, which is significant at well beyond the .01 level of confidence. The mean latency for the last twenty trials of the 18 animals which completed their test series under the high drive was 41.33 seconds. The mean latency for the last twenty trials of the 18 animals which completed their test series under the low drive was 45.32. This difference yields a t of .48 which is not significant. 43 2. (Amplitude. An analagous circumstance to that.reported for latency scores exists in the case of amplitude. .A comparison made between amp- litudes for high drive and low drive animals, which began their test series under these levels, revealed a.mean difference of 230.83 degrees. The mean difference in amplitude scores between the high or low drive level on the terminal 20 trial test series was only 7.17.4 The largest of these two differences was tested by a t. This re- sulted in a standard error of 154.7 and a t of 1.49, which is not sig- nificant. Since the distributions of amplitudes for initial and terminal test series are of approximately the same form, there is no doubt but what the smaller difference (7.17) would be far less significant than the larger (230.83), and a t test for this difference was, therefore, not computed. C. ACTIVITY LEVEL AND PERFORMANCE 1. lam of mm lean]. on: W .m. With reference to activity level, a number of interesting comparisons may be made. Figure 5 illustrates the fact that throughout the experiment there is a consistent trend towards an increase in activity levels, regardless of drive level. This curve is based on the total activity for each of the four experimental days. Since high and low drives contributed equally to the total activity on each of these four days, the effect of differential drive is to a large extent counterbalanced out. ..¥ 4This mean score is based on the sum of the amplitudes for each of the 18 animals contributing to that score. The amplitudes thus summed are the result of each animal's performance on a series of 20 trials. 2. 1mm and lbs 3129 of Wan. Table 9 compares the mean activ- ities for animals, which were run under hunger drives as compared with animals run under thirst drives. For convenience, comparisons were made between matched blocks of 12 animals. By this means, each of the 12 animals run under water deprivation could be matched with a corresponding animal run under food deprivation and a t test applied to the data. Be- cause the thirst group consisted of 12 animals as compared with 24 animals in the hunger groups, it was necessary to match every thirsty animal with two different hungry animals. It is these data which are summarized in table 9. To test the degree to which activity levels for hungry and thirsty animals were matched, it would be necessary to derive four separate t's from the data. If no significant differences existed, one could reason- ably assert that type and length of deprivation were matched in terms of activity levels. High and low drives for food and water deprivation in the present study were empirically "guessed at." Ideally, to test the effect of reward on performance these should have been perfectly matched. The application of a t test to the activity levels of hungry vs. thirsty animals is one of the ways in.which the degree of the match can be estimated. The largest difference was found between one of the groups of hungry animals run under high drive and its control thrist group. The difference 50.0 yielded a t of .38, which is, of course, without significance. .BLE 9 ACTIVITY LFVELS 'OR HLJGRY AND THIBSTY ANTMALS COMPARED TN TEENS 0F REINFORCEMENT AND HOURS OF DEPRTVATTON quantitv Fuentity ‘5 -_ I- “ r O Reinforcement Hunger Reinforcement rhlrSt .32 gms. 628.5 (578.5 High Drive .9 0c ( .08 gms. 580.9 (578.5 .32 gms. 334.2 (493.3 Low Drive .2 cc ( (. 46 3. mm and She Mi We Animals in the present study were run under two values of drive. These have been called for convenience, high drive and low drive. It is assumed, here, that period of deprivation is directly related to the strength of drive. It is further assumed that activity level reflects the strength of drive which results frdm deprivation. One test of this is to compare the activity levels of the so-called high drive (long deprivation) animals with the low drive (short deprivation) animals, neglecting the type or amount of reward administered. Since all animals were run under both levels and counterbalanced, this may be done by employing a.matched t. The two activity measures taken at low drive were summed, and the two activity'measures taken.at high drive were summed. A matched t, based on the difference between the means of 197.93, gives a t of 3.43, which is significant at beyond the .01 level of confidence. This indicates that high drive (long deprivation) animals give significartly higher activity level scores than do low drive or short deprivation animals. 4. ‘Asiizitzmsnd.Laigngz. It is interesting to compare latency and activity level in an effort to estimate the degree to which activity alone will predict the behavior of an animal. Figure 6, which is self-explanatory, illustrates the general trend of the data. The values on this graph were obtained by lumping all of the animals together, regardless of deprivation period, and considering them only in terms of activity level and latency. The animals were grouped by activity counts. Each 100 counts separated a new group. The N of each group varied, of course, and the latency for any given activity group was considered to be the mean of that group. It will be noted from the graph that the activity groups, which are based on 4,7 HUNGER uso ACTIVITY vs. LATENCY I20 60 50 4o LATENCY 13‘) e e 20 IO 200 400 600 800 IOOO IZOO ACTIVITY FIGURE 6. ITFUS..ATES THE DECLINE 0F LATENCY WITH INCREASED ACTIVITY small Us, tend to conform to the notion that more active animals will respond faster than less active animals, whereas prediction breaks down in the middle of the activity range. Pearsonian rs were computed on the latency vs. activity data for high drive and low drive animals. in r of -.236 was obtained for high drive animals, and an r of -.l.28 was obtained for low drive animals. Homogeneity of variance was checked between these tvm groups and an 3:2 equal to 1.1.8 which was obtained is not significant. Having established the fact that the samples were drawn from a common population, an estimate was obtained of the combined r. This equaled -.314, which when tested against the null hypothesis is significant at beyond the . 01 level of confidence. A second test of the prediction value of activity levels was also undertaken. Two frequency distributions of latencies were established. One of these was for the scores of animals under long deprivation and the other was for the scores of animals under the short deprivation period. A median activity level was established for both distributions. For the short deprivation period, this median activity was 1000, and for the long deprivation period this median activity level was 11.42.86. In each case, scores exceeding the median of their distributions were called "high drive" scores (regardless of type or amount of deprivation) and scores falling below this median were called "low drive" scores, again regardless of type or amount of deprivation. For long and short depri- vation periods, latency scores were thus dichotomised and could be tested with a matched t. 49 In the case of low drive or short deprivation scores, there was a difference between the high activity and low activity animals of 27.22. This difference, when tested, yielded a t for related.measures of 1.15 which was not significant. The difference in the case of high drive or long deprivation scores amounted to only 5.34, and in view of the simi- larity of the distribution of the data in this case and that Just re- ported, it seemed.unnecessary to statistically test the difference. D. EXTINCTION 1. Ember of We is Extinction. Each of the six major groups of animals was equally divided. One half of the animals in each group was extinguished under the low drive. A simple analysis of variance was used to test the difference between these two papulations. The resulting design is given in table 10. Since the within group mean square exceeds the between group mean square, there is no significant difference between animals extinguished under a high drive and animals extinguished under a low drive. The number of degrees of freedom, it will be noted, is 32 rather than 35. This results from the fact that three animals were eliminated from consideration. These animals were extinguished, by error, to a criterion of two, rather than three minutes. One of these animals was from the low deprivation group and two were from the high deprivation group. Although the difference was not significant a mean difference existed. between the groups of 7.9 responses. The high drive animals required a mean of 40.2 responses to extinction, whereas the low drive animals required a mean of 32.3 responses. . ... anti. main \1 .. ....fla.....r.. met... we fit“ ,.........%e g TABLE 10 ANALYSIS CF VARIANCE DESIGN T0 TEST THE DIFFERENCE IN NUMBER (F RESPQWSFS T0 EXTINCTION BETWEEN ANIMALS EXTINGUISHED UNDER HIGH DRIVE AND THOSE EXTINGUISHED UNDER L04! DRIVE 50 Source 8.3. df M.S. Between 335.09 1 335.09 Within 16,285.28 31 525.33 Total 16,620.37 32 51 2. m 5.11 m: Latency. The minimum number of trials, which any animal required to complete extinction, was six. Comparisons of the latencies on the first six extinction trials for all thirty-six animals were made and the results are shown in figures 7, 8, 9, and 10. Figure 7 plots the sum of the latencies for all 18 animals run under each of the two'drive levels. Figure 8 is a similar graph based, however, on the median latency for all 18 animals run on each of the two drive levels. Both groups show an initial drop in reaponse time and a gradual increase in latency thereafter. 3. m m m: Me. A similar analysis of the extinction data may be made in the case of amplitude. In figure 9, the total amplitude of the response of the 18 animals extinguished under high drive and the 18 animals extinguished under low drive are plotted. High drive animals show a sharp increase in amplitude on trial two and then a gradual decline . Figure 10 is a graph plotted with the same data but based on the median latency for the 18 animals in each of the two drive conditions. Beoause medians are not influenced by marked shifts in data, the graphs plotted using the medianswrather than those based on sums-- is probably the most satisfactory for indicateing the general trend of the data. In general, the curves for high and low drives parallel one another both in the case of latency and amplitude. t... .....\.. n3... is... L3H . . Ii RESPONSE LATENCY 52 LATENCY 3.. ----- - LOW DRIVE ”.0 300 HIGH DRIVE ’9’” 250 200 I50 I00 50 I 2 3 4 5‘6 FIRST SIX EXTINCTION TRIALS FIGURE 7. ILLUSTRATES THE DECLINE OF THE MEAN LATENCY ON THE INITIAL EXTINCTION TRIiLS 53 MEDIAN LATENCY 10 53 / ./ LON DRIVE —————— / I ‘NCY 8 " HIGH DRIVE -— ’ l 6 MEDIAN IJATEIICY 4:- 1 2 3 h 5 6 FIRST SIx EXTINCTION TRIALS FIGURE 8. ILLUSTRATES THE DECLINE OF THE MEDIIN LATENCY ON THE INITIAL EXTINCTION TRIILS 51+ FORCE I200 ”00 "n" LOW DRIVE — RICH DRIVE I000 900 800 FORCE 0F RESPONSE 700 600 500 I 2 3 4 5 6 FIRST SIX EXTINCTION TRIALS FIGURE 9. IJWWJ: TN: RISE 01“ new IIIDLITUDE ON THE INITIAL E‘ITWTFC‘N TRIALS IN IIII: use: (I? HIGH DRIVE ANIMALS MEDIAN ALIP LI TUDE LOW DRIVE -' - - " " HIGH DRIVE — 14.0 50 [:1 9 EH.‘ a 2 91 o 2 10 FIGURE 10. FIRST SIX EXTINCTION TRIALS ILLUSTRATES THE IEDIILN RISE IN AMPLITUDE ON THE INITIAL EXTINCTION TRIALS IN THE CASE CF BOTH HIGH DRIVE AND LOW DRIVE .SINIMIIII‘S 55 56 v1. W A. sEr AND THE SIZE OF REWARD In the introduction it was pointed out that sEr is believed to be a function of a number of factors, one of which has been called the incentive component, K. K is held to be a negatively accelerated increasing monotonic function of the weight of food given as a reinforce- ment. In the present study, despite a weight-ratio of approximately 41:, animals rewarded with the larger pellets failed to respond signifi- cantly faster to the push panel than animals rewarded with the smaller pellets. An inspection of table 1. reveals some interesting comparisons. Here it can be noted that small reward pellet animals are superior to large reward pellet animals in every cell except one, Low drive (Low to High). This implies that differences in perfomance, resulting from differential reward, depend not only upon the size of reward but also upon the m1 drive level under which animals are tested and the strength of the habit at the time of testing. Thus, one factor-such as habit-nay mask the influence of other factors. This is illustrated in the case of animals shifted from a high to a low drive. Both reward groups show a m in response times as the animals are shifted from a high to a low drive, and this happens, despite the fact that for all animals combined high drive performance is signifi- cantly superior to low drive performance. 57 It might be asserted that animals which are run under a low drive the last two days are actually running under a low drive pm a residual motivation, resulting from two test days under long deprivation. Il‘his seems unlikely, for in this two-day period, these animals are satiated no less than four times,.so that it would certainly seem that by the fourth day their running times would be somewhat depressed. Some light is shed on this problem by reference to the curve illus- trating the general rise in activity level over the four test days (see figure 5). ‘With.reference to this curve, it'may be asked: Does this increase in activity result from a residual drive or is it the result of some learned anticipatory factor? A partial answer to this ‘question can be obtained by noting that the activity measures for the high drive animals show a smaller increase between day, one and day two than do low drive animals. These low drive animals increase their activity count by a mean of 27.9, while high drive animals increase their activity count by a mean of only 12.6. This certainly cannot be accounted for in terms of a residual drive, for if a residual drive were building up, we would certainly expect it to be reflected in the activity of those animals undergoing the most marked physiological deprivation. It thus appears that while deprivation and other similar functions, such as size of reward, do influence performance, the effects of these are often masked by strong learning. It follows from this that in the early stages of learning, the effects of motivation, K, and possibly other 58 factors as well will be most clearly distinguishable. To a large extent, the data gathered in the present study support this hypothesis. The analysis of variance relative to these data indicates that the adapted procedure constituted a significant independent variable. This is an important finding, for an analysis of the data reveals that this difference results primarily from the longer latencies exhibited by animals which begin their test series under a low drive. Assume, for example, that no measurements had been taken late in the series. In this case, marked differences would undoubtedly have been found to exist between the reward groups. In other words, we have evidence here to suggest that the point in the test series at which measurements are taken may determine whether or not differences ever become apparent. All too often this important con- sideration is neglected in the psychological literature. we are inclined to assume that because differences exist utilizing one experimental pro- cedure, they will necessarily be found when another procedure is adepted or vice versa. In connection with the failure to uncover differences between large and small pellets, it should be pointed out that this is somewhat in Oppoation to our theoretical expectancies. Table 4 indicates that such differences as do exist are in the direction which would be predicted within a Hullian framework, however. Several factors may account for the lack of significant differences: (1) The possible dominance of er in the equation for sEr. This has already been considered. 59 (2) The ratio of large reward to small reward may have been insuf- ficient to permit a full realization of the potential differences in response time (39). (3) The measure may have been too asymtotic. If, fer example, the animals had had to run down a 20' runway as in the Crespi study (5), the differences might have been "stretched out" and thereby made more readily apparent, and (4) It may be that in the present situation quantity of reinforce- ment actually made no difference at all. B. sEr AND THE TYPE OF REWARD One of the aims of the present study was to determine a quantity of food reward which equalled a fixed amount of water reward. Actually, the study indicates that no significant differences exist as the result of the amount of reward administered. Rather marked differences are present, however. An inspection of table 4 indicates that in all but one of the cells the small food reward most nearly approximates the water reward. It would therefore seem that in the larger study, which is to be conducted, small pellets and five seconds drinking time would be roughly equivalent in reinfoching value. 0. DEPRIVATION AND sEr l. Latency. The data tested with reference to latencies and deprivation tend in part to confirm out theoretical expectations. But, as we inspect 60 the data, we are struck by the fact that whether or not differences arise is again to a large extent a function of the point in the test series at which comparisons are made. Reference to table 4 and the section entitled, ”The Effect of Differ- ential Deprivation on Performance," will illustrate this fact. marked and significant differences occur between high and low drive animals up to 20 trials after the initial 30 training trials. These differences, however, are observed to be absent when latencies are compared beyond the 50th trial. Again it appears that the peculiar manifestation of the effect of a significant variable, such as drive, may depend to a large extent upon the strength of the habit at the time that measurements are made. 2. Amplitude. Deprivation, in the present study, appears to be totally unrelated to the amplitude of the response measured. This is, of course, in disagreement with the generally accepted Hullian beliefs concerning the amplitude of the response and sEr. It should be noted, however, that the studies cited by Hull to support the relationship of amplitude to sEr involve autonomic rather than skeletal response measures (9) (10). mnmmmmmmmmmudomm mmwmmmmmmmmmm. A difference of 230.83 was observed between high and low drives on the first series of 20 trials as compared with a difference of only 7.17 on the second twenty trial series. 61 D. EXTINCTION l. n‘as a Significant.lnd1§atgx‘Qf‘aEr. A second major point of con- ‘flict with contemporary Hullian theory exists with reference to the number of responses to extinction. According to the evidence collected in the present study, there is no basis for asserting that n is dependent in any way upon the drive level under which an animal is extinguished. n, in fact, appears to be completely unrelated to any variable which is controllable in the experimental situation. The camful observation of animals under extinction and the correlated data reveals that beyond the first few trials, few, if any, really con- sistent generalizations can be drawn. Frequently, animals will come to a point where responses are delayed for as long as 160 or 170 seconds, and then, quite suddenly, there will occur a long flurry of reaponses, one immediately following another. Most individual animals show no tendency towards a gradual decrease in latency. Especially those requiring a large number of responses to extinction. In fact, it quite often occured that the criterion latency came at the most unexpected points. Many animals would try to reach and to epen the door through which they were admitted to the box, and other animals would continue to scratch themselves fer the full three- minute criterion period. These observations, it seems to the author, can only be accounted for in terms of behavioral exploration and relearning. What has been said about latency applies to the amplitude of the responses made, as well. Beyond the first several trials, no consistent trends are observable. . a . < < . - h z z < - z . . z \ < : G ‘ D t t . - . < - Q 62 2. Ihsinsmasnfnfanmmmlafrmd. me ofthemost interesting observations which can be made with reference to the extinc- tion data concerns the marked drOp in the median latency and/ i1; 8libs median amplitude on the second extinction trial. On the first extinction trial, animals have never experienced Opening the door and the finding of no food, but on the second trial, this is not the case. Since this con- stitutes an instance of what is commonly called, "frustration,” we are led to ask: Is this drop in latency and the increase in amplitude on trial two a function of "frustration,” and if so, how does "frustration" enter the equation for sEr? An inspection of the four figures, illustrating the trend of this data (figures 7, 8, 9, and 10) reveals several interesting and possibly significant facts. (1) ‘While high drive and low drive curves exhibit a marked simil- arity in form, in virtually every instance the point on the curves drawn for high drive are superior (perfOrmance-wise) to those drawn for low drive. (2) Latency scores (especially in the case of high drive animals) approach a lhmiting asymptote, so that it is difficult to test the differences between trials one and two and draw meaningful conclusions. This is not true, however, in the case of amplitude scores, and (3) ‘At the time of this second extinction trial, animals have experienced 70 rewarded trials and 1x unrewarded trials in the experi- mental situation. The marked drop in latency and rise in amplitude following the first extinction trial certainly cannot be accounted for ......n? z. ......z- ... . 39.. .. .. . 12...... :1... . ... ... ..n.. ..demt. ....I......,. ..J....,... A14... ......w. a... fawn/u. NHL-r}? u/t».¢.— 4. Hr n?.....b”|.l.1\1.r§.h§l..E I L .usu.” ure‘fluwnfl .i 63 in terms of a sudden rise in sEr. It seems more likely that there is, in this instance, a.marked increase in drive, resulting from.the addition of "frustration." E. ACTIVITY LEVEL AS AN INDEPENDENT MEASURE The consistent rise in total activity level which is noted on the four test days mey'be accounted for in one of two ways: either a residual drive (tissue need) is accumulating or else some learned "anticipatory factor” is Operating in the situation. The first assumption may be tested by comparing total activities for high drive days one and two as compared with low drive days one and two. This we have already done, and it was pointed out that the increase fer low drive animals was almost nine times that for high drive animals. If the increase in activity was the result of a high residual drive building up, it would certainly seem that his residual drive would be greatest on high drive days. But, such is apparently not the case. It seems more likely, than, that some learned anticipatory factor is Operating here to increase activity. In one sense, this increase may reflect a change of.nctivation as well, but the change-it is important to note-is to be accounted fer in terms of learning and not tissue need. With reference to activity level and performance, few generalizations can be made. There appeared to be a genuinely significant difference between activity levels taken at high drive and those taken at low drive, but the prediction of performance on the basis of activity-either at the individual or group level-was limited. A significant correlation 64 between activity level and latency was established, tending to verify the graphic relationship illustrated in figures 7, 8, and 9. While gross trends such as those observable in figures 7, 8, 9, and 10 are detectable, existing differences are overshadowed by the tendency for the vast majority of the data to center about a common range in activity level. Thus, when median latencies are established to separate arbitrary ”high" and ”low" drive levels, latencies cluster about the line of demarcation and tend to "flatten" the curve so that while marked differences may exist at the extremes, such differences are minimized and statistical differences do not appear. 65 VII. W The present study was conducted in order to establish incentive values for food and water, and in an effert to determine the feasibility of quantifying the drive construct once such values were known. A pushepanel apparatus was constructed in which activity levels could be measured simultaneously with response amplitude and latency. Thirty-six male, albino rats were divided into two major groups, both of which were subdivided again into three groups. Small Food Reward c) Mediumeater Reward 1. High Drive: (a; Large Food Reward b b) Small Food Reward c) Mbdiumfiwater Reward 2. Low Drive: is) Large Food Reward Each of the 36 animals was habituated to the box, assigned to one of the subdivisions, trained to Open the pushrpanel for either food or water, and then tested fer a total of 40 trials, 20 trials under a high drive and 20 trials under a low drive. Half of the animals began their test series under a high drive and half began their test series under a low drive in order to counterbalance the trials. Activity level for six minutes before the exposure of the pushepanel, and the latency and amplitude of each response was recorded. At the close of the test series, all animals were extinguished under either high drive or low drive. R) \I) \I/)) The results were as follows: 1. .Lstencz: Amount or type of reinforcement was not a significant variable with respect to latency. The incentive value of small food reward, however, more nearly matched the incentive value of the amount ' of water employed. Such differences as do exist are largely confined to the first half of the test series. ‘With reference to drive level early in the test series latency appears definitely to be a function of the drive level under which it is measured. This does not hold true for differences measured late in the series. 2. W Legal: Activity level, in the present study, offered some promise as an independent measure of drive. Individual and group predictions of performance based on activity are, however, of limited reliability. Some trends are observable. Activity shows a consistent upward trend throughout the test series. This, it was pointed out, cannot be accounted for in terms of some generalised increasing drive but seems to be a consequence of learned anticipation. A significant negative correlation was obtained between activity level and latency. Curves drawn comparing activity level with latency indicate a trend toward decreased latency with increased activity. This is particularly true of extremely high or low activity animals. An effort was made to test the difference between men and low activity animals using a t test, but no significant differences were detected. There was, however, a significant difference between activity levels taken following long deprivation and those taken following a short deprivation. Type of reinforcement was unrelated to activity. 67 3. Amplitude: The amplitude of the response as measured in the present study did not prove to be related to either the amount or type of reinforcement, or to the amount of deprivation. 4. .Extinctign: No difference was feund between animals extinguished under high drive and those extinguished under low drive in number of responses to extinction. The lack of trends in the extinction data after the first few trials was discussed. Some drop in latency and rise in amplitude was noted on the second extinction trial, but this did not prove to be significant. The lack of significance may in the case of the latencies arise from the asymptotic level of the response. 5. One or the significant findings of the study was the discovery that differences often appear to be a consequence of the point in the test series at which measurements are taken, rather than a simple func- tion of some variable such as drive or reinforcement. Habit in this respect, appears to be the dominant factor in the determination of sEr. Recommendations for further research in the area of motivation: 1. In order to insure that animals will respond on every prescribed trial at low drive levels, the number of training trials should be increased to between 50 and 100. The number of test trials administered at each of the various drive levels can then be reduced from 20 to a much smaller number. 2. While there was no significant difference in the effect on performance of small reward, large reward, or water reward, water reward was most nearly matched by the small food reward. Any future study aimed .....t. .... ....... fihfllffiaJsf . . . . ..... Dr. .....‘.... .rt. .-...1 W7. (....I...... .. , bud. fnl‘» at the quantification of the drive construct, should adopt, therefOre, the two incentive values for the different types of deprivation which are most nearly equal. 3. Certain changes in apparatus are recommended: (1) The large guillotine door should be moved nearer the wall which contains the push- panel in order to prevent animals from retreating into the intervening space. (2) The ball should be removed from.the center of the false floor in order to obtain a more accurate activity count. l. 2. 3. 4. 5. 7. 9. 10. 12. 69 BIBLImRAPHY References Anderson, J. E., and Smith, A. H., 1926, "The Effect of Quantitative and Qualitative Stunting upon Maze Learning in the Hhite Bat," L m. mug]... .6. 337-359. Broman, L. G. , 1943a, "The Effect of Controlled Temperatures upon the Spontaneous Activity Rhythms of the Albino Rat,” .1. m. 2.991., 3‘, 477-4890 Browman, L. G. , 191.1. , "Modified Spontaneous Activity Rhythms in Rate," As- I. Rania]... 1A2. 633-637. Covles, J. T. , and Nissan, H. $1., 1937, "Reward Expectancy in Delayed Responses in Chimpmeea.” J. 9mm 2mm... 21.. 345-358. Crespi, L. P. , 1942, "Quantitative Variation of Incentive and Per- formance in the White Bat,” W. 31].]... 39: 442.443. Finan, J. L., 1940, ”Quantitative Studies in Motivation. I. Strength of Conditioning in Rats under Varying Degrees of Hunger," .1. mm. 23191191., 22. 119-131.. Grindley, G. C. , 1929.1930, uiliixperiments on the Influence of the Amount of Reward on Learning in Young Chickens," Brit. 1. We, 29, 173-180. Heron, W. T. , 191.0, "The Behavior of Active and Inactive Rats in kperinental Extinction and Discrimination Problems ," W. Ban. , L, 23.31. Hull. 0. Lo. 1943. W 21 W. New York: Appleton- Century. Hull, Clark D., 1951, W31 W. New Haven: Yale University Press. Hunt, J. McV., and Schlosherg, 8., 1939, "General Activity in the Male Whit. ht). 1- 2m- m" 23: 23‘38° Kendler, H. H. , 1945a, "Drive Interaction: II. Experimntal Analysis of the Role of Drive in Learning Theory,” J. m. W» 3.5, 188-198. . 1"!" 15. 16. 17. 19. 20. 21. 22. 23. 25. 26. 27. 70 Kinder, E. P., 1927, "A Study of the Nest-Building Activity of the Albino mt," Jo Me Me, a, 117-161. Meyer, 1951, ”The Effect of Differential Rewards on Discrimination Reversal Learning by Monkeys ," .I. m. W. , 5.1, 268-274. Miller, B. E., 1951, Chap. 13. Iearnable Drives and Rewards. Stevens. 3. 3., Editor. Handbook .91 W W. New York: John Wiley and Sons. Munn, N. L., and Collins, 11., 1936, "Discrimination of Red by m“ Rats," :10 We W03. ‘39 72‘87° Mn. N. I». 1950. W of W.m gains at. Boston: Houghton Mifflin Company 53-83. Reed, J. D., 1947, "Spontaneous Activity of Animals," PM. an]... M, 393-412. Reynolds, B. , 191.9, “Ihe Acquisition of a Black-White Discrimination Habit under the Levels of Reinforcement,” J. M. W” 32, 760-769. Reynolds, 3., 1950, "Acquisition of a Simple Spatial Discrimination As a Function of the Amount of Reinforce-lent," .I. m. 2:131:91... ‘9, 152-1610 Richter, C. P., 1922, "A Behavioristic Study of the Activity of the Rat." m. 2mm.m.. 1. 2. Richter, C. P., 1927, ”Animal Behavior and Internal Drives," m. m. 3.91... 2’ 307“3430 Rundquist, E. A., and Heron, W. ‘12., 1935, ”Spontaneous Activity and mac Rani-118’. Jo We We, 12, 297.311e Shirley, 14., 1928c, "Studies in Activity. III. The Influence of Activity to Maze Learning and to Brain Weight,” .1. 99.9. mg]... a, 187-195 e Siegel, P. S., 1946a, I'A Simple Electronic Device for the Measure- ment of Gross Bodily Activity of Small Manuals," 1. No: 2.1, 227.236e Siegel, P. S., 194.61), ”Activity Level as a Function of Physically Worced Inaction,” J. W” 21, 285-291. Skinner, B. F. , 1933a, "The Measurement of 'Spontaneous Activity, "' 1. Gen. W0: 2: 3'23. U I C O 1 0 I l I I Q I O ' O . n . , . 1 n H I . t I\ IV tn 1 Q !\ '\ es e\ . u. I a\ I i I e a I l\ l\ -\ l I\ 1 ex a e . . 2 e e t I i I. I h ‘ e ..\ O I I as V\ A\ a\ I I I I\ . n I I t n t .x .\ v I a it . v I C Q ‘ I\ la . C I Q . l e\ s . C I a 9 Q n t. f\ O \ fix l «x n. .\ I I l t I V e o . l v v e I . l ..\ Q .\ . i . . n . .\ n a n . t e \ l\ \ a 0 1 I ax .\ t. .\ n e I .x e e t A C . . . n 1 u . - I -\ I .\ . n n» o ‘ P» D . . t e . . . u ‘ 9. I . _ . .\ . \ n I e i 4 t .x u q . . Q o . t 1 '\ I\ I. 28. Slonaker, J. R., 1925a, ”Analysis of Daily Activity of the Albino Bat.” Am. 1. mm]... 13. 485’503- 29. Strassburger, R. 0., 1950, "Resistance to Extinction of a Conditional Operant as Related to Drive Level at Reinforcement," J. m. W. , A9. 473-487. 30. Tattle, W. U., and Dykshorn, S., 1929, "A Comparison of Spontaneous Activity of the Albino Hat with the Ability to Learn, with Special Reference to the Effect of Castration and Ovariectow on ihese Processes,“ W. 2121. , 2, 157-167. 31. Underwood, Benton J ., 1949, W W. New York: Appleton-Century-Croft, Inc. 32. Wald, G., and Jackson, D., 191.4. . "Activity and Nutritional Deprivation,” 2m. in. AM. an... M0: 39. 255-263. 33. Warden, c. J., 1931. mm mm mm. m m.m. New York: Columbia Univ. Press. 31.. Webb, W. B. , 1949, " The Motivational Aspect of An Irrelevant Drive in the Behavior of the White Rat,“ J. M. W" 32, 1-14. 35. Wolfe, J. B., and Kaplon, M. D. , 1941, "Effect of Amount of Reward and Consumative Activity on learning in Chickens," .1. gm. 25151191., 31, 353.361» 36. Young, P. 1., 1936, W of W. New York: Wiley. 37. Young. P- 1‘» 1951. We: of Asian]. main W W. Stone, Calvin P., Editor. New York: Prentice-Hall, 2-109. 38. Pitts, P. 14., 1940a, "The Effect of a Large and a Small Reward as Indicated by the Resistance-to-Extinction Curve for the Rat,” W. 3311.. 3.2, 429-430. 39. Zeamsn, D., 1948, ”Response Latency as a Function of the Amount of mmorcemnte” As. We 3: 238- . . pt . -\ _. -\ - Ix . r A\ . I\ l\ -\ Q . A .\ ah !\ . I .. .k e u . A . . e at . t A I Vs .r\ I d , . .\ . . a . e \ I I v\ u\ I\ W\ l\ I\ . n u\ . .1 . I i .\ o\ ( I I at t a I r .\ . i . _. t I\ I n\ . ‘ ’\ _ . . t . \ n\ h l .\ . e .\ I . i t f .\ r .. .x 1 . . t . c . i 1 e.\ x i C v\ . . . . . K . u . 1 . J o a i . a f\ . - Jul 15 ad JAN 5” 1962 b IIIIIIIIIIIII w ucmumm17mmmfimflflmmftmflwhim/W 31 000000000000