Citation
Motivation, Krech hypothesis behavior, and adaptability

Material Information

Title:
Motivation, Krech hypothesis behavior, and adaptability
Creator:
Procter, Robert Louis, 1928-
Publication Date:
Language:
English
Physical Description:
1 online resource (54 leaves) : ill. ; 1962.

Subjects

Subjects / Keywords:
Alleys ( jstor )
Body weight changes ( jstor )
Experimentation ( jstor )
Learning ( jstor )
Learning motivation ( jstor )
Mazes ( jstor )
Motivation ( jstor )
Rats ( jstor )
Water deprivation ( jstor )
Water tables ( jstor )
Animal behavior ( fast )
Animal intelligence ( fast )
Dissertations, Academic -- Psychology -- UF ( lcsh )
Psychology thesis Ph. D ( lcsh )
Psychology, Comparative ( fast )
Rats ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 48-50.
General Note:
Manuscript copy.
General Note:
Thesis - University of Florida.
General Note:
Vita.
Statement of Responsibility:
by Robert Louis Procter

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
This item is presumed in the public domain according to the terms of the Retrospective Dissertation Scanning (RDS) policy, which may be viewed at http://ufdc.ufl.edu/AA00007596/00001. The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
021320580 ( ALEPH )
885359058 ( OCLC )

Downloads

This item has the following downloads:


Full Text













MOTIVATION, KRECH HYPOTHESIS BEHAVIOR, AND ADAPTABILITY













By

ROBERT L. PROCTER













A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY











UNIVERSITY OF FLORIDA June, 1962












ACKNOW LEDGENTS


The author wishes to express his gratitude to the members of his committee, Prof. Rolland H. Waters, Chairman, Prof. Richard J. Anderson, Prof. Bradford N. Bunnell, Prof. James C. Dixon, Prof. Herbert D. Kimmel, Prof. J. Milan Kolarik, and Prof. Zareh M. Pirenian. He is especially appreciative of the many hours given by Prof. Waters and Prof. Bunnell. The clarification of learning theory by Prof. Judson S. Brown and some invaluable assistance from Mr. William F. Stone are also gratefully acknowledged.




















ii













TABLE O NTENTS
Page
AGNOWLEDGNTS . . i

LIST Of TABLES. . .. iv

LIST OF FIGURES. . . y

ORAPTER

I. INTRODUOTION. . 1

II. PROCEDURE. . 20
III RESULTS . . 25

IV, DISCUSSION. ... ..... 41

V. SUMMIARY . . 46

BIBLIOGRAP. . . 48

APPEDIX. 0 511 BIOGRAPHICAL SKETOH . 54



















iii












LIST OF TABLES

Table Page

I. The Krechevsky Trial Order for the Position of Open Doors and Lighted Alleys on the
Twelve Daily Trials in the Maze. .. 5

2. Mean Running Time in Seconds in the Training Trials. . 26
3. Mean Running Time in Seconds in the
hypothesis Mase . 27

4. Analysis of Variance on the Effeots of Type and Level of Deprivation on Running Times in
the ypothesis Maze . 28

5. Mean Per Cent of Body Weight Change Over the Course of the Experiment. ,.... 29

6, Comparison of the Moderate Food Deprived and Severe Water Deprived Groups for
Frequencies of Dominant hypothesis Type 33

7. Food and Water Deprived Animals Matched for Equivalent Body Weight Change .. 34

8. Type and Frequency of Hypothesis Behavior 36
9. Per Cent Body Weight Change, Type and Occurrence of Hypotheses, and Trials to Solve
the Adaptability Problems ... 52











iv












LIST OF FIGURES
Figure Page
1. Floor Plan of the Kreshesky hypothesis
Haze *. 0 0 0 0 0 0 0 0 0 0 0 0 4

2. Nmaber and ype of Hypotheses ormed as a

tneotion of Body Weight. .. ... 30
3. The Relation Between Body Weight Change and

Average Number of Trials to Solve Both
Adaptability Problems. . 40

























w












CHAPTER I


INTRODUCTION


This study was primarily concerned with determining how hypothesis behavior in the hooded rat varies with changes in motivation. It was also concerned with finding out what relation such hypotheses have to adaptive behavior, e*g., are animal that show certain types and numbers of hypotheses more adaptable or flexible than others? And finally, it was concerned with the relation between motivation and adaptability.
Hypothesis behavior is a term first used by
Kreohevsky (1932a) to describe the apparently non-random, systematic attempts of rate to solve an insoluble linear mase problem. It is necessary to maintain a careful distinction between the behavior of the animal in the experimental situation and the implications of the experimental results. Rpothesis behavior refers to a particular type of behavior which occurs in a specific situation. The behavior is a pattern of choices made in relation to specific oues in an insoluble discrimination mase. When the pattern of choices differs from a random


1








2

level by a certain amount, hypotheses are said to be formed. Hypothesis behavior and hypotheses are terms which have meaning only when the apparatus, the cues, the pattern of choices, and the statistical level are understood.
Hamilton (1911) noticed in the course of a trial and error learning experiment that rats and other animals make apparently systematic attempts at solving problems that cannot be solved; they may go consistently to the right, or left, or to the last place found successful, and so on. Lashley (1929) also noted this feature of behavior in discrimination problems. He indicated that while there was no experimental technique available at that time to investigate this behavior, these "attempted solutions" probably represented a significant aspect of the learning process. Lashley's observations served as as at least a partial basis for Krechevsky's investigations; Krechevsky's experiments were designed to provide the lacking experimental technique. In one of his earliest published works he states, "The data from the present experiment have been examined in the light of Lashley's suggestion and an attempt has been made to devise a method for the objective determination of the validity of that suggestion." Krechevsky concluded from this study that "...it is shown quite definitely in the presolution period that. the animal is engaged in bringing to perfection various










attempted solutions.... In the light of the evidence presented here it is suggested that the helter-skelter unorganized trial and error response as a description of the early part of the learning process is invalid, and that we must change our description of the learning prooess so as to recognise the existence o of organied and systematic responses at all stages of the process." (1932a, p. 43)
reo hevsky's interpretations were directed against the prevalent behaviorism which used trial and error learning and later the conditioned reflex as a model. Boring (1950) characterized both of these models as eliminating the need for integrative principles such as organization or insight. Tolman, under whom Kreehevsk was studying for his doctorate, was then developing his theory of purposive behaviorism. This theory emphasized the cognitive, intellectually organized aspects of learning. Ereohevsky's work with hypothesis behavior served both as a fundamental tenet of the Tolmanian system and as experimental evidence against an an anti-insight behaviorism.
[rechevsky's 1932b experiment formed the basis for
all the later work in this area. The maze which he designed was a four unit discrimination apparatus which could not be solved (see Figure 1). The animal was confronted by four ohoice-points between the start box and the goal box. A














Start b
Curtain Swinging Door Goal box
Fig. 1.-Floor Plan of the Krechevsky Hypothesis Maze.
This diagram shows the maze as it appears on trial 7.
An animal following a visual-dark hypothesis would
choose alleys in the order left, left, right, right.




ooreetion method was used at each choioe-point and the reward was available in the goal box at the completion of every trial regardless of the correctness of the animal's four choices. Kreohevsky used a 23-hour food deprivation schedule; a cube of bread soaked in milk was used as the incentive. At each choice-point there were two alleys, a right and a left, between which the animal could choose. Over each alley was an electric light, but only one side was illuminated. The animal might choose on the basis of patial cues (right, left) or visual cues (light, dark). A curtain at the end of each alley prevented the animal from seeing a swinging door which could be looked to either side, thus preventing the animal from using one or the other alleys. One alley of each of the four units was always locked. When the animal chose an alley which had an open door, it went through to the next unit. When the











animal chose an alley with a looked door, it was forced to reverse its course and go through the alley on the other side.
The animal was given twelve trials each day. On
every trial the lights and doors were shited in such a way that's 1) the lights and open doors were on either side of the ohoice-point equally often, and, 2) the use by the animal of any of the eight hypotheses which Krechevsky identified would result in encountering the same number of open and looked doors. (See Table 1 for the trial order positions of lights and doors.*)


TABLE 1

THE K HESKY TRIAL ORDER FOR THE POSITION OF OPEN DOORS AND LIGHTED ALLEYS ON THE TWELVE DAILY TRIALS IN THE HAZE

Trial 1 2 3 4 Door Open L L R L R R R L R R L R L L RR Light On R L R L R L R R L R R R L R L L Trial 5 6 7 8 Door Open RRL L RLL L R L L R LR LL Light R RL L LLRR L R L RLLL Trial 9 10 11 12 Door Open L R R L R L RR L L LR L R R R Light On L LLR L R R L R L RR L R L R








6

To major types of hypotheses were distinguished,
those based on visual Cues and those based on spatial cues.

Among the visual hypotheses, an animal may systematically
choose the lighted alleys (a "Light Hypothesis") or the dark alleys (a "Dark ypothesi") Sometimes an animal tends to choose an alley in teams of the similarity of its visual characteristics to the immediately preceding
correctt" alley (e.g., choosing a lighted alley if the
previous lighted alley had proved correct). This
behavior is termed a "Perseverative Visual Hypothesis."
Conversely the rat may show a. "Alternating Visual
Hypothesis
The other general categories of hypotheses spatial
also has four possibilities.., there are "Right", "Left", "Perseverative Spatial" and "Alternating
Spatial" hypotheses. (Rosensweig, et al. 1958, p. 375)

It is possible to invent any number of systematic patterns of choice hypotheses for the animal, but fKrehevsky compared the animal's choices for correspondence to these eie eight. The equivalent of the Spatial hypotheses,
-Right, -Left, and -Perseverative, and the Visual hypotheses,
-Light and -Dark, had already been noted by previous investigators. The other hypotheses appear to be logical extensions of these five.
If the animals made completely random, unsystematic responses at each of the 48 choice-points (four choices on each of twelve trials), they would choose on the average 12 right-lighted, 12 right-darkened, 12 left-lighted, and 12 left-darkened alleys in some random order. A theoretical random animal would be expected to have an average of 24








7

choices to each of the four cues (Left, Right, Light, Dark). Krechevsky defined non-random, systematic behavior his hypothesis behavior as being present when the animal chose one of the eight alternatives with a frequency significantly greater than the theoretical random mean of 24. The level of significance which he chose was three standard deviations above the mean. This amounts to 34.39 choices out of 48.
With this criterion, Krechevsky found that most of his animals formed hypotheses and from these early studies he concluded that "...the presence of systematic forms of behavior in such a situation is to be interpreted to mean that these systems were determined not as something forced ab extra by the situation, but as something originating from the animal himself" (1932b, p. 45). He suggested that "...descriptions of the lower animals' behavior as consisting of "stereotyped," "haphazard," "non-insightful" responses are to be attributed not to a lack of insight on the animals' part but rather to a lack of insight on the experimenter's part" (1932b, p. 63).
Over the next several years, Krechevsky continued
his investigation of hypothesis behavior. In one of these studies (1933b) he used animals from Tryon's "bright" and "dull" strains and found that the "bright" animals had more spatial hypotheses and the "dull" animals had more visual hypotheses. Another study examined some of the conditions








8

under which animals give up or persist in hypotheses. Rats which had formed hypotheses were divided into three groups. One group then found the maze solvable in conformance with the hypothesis each animal was exhibiting; the second group had the maze solvable for a different hypothesis; a third group had the incentive removed as soon as an hypothesis was demonstrated. The first group persisted in their original hypotheses, the second group changed hypotheses to the correct solution, and in the third group more than fifty per cent of the animals refused to run at all. This study was integrated with a further exposition of Tolman's theory and Krechevsky's most important conclusion was that "...if any behavior-act can be established as docile and purposive, 'hypothesis'-behavior is definitely so" (1933a, p. 442).
Krechevsk. began a series of investigations on the
relation of hypothesis behavior to the structure and function of the brain (1935, 1937a, b, c). He found that cortical lesions decreased the number of different hypotheses an animal might use, but increased the number of days n n animal would use the same hypothesis. He also found evidence that this behavior was not related in simple fashion to the amount of cortical damage, but to both the amount and locus of damage. One area (occipital lobe) he called V because few animals made visual hypotheses when it was damaged. Another area (somesthetic) was identified as S, because of










a similar effect on spatial hypotheses. Damage to both areas tended to result in spatial hypotheses.
Krechevsk published no more articles on hypothesis behavior until recent years. Hypothesis behavior served as the prototype for other Tolmanian concepts such as meansend-readiness, and Krechevky began experimentation in other areas. Also, while hypothesis behavior was directly related to the controversy over the continuous or discontinuous nature of learning, the hypothesis behavior maze was not readily adaptable to investigating this problem. Kreohevsky designed other pieces of apparatus for this purpose.
The only critique of hypothesis behavior which
exists in the psychological literature is that provided by Within (1942). Witkin noted that hypothesis behavior is not typical of solvable problem situations, but rather of situations for which no solution can be found. He found the same type of behavior in the Kreshevsky hypothesis mase when the doors were completely removed (free-choice) as when the doors were looked according to Kreehevaky's insoluble pattern; the behavior was not found when a solvable pattern was used. He questioned the adaptive or purposive nature of hypotheses, since animals shifted hypotheses in a free-choice situation although the first hypothesis or no hypothesis at all was equally adaptive.








10
He also questioned Krechevsky's interpretation of the cortical lesion studies. "Habits having a far simpler basis than 'hypotheses' cannot be performed after extensive cortical lesion. Yet every animal submitted to decortication showed systematic behavior in linear 'insoluble' situations, pointing to the relatively simple basis of these responses" (pp. 566-567). Within suggests that the changes in behavior resulting from differential locus of lesions may be the result, essentially, of removing from the animal the capacity to receive particular sensory cues within the maze. Within concludes, "The 'hypotheses' concept and the system of which it is a part arose in opposition to the crudely mechanistic conceptions which have gained prominence.... As opposed to such piecemeal conceptions, the 'hypotheses' concept attempted to present the learning process as orderly, organized, and predictable, and pictured the learner as a more active participant in determining the character of the final learned habit. This...is a very desirable end, but since..the 'hypotheses' concept swings to an extreme where it is beyond the bounds of evidence, it constitutes a very weak kind of opposition..." (p. 567).
Spence also had been critical of Krechevsky's
interpretations, "Oontrary to the belief of certain writers, ... there has been no disagreement concerning the behavioral facts.... Agreement ceases however, over attempts at further










interpretation of the phenomena" (1945, p. 253). In an early theoretical paper on the nature of discrimination learning, Spence (1936) demonstrated that perseverative behavior similar to hypotheses was compatible with the Hull-Spence system. Hypotheses were considered as a phenomenon dependent on learning processes before the correct learned response became manifest; as such, it is a phenomenon characteristic of a pro-solution period. At a later time Spence distinguished between this perseverative pre-solution behavior and the hypotheses which Kreohevsky found in an insoluble mase, "...these pre-solution phenomena appear to be a typical example of what has been described as trial and error learning, while hypotheses are far from what...(I understand)...by the terms insightful and intelligent. Only persistent non-adaptive responses can attain the distinction of being hypotheses-for, in order to classify as a hypothesis, a response, although ineffective, must continue to be persisted in a certain minimum number of times. A maladaptive act which is speedily (intelligently?) abandoned cannot ever be a hypothesis" (1940, p. 287). Both forms of perseverative behavior are, for Spence, predictable from his theoretical framework. As a pre-solution phenomenon, the hypothesis is expected to be replaced by the correct response without any intermediary hypotheses. In the insoluble problem situation,








12

the development of hypotheses (or the complete nonoccurrence of them), shifts between hypotheses, and the persistence of hypotheses, are all theoretically explainable hat ho providing certain assumptions regarding initial response strengths and reinforcemnts can be made. Unfortunately, the Krechersky hypothesis maze does not lend itself to an empirical test of the Spence position.
There was no further work on hypothesis behavior until 1954 when Erech Rosenzweig, and Bennett began publishing the results of a series of experiments which correlated brain chemistry with adaptability. Despite the earlier criticisms of hypothesis behavior, this technique was used as the measure of adaptability. Two changes were made in this measure. First, a reduced level of significance was used as a criterion of hypotheses (33 rather than 35 choices of the 48 occurring each day), and second, Krechevsky's (1933) spatial-visual Preference Score was also used. This Preference Score provided a continuous scale of relative preference for either visual or spatial modalities within each animal. It is computed in this fashions 1) the number of choices in each of the eight hypothesis alternatives is determined for an animal for a


reerstky changed his name to reh.








13

given day; 2) the deviation in per cent of these eight numbers from fifty per cent is computed; 3) the percentages based on spatial hypotheses are given positive signs, the visual percentages are given negative signs, and the eight percentages are then algebraically added.
In a later study by Rosenzweig, et al. (1958), the results indicated that the amount of cholinesterase (ChE) significantly increased from the visual area (V) to the somesthetic area (S) to the motor area (M); the amount in all areas declined with age. When visual preference animals were compared to those with spatial preferences, the ChE rate of decline was found to be very rapid. The authors were using GhE as a measure of acetylcholine (ACh) metabolism, and inferred from these results that AOh metabolism is related to adaptive behavior patterns.
Such an inference necessitates two assumptions. One concerns the Ch_-ACh relation and is a chemical problem; the relation of hypothesis behavior to adaptability is properly a psychological problem which the authors justify as follows:
...we believe that animals that show a spatial
Preference Score are more adaptive than animals
showing a visual Preference Score. We make this interpretation for the following reasons: (1) As we have already pointed out, our test is designed
to measure the animal's perceptual selectivity.
(2) We next assume that adaptive behavior is correlated with the ability of the animal to "pay
attention to" various stimulus aspects of its








14

environment when confronted with a problem. Thus,
for example, Tolman stresses the importance of
determining the "...cohditions which favor relatively rapid shifts in the dimension of discrimination of a
sig or of a significate..." in understanding
differential effectiveness of performance. (3) In our mase, under the conditions of training that we
have used, almost all animals show a light-going
preference on their first few trials... Achievin a spatial Preference Score therefore requires that an animal ignore the dominant illumination cue and pay
attention to the less obvious cue of location in
space. In other words, an animal that develops a
spatial preference shows readier "shifts in discrimination" than an animal whose behavior remains controlled by the visual cues. (1958, p. 389)

While this is an attractive line of reasoning, there are certain disturbing features. Whether a continuous Preference Score or the discrete category, hypotheses, is used, the criticisms of Witkin and Spence remain pertinent and have not been answered. In addition, very little is known about hypothesis behavior and the variables affecting it; the phenomenon has been a theoretical pawn and there have been no investigations since Witkin's efforts twenty years earlier. Despite the occasional criticisms and the lack of a sound empirical foundation, Krech treats hypothesis behavior as definitely adaptive, cognitive, attempted solutions. He receives some support for this position from standard referference works which review his work; hypothesis behavior is unoritically treated as evidence for attempted solutions on the part of the rat (see Munn, 1950, pp. 250253, 333; Stevens, 1951, pp. 312, 744, 774-776; Osgood, 1953, pp. 445-446; ilgard, 1956, p. 201).








15
The results of experiments are sometimes discussed with reference to hypotheses. Petrinovitch and Bolles investigated the effects of kind of deprivation on learning. Using hungry and thirsty rate in a single-unit-T-maze, they found that when the goal was alternated on every trial the food-deprived animals were better, but when the goal remained fixed, the water-deprived animals were superior. They related their findings to Krechevsky's (1937) report that.normal animals alternate and vary their hypotheses. The authors suggested "In the light of the present experiment, it is possible that his data do not pertain to 'normal' animals, but only to normal hungry animals. In general, our results suggest that in any study of variability or stereotypy in behavior, the nature of the 'drive' used must be carefully considered" (1954, p. 452).
Cognitive theorists have been attaching increased
importance to drive states. Tolman (1949) has modified his earlier position and suggested that drive reduction may be necessary for some types of learning. Similarly, Krech no longer holds to a strict cognitive theory of learning. He writes "It is quite obvious, if you examine many of the theories extant in psychology today, that these theories do not rest content with motivational constructs only, or with cognitive constructs only, but require both sets of constructs.... The belief in the interaction among the








16

hypothesized cognitive and motivational processes is certainly one of the most outstanding,..urrent trends in... theories today (1951, p. 114).
The effect on hypotheses of manipulating drive remains unknown. Motivational variables are known to facilitate or decrease performance levels, but it is difficult to apply the motivational literature to hypothesis behavior for several reasons. First, most of the experiments are more applicable to motivational phenomena or theory. Second, the general nature of hypothesi behavior is unknown. And third, data in hypothesis behavior experiments are not directly comparable to data in the usual learning experiment with only one solution.
The present study was directly concerned with these problems. It was designed to determine empirically the relationship between the hypotheses formed by an animal and another measure of that animal* adaptability; it was also designed to explore the effects of motivation on hypothesis behavior and adaptability.
The following researc hypotheses were advanced: 1) The type of hypotheses and the frequency with which
they are formed is a function of both the level of
motivation and the type of motivation.
Three major characteristics of hypothesis behavior could vary as a function of motivational level these may be








17
stated in question form:
a) Do animals at a low motivational level form
fewer, or more, hypotheses than animals more
highly motivated?
b) Do animals characteristically form visual
hypotheses at one level of motivation, and
spatial hypotheses at another?
c) Is motivational level related to shifts between
types of hypotheses?
Petrinovitch and Bolles had suggested that food deprivation might lead to more variation in hypothesis behavior than water deprivation. Accordingly, when level of deprivation is held constant, the following statement will be tested: 2) Do food deprived animals show more shifts between
hypotheses than water deprived animals.
The establishment of equivalent levels of motivation in two different modalities, food and water, presents difficulties in both laboratory technique and research design. True equivalence is impossible to attain since hunger and thirst are different experiences. An effective equivalence may possibly be approached by using equivalent operational definitions. The more usual approach of using hours of deprivation would not prove to be a satisfactory procedure; Stellar and Hill (1945) observed that rats may adjust their water consumption, within a limited time of access, to fit








18

their deprivational state. Brown (1961, pp. 71-74) presented data which leads to the same conclusion. The method which was used in the present study determined the d libitum food or water intake for a pilot group of animals and then provided that fixed percentage of such an amount which would result in equivalent running speeds in a straight runway. The experimental animals were also weighed Just before this limited amount deprivation schedule was started, and, immediately after it ended; this measure provided an additional means of defining motivational level. 3) The adaptability of animals is related to the type and
frequency of the hypotheses which they display.
Rosenzweig, and Bennett (1958) have argued that spatial preferences and hypotheses, and shifts of preference and hypotheses, represent more adaptable behavior since the animal has given up responding to an initial (apparently more dominant) aspect of the situation, The measure of adaptability used in the present study involved a similar logic; the animal was presented with the doors and lights of the hypotheses mase arranged so that one of the hypothesis alternatives was solvable. When criterion was reached, the maze was adjusted for solution of a seond hypothesis alternative. Adaptability was thus equated with the number of trials to adopt new modes of response as they became more efficient. The appropriate questions aret








19

a) Do animals that demonstrate spatial hypotheses
and preferences take fewer trials to solve the
problems than animals with visual hypotheses and
preferences?
b) Do animals that have many hypotheses take fewer
trials to solve the problems than animals with
few or no hypotheses?
c) Do animals that shift hypotheses and preferences
take fewer trials to solve the problems than
nimals that do not shift?
4) the adaptability of the animals is related to the type
and level of motivation.
The Terkes-Dodson (1908) law holds that optimal sotivatioal level i is nversely related to problem difficulty; in the present experiment problem difficulty was not effectively manipulated, therefore, according to t his law an optimal level of motivation should be apparent. The positions of Duffy (1951, 1957) and Malmo (1957), are also consistent with this. Brown (1961, p. 94), however, found equivocal results in the area of discrimination learning and concluded that further studies were "urgently needed," The relation. ship can be posed as the following questions
a) Does a linear relation exist between motivational
level and number of trials to solve the problems
such that increased motivational level results in
fewer trials?











CHAPTER II


PROCEDURE


The general experimental procedure as described by Krechevsky (1932b), and Rosenzweig, Krech, and Bennett (1958) was followed with certain exceptions. These exceptions were a difference in kind and amount of incentive, a slight modification of the maze, and a slightly higher criterion for determining hypotheses. The criterion was placed at 36 choices out of 48 for the following reasons Since each animal had a number of days on which to be scored for hypotheses, the level of significance should be adjusted accordingly. Similarly, the scoring of each animal's daily responses for hypotheses increased the likelihood of finding hypotheses in direct proportion to the number of alternative hypotheses. The four alternating and perseverating hypotheses were not scored so the level of significance was not adjusted to include these data in the determination of hypotheses. Significance levels were adjusted to remain for each animal at P ( .01 (see Ryan, 1959) in consideration of nine days opportunity to form an hypothesis and four scorings of each day's behavior (spatial -right, -left, and visual -light, -dark).

20









21

Subjects. Subjects were 32 naive male hooded rats, varying in age from 92 to 127 days at the start of the experiment. Of this group, one animal was lost from the experiment due to escape from the cage and an undeterminable amount of eating. All animals were housed individually. Water-deprived animals had ad libitum Purina rat pellets; food-deprived animals had ad libium water. They were maintained on a 12 hour light-dark cycle in an air conditioned room which had a mean temperature of approximately 780 F.
APparatus. The standard Erechevsky hypothesis maze was used, modified with five guillotine doors to prevent retracing among the start box, goal box, and discrimination units. The top was covered with window screening painted with aluminum paint to reflect the interior light of the maze and reduce room cues. Determining each animal's choices by mirrors proved unsatisfactory, and direct observation through the screen was used. The maze was placed on a table at the side of a small room. The only light came from the apparatus itself and a 74 watt bulb suspended over the goal box. This light permitted the experimenter to observe when the animal had reached the goal box and to record the animals' responses. The straight








22

runway which was used in the pretraining period was similar to the hypothesis mase but with the center sections removed so that no alleys were formed. The guillotine doors were retained and curtains were hung at the three center doors to provide habituation for the animals.
Procedure. Animals were treated in groups of about eight, with random assignment to each condition. The animals assigned to the food-deprivation category (PD) were placed on s4 libitum powdered Purina rat chow and water schedules; animals in the water-deprivation category (BD) received the same treatment except they were given pellets rather than powdered food. After a four-day adjustment period, measurements of each animal's food or water (as appropriate) consumption was taken for 14 days. At the end of this time the average daily consumption was computed and the animals were weighed.
A pilot study had indicated that 65 per cent of the A libitum food intake and 70 per cent of the ad libitum water intake would result in approximately equivalent running times in a straight runway; similar results were found for a 45 per cent food and a 50 per cent water level. Therefore those four levels were adopted. The animals were placed on the appropriate schedule one day prior to the start of the pretraining trials and remained on these diets until the conclusion of the experiment.








23

Pretraing. The animals were given 5 trials the
first day, 5 trials the second day, and then 10 trials each day for nine successive days; the animals were rmn every 24 hours. The incentive was approximately 0.1 gram of powdered chow or 0.1 c.c. of water, as appropriate. The incentive was taken from each animal's daily ration.
othesis trials. The animals were given 6 trials on both the first and second days, and 12 trials each of the successive eight days. The Krechevsky order (see Table 1) was followed. After the final day's trialsa within the hypothesis mase were completed, the hypotheses for each animal were computed. Weights were obtained for each animal.
Adaptability trials. On the basis of each animal's hypotheses and membership in the four experimental roups, a decision was made as to which of two visual problems and two spatial problems were to be solved. The criterion for solution was the same as for hypotheses; however, whereas hypothes were computed on the basis of an animal's daily performance, these trials were counted serially without respect to the termination of each days trials; such a procedure is necessary to prevent grossly unequivalent amounts of practice. An animal was considered to have reached criterion when it first attained 36 out of any 48 consecutive choices which were correct; animals that reached criterion on n the first, second, or third choice-points of a









24

trial were allowed to finish that trial and attain the incentive.
Roughly half of the animals which had made visual hypotheses were assigned spatial problems first; spatial hypothesis animals were treated similarly. When animals had made both types of hypotheses, or none, the assignment was made on the basis of the dominant preference. In all asesa, the assignment of an problem within either stimulus modality was opposite to that animals dominant preference.

The number of daily trials and the incentives

remained the same as in the latter part of the hypothesis ma eeen. A e experiment. At the conclusion of the first problem, each animal was immediately shifted to the second problem on the next trial. The scores were each animal's total number of trials to solve both problems.












CHAPTER III


RESULTS


This study was concerned with certain relationships between motivation, hypothesis behavior, and adaptability. Two semi-independent measures of motivation were used. First there was a deprivation schedule providing for restricted amounts of food (PD) or water (WD) at two different levels, moderate and severe. Second, a measurement of body weight change (BWO) over the course of the experiment was obtained.
Since the experimental design required equivalent levels of motivation in the _D and WD groups for testing certain hypotheses, evidence concerning this equivalence will be presented first. Percentages of d libitum food and water had been chosen to obtain equivalent running times. Speed of running was thus used as a criterion of equivalence. The concept of equivalence can be assumed to have been validly met if it can be shown that the deprivation schedules establish equivalent running times within each level of deprivation.
There are two sources of running time data in the
experiment, one from the training trials and the other from 25








26

the trials in the Krech apparatus. Response time data were available from 19 animals during the training trials to use as a check against this method. These data, presented in Table 2, showed that only a slight degree of equivalence existed between each of the moderate and each of the severe groups. These data were obtained in a straight runway and were therefore similar to the conditions of the pilot group. However, it is of more pertinence to determine whether equivalent running speeds existed at the time of the hypothesis behavior trials.
TABLE 2
EAN RUNNING TIM IN SECONDS IN THE TRAINING TRIALS
(Days 3 through 10, N = 19)
Deprivation
Level Food Water Moderate 23.9 17.7 Severe 7.6 15.6

Response time data for a sample of 20 animals in
the hypothesis maze are presented in Table 3. It will be noted that the moderate FD and severe WD groups were approximately equivalent, but the other groups occupy extreme positions. This indicated that type of deprivation, food or water, was possibly confounded with level of motivation. Analysis of variance established that such










TABLE 3
MEA RUNNING TIME IN SECONDS IN THE HYPOTHESIS MAZE
(Days 1-9, N = 20)
Level Food WVaer Moderate 32.1 49.2 Severe 20.4 35.2

confounding did take place. According to the original experimental procedure, a significant difference in running time should be found for the level of motivation variable (since the restricted diets were deliberately chosen to produce different running times). As the summary in Table 4 indicates, animals differed significantly (P < .005) in level of motivation as measured by running time level. However, these restricted diets were also deliberately chosen with the intent of producing no significant differonce between running times on the type of deprivation variable. Here, unfortunately, they also differed and this difference is significant (P < .001). Clearly, testing the experimental hypotheses relating to the food-water variable would introduce motivational level as a contaminating factor.








28
TALE 4
ANALYSIS OF VARIANCE ON THE EFFECTS OF TYPE AND
LEVEL OF DEPRIVATION ON RUNMNG TIM~ IN TE RYPOESIS AZE
(I' a log I transformation)

Source df. M.S. P
Type (Food-Water) 1 .230 20.35 .001 Level (Moderate-Severe) 1 .138 12.21 .005
Type x Level 1 .004 .35
Error 16 .011

Total 19


Another indication that equivalence did not exist was found when the mean BC was computed for each of the calls. Table 3 with running time data and Table 5 with mean body weight changes reflect the same general trend. This suggests that the two measures are correlated. This turned out to be true, with a rank order correlation of S *680 (P < .01). Because of this correlation, per cent of BWC, rather than the dichotomous loderate-Severe categories, was used as the indicator of motivational level in further analysis of the data.








29
TABLE 5
MEAN PF CENT OF BODY WEIGHT CHANGE OVER
THE CORE OF THE EXERIMENT. (N = 31)


Level lood WVater Moderate -13.1 -1.8 Severe -28.9 -8.8

To summarise the results obtained so far,
1. The desired two equivalent levels of motivation were not attained, since type of motivation was contaminated by level of motivation.
2. Per cent of BWO was significantly correlated with running time and was used as the independent variable representing moatvaional level.


Tests of the Rypotheses

A. The effect of varying otvational level.
The hypothesis was advanced in the introduction that the hypothesis-behavior of the animals would be a function of level of motivation. It is apparent from Figure 2 that varying motivational level induces major changes in both the gross amount and type of hypothesis behavior, spatial
(8) or visual (V). The more negative the BWC, the more hypotheses the animal forms.









30





7- N=4








5
U

S/ Spatial
o








S2- N8



1
Visual



+10-0 0- -10 -10- -20 -20- -30 -30- -40
% change in body weight

Fig.2.-Number and type of hypotheses formed as a function
of body weight (totals of all hypotheses within a class interval, divided by the number of subjects,
N, of the class.)









The form of the data and the nature of the inquiry lent itself in this case and in many of the tests to follow to the use of the%2 test of independence. The data are usually in frequency form and also in dichotomous categories such as I or hypotheses. aven where dichotomous categories do not exist it was meaningful to create at times such classes as above or below the median. The Yates correction of the formula was used (MoNemar, 1955, formula 86a). The null hypothesis (Ho) was rejected when(2. 05,
1 d.f* 3.84 was surpassed.
#I Animals do not differ in the number of hypotheses
formed as a function of being above or below the
median in BWO. (N -= 1).
Results 2 a 5.59 and the Ho is rejected. Another indication of this relationship is given by I which has a value of .663 (P < .001). (The F ratio between'r and r 1.057 and linear regression may be assumed). It is concluded that the number of hypotheses formed is a function of BWCI more bypotheses are formed with severe losses of body weight and fever hypotheses are formed with moderate increases of weight.

Hos That animals do not differ in formin 8 or V
hypotheses predominantly as a function of being
above or below the median in B8C.








32

In this test, N = 28 and animals forming both S and V hypotheses were included providing one type occurred with greater frequency. Eight animals were included from this mixed category and one was excluded. Result: 12 = 5.89 and the Ho is rejected. It is concluded that S hypotheses predominate with major negative BWO, and V hypotheses predominate with slight increases or decreases in body weight.
Ho: That animals do not differ in shifting or not
shifting hypotheses as a function of being
above or below the median in BWO.
Since 29 of the 31 animals formed hypotheses, N = 29. However, of these, only nine animals shifted hypotheses. Results X 2 = 0.46 and the Ho cannot be rejected. It should be noted from Figure 2 and from the results presented above that S hypotheses increase while V hypotheses decrease as the animals lose weight. Since shifting hypotheses involves by definition both V and S hypotheses, some restriction of this shifting within the more moderate levels of BWC might be expected. Inspection of the data in the Appendix, Table 9, reveals that six of these nine animals are above the median BWO. There are thus empirical grounds cautioning against acceptance of this Ho.








33
B. Te effect of varying food and water.
The contamination of the tye of deprivation by the level of deprivation prevented an adequate test of the effect of food and water on hypothesis behavior. Some indication of the effect may be gained by constructing roughly equivalent groups. Two such groups are the moderate FD and severe WD from the original design (note their similarity in Tables 3 and 5). Table 6 presents a comparison of these two groups and it will be seen that there is little difference in the frequencies.

TABLE 6
COMPARISO OF TE MODERATE FOOD DEPRIVED
AND SEVERE VATER DEPRIVED GROUPS FOR
FREQUENIES OF DOMINANT HTPOTHESIS TYPE
Reference fe Water

Spatial 5 5 Visual 2 3

Hatched groups can also be composed by choosing
animals with nearly equivalent BWO. Unfortunately, these are too few to warrant statistical test but the data are pre. seated in Table 7 for inspection. There is a slight indication that food deprivation may more likely be associated with 8 Bypothesis formation than water deprivation.










TABLE 7
FOOD AND WATER DEPRIVED ANIMALS MATCHED FOR
EQUIVALENT BODY WEIGHT CHANGE


BW Number and othi Trials to Solve rood Vae od V oIWater AP Water

-5.7 + 5.8 3 o 0 1 38 36
5.5 4.7 3 1 0 2 21 35
7.9 7,7 1 3 2 1 33 27
-18.5 -16.8 2 0 3 0 25 23
-31.2 -31.4 6 0 5 0 33 36


It will be shown in section 0 that 9s0 was not
significantly related to the shifting of hypotheses. Since it therefore seems probable that motivation is not a contaminating factor, we may test the food and water deprivation variable on the shifting of hypotheses. The resultingX2 .016 and the Ho is not rejected. The effect of this variable on the AP trials was also tested and 12 .0039 and the Ho is not rejected. It is concluded that type of deprivation was related neither to the shifting of hypotheses nor to the ease of solving the adaptability problems. It was not possible to test the influence of deprivation type on the kind of hypotheses formed.








35

G. The relation of hypothsis behavior to adaptabilit.
Adaptive behavior was measured by the total number of trials to solve two problems. These problems (AP) were arrangements of the doors of the rech apparatus such that one of the four hypotheses S right (SR) or left (SL), V light (VL) or dark (VD), was a correct solution, Each animal was presented with one of the S problems and one of the V problems, the order of presentation (S or V first) was determined by the experimenter who assigned roughly equal amounts of the orders to the various categories of hypothesis-forming animals.
It is possible that order effects exist in the
presentation of these problems. If an animal had shown an SR preference or hypotheses the use of problems SL and D in that order may have been more, or less, intrinsically difficult than the order VD, L.
A second source of order effect might be found in the choice of the particular cue of each modality; that is, an SR (or VD) problem also might be more or less intrinsically difficult than SL (or ) problems. However, this second order effect was controlled by giving all animals problems which differed from their last demonstrated preference (as revealed by the largest number of choices witthi he R, SL, VL, VD categories). If an animal had shown SR and VD preferences during the hypothesis trials,








36

the problems presented would be SL and VL, although not necessarily in that order. Thus, the first type of order effect varied while the second type was partially controlled.
An indication of the effect of this first type of order can be obtained from the point-biserial correlation between order (first problem the a, first problem opposite demonstrated preference) and the distribution of number of trials on the AP. This correlation has the value rPb a .013 which of course is not significant. When tested by(2, (Same-Opposite x More-Less median trials an the AP), X2 .028 and is not significant. Therefore it will be assumed that an order effect does not exist in the analysis of the data to follow.
Some statistics of the nimals' gross behavior are presented in Table 8. It will be noted that more animals formed hypotheses and more frequently, than any other TABEI 8
TYPE AND FREQUENCY OF HYPOTHESIS BEHAVIOR

mber of Type of Hypothesis Meq Number of Animals Formed Days Persisted
14 8 only 4.57 6 V only 2.67 9 Both 8 and V 4.11
2 none --








37

category. This differs from Krech's finding and will be discussed in the next chapter. However, on the basis of his argument for adaptability there appears to be a logical basis for inferring that animals which form V hypotheses are more adaptive since S hypotheses are more "popular." The California group had stated that on the first five or six trials their animals displayed a decided preference for the lighted alleys (Rosenzweig, et al., 1958). Accordingly, the data from these trials were analyzed. The Spatial minus Visual Preference score was computed for the first six trials and found to average -2.4 per cent. This indicates
that there may be a very slight general orientation of the animals to the light cues, light or dark. When the individual visual preferences are examined it is found that 13 preferred the lighted alleys to some degree. Thirteen of the animals also exhibited a preference for the darkened alleys, and for five of the animals the preferences were
equal. When the data were examined for the most marked preference, S or V, it was found that 16 animals preferred 8, 14 preferred V, and with one animal the choices were tied. Thus, it is not clear which category of hypotheses Krech would identify as the more adaptive on the basis of his line of argument. All alternatives will be tested.
Ho: Animals forming only S hypotheses and all
animals forming V hypotheses do not differ in








T8
respect to being above or below the median in
AP trials.
Result: e 2 a .281 and the Ho is not rejected. A test of the same type on V vs. S plus S and V frequencies yields ai 2 = .0023, and it is concluded that there is no evidence to identify either a V or S hypothesis factor with adaptability. The test may be applied to animals that shift hypotheses. (In this case, all animals that shifted hypotheses did so from one modality to another. None of the animals shifted within modalities.) TheX^A2 = .016 and the Ho is not rejected. It is concluded that there is no evidence to identify the shifting or non-shifting of hypotheses with adaptive behavior.
It may be argued that the more hypotheses an animal forms, the more adaptive that animal is. This proposition was tested by dividing the animals into categories above or below the median total number of hypotheses formed and above or below median number of trials an the AP. ^ 2 = .152 and the hypothesis is not rejected. It is concluded that there was no experimental basis for identifying frequency of hypothesis formation with adaptability.
In the recent publications of the California group, the implications of hypothesis behavior have been retained but the measure more usually relied on for the interpretation of the data has been the Spatial-Visual Preference Score
(PS). The authors have equated a S preference as being more







39

innately adaptive. This may be tested by a correlation between each animal's PS and the trials on the AP. A significant negative correlation would indicate that V preferences are associated with few trials on the AP; positive correlations would associate S preferences with ease of solution. The correlation (r) between these two measures was -.013; this is taken as indicating no relation existed between an animal's PS and its adaptability. When the number of shifts in algebraic sign of the PS are correlated with the AP trials, r -.229 which is not significant (P > .10).
To summarize the results of this section, no evidence was found in the present study to identify either type or amount or method of measuring hypothesis behavior with the measure of adaptive behavior (AP). D. The effect of level of motivation on adaptabilit.
As will be noted from igure 3, there is a general
U-shaped relation between motivational level and AP trials; an optimal level appears to exist between a -10 per cent and -20 per cent BWC. The extent of this relation is given by the coefficient I which is .486 (P > .10) and by r -.275 (P > .10). This also results in an insignificant
2. It is concluded that there is little statistical evidence for an optimal level of motivation for solving the AP, although a slight U-shaped relation is present.









40










40.




35

0
$4
0











+10-0 O- -10 -10- -20 -20- -30 -30 -40
% change in body weight

Fig.3.-The relation between body weight change and
average number of trials to solve both
adaptability problems.











ORAPTER IV

DISCUSSION


Consideration of the Experimental Hypotheses

A. Motivation and =yothesis behavior. One research hypothesis advanced earlier in this paper stated that the frequency, types, and shifts of hypothesis behavior would be some function of motivational level. The results have shown: (1) that an increase in motivational level, as measured by body weight change, was accompanied by an increase in the total number of days a hypothesis was exhibited; (2) that, while S and V hypotheses occurred with approximately equal frequency at the lower levels of motivation, V types declined but S types increased as motivational level increased; (3) that no statistically significant relation between motivational level and shifting of hypotheses was present, although six of the nine animals that did exhibit shifts were below the median in amount of body weight lost.
The effect of motivational level on hypothesis
behavior had not been investigated prior to this study, and in neither the prior work of Krechevsky, nor in his more current work with the California group, has there been any 41










indication that these effects might exist. Particular forms of hypothesis behavior (S preferences and shifts in preference) were assumed by the California group to represent more adaptive behavior than V preferences and lack of shift in preference. Since shifts predominate at lower levels while S hypotheses predominate at higher levels of motivation, the results are apparently inconsistent with these assumptions. Rather, the present results indicate that such shifts in hypothesis behavior should more parsimoniously be related to motivational levels rather than to the relative adaptive level of some particular form of hypothesis behavior or the "adaptability" of the animal.
A second hypothesis stated that food and water
deprivation would lead to different kinds of hypothesis behavior. This hypothesis proved to be untestable since the attempt to establish the required operationally equivalent motivational levels (measured by running speed and weight change) failed. This failure may be attributed to three causes. First, too small a sample of animals was used in the pilot study. Second, although it was not apparent from the pilot study, it now seems that the use of a percentage of ad libitum food or water intake does not produce sufficiently homogeneous running times or weight changes. Third, the relatively long period (about a month) of the deprivation schedule may have led to different effects within the food and water conditions.








43

The method which was utilized (reduced percentage of ad libitum intake) was an attempt to overcome certain difficulties of the traditional limited time schedules (water deprived rats may adjust their water intake to the amount of time available) and limited amount schedules (intakes and nutritional requirements vary partly as a function of age, body size, and metabolism). It now appears that sufficiently precise control would be obtained by reducing the body weight of each animal to some predetermined level before the start of the experimental program; this body weight level could be maintained by making daily adjustments in each animal's daily food or water ration.

B. Rypothsis behavior and aptability. A third experimental hypothesis asserted a positive relationship between type and frequency of hypothesis behavior and adaptability (as measured by the trials to solve the adaptability problems, A?). In the present study no relation was found between the AP measure and total number of hypotheses formed, shifts in hypotheses, or types of hypotheses. This is a result which is in apparent opposition to the assumptions of Krech and his associates since a particular type of hypothesis behavior (spatial) and shifts in hypothesis behavior had been identified as "adaptive." There has not been any previous empirical evidence comparing hypothesis behavior with some other measure of adaptability.










These results may be interpreted to mean that
hypothesis behavior is not a valid measure of adaptability. However, it is possible that the measure of adaptability, AP, used in the present study, is invalid. This is a crucial question, but since there is little precedent and no consensually agreed upon definition of adaptive behavior in the psychological literature, this point must at present be settled on logical grounds. Hypothesis behavior has been the major, but seldom used, convention for assessing this variable. While other measures such as the HebbWilliams apparatus might have been used, the AP measure was chosen because it uses the same apparatus, procedure, criteria, and perceptual-motor capacities equally important, its logic is congruent with Krech's reasoning that hypotheses are adaptive. It differs essentially in permitting a solution to take place. There exists such a high degree of similarity between hypothesis behavior and the AP measure, that the lack of any apparent relation between the two raises serious doubts as to the "adaptive" nature of hypothesis behavior. It also raises the question as to what hypothesis behavior is.

C. Motivation and trials to solve the two adaptability problems. A fourth hypothesis was advanced that adaptability would be related to motivational level. The results were








45

inconclusive; a U-shaped relation indicating an optimal motivational level (between -10 and -20 per cent BWO) was found but statistically significant correlations were not obtained. The concept of an optimal motivational level dates back to an original formulation by Yerkes and odson (1908). It has been found to be generally true in such studies as the Columbia obstruction box experiments (Warden, 1931), and more recently arousal level theorists have postulated a similar relation (Freeman, 1940; Duffy, 1957).

D. Conclusions. The relation of various hypothesis behavior measures to the solution of two problems has been examined and no support was found for the assumption that a particular type of hypothesis behavior can be considered intrinsically more adaptive than some other type. It has also been shown that the hypothesis behavior exhibited by an animal is a function of the animal's motivational level. It was therefore considered more parsimonious to relate these hypothesis behavior characteristics to motivational level rather than to assumed levels of adaptability. It was not possible to test the effects of food and water deprivation on hypothesis behavior since the attempt to establish operationally equivalent motivational levels in these categories was not successful. Finally, some evidence of an optimal level of motivation for performance was obtained but the evidence was not statistically significant.












CHAPTER V

SUMMARY

This study investigated relationships between
Krech's various measures of hypothesis behavior in the rat and adaptability, motivational level and hypothesis behavior, and motivational level and adaptability.
Thirty-one hooded rats were placed on limited foodor water-diets. After receiving trials to determine hypothesis behavior in the Krech insoluble maze, the animals were tested for adaptability by number of trials to solve two problems in the same apparatus.
The results indicated that:
1. There was no apparent relation between type, duration, or shifts in Krech's hypothesis behavior and the measure of adaptability.
2. The type of hypothesis behavior and its duration was a function of motivational level; hypothesis behavior which utilized light cues gradually diminished with increased body weight loss while hypothesis behavior based on spatial cues rapidly increased.

3. There was slight evidence of a curvilinear
relation of adaptability to motivational level but it was not statistically significant.

46








47

4. It was not possible to determine differences in hypothesis behavior as a function of type of deprivation because of an inability to establish equivalent motivational levels.
The results were interpreted as evidence against
Krech's assumptions that hypothesis behavior is adaptive. A motivational interpretation was considered to be the most parsimonious organization of the data.












BIBLIOGRAPHY


Boring, E. G. A histo of e erimental apcholo
(2nd ed.) rew York: Appleton-entury- rofts, 1950.
Brown, J. 8. Motivation New York:s oGraw-Hill, 1961.
Duffy, E. The concept of energy mobilization, Psychol.
Rev., 1951, 8, 30-40.
The psychological significance of the concept
of 'arousal" or "activation." Psychol. Rev., 1957,
64., 265-275.

Freeman, G. L. The relationship between performance
level and bodily activity level. J. x. Psychol,,
1940, 26, 602-608.
Hamilton, G. V. A study of trial and error reactions in
mammals. J. ALt. Behav., 1911, 33-66.
Hilgard, E. R. Theories of learning. New York: AppletonCentury-Crofts, 1956.
Krech, D. Cognition and motivation in psychological
theory. In W. Dennis (Ed.), Current trends in
holical theory. Pittsburghs Urv. Pits urgh
Press, 1951 PP 111-139.
., Rosenzweig, H. R., Bennett, E. L., & Krueckel, B.
Mnsyme concentrations in the brain and adjustive behavior patterns. Science, 1954, 120, 994-996.
Krechevsky, I. "Hypotheses" versus "chance" in the presolution period in sensory discrimination learning,
niv. Galif. Puht. Psychol., 1932, 27-44.
The genesis of "hypotheses" in rats. Univ.
allf.. Pbl. Psychol., 1932, 6, 45-64.
The docile nature of "hypotheses." J. 9omg,
Pc0ol1, 1933, 1, 429-443.
Hereditary nature of "hypotheses," J. omp,.
Psychol., 1933, 16, 99-116.
48








49

Krechevsky, I. Brain mechanisms and "bypotheses." T.
eom. Psychol., 1935, 1 425-468.
Brain mechanisms and variability I. Variability
within a means-end-readiness. J. comp. Psychol., 1937,
E, 121-138.
Brain mechanisms and variability: II. Variability
where no lea-r~ng is involved. J. com. Psychol., 1397,
139-163.
Brain mechanisms and variability: III. Limita. tion s of the effect of cortical injury upon variability.
*J. comp. Psychol., 1937, 23, 351-364.
Lashley, Karl S. Brain mechanisms and intelligence.
Chicago: Univ. Chicago Press, 1929.
Malmo R. B. Anxiety and behavior arousal. Pshol. Rev.,
1457, 64, 276-287.

McNemar, Q. Psychological statistics. (2nd ed.) New York:
Wiley, 1955.
Munn, N. L. Handbook of psychological research on the rat.
Boston: Houghton-Mffiin, .1950.
Osgood, C. E. Method and theory in exerimental pschology.
New Yorks: Iord University Press, 1953.
Petrinovitch, L., & Bolles, R. Deprivation states and
behavioral attributes. J. gom. physio1. Psychol.,
1954, E, 450-453.
Rosenzweig, M. R., Krech, D., & Bennett, E. L. Brain
chemistry and adaptive behavior. In H. F. Harlow & C. N. Woolsey. Biological and biochemical bases of
behavior. Madison University of Wisconsin Press,
7 -67-400o.
Ryan, T. A. Multiple comparisons in psychological research.
Psychol. Bul., 1959, 4, 26-47.
Spence, K. W. The nature of discrimination learning in
animals. Psychol. Rev., 1936, 4_, 427-449.
rsus noncontinuoun veursu non-conious interpretations
of discrimination learning. Psychol. Rev., 194-0, +7,
271-288.









50

Spence, K. W. An experimental test of the continuity and
non-continuity theories of discrimination learning.
J*. 2esychol., 1945, J, 253-266.
Stellar, E. & Hill, J. H. The rat's rate of drinking as
a function of water deprivation. J. cm. physiol.
Psychol., 1952, 4, 96-102.
Stevens, S. S. (Ad.), Handbook of experimental psychology.
New York: Wiley, 1951.
Tolman, E. C. There is more than one kind of learning.
P~sychol. Rev., 1949, Sj, 144-155.
Warden, 0. J. Animal motivation studies. The albino rat.
New York: Columbia Univ. Press, 1931.
Witkin, H. A. 'Hypotheses' in rats: An experimental
critique. III. Summary evaluation of the hypotheses
concept. Psychol. Rev., 1942, 4# 541-568.
Yerkes, R. M., & Dodson, J. D. The relation of strength
of stimulus to rapidity of habit formation. J. co.
Neurol. Psychol., 1908, 18, 4.59482.


























APPENDII









52



TABLE 9

PER CENT BODY WEIGHT CHANGE, TYPE AND OCCURRENCE
OF HYPOTHESESo AND TRIALS TO SOLVE THE ADAPTABILITY PROBLEMS


Per Cent Deprivation
Body Condition Order of Presentation Weight Food (F) and Hypothesis Formed and Trials to Solve Change Water (W) on Day Number Adaptability Problems V S V
+ 9.7 70 W VD 3; SR 7 19 23 + 5.8 70 W YD 8 21 15 + 5.7 65 F SL 1, 2, 3 18 20 + 3.1 70 W VD 5, 6, 7, 8, 9 12 12 + 2.4 70 W none 13 12 + 2. 2 70 W SL 6, 7, 8, 9 30 18 + 1.7 70 W SR 5, 6; VD 7,8, 9 18 21 + 0.9 50 W none 12 15
1.6 50 W SR 2, 7 13 13
2.4 50 W SL 5, 6 13 14
2.7 70 W VD 4, 5; SR 9 25 11
4.7 50 W VD 5, 6 22 13
5.5 65 F SR 5, 8, 9; VD 7 11 10
6.8 50 W VD 7,8,9 11 13
7.5 50 W VD 4, 5, 8, 9 24 15
7.7 70 W VD Z; SR 6, 9 14 13
7.9 65 F D 3,7, 9; SL 4 20 13
-11.7 65 F SR 3, 5, 6, 7; VD 9 12 15









53
Table 9 Cont.

Per Cent Deprivation
Body Condition Order of Presentation Weight Food (F) and Hypothesis Formed and Trials to Solve Change Water (W) on Day Number Adaptability Problems V S V
-16.8 S0 W SL 3,4, 11 12
-18.5 65 F 6W 6, 8 13 12
-20.5 65 F VD 8 12 16
-21.7 65 F SL 4, 5, 67, 8, 9 12 13
-23.1 45 F SR 5, 6, 7,8, 9 18 18
-23.5 65 F VD 3; SR 1. 7. 8, 9 16 17
-25.3 45 F SL 2, 3,4,6,7,8 10 14
-26.6 45 F SL, 4,5, 6, 7, 8,9 22 11
-28.9 4S F SR 3,4,5,6; VL 7 18 13
-31.2 45 F SL 4,S 6, 6.7,8,9 16 17
-31.4 50 W SR 3, 5, 7, 8,9 18 18
-33.6 45 F SL 4, 5, 6, 7, 8, 9 12 13
-33.7 45 F $L 2, 3, 4, 5, 6, 7, 8, 9 12 12
,, ,- .... .. .. .. ..











BIOGRAPHICAL SKETOR

Robert Louis Procter was born March 26, 1928 in Washington, D. C. The son of a Naval Officer, he lived in Virginia, New York, California, and Massachusetts until his father's retirement in Vermont in 1937. He graduated from Springfield (Vt.) High School in 1945 and was accepted into the Army Specialized Reserve Training Program and assigned to Massachusetts State College. He was later transferred to Korea and then discharged from the Army in 1947. He was admitted to the University of Missouri in 1948 and received his B.A. degree in 1951 and his M.A. degree in psychology in 1953. He returned to the Army in that year and was discharged in 1955. He entered the University of Florida in 1955 and held a series of graduate assistantships, primarily in clinical and comparative psychology. He was also employed at Sunland Training Center, Gainesville, Florida, and Rome (New York) State School, where he was Senior Clinical Psychologist. His
internship was at the Veterans Administration Hospitals in Gulfport and Biloxi, Mississippi. He is a member of Psi Chi, psychology honorary, and the Florida Psychological Association.


54












This dissertaion was prepared under the direction of the chairman of the candidate's supervisory committee and has been approved by all members of that committee. It was submitted to the Dean of the College of Arts and Sciences and to the Graduate Council, and was approved as partial fulfillment of the requirements for the degree of Doctor of Philosophy.


June 11, 1962



Dean, COllege of Arts dnd Sciences



Dean, Graduate School

Supervisory Committee a




Full Text
13
given day; 2) the deviation in per cent of these eight
numbers from fifty per cent is computed; 3) the percentages
based on spatial hypotheses are given positive signs, the
visual percentages are given negative signs, and the eight
percentages are then algebraically added.
In a later study by Rosenzweig, et al. (1958), the
results indicated that the amount of cholinesterase (ChE)
significantly increased from the visual area (V) to the
somesthetic area (S) to the motor area (M); the amount in
all areas declined with age. When visual preference
animals were compared to those with spatial preferences, the
ChE rate of decline was found to be very rapid. The authors
were using ChE as a measure of acetylcholine (ACh)
metabolism, and inferred from these results that ACh
metabolism is related to adaptive behavior patterns.
Such an inference necessitates two assumptions. One
concerns the ChS-ACh relation and is a chemical problem;
the relation of hypothesis behavior to adaptability is
properly a psychological problem which the authors justify
as follows!
...we believe that animals that show a spatial
Preference Score are more adaptive than animals
showing a visual Preference Score. We make this
interpretation for the following reasons: (1) As
we have already pointed out, our test is designed
to measure the animal's perceptual selectivity.
(2) We next assume that adaptive behavior is cor
related with the ability of the animal to "pay
attention to" various stimulus aspects of its


37
category. This differs from Kreoh*s finding and will he
discussed in the next chapter. However, on the basis of
his argument for adaptability there appears to be a logical
basis for inferring that animals which form V hypotheses
are more adaptive since S hypotheses are more "popular."
The California group had stated that on the first five or
six trials their animals displayed a decided preference for
the lighted alleys (Rosenzweig, et al.. 1958). Accordingly,
the data from these trials t/ere analyzed. The Spatial minus
Visual Preference score was computed for the first six
trials and found to average -2.4 per cent. This indicates
that there may be a very slight general orientation of the
animals to the light cues, light or dark. When the
individual visual preferences are examined it is found that
13 preferred the lighted alleys to some degree. Thirteen
of the animals also exhibited a preference for the darkened
alleys, and for five of the animals the preferences were
equal. When the data were examined for the most marked
preference, S or V, it was found that 16 animals preferred
S, 14 preferred V, and with one animal the choices were
tied. Thus, it is not clear which category of hypotheses
Krech would identify as the more adaptive on the basis of
his line of argument. All alternatives will be tested.
Hot Animals forming only S hypotheses and all
animals forming V hypotheses do not differ in


8
under which, animals give up or persist in hypotheses. Hats
which had formed hypotheses were divided into three groups.
One group then found the maze solvable in conformance with
the hypothesis each animal was exhibiting; the second group
had the maze solvable for a different hypothesis; a third
group had the incentive removed as soon as an hypothesis
was demonstrated. The first group persisted in their
original hypotheses, the second group changed hypotheses to
the correct solution, and in the third group more than fifty
per cent of the animals refused to run at all. This study
was integrated with a further exposition of Tolman's theory
and Krechevsky*s most important conclusion was that "...if
any behavior-act can be established as docile and purposive,
hypothesis-behavior is definitely so (1933a, p. 442),
Krechevsky began a series of investigations on the
relation of hypothesis behavior to the structure and function
of the brain (1935 1937a, b, c). He found that cortical
lesions decreased the number of different hypotheses an
animal might use, but increased the number of days an animal
would use the same hypothesis. He also found evidence that
this behavior was not related in simple fashion to the
amount of cortical damage, but to both the amount and locus
of damage. One area (occipital lobe) he called V because
few animals made visual hypotheses when it was damaged.
Another area (somesthetic) was identified as S, because of


26
tlie trials in the Krech apparatus. Response time data were
available from 19 animals during the training trials to use
as a check against this method. These data, presented in
Table 2, showed that only a slight degree of equivalence
existed between each of the moderate and each of the severe
groups. These data were obtained in a straight runway and
were therefore similar to the conditions of the pilot group.
However, it is of more pertinence to determine whether
equivalent running speeds existed at the time of the
hypothesis behavior trials.
TABLE 2
MEAN RUNNING TIME IN SECONDS IN THE
TRAINING TRIALS
(Days 3 through 10, N 19)
Deprivation
Level
Pood
Water
Moderate
23.9
17.7
Severe
7.6
15.6
Response time data for a sample of 20 animals in
the hypothesis maze are presented in Table 3. It will be
noted that the moderate PD and severe WD groups were
approximately equivalent, but the other groups occupy
extreme positions. This indicated that type of deprivation,
food or water, was possibly confounded with level of
motivation. Analysis of variance established that such


43
The method which was utilised (reduced percentage
of ad libitum intake) was an attempt to overcome certain
difficulties of the traditional limited time schedules
(water deprived rats may adjust their water intake to the
amount of time available) and limited amount schedules
(intakes and nutritional requirements vary partly as a
function of age, body size, and metabolism). It now appears
that sufficiently precise control would be obtained by
reducing the body weight of each animal to some predetermined
level before the start of the experimental program; this
body weight level could be maintained by making daily
adjustments in each animal*s daily food or water ration.
B. Hypothesis behavior and adaptability. A third
experimental hypothesis asserted a positive relationship
between type and frequency of hypothesis behavior and
adaptability (as measured by the trials to solve the
adaptability problems, A?). In the present study no
relation was found between the AP measure and total number
of hypotheses formed, shifts in hypotheses, or types of
hypotheses. This is a result which is in apparent oppo
sition to the assumptions of Krech and his associates since
a particular type of hypothesis behavior (spatial) and shifts
in hypothesis behavior had been identified as "adaptive."
There has not been any previous empirical evidence comparing
hypothesis behavior with some other measure of adaptability


APPENDIX


47
4. It was not possible to determine differences in
hypothesis behavior as a function of type of deprivation
because of an inability to establish equivalent motivational
levels*
The results were interpreted as evidence against
Krech's assumptions that hypothesis behavior is adaptive.
A motivational interpretation was considered to be the most
parsimonious organization of the data.


TABLE OP CONTENTS
Page
ACKNOWLEDGMENTS ...... ....... ii
LIST OP TABLES iv
LIST OP PIGURES y
CHAPTER
I.INTRODUCTION 1
II.PROCEDURE ......... 20
III,RESULTS 25
IV.DISCUSSION. ... 41
V.SUMMARY .................. 46
BIBLIOGRAPHY, . . . 48
APPENDIX 51
BIOGRAPHICAL SKETCH 54
iii


7
choices to each of the four cues (Left, Right, Light, Dark).
Krechevsky defined non-random, systematic behavior his
hypothesis behavior as being present when the animal chose
one of the eight alternatives with a frequency significantly
greater than the theoretical random mean of 24. The level
of significance which he chose was three standard deviations
above the mean. This amounts to 34.39 choices out of 48.
With this criterion, Krechevsky found that most of
his animals formed hypotheses and from these early studies
he concluded that "...the presence of systematic forms of
behavior in such a situation is to be interpreted to mean
that these systems were determined not as something forced
ab extra by the situation, but as something originating
from the animal himself" (1932b, p. 45). He suggested
that "...descriptions of the lower animals* behavior as
consisting of "stereotyped," "haphazard," "non-insightful"
responses are to be attributed not to a lack of insight on
*
the animals' part but rather to a lack of insight on the
experimenter's part" (1932b, p. 63).
Over the next several years, Krechevsky continued
his investigation of hypothesis behavior. In one of these
studies (1933b) he used animals from Tryon's "bright" and
"dull" strains and found that the "bright" animals had more
spatial hypotheses and the "dull" animals had more visual
hypotheses. Another study examined some of the conditions


CHAPTER III
RESULTS
This study was concerned with certain relationships
between motivation* hypothesis behavior* and adaptability.
Two semi-independent measures of motivation were used.
First there was a deprivation schedule providing for
restricted amounts of food (FD) or water (WD) at two
different levels, moderate and severe. Second, a measure
ment of body weight change (BWC) over the course of the
experiment was obtained.
Since the experimental design required equivalent
levels of motivation in the FD and VD groups for testing
certain hypotheses, evidence concerning this equivalence
will be presented first. Percentages of ad libitum food
and water had been chosen to obtain equivalent running
times. Speed of manning was thus used as a criterion of
equivalence. The concept of equivalence can be assumed to
have been validly met if it can be shown that the
deprivation schedules establish equivalent running times
within each level of deprivation.
There are two sources of running time data in the
experiment, one from the training trials and the other from
25


27
TABLE 3
MEAN RUNNING TIME
IN
SECONDS
IN THE
HYPOTHESIS
MAZE
(Days 1-9,
N
- 20)
Level
Pood
Water
Moderate
32.1
49.2
Severe
20.4
35.2
confounding did take place. According to the original
experimental procedure, a significant difference in running
time should be found for the level of motivation variable
(since the restricted diets were deliberately chosen to
produce different running times). As the summary in Table
4 indicates, animals differed significantly (P < .005) in
level of motivation as measured by running time level.
However, these restricted diets were also deliberately
chosen with the intent of producing no significant differ
ence between running times on the type of deprivation
variable. Here, unfortunately, they also differed and this
difference is significant (P < .001). Clearly, testing the
experimental hypotheses relating to the food-water variable
would introduce motivational level as a contaminating
factor.


MOTIVATION, KRECH HYPOTHESIS
BEHAVIOR, AND ADAPTABILITY
By
ROBERT L. PROCTER
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
June, 1962


11
interpretation of the phenomena (19^5 p. 253)* In an
early theoretical paper on the nature of discrimination
learning, Spence (1936) demonstrated that perseverative
behavior similar to hypotheses was compatible with the
Hull-Spence system. Hypotheses were considered as a
phenomenon dependent on learning processes before the
correct learned response became manifest; as such, it is a
phenomenon characteristic of a pre-solution period. At a
later time Spence distinguished between this perseverative
pre-solution behavior and the hypotheses which Krechevsky
found in an insoluble maze, "...these pre-solution phenomena
appear to be a typical example of what has been described
as trial and error learning, while hypotheses are far from
what...(I understand)...by the terms insightful and
intelligent. Only persistent non-adaptive responses can
attain the distinction of being hypothesesfor, in order
to classify as a hypothesis, a response, although
ineffective, must continue to be persisted in a certain
minimum number of times. A maladaptive act which is
speedily (intelligently?) abandoned cannot ever be a
hypothesis" (1940, p* 287). Both forms of perseverative
behavior are, for Spence, predictable from his theoretical
framework. As a pre-solution phenomenon, the hypothesis is
expected to be replaced by the correct response without any
intermediary hypotheses. In the insoluble problem situation,


44
These results may be interpreted to mean that
hypothesis behavior is not a valid measure of adaptability.
However, it is possible that the measure of adaptability,
AP, used in the present study, is invalid. This is a
crucial question, but since there is little precedent and
no consensually agreed upon definition of adaptive behavior
in the psychological literature, this point must at present
be settled on logical grounds. Hypothesis behavior has
been the major, but seldom used, convention for assessing
this variable. While other measures such as the Hebb-
Williams apparatus might have been used, the AP measure
was chosen because it uses the same apparatus, procedure,
criteria, and perceptual-motor capacitiesj equally
important, its logic is congruent with Krech's reasoning
that hypotheses are adaptive. It differs essentially in
permitting a solution to take place. There exists such a
high degree of similarity between hypothesis behavior and
the AP measure, that the lack of any apparent relation
between the two raises serious doubts as to the "adaptive
nature of hypothesis behavior. It also raises the question
as to what hypothesis behavior is.
C. Motivation and trials to solve the two adaptability
problems. A fourth hypothesis was advanced that adaptability
would be related to motivational level. The results were


animal chose an alley with, a locked door, it was forced to
reverse its course and go through the alley on the other
5
side.
The animal was given twelve trials each day. On
every trial the lights and doors were shifted in such a way
that: 1) the lights and open doors were on either side of
the choice-point equally often, and, 2) the use by the
animal of any of the eight hypotheses which Krechevsky
identified would result in encountering the same number of
open and locked doors. (See Table 1 for the trial order
positions of lights and doors.)
TABLE 1
THE KRECHEVSKY TRIAL ORDER FOR THE POSITION OF OPEN
DOORS AND LIGHTED ALLEYS ON THE TWELVE DAILY TRIALS
IN THE MAZE
Trial
1
2
3
4
Door Open
L L R L
R R R L
R R L
R
L
L R
R
Light On
R
L R L
R L R R
L R R
R
L
R L
L
Trial
5
6
7
8
Door Open
R
R L L
R L L L
R L L
R
L
R L L
Light On
R
R L L
L L R R
R R L
L
R
L L
L
Trial
9
10
11
12
Door Open
L
R R L
R L R R
L L L
R
L
R R
R
Light On
L
L L R
L R R L
R L R
R
L
R L
R


28
TABLE 4
ANALYSIS OF VARIANCE ON THE EFFECTS OF TYPE AND
LEVEL OF DEPRIVATION ON RUNNING TIMES IN THE
HYPOTHESIS MAZE
(X* log X transformation)
Source
d.f.
M.S.
F
P<
Type (Food-Water)
1
.230
20.35
.001
Level (Moderate-Severe)
1
.138
12.21
.005
Type x Level
1
.004
55
Error
16
.011
Total
19
Another indication that equivalence did not exist
was found when the mean BVC was computed for each of the
cells. Table 3 with running time data and Table 5 with
mean body weight changes reflect the same general trend.
This suggests that the two measures are correlated. This
turned out to be true, with a rank order correlation of
.680 (P < .01). Because of this correlation, per cent
of BVC, rather than the dichotomous Moderate-Severe
categories, was used as the indicator of motivational level
in further analysis of the data.


LIST OF TABLES
Table Page
1. The Krechevsky Trial Order for the Position
of Open Doors and Lighted Alleys on the
(Twelve Daily Trials in the Haze *. 5
2. Hean Sunning Time in Seconds in the Training
Trials 26
3. Hean Running Time in Seconds in the
Hypothesis Haze * 27
4. Analysis of Variance on the Effects of Type
and Level of Deprivation on Running Times in
the Hypothesis Haze 28
5. He an Per Cent of Body Weight Change Over the
Course of the Experiment* 29
6. Comparison of the Hoderate Food Deprived and
Severe Water Deprived Groups for
Frequencies of Dominant Hypothesis Type ... 33
7. Food and Water Deprived Animals Hatched for
Equivalent Body Weight Change 34
8. Otype and Frequency of Hypothesis Behavior 36
9. Per Cent Body Weight Change, Type and
Occurrence of Hypotheses, and Trials to Solve
the Adaptability Problems 52
iv


9
a similar effect on spatial hypotheses. Damage to both
areas tended to result in spatial hypotheses,
Krechevsky published no more articles on hypothesis
behavior until recent years, Hypothesis behavior served as
the prototype for other Tolmanian concepts such as means-
end-readiness, and Krechevsky began experimentation in
other areas. Also, while hypothesis behavior was directly
related to the controversy over the continuous or
discontinuous nature of learning, the hypothesis behavior
maze was not readily adaptable to investigating this
problem. Krechevsky designed other pieces of apparatus for
this purpose.
The only critique of hypothesis behavior which
exists in the psychological literature is that provided by
Vitkin (194-2). Vitkin noted that hypothesis behavior is
not typical of solvable problem situations, but rather of
situations for which no solution can be found. He found
the same type of behavior in the Krechevsky hypothesis
maze when the doors were completely removed (free-choice)
as when the doors were locked according to Krechevskys
insoluble pattern; the behavior was not found when a
solvable pattern was used. He questioned the adaptive or
purposive nature of hypotheses, since animals shifted
hypotheses in a free-choice situation although the first
hypothesis or no hypothesis at all was equally adaptive.


59
innately adaptive. This may be tested by a correlation
between each animal's PS and the trials on the AP. A
significant negative correlation would indicate that V
preferences are associated with few trials on the AP;
positive correlations would associate S preferences with
ease of solution. The correlation (r ) between these two
measures was -.013; this is taken as indicating no relation
existed between an animal's PS and its adaptability. When
the number of shifts in algebraic sign of the PS are cor
related with the AP trials, r -.229 which is not signifi
cant (P > .10).
To summarize the results of this section, no evidence
was found in the present study to identify either type or
amount or method of measuring hypothesis behavior with the
measure of adaptive behavior (AP).
D. The effect of level of motivation on adaptability.
As will be noted from Figure 3* there is a general
U-shaped relation between motivational level and AP trials;
an optimal level appears to exist between a -10 per cent
and -20 per cent BWO. The extent of this relation is given
by the coefficient which is .486 (P > .10) and by
I* ** -275 (P > .10). This also results in an insignificant
It is concluded that there is little statistical
evidence for an optimal level of motivation for solving the
AP, although a slight U-shaped relation is present.


50
Spence, K. W. An experimental test of the continuity and
non-continuity theories of discrimination learning.
J. exE- Psychol.. 1945* 255-266.
Stellar, E., & Hill, J. H. (She rat's rate of drinking as
a function of water deprivation. J. comp, physiol.
Psychol.. 1952, 4, 96-102.
Stevens, S. S. (Ed.), Handbook of experimental psychology.
Hew York Wiley, l9i?l. irTnrT'r,nnnTrl,'IM Tlnrn Tr'mrrr.mmT
Tolman, E. C. There is more than one kind of learning.
Psychol. Rev.. 1949, 6, 144-155.
Warden, 0. J. Animal motivation studies. The albino rat.
New York: Columbia tfniv. Press, l93l*
Witkin, H. A. 'Hypotheses' in rats: An experimental
critique. III. Summary evaluation of the hypotheses
concept. Psychol. Rev., 1942, 4£l, 541-568.
Yerkes, R. M., & Dodson, J. D. The relation of strength
of stimulus to rapidity of habit formation. J. comp.
Neurol. Psychol.. 1908, 18, 459-482.


38
respect to being above or below the median in
AP trials.
A
Result: .281 and the Ho is not rejected. A test of
the same type on V vs. S plus S and V frequencies yields
a*X2 .0023, and it is concluded that there is no evidence
to identify either a Y or S hypothesis factor with
adaptability. The test may be applied to animals that
shift hypotheses. (In this case, all animals that shifted
hypotheses did so from one modality to another. Hone of
A
the animals shifted within modalities.) The'X^ .016 and
the Ho is not rejected. It is concluded that there is no
evidence to identify the shifting or non-shifting of
hypotheses with adaptive behavior.
It may be argued that the more hypotheses an animal
forms, the more adaptive that animal is. This proposition
was tested by dividing the animals into categories above
or below the median total number of hypotheses formed and
above or below median number of trials on the AP. 2 .152
and the hypothesis is not rejected. It is concluded that
there was no experimental basis for identifying frequency
of hypothesis formation with adaptability.
In the recent publications of the California group,
the implications of hypothesis behavior have been retained
but the measure more usually relied on for the interpretation
of the data has been the Spatial-Visual Preference Score
(PS). The authors have equated a S preference as being more


LIST OF FIGURES
Page
Figure
1 Floor Plan of the Krechevsky Hypothesis
naze 4
2. Number and Type of Hypotheses Formed as a
Function of Body Weight. 30
3. The Relation Between Body Weight Change and
Average Number of Trials to Solve Both
Adaptability Problems. 40
v


52
TABLE 9
PER CENT BODY WEIGHT CHANGE, TYPE AND OCCURRENCE
OF HYPOTHESES, AND TRIALS TO SOLVE THE
ADAPTABILITY PROBLEMS
Per Cent
Body-
Weight
Change
Deprivation
Condition
Food (F) and
Water (W)
Hypothesis Formed
on Day Number
Order of Presentation
and Trials to Solve
Adaptability Problems
V
S
V
+ 9.7
70 W
VD 3; SR 7
19
23
+ 5.8
70 W
VD 8
21
15
+ 5.7
65 F
SL 1, 2, 3
18
20
+ 3.1
70 W
VD 5, 6, 7. 8. 9
12
12
+ 2.4
70 W
none
13
12
+ 2.2
70 W
SL 6, 7. 8, 9
30
18
+ 1.7
70 W
SR 5, 6} VD 7, 8, 9
18
21
+ 0.9
50 W
none
12
15
- 1.6
50 W
SR 2, 7
13
13
-2.4
50 W
SL 5,6
13
14
-2.7
70 W
VD 4, 5; SR 9
25
11
-4.7
50 W
VD 5, 6
22
13
- 5.5
65 F
SR 5, 8, 9; VD 7
11
10
- 6. 8
50 W
VD 7, 8, 9
11
13
-7.5
50 W
VD 4, 5, 8, 9
24
15
-7.7
70 W
VD 2; SR 6, 9
14
13
- 7.9
65 F
VD 3, 7, 9; SL 4
20
13
-11.7
65 F
SR 3, 5, 6, 7; VD 9
12
15


49
Krechevsky, I. 3rain mechanisms and ^hypotheses.M J.
comp. Psychol., 1935 12 425-468.
. Brain mechanisms and variability: I. Variability
wITEin a means-end-readiness. J. comp. Psychol.. 1937.
2t 121-138.
. Brain mechanisms and variability: II. Variability
where no learning is involved. J. comp. Psychol., 1397*
22, 139-163.
Brain mechanisms and variability: III. Limita-
tions of the effect of cortical injury upon variability,
j. comp. Psychol.. 1937, 22 351-364.
Lashley, Karl S. Brain mechanisms and intelligence.
Chicago: Univ. Chicago Press, 19-9.
Malmo. R. B. Anxiety and behavior arousal. Psychol. Rev..
1957, 64, 276-287.
McNemar, Q. Psychological statistics. (2nd ed.) New York:
Wiley, 1955.
nmmiolion, w th9 ra
Osgood, C. E. Method and theory in experimental psychology.
New York: Oxford University Press, 1953.
Petrinovitch, L., & Bolles, R. Deprivation states and
behavioral attributes. J. comp, physiol. Psychol,.
1954 42, 450-453.
Rosensweig, M. R., Krech, D.,J& Bennett, E. L. Brain
chemistry and adaptive behavior. In H. P. Harlow &
C. N. Woolsey. Biological and biochemical bases of
behavior. Madison:University of Wisconsin Press,
567-400.
Ryan, T. A. Multiple comparisons in psychological research.
Psychol. Bull.. 1959, 6, 26-47.
Spence, K. W. The nature of discrimination learning in
animals. Psychol. Rev.. 1936, 42, 427-449.
Continuous versus non-continuous interpretations
of discrimination learning. Psychol. Rev.. 1940, 47,
271288.


CHAPTER V
SUMMARY
This study investigated relationships "between:
Krech's various measures of hypothesis behavior in the rat
and adaptability, motivational level and hypothesis
behavior, and motivational level and adaptability.
Thirty-one hooded rats were placed on limited food-
or water-diets. After receiving trials to determine
hypothesis behavior in the Krech insoluble maze, the
animals were tested for adaptability by number of trials
to solve two problems in the same apparatus.
The results indicated that:
1. There was no apparent relation between type,
duration, or shifts in Krechs hypothesis behavior and
the measure of adaptability.
2. The type of hypothesis behavior and its duration
was a function of motivational level; hypothesis behavior
which utilized light cues gradually diminished with
increased body weight loss while hypothesis behavior based
on spatial cues rapidly increased.
3. There was slight evidence of a curvilinear
relation of adaptability to motivational level but it was
not statistically significant.
46


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EKLPZMH6Q_98E0J8 INGEST_TIME 2014-11-17T20:59:23Z PACKAGE AA00026380_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


14
environment when confronted with a problem. Thus,
for example, Tolman stresses the importance of
determining the ,..conditions which favor relatively
rapid shifts in the dimension of discrimination of a
sign or of a signifcate.." in understanding
differential effectiveness of performance. (3) In
our maze, under the conditions of training that we
have used, almost all animals show a light-going
preference on their first few trials... Achieving a
spatial Preference Score therefore requires that an
animal ignore the dominant illumination cue and pay
attention to the less obvious cue of location in
space. In other words, an animal that develops a
spatial preference shows readier "shifts in discrimi
nation" than an animal whose behavior remains
controlled by the visual cues. (1958, p. 389)
While this is an attractive line of reasoning, there
are certain disturbing features. Whether a continuous
Preference Score or the discrete category, hypotheses, is
used, the criticisms of Witkin and Spence remain pertinent
and have not been answered. In addition, very little is
known about hypothesis behavior and the variables affecting
it; the phenomenon has been a theoretical pawn and there
have been no investigations since Witkin*s efforts twenty
years earlier. Despite the occasional criticisms and the
lack of a sound empirical foundation, Krech treats hypothesis
behavior as definitely adaptive, cognitive, attempted
solutions. He receives some support for this position from
standard reference works which review his work; hypothesis
behavior is uncritically treated as evidence for attempted
solutions on the part of the rat (see Munn, 1950, pp. 250-
253, 333? Stevens, 1951, pp. 312, 744, 774-776; Osgood,
1953* PP. 445-446; Hilgard, 1956, p. 201).


4-5
inconclusive; a U-shaped relation indicating an optimal
motivational level (between -10 and -20 per cent BWC) was
found but statistically significant correlations were not
obtained. The concept of an optimal motivational level
dates back to an original formulation by Yerkes and Dodson
(1908). It has been found to be generally true in such
studies as the Columbia obstruction box experiments (Warden,
1951), and more recently arousal level theorists have
postulated a similar relation (Freeman, 194-0; Duffy, 1957).
* *
D. Conclusions. The relation of various hypothesis
behavior measures to the solution of two problems has been
examined and no support was found for the assumption that a
particular type of hypothesis behavior can be considered
intrinsically more adaptive than some other type. It has
also been shown that the hypothesis behavior exhibited by
an animal is a function of the animal's motivational level.
It was therefore considered more parsimonious to relate
these hypothesis behavior characteristics to motivational
level rather than to assumed levels of adaptability. It
was not possible to test the effects of food and water
deprivation on hypothesis behavior since the attempt to
establish operationally equivalent motivational levels in
these categories was not successful. Finally, some evidence
of an optimal level of motivation for performance was
obtained but the evidence was not statistically significant.


AOraOW&BDOMEireS
The author wishes to express his gratitude to the
members of his committee, Prof, Holland H. Waters,
Chairman, Prof. Richard J. Anderson, Prof. Bradford N.
Bunnell, Prof. James C. Dixon, Prof. Herbert D. Kimmel,
Prof. J. Milan Kolarik, and Prof. Zareh M. Pirenian. He
is especially appreciative of the many hours given by
Prof. Waters and Prof. Bunnell. The clarification of
learning theory by Prof. Judson S. Brown and some invaluable
assistance from Mr. William P. Stone are also gratefully
acknowledged.
ii


16
hypothesized cognitive and motivational processes is
certainly one of the most outstanding...current trends in...
theories today (1951* p* 114).
The effect on hypotheses of manipulating drive
remains unknown. Motivational variables are known to
facilitate or decrease performance levels, but it is
difficult to apply the motivational literature to hypothesis
behavior for several reasons. First, most of the experi
ments are more applicable to motivational phenomena or
theory. Second, the general nature of hypothesis behavior
is unknown. And third, data in hypothesis behavior
experiments are not directly comparable to data in the
usual learning experiment with only one solution.
The present study was directly concerned with these
problems. It was designed to determine empirically the
relationship between the hypotheses formed by an animal and
another measure of that animal's adaptability; it was also
designed to explore the effects of motivation on hypothesis
behavior and adaptability.
The following research hypotheses were advanced:
1) The type of hypotheses and the frequency with which
they are formed is a function of both the level of
motivation and the type of motivation.
Three major characteristics of hypothesis behavior could
vary as a function of motivational level; these may be


52
In this test, N > 28 and animals forming both S and V
hypotheses were included providing one type occurred with
greater frequency. Bight animals were included from this
mixed category and one was excluded.
Result: 5.89 and the Ho is rejected. It is concluded
that S hypotheses predominate with major negative BWC, and
V hypotheses predominate with slight increases or decreases
in body weight.
Ho: That animals do not differ in shifting or not
shifting hypotheses as a function of being
above or below the median in BWC.
Since 29 of the 31 animals formed hypotheses, N 29.
However, of these, only nine animals shifted hypotheses.
p
Results: Y a 0.46 and the Ho cannot be rejected. It
should be noted from Figure 2 and from the results pre
sented above that S hypotheses increase while V hypotheses
decrease as the animals lose weight. Since shifting
hypotheses involves by definition both V and S hypotheses,
some restriction of this shifting within the more moderate
levels of BWC might be expected. Inspection of the data
in the Appendix, Table 9* reveals that six of these nine
animals are above the median BWC. There are thus empirical
grounds cautioning against acceptance of this Ho.


5
attempted solutions.... In the light of the evidence
presented here it is suggested that the helter-skelter
unorganized trial and error response as a description of
the early part of the learning process is invalid, and that
we must change our description of the learning process so
as to recognize the existence of organized and systematic
responses at all stages of the process. (1932a, p. 43)
Krechevsky's interpretations were directed against
the prevalent behaviorism which used trial and error
learning and later the conditioned reflex as a model.
Boring (1950) characterized both of these models as elimi
nating the need for integrative principles such as
organization or insight. Tolman, under whom Krechevsky
was studying for his doctorate, was then developing his
theory of purposive behaviorism. This theory emphasized
the cognitive, intellectually organized aspects of learning.
Krechevsky's work with hypothesis behavior served both as
a fundamental tenet of the Tolmanian system and as
experimental evidence against an anti-insight behaviorism.
Krechevsky's 1932b experiment formed the basis for
all the later work in this area. The maze which he designed
was a four unit discrimination apparatus which could not be
solved (see Figure 1). The animal was confronted by four
choice-points between the start box and the goal box. A


CHAPTER I
INTRODUCTION
This study was primarily concerned with determining
how hypothesis behavior in the hooded rat varies with
changes in motivation. It was also concerned with finding
out what relation such hypotheses have to adaptive behavior,
e.g., are animals that show certain types and numbers of
hypotheses more adaptable or flexible than others? And
finally, it was concerned with the relation between
motivation and adaptability.
Hypothesis behavior is a term first used by
Krechevsky (1932a) to describe the apparently non-random,
systematic attempts of rats to solve an insoluble linear
maze problem. It is necessary to maintain a careful
distinction between the behavior of the animal in the
experimental situation and the implications of the
experimental results. Hypothesis behavior refers to a
particular type of behavior which occurs in a specific
situation. The behavior is a pattern of choices made in
relation to specific cues in an insoluble discrimination
maze. When the pattern of choices differs from a random
1


21
Subjects> Subjects were 32 naive male hooded rats,
varying in age from 92 to 127 days at the start of the
experiment* Of this group, one animal was lost from the
experiment due to escape from the cage and an undeterminable
amount of eating. All animals were housed individually.
Water-deprived animals had ad libitum Purina rat pellets;
food-deprived animals had ad libitum water. They were
maintained on a 12 hour light-dark cycle in an air
conditioned room which had a mean temperature of approxi
mately 78 F.
Apparatus. The standard Krechevsky hypothesis maze
was used, modified with five guillotine doors to prevent
retracing among the start box, goal box, and discrimination
units. The top was covered with window screening painted
with aluminum paint to reflect the interior light of the
maze and reduce room cues. Determining each animal's
choices by mirrors proved unsatisfactory, and direct
observation through the screen was used. The maze was
placed on a table at the side of a small room. The only
light came from the apparatus itself and a 7# watt bulb
suspended over the goal box. This light permitted the
experimenter to observe when the animal had reached the
goal box and to record the animals' responses. The straight


19
a) Do animals that demonstrate spatial hypotheses
and preferences take fewer trials to solve the
problems than animals with visual hypotheses and
preferences?
b) Do animals that have many hypotheses take fewer
trials to solve the problems than animals with
few or no hypotheses?
c) Do animals that shift hypotheses and preferences
take fewer trials to solve the problems than
animals that do not shift?
4) The adaptability of the animals is related to the type
and level of motivation.
The Yerkes-Dodson (1908) law holds that optimal motivational
level is inversely related to problem difficulty; in the
present experiment problem difficulty was not effectively
manipulated, therefore, according to this law an optimal
level of motivation should be apparent. The positions of
Duffy (1951 1957) and Malmo (1957)* are also consistent
with this. Brown (1961, p. 9*0* however, found equivocal
results in the area of discrimination learning and concluded
that further studies were "urgently needed." The relation
ship can be posed as the following question*
a) Does a linear relation exist between motivational
level and number of trials to solve the problems
such that increased motivational level results in
fewer trials?


CHAPTER II
PROCEDURE
The general experimental procedure as described by
Krechevsky (1932b)# and Rosenzweig, Krech, and Bennett
(1958) was followed with certain exceptions. These
exceptions were a difference in kind and amount of incentive,
a slight modification of the maze, and a slightly higher
criterion for determining hypotheses. The criterion was
placed at 36 choices out of 48 for the following reasons
Since each animal had a number of days on which to be
scored for hypotheses, the level of significance should be
adjusted accordingly. Similarly, the scoring of each
animal's daily responses for hypotheses increased the
likelihood of finding hypotheses in direct proportion to
the number of alternative hypotheses. The four alternating
and perseverating hypotheses were not scored so the level
of significance was not adjusted to include these data in
the determination of hypotheses. Significance levels
were adjusted to remain for each animal at P < .01 (see
Ryan, 1959) in consideration of nine days opportunity to
form an hypothesis and four scorings of each day's behavior
(spatial -right, -left, and visual -light, -dark).
20


BIOGRAPHICAL SKETCH
Robert Louis Procter was born March 26, 1928 in
Washington, D. C. The son of a Naval Officer, he lived
in Virginia, New York, California, and Massachusetts until
his father's retirement in Vermont in 1937. He graduated
from Springfield (Vt.) High School in 194-5 and was accepted
into the Army Specialized Reserve Training Program and
assigned to Massachusetts State College. He was later
transferred to Korea and then discharged from the Army in
194-7 He was admitted to the University of Missouri in
194-8 and received his B.A. degree in 1951 and his M.A.
degree in psychology in 1953* He returned to the Army in
that year and was discharged in 1955. He entered the
University of Florida in 1955 and held a series of graduate
assistantships, primarily in clinical and comparative
psychology. He was also employed at Sunland Training
Center, Gainesville, Florida, and Rome (New York) State
School, where he was Senior Clinical Psychologist. His
internship was at the Veterans Administration Hospitals in
Gulfport and Biloxi, Mississippi. He is a member of
Psi Chi, psychology honorary, and the Florida Psychological
Association.
54


BIBLIOGRAPHY
Boring, E. G, A history of experimental psychology.
(2nd od.) Hew York* Applepon-Century-Crofts, 1950.
Brown, J. S. Motivation. New York: McGraw-Hill, 1961.
Duffy. E. The concept of energy mobilization. Psychol,
Rev., 1951, 8, 50-40.
The psychological significance of the concept
of harousal" or "activation." Psychol. Rev.. 1957.
64, 265-275.
Freeman, G. L. The relationship between performance
level and bodily activity level, J. exp. Psychol..
1940, 26, 602-608.
Hamilton, G. V. A study of trial and error reactions in
mammals. J. Anim. Behav.. 1911, 2 53-66.
Hilgard, E. R. Theories of learning. New York: Appleton-
Century-Crofts,
Krech, D. Cognition and motivation in psychological
theory. In W. Dennis (Ed.), Current trends in
psychological theory. Pittsburgh: Univ. Pittsburgh
Press, 1951. £p. lll-l39.
Rosenzweig, M. R., Bennett, E. L., & Krueckel, B.
Enzyme concentrations in the brain and adjustive
behavior patterns. Science. 1954, 120. 994-996.
Krechevsky, I. "Hypotheses" versus "chance" in the pre
solution period in sensory discrimination learning.
Univ. Calif. Publ. Psychol.. 1952, 6, 27-44.
The genesis of "hypotheses" in rats. Univ.
Calif. Publ. Psychol.. 1932, 6, 45-64.
. The docile nature of "hypotheses." J. comp.
Psychol.. 1933, 1, 429-443.
. Hereditary nature of "hypotheses." J. conro.
Piychol.. 1933, 16, 99-116.


This dissertaion was prepared under the direction
of the chairman of the candidate's supervisory committee
and has been approved by all members of that committee.
It was submitted to the Dean of the College of Arts and
Sciences and to the Graduate Council, and was approved as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
June 11, 1962
Dean, Graduate School
Supervisory Committee*


17
stated in question form:
a) Do animals at a low motivational level form
fewer, or more, hypotheses than animals more
highly motivated?
b) Do animals characteristically form visual
hypotheses at one level of motivation, and
spatial hypotheses at another?
c) Is motivational level related to shifts between
types of hypotheses?
Petrinovitch and Bolles had suggested that food deprivation
might lead to more variation in hypothesis behavior than
water deprivation. Accordingly, when level of deprivation
is held constant, the following statement will be tested:
2) Do food deprived animals show more shifts between
hypotheses than water deprived animals.
The establishment of equivalent levels of motivation in two
different modalities, food and water, presents difficulties
in both laboratory technique and research design. True
equivalence is impossible to attain since hunger and thirst
are different experiences. An effective equivalence may
possibly be approached by using equivalent operational
definitions. The more usual approach of using hours of
deprivation would not prove to be a satisfactory procedure;
Stellar and Hill (194-5) observed that rats may adjust their
water consumption, within a limited time of access, to fit


6
Two major types of hypotheses were distinguished,
those based on visual cues and those based on spatial cues.
Among the visual hypotheses, an animal may systematically
choose the lighted alleys (a "Light Hypothesis") or the
dark alleys (a "Dark Hypothesis"). Sometimes an animal
tends to choose an alley in terms of the similarity of
its visual characteristics to the immediately preceding
"correct" alley (e.g., choosing a lighted alley if the
previous lighted alley had proved correct). This
behavior is termed a "Perseverative Visual Hypothesis."
Conversely, the rat may show an "Alternating Visual
Hypothesis"...
The other general categories of hypotheses spatial -
also has four possibilities... there are "Right",
"Left", "Perseverative Spatial", and "Alternating
Spatial" hypotheses. (Rosenzweig, et al. 1958, p. 575)
It is possible to invent any number of systematic
patterns of choice hypotheses for the animal, but
Krechevsky compared the animal's choices for correspondence
to these eight. The equivalent of the Spatial hypotheses,
-Right, -Left, and -Perseverative, and the Visual hypotheses,
-Light and -Dark, had already been noted by previous
investigators. The other hypotheses appear to be logical
extensions of these five.
If the animals made completely random, unsystematic
responses at each of the 4-8 choice-points (four choices on
each of twelve trials), they would choose on the average -
12 right-lighted, 12 right-darkened, 12 left-lighted, and
12 left-darkened alleys in some random order. A theoretical
random animal would be expected to have an average of 24


18
their deprivational state. Brown (1961, pp. 71-74)
presented data which leads to the same conclusion. The
method which was used in the present study determined the
ad libitum food or water intake for a pilot group of animals
and then provided that fixed percentage of such an amount
which would result in equivalent running speeds in a
straight runway. The experimental animals were also weighed
just before this limited amount deprivation schedule was
started, and, immediately after it ended; this measure
provided an additional means of defining motivational level.
3) The adaptability of animals is related to the type and
frequency of the hypotheses which they display.
Rosenzweig, Krech, and Bennett (1958) have argued that
spatial preferences and hypotheses, and shifts of preference
and hypotheses, represent more adaptable behavior since the
animal has given up responding to an initial (apparently
more dominant) aspect of the situation. The measure of
adaptability used in the present study involved a similar
logic; the animal was presented with the doors and lights
of the hypotheses maze arranged so that one of the
hypothesis alternatives was solvable. When criterion was
reached, the maze was adjusted for solution of a second
hypothesis alternative. Adaptability was thus equated
with the number of trials to adopt new modes of response as
they became more efficient. The appropriate questions are*



PAGE 1

MOTIVATION, KRECH HYPOTHESIS BEHAVIOR, AND ADAPTABILITY By ROBERT L. PROCTER A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA June, 1962

PAGE 2

AOKNOWLEDGIMirS The author wishes to express his gratitude to the members of his committee, Prof. Holland H. V/aters, Chairman, Prof. Richard J. Anderson, Prof, Bradford N. Bunnell, Prof. James C. Dixon, Prof. Herbert D. Kimmel, Prof. J. Milan Kolarik, and Prof. Zareh M, Pirenian. He Is especially appreciative of the many hours given by Prof, Waters and Prof. Bunnell. The clarification of learning theory by Prof. Judson S. Brown and some invaluable assistance from Mr. William F. Stone are also gratefully acknowledged. ii

PAGE 3

TABLS OF 00HTMT3 Page ACKNOWLEDGMBNTS 11 LIST OF TABLES. lir LIST OP FIGURES y CHAPTEH INTRODUCTION 1 II, PROCEDURE 20 III. RESULTS 25 xy, DISCUSSION ta V. SUMMARY ^6 BIBLIOGRAPHT 48 APPENDIX 51 BIOGRAPHICAL SKETCH • ^ iii

PAGE 4

LIST OF TABLES Table Pag 1 The Krechevsky Trial Oi^der for the Position .% of Open Doors aad Lighted Alleys on the: Twelve Daily Trials in the Maze 5 2. Mean Running Time in Seconds in the Training Trials .# ^ # • ,ii 3 Mean Running Time in Seconds in the Hiirpothesis Maze 27 ^. Analysis of Variance on the Effects of Type and Level of Deprivation on Running Times in • the Hypothesis Haze .•.. S6 5t Mean Per Cent of Body Weight Change Over the ^ Course of the Experiment ^ 6. Comparison of the Moderate Food Deprived and Severe Water Deprived Groups for Frequencies of Dominant Hypothesis Type '4, 7. Pood and Water Deprived Animals Matched for Equivalent Body Weight Change ....... i fffk'^" 8. 'Sjpe and Frequency of Hypothesis Behavior 36 9. Per Cent Body Weight Change, Type and Occurrence of Hypotheses, and Trials to Solve the Adaptability Problems • • 5(2

PAGE 5

LIST OF FIGUHSS i ^ yigure ^ Page ; !• Floor Plan of tke KreclaeTsky Hypothesia Maze ; 2 ITumber azid Type of Hypotheses Formed as a i> V. function of Body Weight. # • 50 5, The Relation Between Body Weight Change and Average Number of Trials to Solve Botli Adaptability Problems 40

PAGE 6

CHAPTER I INTRODUCTION This study was primarily concerned with determining how hypothesis behavior in the hooded rat varies with changes in motivation. It was also concerned with finding out what relation such hypotheses have to adaptive behavior, e.g., are animals that show certain types and n\mbers of hypotheses more adaptable or flexible than others? And finally, it was concerned with the relation between motivation and adaptability. H^othesis behavior is a term first used by Krechevsky (1932a) to describe the apparently non-random, systematic attempts of rats to solve an insoluble linear maze problem. It is necessary to maintain a careful distinction between the behavior of the animal in the experimental situation and the implications of the experimental results. Hypothesis behavior refers to a particular type of behavior which occurs in a specific situation. The behavior is a pattern of choices made in relation to specific cues in an insoluble discrimination maze. When the pattern of choices differs from a random

PAGE 7

level by a certain amount, hypotheses are said to be formed. Hypothesis behavior and hypotheses are terms which have meaning only when the apparatus, the cues, the pattern of choices, and the statistical level are understood. Hamilton (1911) noticed in the course of a trial and error learning experiment that rats and other animals make apparently systematic attempts at solving problems that cannot be solved; they may go consistently to the right, or left, or to the last place found successful, and so on. Lashley (1929) also noted this feature of behavior in discrimination problems. He indicated that while there was no experimental technique available at that time to investigate this behavior, these "attempted solutions" probably represented a significant aspect of the learning process. Lashley' s observations served as at least a partial basis for Krechevsky's investigations; Krechevsky's experiments were designed to provide the lacking experimental technique. In one of his earliest published works he states, "The data from the present experiment have been examined in the light of Lashley' s suggestion and an attempt has been made to devise a method for the objective determination of the validity of that suggestion." Krechevsky concluded from this study that "...it is shown quite definitely in the presolution period that... the animal is engaged in bringing to perfection various

PAGE 8

attempted solutions.... In the light of the evidence presented here it is suggested that the helter-skelter imorganized trial and error response as a description of the early part of the learning process is invalid, and that we must change our description of the learning process so as to recognize the existence of organized and systematic responses at all stages of the process." (1952a, p. 4-3) Krechevsky's interpretations were directed against the prevalent behaviorism which used trial and error learning and later the conditioned reflex as a model. Boring (1950) characterised both of these models as eliminating the need for integrative principles such as organization or insight. Tolman, under vrhom Krechevsky was studying for his doctorate, was then developing his theory of purposive behaviorism. This theory emphasized the cognitive, intellectually organized aspects of learning. Krechevsky's work with hypothesis behavior served both as a fundamental tenet of the Tolmanian system and as e25)erimental evidence against an anti-insight behaviorism. Krechevsky's 1952b experiment formed the basis for all the later work in this area. The maze which he designed was a four unit discrimination apparatus which could not be solved (see Figure 1). The animal was confronted by four choice-points between the start box and the goal box. A i

PAGE 9

4 start b ^ Curtain Swinging Door Goal box Fig. 1. -Floor Plan of the Krechevsky Hypothesis Maze. This diagram shows the maze as it appears on trial 7< An animal following a visual -dark hypothesis would choose alleys in the order left, left, right, right. correction method was used at each choice-point and the reward was available in the goal box at the completion of every trial regardless of the correctness of the animal's four choices. Krechevsky used a 25-hour food deprivation schedule; a cube of bread soaked in milk was used as the incentive. At each choice-point there were two alleys, a ri^t and a left, between which the animal could choose. Over each alley was an electric light, but only one side was illuminated. IThe animal might choose on the basis of spatial cues (right, left) or visual cues (light, dark). A curtain at the end of each alley prevented the animal from seeing a swinging door which could be locked to either side, thus preventing the animal from using one or the other alleys. One alley of each of the four units was always locked. When the animal chose an alley which had an open door, it went through to the next xmit. When the

PAGE 10

animal chose an alley with a locked door, it was forced to reverse its course and go through the alley on the other side. !Ehe animal was given twelve trials each day. On every trial the lights and doors were shifted in such a way that: 1) the lights and open doors were on either side of the choice-point equally often, and, 2) the use the animal of any of the eight hypotheses which Krechevsky identified would result in encountering the same number of open and locked doors. (See Table 1 for the trial order positions of lights and doors.) TABLE 1 THE KRBCHBVSKT THIAL ORDER FOR THE POSITION OF OPBH DOORS AND LIGHTED ALLEYS ON THE TWELVE DAILY TRIALS IN THE MAZE Trial Door Open Light On 1 L L R L R L R L 2 R R R L R L R R 5 R R L R L R R R 4L L R H L R L L Trial Door Open Light On 5 R R L L R R L L 6 R L L L L L R R 7 R L L R R R L L 8 L R L L R L L L Trial Door Open Light On 9 L R R L L L L H 10 R L R R L R R L 11 L L L R R L R R 12 L R R R L R L R

PAGE 11

• i Two major types of hypotheses were distinguished, those based on visual cues and those based on spatial cues. Among the visual hypotheses, an animal may systematically choose the lighted alleys (a "Light Hypothesis") or the dark alleys (a "Dark Hypothesis"). Sometimes an animal tends to choose an alley in terms of the similarity of its visual characteristics to the immediately preceding "correct" alley (e.g., choosing a lighted alley if the previous lighted alley had proved correct). This behavior is termed a "Perseverative Visual I^ypothesis. Conversely, the rat may show an "Alternating Visual Hypothesis". • The other general categories of hypotheses spatial also has four possibilities... there are "Right", "Left", "Perseverative Spatial", and "Alternating Spatial" hypotheses. (Rosenzweig, et al. 1958, p. 57$) It is possible to invent any number of systematic patterns of choice hypotheses for the animal, but Krechevsky compared the animal's choices for correspondence to these eight. The equivalent of the Spatial hypotheses, -Right, -Left, and -Perseverative, and the Visual hypotheses, -Light and -Dark, had already been noted by previous investigators. The other hypotheses appear to be logical e3Ctensions of these five. If the animals made completely random, unsystematic responses at each of the 48 choice-points (four choices on each of twelve trials), they would choose on the average 12 right-lighted, 12 right-darkened, 12 left-lighted, and 12 left-darkened alleys in some random order, A theoretical random animal would be expected to have an average of 2'^

PAGE 12

choices to each, of the four cues (Left, Right, Light, Dark), Krechevsky defined non-random, systematic behavior his hypothesis behavior as being present when the animal chose one of the eight alternatives with a frequency significantly greater than the theoretical random mean of 2^. The level of significance which he chose was three standard deviations above the mean. This amounts to 5^39 choices out of 48 With this criterion, Erechevsky found that most of his animals formed hypotheses and from these early studies he concluded that ",,,the presence of systematic forms of behavior in such a situation is to be interpreted to mean that these systems were determined not as something forced ab extra by the situation, but as something originating from the animal himself" (1932b, p, ^5). He suggested that "...descriptions of the lower animals* behavior as consisting of "stereotyped," "haphazard," "non-insightful" responses are to be attributed not to a lack of insight on the animals' part but rather to a lack of insight on the experimenter's part" (1932b, p. 63). Over the next several years, Krechevsky continued his investigation of hypothesis behavior. In one of these studies (1933b) he used animals from Tryon's "bright" and "dull" strains and found that the "bright" animals had more spatial hypotheses and the "dull" animals had more visual hypotheses. Another study examined some of the conditions

PAGE 13

under wMcli animals give up or persist in hypotheses. Hats which had formed h;7potheses were divided into three groups. One group then found the maze solvable in conformance with the hypothesis each animal was exhibiting; the second group had the maze solvable for a different hypothesis; a third group had the incentive removed as soon as an hypothesis was demonstrated. The first group persisted in their original hypotheses, the second group changed hypotheses to the correct solution, and in the third group more than fifty per cent of the animals refused to run at all. This study was integrated with a further exposition of Tolmaix's theory and Krechevsky*s most important conclusion *as tliat *...if any behavior-act can be established as docile and purposive, •hypothesis '-behavior is definitely so" (1935a, p. -4^2) • ^rechevsky began a series of investigations on the relation of hypothesis behavior to the structure and function of the brain (1935t 1937a, b, c). He found that cortical lesions decreased the number of different hypotheses an animal might use, but increased the number of days an animal would use the same hypothesis. He also found evidence that this behavior was not related in simple fashion to the amount of cortical damage, but to both the amount and locus of damage. One area (occipital lobe) he called V because few animals made visual hypotheses when it was damaged. Another area (somesthetic) was identified as S, because of

PAGE 14

a similar effect on spatial hypotheses. Damage to botli areas tended to result in spatial hypotheses. Krechevsky published no more articles on hypothesis behavior until recent years. Hn>0'tliesi3 behavior served as the prototype for other Tolmanian concepts such, as meansend-readiness, and Krechevsky began experimentation in other areas. Also, while hypothesis behavior was directly related to the controversy over the continuous or discontinuous nature of learning, the hypotliesis behavior maze was not readily adaptable to investigating this problem. Krechevsky designed other pieces of apparatus for this purpose. The only critique of hypothesis behavior which exists in the psychological literature is that provided by Uitkin (19^2). Witkin noted that hypothesis behavior is not typical of solvable problem situations, but rather of situations for which no solution can be found. He found the same type of behavior in the Krechevsky hypothesis maze when the doors were completely removed (free-choice) as when the doors were locked according to Krech-evsky's insoluble pattern; the behavior was not found when a solvable pattern was used. He questioned the adaptive or pui^osive nature of hypotheses, since animals shifted hypotheses in a free-choice situation although the first hypothesis or no hypothesis at all was equally adaptive.

PAGE 15

10 Ho also questioned Krechevsky's inteirpretation of the cortical lesion studies* "Habits having a far simpler "basis than 'hypotheses* cannot be performed after extensive cortical lesion. Yet every animal submitted to decorti^ cation showed systematic behavior in linear 'insoluble' situations, pointing to the relatively simple basis of these responses" (pp. 566-567). Witkin suggests that the chaages in behavior resulting from differential locus of lesions may be the result, essentially, of removing from the animal the capacity to receive particular sensory cues within the ma25e. Witkin concludes, "The 'hypotheses' concept and the system of which it is a part arose in opposition to the crudely mechanistic conceptions which have gained prominence...* As opposed to such piecemeal conceptions, the 'hypotheses' concept attempted to present the learning process as orderly, organized, and predictable, and pictured the learner as a more active participant in determining the character of the final learned habit. This... is a very desirable end, but since... the 'hypotheses' concept swings to an extreme where it is beyond the bounds of evidence, it constitutes a very weak kind of opposition..." (p. 567). Spence also had been critical of Erechevsky's interpretations, "Contrary to the belief of certain writers, ... there has been no disagreement concerning the behavioral facts.... Agreement ceases however, over attempts at further

PAGE 16

11 interpretation of the phenomena" (19^5 P 255) • In an early theoretical paper on the nature of discrimination learning, Spence (1956) demonstrated that perseverative behavior similar to hypotheses was compatible with the Hull-Spence system. Hypotheses were considered as a phenomenon dependent on learning processes before the correct learned response became manifest; as such, it is a phenomenon characteristic of a pre-solution period. At a later time Spence distinguished between this perseverativ* pre~solution behavior and the hypotheses which Krechevsky found in an insoluble maze, "...these pre-solution phenomena appear to be a typical example of what has been described as trial and error learning, while hypotheses are far from what... (I Tinderstand) .by the terms insightful and intelligent. Only persistent non-adaptive responses caa attain the distinction of being hypotheses — for, in order to classify as a hypothesis, a response, although ineffective, must continue to be persisted in a certain minimum number of times. A maladaptive act which is speedily (intelligently?) abandoned cannot ever be a hypothesis" (19^, p. 287). Both forms of perseverative behavior are, for Spence, predictable from his theoretical framework. As a pre-solution phenomenon, the hypothesis is expected to be replaced by the correct response without any intermediary hypotheses. In the insoluble problem situation,

PAGE 17

' 12 the development of hypotheses (or the complete nonoccurrence of them), shifts between loypotheses, and the persistence of hypotheses, are all theoretically explainable ?ost hoc providing certain assumptions regarding initial response strengths and reinforcements can be made. Unfortunately, the ErechevsJsy hypothesis maze does not lend itself to an empirical test of the Spence position. There was no further work on hypothesis behavior until 195^ when Krech^, Hosenzweig, and Bennett began publishing the results of a series of eacperiments which correlated brain chemistry with adaptability. Despite the earlier criticisms of hypothesis behavior, this technique was used as the measure of adaptability. a?wo changes were made in this measure. First, a reduced level of significance v:ae used as a criterion of hypotheses (33 rather than 55 choices of the ^8 occurring each day), and second^ Krechevsky's (1933) spatial-visual Preference Score was also used. This Preference Score provided a continuous scale of relative preference for either visual or spatial modalities \d.thin each animal. It is computed in this fashion t 1) the number of choices in each of the eight hypothesis alternatives is determined for an animal for a ^ .' .-• Krechevsky changed his name to jKrech,

PAGE 18

13 given day; 2) the deviation in per cent of these ei^t numbers from fifty per cent is computed; 3) the percentages based on spatial hypotheses are g-iven positive signs, the visual percentages are given negative signs, and the eight percentages are then algebraically added. In a later study by Rosenzweig, et al (1958), the results indicated that the amoujit of cholinesterase (ChE) significantly increased from the visual area (Y) to the somesthetic area (S) to the motor area (M); the amount in all areas declined with age. When visual preference animals were compared to those with spatial preferences, the OhE rate of decline was found to be very rapid. The authors were using OhE as a measure of acetylcholine (ACh) metabolism, and inferred from these results that AOh metabolism is related to adaptive behavior patterns. Such an inference necessitates two assumptions. Out concerns the OhE-ACh relation and is a chemical problem; the relation of hypothesis behavior to adaptability is properly a psychological problem which the authors justify as follows I I .. ...we believe that animals that show a spatial Preference Score are more adaptive than animals showing a visual Preference Score. Ve make this interpretation for the following reasons: (1) As we have already pointed out, our test is designed to measure the animal's perceptual selectivity, (2) We next assume that adaptive behavior is correlated with the ability of the animal to "pay attention to" various stimulus aspects of its

PAGE 19

14 environment when confronted with a problem. Thus, for example, Tolman stresses the importance of determining the .cohditions which favor relatively rapid shifts in the dimension of discrimination of a sign or of a significate. in imderstanding differential effectiveness of performance. X3) In our maze, under the conditions of training that we have used, almost all animals show a light-going preference on their first few trials... Achieving a spatial Preference Score therefore requires that an animal ignore the dominant illumination cue and pay attention to the less obvious cue of location in space. In other words, an animal that develops a spatial preference shows readier "shifts in discrimination" than an animal whose behavior remains controlled by the visual cues. (1958, p. 589) While this is an attractive line of reasoning, there are certain disturbing features. Whether a continuous Preference Score or the discrete category, hypotheses, is used, the criticisms of Witkin and Spence remain pertinent and have not been answered. In addition, very little is known about hypothesis behavior and the variables affecting it J the phenomenon has been a theoretical pawn and there have been no investigations since Witkin's efforts twenty years earlier. Despite the occasional criticisms and the lack of a sound empirical foundation, Krech treats hypothesis behavior as definitely adaptive, cognitive, attempted solutions. He receives some support for this position from standard reference works which review his work; hypothesis behavior is uncritically treated as evidence for attempted solutions on the part of the rat (see Munn, 1950, pp. 250253, 535? Stevens, 1951, pp. 312, 744, 774-776; Osgood, 1953, pp. 445-446; Hilgard, 1956, p. 201).

PAGE 20

I 15 The results of experiments are sometimes discussed with reference to hypotheses, Petrinovitch and Bolles investigated the effects of kind of deprivation on learning. Using hungry and thirsty rats in a single-unit-T-maze, they found that when the goal wsis alternated on every trial the food-deprived animals were hotter, but when the goal remained fixed, the water-deprived animals were superior. They related their findings to Krechevsky's (1957) report that normal animals alternate and vary their hypotheses. The authors suggested "In the light of the present experiment, it is possihle that his data do not pertain to •normal* animals, but only to normal hungry animals. In general, our results suggest that in any study of variability or stereotypy in behavior, the nature of the *drive' used must be carefully considered" (195^ p. ^52), Cognitive theorists have been attaching increased importance to drive states, Tolman (19^4-9) has modified his earlier position and suggested that drive reduction may be necessary for some types of learning. Similarly, Krech no longer holds to a strict cognitive theory of learning. He writes "It is quite obvious, if you examine many of the theories extant in psychology today, that these theories do not rest content with motivational constructs only, or with cognitive constructs only, but require both sets of constructs.... The belief in the interaction among the

PAGE 21

16 bypothesized cognitive and motivational processes is certainly one of the most outstanding. .current trends in... theories today" (1951f p. 11^). The effect on hypotheses of manipulating drive remains unknown. Motivational variables are kno^'jn to facilitate or decrease performance levels, but it is difficult to apply the motivational literature to hypothesis behavior for several reasons. First, most of the experiments are more applicable to motivational phenomena or theory. Second, the general nature of hypothesis behavior is unknown. And third, data in hypothesis behavior experiments are not directly comparable to data in the usual learning experiment with only one solution. The present study was directly concerned with these problems. It was designed to determine empirically the relationship between the hypotheses formed by an animal and another measure of that animal's adaptability; it was also designed to explore the effects of motivation on hypothesis behavior and adaptability. The following research hypotheses were advanced: 1) The type of hypotheses and the frequency with which they are formed is a function of both the level of motivation and the type of motivation. Three major characteristics of hypothesis behavior could vary as a function of motivational level; these may be

PAGE 22

. I 17 stated In (question form} a) Do animals at a lov motivational level form fewer, or more, hypotheses than animals more highly motivated? > ^-Z b) Do animals characteristically form visual hypotheses at one level of motivation, and spatial hypotheses at another? c) Is motivational level related to shifts "between types of hypotheses? Petrinovitch and Bolles had suggested that food deprivation might lead to more variation in hypothesis behavior than water deprivation. Accordingly, when level of deprivation is held constant, the following statement will be tested: 2) Do food deprived animals show more shifts between hypotheses than water deprived animals. Ifhe establishment of equivalent levels of motivation in two different modalities, food and water, presents difficulties in both laboratory technique and research design. True equivalence is impossible to attain since hunger and thirst are different experiences. An effective equivalence may possibly be approached by using equivalent operational definitions. The more usual approach of using hours of deprivation would not prove to be a satisfactory procedure; Stellar and Hill (19^5) observed that rats may adjust their water consumption, within a limited time of access, to fit

PAGE 23

18 their dprirational state. Brown (1961, pp. 71-74) presented data which leads to the same conclusion. The method which was used in the present study determined the ad libitum food or water intake for a pilot group of animals and then provided that fixed percentage of such an amount which would result in equivalent running speeds in a straight runway. The experimental animals were also weighed Just hef ore this limited amount deprivation schedule was started, and, immediately after it ended; this measure provided an additional means of defining motivational level. 3) The adaptability of animals is related to the type and frequency of the hypotheses which they display^,Rosenzweig, Krech, and Bennett (1958) have argued that spatial preferences and hypotheses, and shifts of preference and hypotheses, represent more adaptable behavior since the animal has given up responding to an initial (apparently more dominant) aspect of the situation. The measure of adaptability used in the present study involved a similar logic; the animal was presented with the doors and lights of the hypotheses maze arranged so that one of the hypothesis alternatives was solvable. When criterion was reached, the maze was adjusted for solution of a second hypothesis alternative. Adaptability was thus equated with the number of trials to adopt new modes of response as they became more efficient. The appropriate questions are: I

PAGE 24

19 a) Do animals that demonstrate spatial hypotheses and preferences take fewer trials to solve the problems than animals with visual hypotheses and preferences? ^ b) Do animals that have many hypotheses taike fewer trials to solvo the problems than, animals with few or no hypothesest c) Do animals that shift hypotheses and preferences take fewer trials to solve the problems than animals that do not shift? ^) The adaptability of the animals is related to the type and level of motivation. The Terkes-Dodson (1908) law holds that optimal motivational level is inversely related to problem difficulty; in the present experiment problem difficulty was not effectively manipulated, therefore, according to this law an optimal level of motivation should be apparent. The positions of Duffy (1951, 1957) and Malmo (1957), are also consistent with this. Brown (1951, p. 9^)t however, found equivocal results in the area of discriiaination learning and concluded that further studies were "urgently needed." 0?he relationship can be posed as the following question* a) Does a linear relation exist between motivational level and number of trials to solve the problems such that increased motivational level results in fewer trials?

PAGE 25

CHAPTEH II PE0C3DURE The general experimental procedure as described by Krechevsky (1932b), and Rosenzwoig, Krech., and Bennett (1958) was followed with certain exceptions. These exceptions were a difference in kind and amount of incentive, a slight modification of the maze, and a slightly higher criterion for determining hypotheses. The criterion was placed at 56 choices out of ^ for the following reasons Since each animal had a number of days on which to be scored for hypotheses, the level of significance should be adjusted accordingly. Similarly, the scoring of each animal's daily responses for hypotheses increased the likelihood of finding hypotheses in direct proportion to the number of alternative hypotheses. The four alternating and perseverating hypotheses were not scored so the level of significance was not adjusted to include these data in the determination of hypotheses. Significance levels were adjusted to remain for each animal at P < .01 (see Ryan, 1959) in consideration of nine days opportunity to form an hypothesis and four scorings of each day*s behavior (spatial -right, -left, and visual -light, -dark). 20

PAGE 26

21 Subjects Subjects were 32 naive male hooded rats, varying in age from 92 to 12? days at the start of the e3Cperiment Of this group, one animal was lost from the experiment due to escape from the cage and an \inde terminable amount of eating. All animals were housed individually. Water-deprived animals had ad libitum Purina rat pellets; food-deprived animals had ad libitum water. They were maintained on a 12 hour light-dark cycle in an air conditioned room which had a mean temperature of approximately 78" P. Apparatus The standard Krechevsky hypothesis maze was used, modified with five guillotine doors to prevent retracing among the start box, goal box, and discrimination units. The top was covered with window screening painted with aluminum paint to reflect the interior light of the maze and reduce room cues. Deteimining each animal's choices by mirrors proved unsatisfactory, and direct observation throu^ the screen was used. The maze was placed on a table at the side of a small room. The only light came from the apparatus itself and a Tk watt bulb suspended over the goal box. IHiis light permitted the experimenter to observe when the animal had reached the goal box and to record the animals* responses. The straight

PAGE 27

22 runway wMcli was used in the pre training period was similar to the hypothesis maze but with the center sections removed 80 that no alleys were formed. The guillotine doors were retained and curtains were hung at the three center doors to proTide habituation for the animals, Procedure Animals were treated in groups of about eight, with random assignment to each condition. The animals assigned to the food-deprivation category (PD) were placed on ad libitum powdered Purina rat chow and water schedules; animals in the water-deprivation category (WD) received the same treatment except they were given pellets rather than powdered food. After a four-day adjustment period, measurements of each animal's food or water (as appropriate) consumption was taken for 1^:days. At the end of this time the average daily consumption was computed and the animals were weighed, ; >. A pilot study had indicated that 65 per ent of the M libitum food intake and 70 per cent of the ad libitum water intake vjould result in approximately equivalent running times in a straight runway; similar results were found for a ^5 per cent food and a 50 per cent water level. Therefore those four levels were adopted. The animals were placed on the appropriate schedule one day prior to the start of the pretraining trials and remained on these diets \intil the conclusion of the experiment.

PAGE 28

23 Pre training The animals were given 5 trials the first day, 5 trials the second day, and then 10 trials each day for nine successive days} the animals were run every 24 hours. The incentive was approximately 0.1 gram of powdered chow or 0.1 c.c. of water, as appropriate. The incentive was taken from each animal's daily ration. • Sn)othesis trials !Ehe animals were given 6 trials on both the first and second days, and 12 trials each of the successive eight days. The Krechevsky order (see Table 1) was followed. After the final day's trials \d.thin the hypothesis maze were completed, the hypotheses for each animal were computed. Weights were obtained for each animal. Adaptability trials On the basis of each animal's hypothesee and membership in the four experimental groups, ft decision was made as to which of two visual problems and two spatial problems were to be solved. The criterion for solution was the sane as for hypotheses; however, whereas hypotheses were computed on the basis of an animal's daily performance, these trials were counted serially without respect to the termination of each days trials; such a procedure is necessary to prevent grossly unequivalent amounts of practice. An animal was considered to have reached criterion when it first attained 36 out of any 46 consecutive choices which were correct; animals that reached criterion on the first, second, or third choice-points of a i

PAGE 29

24 trial were allowed to finish that trial and attain the incentive. K f. Roughly half of the animals which had made visual hypotheses were assigned spatial problems first; spatial hypothesis animals were treated similarly. When animals had made both types of hypotheses, or none, the assignment vias made on the basis of the dominant preference. In all cases, the assignment of any problem within either stimulus modality was opposite to that animal's dominant preference. I The ntimber of daily trials and the incentives remained the same as in the latter part of the hypothesis maze experiment. At the conclusion of the first problem, each animal was immediately shifted to the second problem on the next trial. The scores were each animal's total number of trials to solve both problems.

PAGE 30

OHAPm III RESISTS IHiis study was concerned witli certain relationsMps between motivation, hypothesis behavior, and adaptability. 2?wo semi-independent measures of motivation were used. First there was a deprivation schedule providing for restricted amoxmts of food (FD) or water (VD) at two different levels, moderate and severe. Second, a measurement of body weight change (BWO) over the course of the experiment was obtained. Since the experimental design required equivalent levels of motivation in the FD and WD groups for testing certain hypotheses, evidence concerning this equivalence will be presented first. Percentages of ad libitum food and water had been chosen to obtain equivalent running times. Speed of running \fas thus used as a criterion of equivalence. The concept of equivalence can be assumed to have been validly met if it can be shown that the deprivation schedules establish equivalent running times within each level of deprivation. There are two sotxrces of running time data in th experiment, one from the training trials and the other from 25

PAGE 31

26 the trials in tinB Krech. apparatus* Response time data were available from 19 animals during the training trials to use as a check against this method. These data, presented in Table 2, showed that onlj a slight degree of equivalence existed between each of the moderate and each of the severe groups. These data were obtained in a straight runway and were therefore similar to the conditions of the pilot group. Hoiirever, it is of more pertinence to determine whether equivalent running speeds existed at the time of the hypothesis behavior trials, TABLE 2 MEAN RUNNING TIME IN SECONDS IN THE TRAINING TRIALS (Bays 3 through 10, N = 19) Deprivation Level Pood Water Moderate 23.9 17.7 Sever* 7.6 15.6 Response time data for a sample of 20 animals in the hypothesis maze are presented in Table 5. It will be noted that the moderate FD and severe WD groups were approximately equivalent, but the other groups occupy extreme positions. This indicated that type of deprivation, food or water, was possibly confounded with level of motivation. Analysis of variance established that such

PAGE 32

27 I ^ TABLE 3 MEAN RUNNING TIME IN SECONDS IN THE HTPOTHBSIS MAZE (Days 1-9, N 20) Level Food Water Moderate 52.1 49.2 Severe 20.4 35.2 confounding did take place. According to the original experimental procedure, a significant difference in running time should he found for the level of motivation variable (since the restricted diets were deliberately chosen to produce different running times). As the summary in Table 4 indicates, animals differed significantly (P < .005) la level of motivation as measured by runnir^ time level. Hovrever, these restricted diets were also deliberately chosen with the intent of producing no significant difference between running times on the type of deprivation variable. Here, unfortunately, they also differed and this difference is significant (P < .001). Clearly, testing the experimental hypotheses relating to the food-water variable would introduce motivational level as a contaminating factor.

PAGE 33

28 i TABLE i\ANALYSIS OF 7AHIAN0B OH 5HB EFFECTS OF TYPE AND LBTEL OF DBPHIVATION ON EUNNING TIMES IN THE • ^ : > ^ ^ HYPOTHESIS MAZE i ; (X' log Z tranaformation) < Source d.f. M.S. F P< Type (Food-Water) 1 .230 20.55 .001 Level (Moderate-Severe) 1 .158 12.21 .005 TTpe X Level 1 .004 .55 Error 16 .011 Total 19 Another indication that equivalence did not exist was found when the mean BV/C was computed for each of the cell£5. Tahle 5 with running time data and Table 5 with mean body weight changes reflect the same general trend. This suggests that the two measures are correlated. This turned out to be true, with a rank order correlation of ^ .680 (P < ,01). Because of this correlation, per cent of BVC, rather than the dichotomous Moderate-Severe categories, was used as the indicator of motivational level in further analysis of the data.

PAGE 34

29 TABLE 5 MBAN PER CENT OF BODY WEIGHT CHANGE OVSR THE COURSE OF THE EXPERIMENT. (N = 31) Dgprlvatlon Level Food Water Moderate -15.1 1.8 Severn 28.9 -8.8 To atmmarisse the results obtained so far, 1. The desired two equivalent levels of motivation were not attained, since type of motivation was contaminated by level of motivation, 2. Per cent of BWC was significantly correlated with running time and was used as the independent variable representing motivational level. A. The effect of varying motivational level 5 ffixe hypothesis was advanced in the introduction that the hypothesis-behavior of the animals would be a function of level of motivation. It is apparent from Figure 2 that varying motivational level induces major changes in both the gross amount and type of hypothesis behavior, spatial (S) or visual (V). The more negative the BWC, the more S hypotheses the animal forms. Tests of the I^jrpothesea

PAGE 35

30 +10-0 0 10 -10 20 -20 30 -50 W % change in body weight Pig.2.-lftiiaber and type of hypotheses formed as a function of body weight (totals of all hypotheses within a class interval, divided by the number of subjects, N, of the class.)

PAGE 36

The form of the data and the nature of the inquiiy lent itself in this case and in many of the tests to follow to the use of the'X test of independence. The data are usually in frequency form and also in dichotomous categories such as S or V hypotheses. Even where dichotomous categories do not exist, it was meaningful to create at times such classes as ahoTe or helow the median. The Yates correction of the fonaula was used (McNemar, 1955, formula 86a). The null hypothesis (Ho) was rejected when'X^.05, 1 d,f, • 584 was surpassed. Hoj Animals do not differ in the number of hypotheses formed as a function of being above or below the median in BWC. (N 3I). Result ; 'X 5.59 and the Ho is rejected. Another indication of this relationship is given by r| which has a value of .663 (P < .001). (The P ratio between tt] and r 1.057 and linear regression may be assumed). It is concluded that the number of hypotheses formed is a function of BUG; more hypothesea are formed with severe losses of body weight and fewer hypotheses are formed with moderate increases of weight. j Ho J That animals do not differ in forming S or V hypotheses predominantly as a function of being above or below the median in BWC.

PAGE 37

32 In this test, N 28 and animals forming both 3 and 7 hypotheses were included providing one type occurred with greater frequency. Bight animals were included from this mixed category and one was excluded. Result : =5*89 and the Ho is rejected. It is concluded that S hypotheses predominate with major negative BVC, and V hypotheses predominate with slight increases or decreases in hody weight. > Ho : That animals do not differ in shifting or not shifting hypotheses as a function of being above or below the median in BWO. Since 29 of the 31 animals formed hypotheses, IT 29. However, of these, only nine animals shifted hypotheses. Results: Y = 0.^6 and the Ho cannot be rejected. It should be noted from Figure 2 and from the results presented above that S hypotheses increase while V hypotheses decrease as the animals lose weight. Since shifting hypotheses involves by definition both V and S hypotheaea, some restriction of this shifting within the more moderate levels of BVC might be expected. Inspection of the data in the Appendix, Table 9, reveals that six of these nine animals are above the median BWO. There are thus empirical grounds cautioning against acceptance of this Ho.

PAGE 38

33 B. The effect of varying food and water The contamination of the type of deprivation by the level of deprivation prevented an adequate test of the effect of food and water on hypothesis behavior. Some indication of the effect may be gained by constructing roughly equivalent groups. Two such groups are the moderate I'D and severe WD from the original design (note their similarity in Tables 3 and 5), Table 6 presents a comparison of these two groups and it will be seen that there is little difference in the frequencies. ,,, ^ ,, j TABLE 6 'y^'^ COMPAHISOIT OF THE MODERATE FOOD DEPRIVED AND SEVERE WATER DEPRIVED GROUPS FOR Preference Spatial Food 5 Water 5 Visual 2 3 Matched groups can also be composed by choosing animals with nearly equivalent BWC. Unfortunately, these are too few to warrant statistical test but the data are presented in Table 7 for inspection. There is a slight indication that food deprivation may more likely be associated with 3 hypothesis formation than water deprivation.

PAGE 39

34 ( TABLE 7 FOOD AND VAT3R DEPRIVED AIOIALS MATCHED FOR BQUIVALEirT BODY WEIGHT CHANGE BWG % Pood V/ater Number and Type Hypothesis Trials to Solve Pood water AP 7 fooA. Water 5.7 5.5 • 7.9 -18.5 31.2 + 5.8 4.7 7.7 -16,8 -31.4 3 3 1 2 6 0 1 3 0 0 0 0 2 5 5 1 2 1 0 0 38 n 33 36 35 27 23 36 It will be shown in section C that BWO was not significantly related to the shifting of hypotheses. Sine* it therefore seems probable that motivation is not a contaminating factor, we may test the food and water deprivation variable on the shifting of hypotheses. The resulting 'X .016 and the Ho is not rejected, fhe effect of this variable on the AP trials was also tested and 2 t .0039 and the Ho is not rejected. It is concluded that type of deprivation was related neither to the shifting of hypotheses nor to the ease of solving the adaptability problems. It was not possible to test the influence of deprivation type on the kind of hypotheses formed.

PAGE 40

ghe relation of hypothftglg behavior to adaptability Adaptive behavior was measured by the total number of trials to solve two problems. These problems (AP) were arrangements of the doors of the Krech apparatus such that one of the four hypotheses S right (SR) or left (SL), V light (7L) or dark (VD), was a correct solution. Bach animal was presented with one of the S problems and one of the y problems, the Qrdr of pres9ntatlon (S or V first) was determined by the experimenter tAlo assigned roughly equal amounts of the orders to the various categories of hypothesis-f omlng animals. > -s. It Is possible that order effects exist in the presentation of these problems. If an animal had shown an SR preference or hypotheses the use of problems SL and TO in that order may have been more, or less, intrinsically difficult than the order TO, SL. A second source of order effect might be found in the choice of the particular cue of each modality; that is, an SR (or TO) problem also might be more or less intrinsically difficult than SL (or VL) problems. However, this second order effect was controlled by giving all animals problems which differed from their last demonstrated preference (as revealed by the largest number of choices within the SR, SL, VL, TO categories). If an animal had shown SR and TO preferences during the hypothesis trials, I

PAGE 41

56 the problems presented would "be SL and VL, altlioush not necessarily in that order. lEhus, the first type of order effect varied while the second type was partially controlled. An indication of the effect of this first type of order can he obtained from the point-bi serial correlation between order (first problem the same first problem opposite demonstrated preference) and the distribution of number of trials on the AP. This correlation has the value Xp^ • .013 which of course is not significant. When tested byX (Same-Opposite x More-Less median trials on the AP), o t m .028 and is not significant. Therefore it will be assumed that an order effect does not exist in the analysis of the data to follow. 1 Some statistics of the animals* gross behavior are presented in Table 8, It will be noted that more animals formed S hypotheses and more frequently, than any other ; TABLE a TIPS AND FREQDBHCy OF HTPOTHESIS BBHAVIOH ITuaber of Animals 1 of Hypothesis Formed Megn Number of Days Persisted 14 S only 4.57 • 7 only 2.67 9 Both S and V 4.11 2 none

PAGE 42

37 category. This differs from Kreoh's finding and will be discussed in the next chapter. However, on the basis of his argument for adaptability there appears to be a logical basis for inferring that animals which form V hypotheses are more adaptive since 3 hypotheses are more "popular." The California group had stated that on the first five or six trials their animals displayed a decided preference for the lighted alleys (Rosenzweig, et_al. 1958). Accordingly, the data from these trials x^ere analyzed. The Spatial minus Visual Preference score was computed for the first six trials and found to average -2.4 per cent. This indicates that there may be a very slight general orientation of the animals to the light cues, light or dark. When the individual visual preferences are examined it is found that 13 preferred the lighted alleys to some degree. Thirteen of the animals also exhibited a preference for the darkened alleys, and for five of the animals the preferences were equal. When the data were examined for the most marked preference, S or V, it was found that 16 animals preferred S, 14 preferred V, and with one animal the choices were tied. Thus, it is not clear which categoiy of hypotheses Krech would identify as the more adaptive on the basis of his line of argument. All alternatives will be tested. Hot Animals forming only S hypotheses and all animals forming V hypotheses do not differ in

PAGE 43

38 respect to being above or below th,e median in AP trials. Result ; .281 and the Ho is not rejected. A test of tbe same type on V vs. S plus S and 7 frequencies yields 'N/ 2 a A .CX)23, and it is concluded that there is no evidence to identify either a V or S hypothesis factor with adaptability. The test may be applied to animals that shift hypotheses. (In this case, all animals that shifted hypotheses did so from one modality to another. None of the animals shifted within modalities.) The'X^ .016 and the Ho is not rejected. It is concluded that there is no evidence to identify the shiftijig or non-shifting of hypotheses with adaptive behavior. It may be argued that the more hypotheses an animal forms, the more adaptive that animal is. This proposition was tested by dividing the animals into categories above or below the median total number of hypotheses formed and above or below median number of trials on the AP. ^ ,152 and the hypothesis is not rejected. It is concluded that there was no experimental basis for identifying frequency of hypothesis formation with adaptability. In the recent publications of the California group, the implications of hypothesis behavior have been retained but the measure more usually relied on for the interpretation of the data has been the Spatial -Visual Preference Score (PS). The authors have equated a S preference as being more

PAGE 44

innately adaptive. OThis may be tested by a correlation between each animal's PS and the trials on the AP. A significant negative correlation would indicate tbat 7 preferences are associated with, few trials on the AP; positive corrdlatlozui would associate S preferences with ease of solution. The correlation ( r ) between these two measures was -.015j this is taJfcen as indicating no relation existed between an animal's PS and its adaptability. When the number of shifts in algebraic sign of the PS are correlated with the AP trials, r -.229 which is not significant (P > .10). To summarize the results of this section, no evidence was found in the present ; study to identify either type or amount or method of measuring hypothesis behavior with the measure of adaptive behavior (AP). B. The effect of level of motivation on adaptability #• As will be noted from Figure 3, there is a general U-shaped relation between motivational level and AP trials; an optimal level appears to exist between a -10 per cent and -20 per cent BVC. The extent of this relation is given by the coefficient ij which is .486 (P > .10) and by r -.275 (P > .10). This also results in an insignificant 'X • It is concluded that there is little statistical evidence for an optimal level of motivation for solving the AP, although a slig^ht U-shaped relation is present.

PAGE 45

40 40-, 03 a •H a 35O U 30L. 25r +10-0 0-10 -10-20 % change in body weight T— -20-30 -30 -40 Fig.3.-^e relation between body weight chancre and average nmaber of trials to solve b^ adaptability problems.

PAGE 46

: OHAPTBH IV : ^ • DISOGSSIOK Consideration of the Experimental Hypotlieata ^ A. Motivation and hypothesis behavior One research hypothesis advanced earlier in this paper stated that the frequency, types, and shifts of hypothesis behavior would be some function of motivational level. The results havt shownj (1) that an increase in motivational level, as measured by body weight change, was accompanied by an increase in the total number of days a hypothesis was exhibited; (2) that, while S and 7 hypotheses occurred with approximately equal frequency at the lower levels of motivation, V types declined but S types increased as motivational level increased; (3) that ao statistically significant relation between motivational level and shifting of hypotheses was present, although six of the nine animals that did exhibit shifts were below the median in amount of body weight lost. ; !Ehe effect of motivational level on hypothesis behavior had not been investigated prior to this study, and in neither the prior work of Krechevsky, nor in his more current work with the California group, has there been any

PAGE 47

^2 indication that these effects might exist. Particular fo3?ms of hypothesis "behavior (S preferences and shifts in preference) were assumed by the California group to represent more adaptive behavior than V preferences and lack of shift in preference. Since shifts predominate at lower levels while 3 hypotheses predominate at higher levels of motlf ration, the results are apparently inconsistent with these assumptions. Rather, the present results indicate that such shifts in hypothesis behavior should more parsimoniously be related to motivational levels rather than to the relative adaptive level of some particular form of hypothesis behavior or the "adaptability" of the animal. A second hypothesis stated that food and water deprivation would lead to different kinds of hypothesis behavior. Thla hypothesis proved to be untestable since the attempt to establish the required operationally equivalent motivational levels (measured by running speed and weight change) failed. This failure may be attributed to three causes. First, too small a sample of animals was used in the pilot study. Second, although it was not apparent from the pilot study, it now seems that the use of a percentage of ad libitum food or water intake does not produce sufficiently homogeneous running times or weight changes, a?hird, the relatively long period (about a month) of the deprivation schedule may have led to different effects within the food and water conditions.

PAGE 48

45 Hie method which was utlllfsed (reduced percentage of ad libitum intalce) was an attempt to overcome certain difficulties of the traditional limited time schedules (water deprived rats may adjust their water intake to the amount of time available) and limited amount schedules (intakes and nutritional requirements vary partly as a function of age, body size, and metabolism). It now appears that sufficiently precise control would be obtained by reducing the body weight of each animal to some predetermined level before the start of the experimental program; this body weight level could be maintained by making daily ad^justments in each animal's daily food or water ration. B. Hypothesis behavior and adaptability A third experimental hypothesis asserted a positive relationship between type and frequency of hypothesis behavior and adaptability (as measured by the trials to solve the adaptability problems, A?). In the present study no relation was found between the AP measure and total number of hypotheses formed, shifts in hypotheses, or types of hypotheses. This is a result which is in apparent oppo• eition to the assumptions of Krech and his associates since it particular type of hypothesis behavior (spatial) and shifts in hypothesis behavior had been identified as "adaptive." There has not been any previous empirical evidence comparing hypothesis behavior with some other measure of adaptability.

PAGE 49

These results may be interpreted to mean that hypothesis behavior is not a valid measure of adaptability. However, it is possible that the measure of adaptability, AP, used in the present study, is invalid. This is a crucial question, but since there is little precedent and no consensually agreed upon definition of adaptive behavior in the psychological literature, this point must at present be settled on logical grounds. Hypothesis behavior has been the major, but seldom used, convention for assessing this variable. While other measures such as the HebbWilliams apparatus might have been used, the AP measure was chosen because it uses the same apparatus, procedure, criteria, and perceptual -motor capacities; equally important, its logic is congruent with Krech*s reasoning that hypotheses are adaptive. It differs essentially in permitting a solution to take place. There exists such a high degree of similarity between hypothesis behavior and the AP measure, that the lack of any apparent relation between the two raises serious doubts as to the "adaptive" nature of hypothesis behavior. It also raises the question as to what hypothesis behavior is. C. Motivation and trials to solve the two adaptability problems A fourth hypothesis was advanced that adaptability would be related to motivational level. The results were

PAGE 50

*5 inconclusive; a U-shaped relation indicating an optimal motirational level (between -10 and -20 per cent BWG) was found but statistically significant correlations were not obtained. The concept of an optimal motivational level dates back to an original formulation by Terkes and Dodson (1908). It has been found to be generally true in such studies as the Columbia obstruction box experiments (Warden, 1931), and more recently arousal level theorists have postulated a similar relation (Freeman, 19^0; Duffy, 1957)*.^ D. Conclusions. The relation of various hypothesis behavior measures to the solution of two problems has been examined and no support was found for the assumption that a particular type of hypothesis behavior can be considered intrinsically more adaptive than some other type. It has also been shown that the hypothesis behavior exhibited by an animal is a function of the animal's motivational level. It was therefore considered more parsimonious to relate these hypothesis behavior characteristics to motivational level rather than to assumed levels of adaptability. It was not possible to test the effects of food and water deprivation on hypothesis behavior since the attempt to establish operationally equivalent motivational levels in these categories was not successful. Finally, some evidence of an optimal level of motivation for performance was obtained but the evidence was not statistically significant.

PAGE 51

, CHAPTER r SUMMAET This study investigated relationships "between: Krech's various measures of hypothesis behavior in the rat and adaptability, motivational level and hypothesis behavior, and motivational level and adaptability. Thirty-one hooded rata were placed on limited foodor water-diets. After receiving trials to determine hypothesis behavior in the Krech insoluble maze, the animals were tested for adaptability by number of trials to solve two problems in the same apparatus^ The results indicated that! 1. There was no apparent relation between type, duration, or shifts in Krechs hypothesis behavior and the measure of adaptability. 2. The type of hypothesis behavior and its duration was a function of motivational level; hypothesis behavior which utilized light cues gradually diminished with increased body weight loss while hypothesis behavior based on spatial cues rapidly increased. 3. There was slight evidence of a curvilinear relation of adaptability to motivational level but it was not statistically significant. 46

PAGE 52

47 It was not possible to determine differences in hypothesis behavior as a function of type of deprivation because of an inability to est^blishi ecjuivalent motivational levels. The results were interpreted as evidence against Erech's assumptions that hypothesis behavior is adaptive, A motivational interpretation was considered to be the most parsimonious organization of the data. I

PAGE 53

BIBLIOGRAPHr < Boring, S. G, A Mstory of experimental paycholoj^y (2nd 3d.) New York: Appleton-Century-Crofts, 1950. Brown, J. S. Motivation New Yorkj McGraw-Hill, 1961. Duffy, E, The concept of energy mobilization. Psychol. Hev. 1951, 30-40. .. The psychological significance of the concept of "arousal" or "activation." Psychol. Hev., 19571 64, 265-275. Freeman, G. L. The relationship between performance level and bodily activity level, J. exp Psychol., 19^, 26, 602-608. — Hamilton, G. 7. A study of trial and error reactions in mammals. J. Anim. Behav 1911, 1, 35-66. Hllgard, E. R. Theories of learning New York: AppletonOentury-Crofts, 1955. Krech, D. Cognition and motivation in psychological theory. In W. Dennis (Ed.), Current trends in psychological theory Pittsburgh: Univ. Pittsburgh Press, 1951. Pp. 111-139. Rosenzwelg, M. R. Bennett, E. L. & Krueckel, B. Enzyme concentrations in the brain and ad^justive behavior patterns. Science 1954, 120, 994-996, Krechevaky, I. "Hjrpothesea" versus "chance" in the presolution period in sensory discrimination learning. Univ Calif. Publ Psychol 1952, 6, 27-44. The genesis of "hypotheses" in rats. Pnlv Calif Publ. Psychol .. 1932, 6, 45-64. The docile nature of "hypotheses." J. comp Psychol .. 1933. 1^, 429-443 Hereditary nature of "hypotheses." J. comp, Psychol., 1933, 16, 99-116, ~ ^ 48

PAGE 54

49 Krechevsky, I. Brain meclianiGEia and "iiypotheses. J. comp Psychol 1935, l^i ^25-468, Brain mechanisms and variability: I, Variability wi thin a means-end-readiness. J. coap Psychol 1937, 2^, 121-138. Brain mechanisms and variability: II. Variability where no learning is involved. J. comp Psychol, 1397, 2^, 159-163. *• Brain mechanisms and variability: III. Limitations of the effect of cortical injury upon variability. J. comp Psychol 1957, 2^, 351-364. Lashley, Karl S. Brain mechanismB and intelligence Chicago: Univ. Chicago Press, 19-9. MalmoR. B. Anxiety and behavior arousal. Psychol Rev 1957, 64, 276-287. McNemar, Q. Psychological statistics (2nd ed.) New York: Wiley, 1955t Munn, sr. L. Handbook o.. ^.^^^^.^^^^ Boston: Houghton-Mifflin, 1950. j^sychological research on the rat Osgood, 0. S. Kethod and th eory in experimental psychology New York: Oxford University Press, 1953. Petrinovitch, L. & Bolles, R. Deprivation states and behavioral attributes. J. comp physi ol. Psychol., 1954, 42, 450-455. Sosenzweig, M. R. Krech, Bennett, E. L. Brain chemistry and adaptive behavior. In H. P. Harlow & C. N. Woolsey. BioloRical and biochemical bases of behavior Madison: University of Wisconsin Press, r55^i 567-400. Ryan, T. A. Multiple comparisons in psychological research. Psychol Bull .. 1959, 26-47. Spence, K. W. The nature of discrimination learning in animals. Psychol Rev., 1936, 4^, 427-449. Continuous versus non-continuous interpretations of discrimination learning. Psychol Rev. 1940, 42, 271—288.

PAGE 55

50 Spence, K. V. An eaqperifflental test of the continuity and non-continuity theories of discrimination learning. J. exp Psychol 19^5, 2I* 255-266. Stellar, £. & Hill, J. H. OJhe rat's rate of drinking as a function of water deprivation. J, comp physiol Paychol .. 1952, ft^, 96-102. Stevens, S. S, (M. ), Handbook of experimental psychology New York: Wiley, l^^TI Tolman, 33. C. There is more than one kind of learning. Psychol Rev., 19^9, 1^^-155. Warden, 0. J. Animal motivation studies* The albino rat. New York: Columbia Univ. Press, 1931. Witkin, H. A. 'Hypotheses' in rats: An esjperimental critique. III. Summary evaluation of the hypotheses concept. Psychol Rev. 19^2, 5^1-568. Yerkes, R. M. & Dodson, J. D. The relation of strength of stimulus to rapidity of habit formation, J. comp Neurol Psychol 1908, 18, 459-^2,

PAGE 56

APPENDIX

PAGE 57

ii TABLE 9 PER CENT BODY WEIGHT CHANGE, TYPE AND OCCURRENCE OF HYPOTHESES, AND TRIALS TO SOLVE THE ADAPTABILITY PROBLEMS Per Cent Deprivation Body Condition Order of Presentation Weight Food (F) and Hypothesis Formed and Trials to Solve Change Water (W) on Day Number Adaptability Problems V S V + 9.7 70 W VD 3; SR 7 19 23 + 5.8 70 W VD 8 l Ii + 5.7 65 F SL I, 2,3 + 3. 1 70 W VD 5, 6, 7, 8, 9 Ii + 2.4 70 W none U ti + 2.2 70 W SL 6, 7, 8, 9 so It + 1.7 70 W SR 5, 6; VD 7, 8. 9 t$ 21 + 0.9 SOW none u If 1.6 • 50 W SR 2, 7 u 13 2.4 [' 50 W SL 5,6 INI, -2.7 70 W VD 4, 5; SR 9 25 a -4.7 50 W VD 5,6 22 13 5.5 65 F SR 5, 8. 9; VD 7 II -6.8 50 W VD 7.8,9 11 13 7. 5 50 W VD 4,5,8,9 24 15 -7.7 70 W VD 2; SR 6. 9 H l> -7.9 65 F VD 3, 7, 9; SL 4 20 13 -11.7 65 F SR 3, 5, 6, 7; VD 9 u H

PAGE 58

Table 9 Cont. 53 Per Cent Deprivaition Body Condition Order of Presentation Weight Food (F) and Hypothesis Formed and Trials to Solve Change Water (W) on Day Niunber Adaptability Problems V S V -16. 8 50 W SL 3. 4. 5 11 12 -18.5 65 F SR 6,8 IS 12 -20.5 65 F VD 8 12 16 ? 1 7 — £ 1 f Aii IT 4. 5, 6, 7. 8. 9 It -23. 1 45 F SR 5, 6, 7. 8, 9 It It -23.5 65 F VD 3; SR 1. 7, 8. 9 16 17 -25.3 45 F SL 2. 3. 4, 6. 7. 8 14 -26. 6 45 F SL 4. 5, 6, 7, 8. 9 22 11 -28.9 45 F SR 3, 4, 5, 6; VL 7 18 13 -31.2 45 F SL 4, 5. 6. 7, 8. 9 ^ It -31.4 50 W SR 3. 5, 7, 8, 9 It 11 -33.6 45 F SL 4. 5, 6. 7, 8. 9 It It -33.7 45 F SL 2. 3. 4. 5, 6, 7, 8. 9 12 12

PAGE 59

.V BIOGSAPHIGAL SKETCH Robert Louis Procter was bom March 26, 1928 in Washington, D. C. The son of a Naval Officer, he lived in Virginia, New York, California, and Massachusetts until his father's retirement in Vermont in 1937. He graduated from Springfield (Vt.) High School in 19^5 and was accepted into the Army Specialized Reserve fUraining Program and assigned to Massachusetts State College. He was later transferred to Korea and then discharged from the Army in 19^7* He was admitted to the University of Missouri in 19^8 and received his B.A. degree in 1951 and his M.A, degree in psychology in 1953. He returned to the Army in that year and was discharged in 1955. He entered the University of Florida in 1955 and held a series of graduate assistantships, primarily in clinical and comparative psychology. He was also employed at Sunland Training Center, Gainesville, Florida, and Rome (New York) State School, where he was Senior Clinical Psychologist. Hi internship was at the Veterans Administration Hospitals in Gulfport and Biloxi, Mississippi. He is a member of Psi Chi, psychology honorary, and the Florida Psychological Association.

PAGE 60

This dissertaion was prepared under the direction of the chairman of the candidate's supervisory committee and has been approved by all members of that committee. It was submitted to the Dean of the College of Arts and Sciences and to the Graduate Council, and was approved as partial fulfillment of the requirements for the degree of Doctor of Philosophy. June 11, 1962 Dean, Graduate School Supervisory Committee i


36
the problems presented would be SL and VL, although not
necessarily in that order, Thus, the first type of order
effect varied while the second type was partially controlled.
An indication of the effect of this first type of
order can be obtained from the point-biserial correlation
between order (first problem the same, first problem
opposite demonstrated preference) and the distribution of
number of trials on the AP. This correlation has the value
r ^ .013 which of course is not significant. When tested
by A (Same-Opposite x More-Less median trials on the AP),
2
"X .028 and is not significant. Therefore it will be
assumed that an order effect does not exist in the analysis
of the data to follow.
Some statistics of the animals* gross behavior are
presented in Table 8, It will be noted that more animals
formed S hypotheses and more frequently, than any other
TABLE 8
TYPE AND FREQUENCY OF HYPOTHESIS BEHAVIOR
Number of
Animals
Type of Hypothesis
Formed
Meqn Number of
Days Persisted
14
S only
4.57
6
V only
2.67
9
Both S and V
4.11
2
none



15
The results of experiments are sometimes discussed
with reference to hypotheses. Petrinovitch and Bolles
investigated the effects of kind of deprivation on learning.
Using hungry and thirsty rats in a single-unit-T-maze, they
found that when the goal was alternated on every trial the
food-deprived animals were better, but when the goal
remained fixed, the water-deprived animals were superior.
They related their findings to Krechevsky's (1957) report
that normal animals alternate and vary their hypotheses.
The authors suggested "In the light of the present experi
ment, it is possible that his data do not pertain to
'normal' animals, but only to normal hungry animals. In
general, our results suggest that in any study of variability
or stereotypy in behavior, the nature of the 'drive' used
must be carefully considered" (1954, p. 452).
Cognitive theorists have been attaching increased
importance to drive states. Tolman (1949) has modified his
earlier position and suggested that drive reduction may be
necessary for some types of learning. Similarly, Krech no
longer holds to a strict cognitive theory of learning. He
writes "It is quite obvious, if you examine many of the
theories extant in psychology today, that these theories do
not rest content with motivational constructs only, or with
cognitive constructs only, but require both sets of
constructs.... The belief in the interaction among the


53
Table 9 Cont.
Per Cent
Body
Weight
Change
Deprivaition
Condition
Food (F) and
Water (W)
Hypothesis Formed
on Day Number
Order of Presentation
and Trials to Solve
Adaptability Problems
V
S
V
16.8
50 W
SL 3, 4, 5
11
12
-18.5
65 F
SR 6,8
13
12
-20.5
65 F
YD 8
12
16
-21.7
65 F
SL 4, 5, 6, 7. 8, 9
12
13
-23. 1
45 F
SR 5, 6, 7, 8, 9
18
18
-23.5
65 F
VD 3; SR 1, 7. 8, 9
16
17
-25.3
45 F
SL 2, 3, 4, 6. 7, 8
10
14
-26. 6
45 F
SL 4, 5, 6, 7, 8, 9
22
11
-28.9
45 F
SR 3,4, 5, 6; VL 7
18
13
-31.2
45 F
SL 4, 5, 6, 7, 8, 9
16
17
-31.4
50 W
SR 3, 5, 7, 8, 9
18
18
-33.6
45 F
SL 4, 5, 6, 7, 8, 9
12
13
-33.7
45 F
SL 2, 3, 4, 5, 6, 7, 8, 9
12
12


10
He also questioned Krechevsky's interpretation of the
cortical lesion studies* Habits having a far simpler basis
than hypotheses* cannot be performed after extensive
cortical lesion* Yet every animal submitted to decorti
cation showed systematic behavior in linear insoluble*
situations, pointing to the relatively simple basis of these
responses (pp* 566-567)* Witkin suggests that the changes
in behavior resulting from differential locus of lesions may
be the result, essentially, of removing from the animal the
capacity to receive particular sensory cues within the maze.
Witkin concludes, "The hypotheses' concept and the system
of which it is a part arose in opposition to the crudely
mechanistic conceptions which have gained prominence...*
As opposed to such piecemeal conceptions, the 'hypotheses'
concept attempted to present the learning process as
orderly, organized, and predictable, and pictured the
learner as a more active participant in determining the
character of the final learned habit. This...is a very
desirable end, but since...the 'hypotheses' concept swings
to an extreme where it is beyond the bounds of evidence, it
constitutes a very weak kind of opposition... (p. 567).
Spence also had been critical of Krechevsky's
interpretations, 0ontrary to the belief of certain writers,
there has been no disagreement concerning the behavioral
facts.... Agreement ceases however, over attempts at further


29
TABLE 5
MEAN PEH CENT OP BODY WEIGHT CHANGE OVER
THE COURSE OP
THE EXPERIMENT.
(H 31)
Deprivation
Level
Pood
Water
Moderate
-13.1
1.8
Severe
-28.9
8.8
To summarize the results obtained so far,
1. The desired two equivalent levels of motivation were not
attained, since type of motivation was contaminated by level
of motivation.
2. Per cent of BWC was significantly correlated with
running time and was used as the independent variable
representing motivational level.
Tests of the Hypotheses
* \
A. The effect of varying motivational level.
The hypothesis was advanced in the introduction that
the hypothesis-behavior of the animals would be a function
of level of motivation. It is apparent from Figure 2 that
varying motivational level induces major changes in both
the gross amount and type of hypothesis behavior, spatial
(S) or visual (V). The more negative the BWC, the more S
hypotheses the animal forms.


31
The form of the data and the nature of the inquiry
lent itself in this case and in many of the tests to
p
follow to the use of the*X test of independence. The
data are usually in frequency form and also in dichotomous
categories such as S or V hypotheses. Even where dichotomous
categories do not exist, it was meaningful to create at times
such classes as above or below the median. The Tates
correction of the formula was used (McNemar, 19551 formula
86a). The null hypothesis (Ho) was rejected when*X 2.05,
1 d.f. 5*84 was surpassed.
Ho; Animals do not differ in the number of hypotheses
formed as a function of being above or below the
median in BWC. (N 31).
p
Result: *X 5.59 and the Ho is rejected. Another indi
cation of this relationship is given by which has a value
of .663 (P < .001). (The F ratio between and r 1.057
and linear regression may be assumed). It is concluded that
the number of hypotheses formed is a function of BWC; more
hypotheses are formed with severe losses of body weight and
fewer hypotheses are formed with moderate increases of
weight.
Ho; That animals do not differ in forming S or V
hypotheses predominantly as a function of being
above or below the median in BWC.


MOTIVATION, KRECH HYPOTHESIS
BEHAVIOR, AND ADAPTABILITY
By
ROBERT L. PROCTER
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
June, 1962

AOraOW&BDOMEireS
The author wishes to express his gratitude to the
members of his committee, Prof, Holland H. Waters,
Chairman, Prof. Richard J. Anderson, Prof. Bradford N.
Bunnell, Prof. James C. Dixon, Prof. Herbert D. Kimmel,
Prof. J. Milan Kolarik, and Prof. Zareh M. Pirenian. He
is especially appreciative of the many hours given by
Prof. Waters and Prof. Bunnell. The clarification of
learning theory by Prof. Judson S. Brown and some invaluable
assistance from Mr. William P. Stone are also gratefully
acknowledged.
ii

TABLE OP CONTENTS
Page
ACKNOWLEDGMENTS ...... ....... ii
LIST OP TABLES iv
LIST OP PIGURES y
CHAPTER
I.INTRODUCTION 1
II.PROCEDURE ......... 20
III,RESULTS 25
IV.DISCUSSION. ... 41
V.SUMMARY .................. 46
BIBLIOGRAPHY, . . . 48
APPENDIX 51
BIOGRAPHICAL SKETCH 54
iii

LIST OF TABLES
Table Page
1. The Krechevsky Trial Order for the Position
of Open Doors and Lighted Alleys on the
(Twelve Daily Trials in the Haze *. 5
2. Hean Sunning Time in Seconds in the Training
Trials 26
3. Hean Running Time in Seconds in the
Hypothesis Haze * 27
4. Analysis of Variance on the Effects of Type
and Level of Deprivation on Running Times in
the Hypothesis Haze 28
5. He an Per Cent of Body Weight Change Over the
Course of the Experiment* 29
6. Comparison of the Hoderate Food Deprived and
Severe Water Deprived Groups for
Frequencies of Dominant Hypothesis Type ... 33
7. Food and Water Deprived Animals Hatched for
Equivalent Body Weight Change 34
8. Otype and Frequency of Hypothesis Behavior 36
9. Per Cent Body Weight Change, Type and
Occurrence of Hypotheses, and Trials to Solve
the Adaptability Problems 52
iv

LIST OF FIGURES
Page
Figure
1 Floor Plan of the Krechevsky Hypothesis
naze 4
2. Number and Type of Hypotheses Formed as a
Function of Body Weight. 30
3. The Relation Between Body Weight Change and
Average Number of Trials to Solve Both
Adaptability Problems. 40
v

CHAPTER I
INTRODUCTION
This study was primarily concerned with determining
how hypothesis behavior in the hooded rat varies with
changes in motivation. It was also concerned with finding
out what relation such hypotheses have to adaptive behavior,
e.g., are animals that show certain types and numbers of
hypotheses more adaptable or flexible than others? And
finally, it was concerned with the relation between
motivation and adaptability.
Hypothesis behavior is a term first used by
Krechevsky (1932a) to describe the apparently non-random,
systematic attempts of rats to solve an insoluble linear
maze problem. It is necessary to maintain a careful
distinction between the behavior of the animal in the
experimental situation and the implications of the
experimental results. Hypothesis behavior refers to a
particular type of behavior which occurs in a specific
situation. The behavior is a pattern of choices made in
relation to specific cues in an insoluble discrimination
maze. When the pattern of choices differs from a random
1

2
level by a certain amount, hypotheses are said to be formed.
Hypothesis behavior and hypotheses are terms which have
meaning only when the apparatus, the cues, the pattern of
choices, and the statistical level are understood.
Hamilton (1911) noticed in the course of a trial and
error learning experiment that rats and other animals make
apparently systematic attempts at solving problems that
cannot be solved; they may go consistently to the right, or
left, or to the last place found successful, and so on.
Lashley (1929) also noted this feature of behavior in
discrimination problems. He indicated that while there was
no experimental technique available at that time to
investigate this behavior, these 'attempted solutions"
probably represented a significant aspect of the learning
process. Lashley's observations served as at least a
partial basis for Krechevsky's investigations; Krechevsky's
experiments were designed to provide the lacking experi
mental technique. In one of his earliest published works
he states, "The data from the present experiment have been
examined in the light of Lashley's suggestion and an attempt
has been made to devise a method for the objective
determination of the validity of that suggestion."
Krechevsky concluded from this study that "...it is shown
quite definitely in the presolution period that...the
animal is engaged in bringing to perfection various

5
attempted solutions.... In the light of the evidence
presented here it is suggested that the helter-skelter
unorganized trial and error response as a description of
the early part of the learning process is invalid, and that
we must change our description of the learning process so
as to recognize the existence of organized and systematic
responses at all stages of the process. (1932a, p. 43)
Krechevsky's interpretations were directed against
the prevalent behaviorism which used trial and error
learning and later the conditioned reflex as a model.
Boring (1950) characterized both of these models as elimi
nating the need for integrative principles such as
organization or insight. Tolman, under whom Krechevsky
was studying for his doctorate, was then developing his
theory of purposive behaviorism. This theory emphasized
the cognitive, intellectually organized aspects of learning.
Krechevsky's work with hypothesis behavior served both as
a fundamental tenet of the Tolmanian system and as
experimental evidence against an anti-insight behaviorism.
Krechevsky's 1932b experiment formed the basis for
all the later work in this area. The maze which he designed
was a four unit discrimination apparatus which could not be
solved (see Figure 1). The animal was confronted by four
choice-points between the start box and the goal box. A

4
Fig. 1.-Floor Plan of the Krechevsky Hypothesis Maze.
This diagram shows the maze as it appears on trial 7
An animal following a visual-dark hypothesis would
choose alleys in the order left, left, right, right.
correction method was used at each choice-point and the
reward was available in the goal box at the completion of
every trial regardless of the correctness of the animal's
four choices. Krechevsky used a 23-hour food deprivation
schedule; a cube of bread soaked in milk was used as the
incentive. At each choice-point there were two alleys, a
right and a left, between which the animal could choose.
Over each alley was an electric light, but only one side
was illuminated. The animal might choose on the basis of
spatial cues (right, left) or visual cues (light, dark).
A curtain at the end of each alley prevented the animal
from seeing a swinging door which could be locked to either
side, thus preventing the animal from using one or the
other alleys. One alley of each of the four units was
always locked. When the animal chose an alley which had
an open door, it went through to the next unit. When the

animal chose an alley with, a locked door, it was forced to
reverse its course and go through the alley on the other
5
side.
The animal was given twelve trials each day. On
every trial the lights and doors were shifted in such a way
that: 1) the lights and open doors were on either side of
the choice-point equally often, and, 2) the use by the
animal of any of the eight hypotheses which Krechevsky
identified would result in encountering the same number of
open and locked doors. (See Table 1 for the trial order
positions of lights and doors.)
TABLE 1
THE KRECHEVSKY TRIAL ORDER FOR THE POSITION OF OPEN
DOORS AND LIGHTED ALLEYS ON THE TWELVE DAILY TRIALS
IN THE MAZE
Trial
1
2
3
4
Door Open
L L R L
R R R L
R R L
R
L
L R
R
Light On
R
L R L
R L R R
L R R
R
L
R L
L
Trial
5
6
7
8
Door Open
R
R L L
R L L L
R L L
R
L
R L L
Light On
R
R L L
L L R R
R R L
L
R
L L
L
Trial
9
10
11
12
Door Open
L
R R L
R L R R
L L L
R
L
R R
R
Light On
L
L L R
L R R L
R L R
R
L
R L
R

6
Two major types of hypotheses were distinguished,
those based on visual cues and those based on spatial cues.
Among the visual hypotheses, an animal may systematically
choose the lighted alleys (a "Light Hypothesis") or the
dark alleys (a "Dark Hypothesis"). Sometimes an animal
tends to choose an alley in terms of the similarity of
its visual characteristics to the immediately preceding
"correct" alley (e.g., choosing a lighted alley if the
previous lighted alley had proved correct). This
behavior is termed a "Perseverative Visual Hypothesis."
Conversely, the rat may show an "Alternating Visual
Hypothesis"...
The other general categories of hypotheses spatial -
also has four possibilities... there are "Right",
"Left", "Perseverative Spatial", and "Alternating
Spatial" hypotheses. (Rosenzweig, et al. 1958, p. 575)
It is possible to invent any number of systematic
patterns of choice hypotheses for the animal, but
Krechevsky compared the animal's choices for correspondence
to these eight. The equivalent of the Spatial hypotheses,
-Right, -Left, and -Perseverative, and the Visual hypotheses,
-Light and -Dark, had already been noted by previous
investigators. The other hypotheses appear to be logical
extensions of these five.
If the animals made completely random, unsystematic
responses at each of the 4-8 choice-points (four choices on
each of twelve trials), they would choose on the average -
12 right-lighted, 12 right-darkened, 12 left-lighted, and
12 left-darkened alleys in some random order. A theoretical
random animal would be expected to have an average of 24

7
choices to each of the four cues (Left, Right, Light, Dark).
Krechevsky defined non-random, systematic behavior his
hypothesis behavior as being present when the animal chose
one of the eight alternatives with a frequency significantly
greater than the theoretical random mean of 24. The level
of significance which he chose was three standard deviations
above the mean. This amounts to 34.39 choices out of 48.
With this criterion, Krechevsky found that most of
his animals formed hypotheses and from these early studies
he concluded that "...the presence of systematic forms of
behavior in such a situation is to be interpreted to mean
that these systems were determined not as something forced
ab extra by the situation, but as something originating
from the animal himself" (1932b, p. 45). He suggested
that "...descriptions of the lower animals* behavior as
consisting of "stereotyped," "haphazard," "non-insightful"
responses are to be attributed not to a lack of insight on
*
the animals' part but rather to a lack of insight on the
experimenter's part" (1932b, p. 63).
Over the next several years, Krechevsky continued
his investigation of hypothesis behavior. In one of these
studies (1933b) he used animals from Tryon's "bright" and
"dull" strains and found that the "bright" animals had more
spatial hypotheses and the "dull" animals had more visual
hypotheses. Another study examined some of the conditions

8
under which, animals give up or persist in hypotheses. Hats
which had formed hypotheses were divided into three groups.
One group then found the maze solvable in conformance with
the hypothesis each animal was exhibiting; the second group
had the maze solvable for a different hypothesis; a third
group had the incentive removed as soon as an hypothesis
was demonstrated. The first group persisted in their
original hypotheses, the second group changed hypotheses to
the correct solution, and in the third group more than fifty
per cent of the animals refused to run at all. This study
was integrated with a further exposition of Tolman's theory
and Krechevsky*s most important conclusion was that "...if
any behavior-act can be established as docile and purposive,
hypothesis-behavior is definitely so (1933a, p. 442),
Krechevsky began a series of investigations on the
relation of hypothesis behavior to the structure and function
of the brain (1935 1937a, b, c). He found that cortical
lesions decreased the number of different hypotheses an
animal might use, but increased the number of days an animal
would use the same hypothesis. He also found evidence that
this behavior was not related in simple fashion to the
amount of cortical damage, but to both the amount and locus
of damage. One area (occipital lobe) he called V because
few animals made visual hypotheses when it was damaged.
Another area (somesthetic) was identified as S, because of

9
a similar effect on spatial hypotheses. Damage to both
areas tended to result in spatial hypotheses,
Krechevsky published no more articles on hypothesis
behavior until recent years, Hypothesis behavior served as
the prototype for other Tolmanian concepts such as means-
end-readiness, and Krechevsky began experimentation in
other areas. Also, while hypothesis behavior was directly
related to the controversy over the continuous or
discontinuous nature of learning, the hypothesis behavior
maze was not readily adaptable to investigating this
problem. Krechevsky designed other pieces of apparatus for
this purpose.
The only critique of hypothesis behavior which
exists in the psychological literature is that provided by
Vitkin (194-2). Vitkin noted that hypothesis behavior is
not typical of solvable problem situations, but rather of
situations for which no solution can be found. He found
the same type of behavior in the Krechevsky hypothesis
maze when the doors were completely removed (free-choice)
as when the doors were locked according to Krechevskys
insoluble pattern; the behavior was not found when a
solvable pattern was used. He questioned the adaptive or
purposive nature of hypotheses, since animals shifted
hypotheses in a free-choice situation although the first
hypothesis or no hypothesis at all was equally adaptive.

10
He also questioned Krechevsky's interpretation of the
cortical lesion studies* Habits having a far simpler basis
than hypotheses* cannot be performed after extensive
cortical lesion* Yet every animal submitted to decorti
cation showed systematic behavior in linear insoluble*
situations, pointing to the relatively simple basis of these
responses (pp* 566-567)* Witkin suggests that the changes
in behavior resulting from differential locus of lesions may
be the result, essentially, of removing from the animal the
capacity to receive particular sensory cues within the maze.
Witkin concludes, "The hypotheses' concept and the system
of which it is a part arose in opposition to the crudely
mechanistic conceptions which have gained prominence...*
As opposed to such piecemeal conceptions, the 'hypotheses'
concept attempted to present the learning process as
orderly, organized, and predictable, and pictured the
learner as a more active participant in determining the
character of the final learned habit. This...is a very
desirable end, but since...the 'hypotheses' concept swings
to an extreme where it is beyond the bounds of evidence, it
constitutes a very weak kind of opposition... (p. 567).
Spence also had been critical of Krechevsky's
interpretations, 0ontrary to the belief of certain writers,
there has been no disagreement concerning the behavioral
facts.... Agreement ceases however, over attempts at further

11
interpretation of the phenomena (19^5 p. 253)* In an
early theoretical paper on the nature of discrimination
learning, Spence (1936) demonstrated that perseverative
behavior similar to hypotheses was compatible with the
Hull-Spence system. Hypotheses were considered as a
phenomenon dependent on learning processes before the
correct learned response became manifest; as such, it is a
phenomenon characteristic of a pre-solution period. At a
later time Spence distinguished between this perseverative
pre-solution behavior and the hypotheses which Krechevsky
found in an insoluble maze, "...these pre-solution phenomena
appear to be a typical example of what has been described
as trial and error learning, while hypotheses are far from
what...(I understand)...by the terms insightful and
intelligent. Only persistent non-adaptive responses can
attain the distinction of being hypothesesfor, in order
to classify as a hypothesis, a response, although
ineffective, must continue to be persisted in a certain
minimum number of times. A maladaptive act which is
speedily (intelligently?) abandoned cannot ever be a
hypothesis" (1940, p* 287). Both forms of perseverative
behavior are, for Spence, predictable from his theoretical
framework. As a pre-solution phenomenon, the hypothesis is
expected to be replaced by the correct response without any
intermediary hypotheses. In the insoluble problem situation,

12
tlie development of hypotheses (or the complete nonoccur-
rence of them), shifts between hypotheses, and the
persistence of hypotheses, are all theoretically explainable
post hoc providing certain assumptions regarding initial
response strengths and reinforcements can be made.
Unfortunately, the Krechevsky hypothesis maze does not lend
itself to an empirical test of the Spence position.
There was no further work on hypothesis behavior
until 1954 when Krech1, Eosenzweig, and Bennett began
publishing the results of a series of experiments which
correlated brain chemistry with adaptability. Despite the
earlier criticisms of hypothesis behavior, this technique
was used as the measure of adaptability. Two changes were
made in this measure. First, a reduced level of signifi
cance was used as a criterion of hypotheses (33 rather than
35 choices of the 48 occurring each day), and second,
Krechevsky*s (1933) spatial-visual Preference Score was
also used. This Preference Score provided a continuous
scale of relative preference for either visual or spatial
modalities \d.thin each animal. It is computed in this
fashion1 1) the number of choices in each of the eight
hypothesis alternatives is determined for an animal for a
T
Krechevsky changed his name to Krech.

13
given day; 2) the deviation in per cent of these eight
numbers from fifty per cent is computed; 3) the percentages
based on spatial hypotheses are given positive signs, the
visual percentages are given negative signs, and the eight
percentages are then algebraically added.
In a later study by Rosenzweig, et al. (1958), the
results indicated that the amount of cholinesterase (ChE)
significantly increased from the visual area (V) to the
somesthetic area (S) to the motor area (M); the amount in
all areas declined with age. When visual preference
animals were compared to those with spatial preferences, the
ChE rate of decline was found to be very rapid. The authors
were using ChE as a measure of acetylcholine (ACh)
metabolism, and inferred from these results that ACh
metabolism is related to adaptive behavior patterns.
Such an inference necessitates two assumptions. One
concerns the ChS-ACh relation and is a chemical problem;
the relation of hypothesis behavior to adaptability is
properly a psychological problem which the authors justify
as follows!
...we believe that animals that show a spatial
Preference Score are more adaptive than animals
showing a visual Preference Score. We make this
interpretation for the following reasons: (1) As
we have already pointed out, our test is designed
to measure the animal's perceptual selectivity.
(2) We next assume that adaptive behavior is cor
related with the ability of the animal to "pay
attention to" various stimulus aspects of its

14
environment when confronted with a problem. Thus,
for example, Tolman stresses the importance of
determining the ,..conditions which favor relatively
rapid shifts in the dimension of discrimination of a
sign or of a signifcate.." in understanding
differential effectiveness of performance. (3) In
our maze, under the conditions of training that we
have used, almost all animals show a light-going
preference on their first few trials... Achieving a
spatial Preference Score therefore requires that an
animal ignore the dominant illumination cue and pay
attention to the less obvious cue of location in
space. In other words, an animal that develops a
spatial preference shows readier "shifts in discrimi
nation" than an animal whose behavior remains
controlled by the visual cues. (1958, p. 389)
While this is an attractive line of reasoning, there
are certain disturbing features. Whether a continuous
Preference Score or the discrete category, hypotheses, is
used, the criticisms of Witkin and Spence remain pertinent
and have not been answered. In addition, very little is
known about hypothesis behavior and the variables affecting
it; the phenomenon has been a theoretical pawn and there
have been no investigations since Witkin*s efforts twenty
years earlier. Despite the occasional criticisms and the
lack of a sound empirical foundation, Krech treats hypothesis
behavior as definitely adaptive, cognitive, attempted
solutions. He receives some support for this position from
standard reference works which review his work; hypothesis
behavior is uncritically treated as evidence for attempted
solutions on the part of the rat (see Munn, 1950, pp. 250-
253, 333? Stevens, 1951, pp. 312, 744, 774-776; Osgood,
1953* PP. 445-446; Hilgard, 1956, p. 201).

15
The results of experiments are sometimes discussed
with reference to hypotheses. Petrinovitch and Bolles
investigated the effects of kind of deprivation on learning.
Using hungry and thirsty rats in a single-unit-T-maze, they
found that when the goal was alternated on every trial the
food-deprived animals were better, but when the goal
remained fixed, the water-deprived animals were superior.
They related their findings to Krechevsky's (1957) report
that normal animals alternate and vary their hypotheses.
The authors suggested "In the light of the present experi
ment, it is possible that his data do not pertain to
'normal' animals, but only to normal hungry animals. In
general, our results suggest that in any study of variability
or stereotypy in behavior, the nature of the 'drive' used
must be carefully considered" (1954, p. 452).
Cognitive theorists have been attaching increased
importance to drive states. Tolman (1949) has modified his
earlier position and suggested that drive reduction may be
necessary for some types of learning. Similarly, Krech no
longer holds to a strict cognitive theory of learning. He
writes "It is quite obvious, if you examine many of the
theories extant in psychology today, that these theories do
not rest content with motivational constructs only, or with
cognitive constructs only, but require both sets of
constructs.... The belief in the interaction among the

16
hypothesized cognitive and motivational processes is
certainly one of the most outstanding...current trends in...
theories today (1951* p* 114).
The effect on hypotheses of manipulating drive
remains unknown. Motivational variables are known to
facilitate or decrease performance levels, but it is
difficult to apply the motivational literature to hypothesis
behavior for several reasons. First, most of the experi
ments are more applicable to motivational phenomena or
theory. Second, the general nature of hypothesis behavior
is unknown. And third, data in hypothesis behavior
experiments are not directly comparable to data in the
usual learning experiment with only one solution.
The present study was directly concerned with these
problems. It was designed to determine empirically the
relationship between the hypotheses formed by an animal and
another measure of that animal's adaptability; it was also
designed to explore the effects of motivation on hypothesis
behavior and adaptability.
The following research hypotheses were advanced:
1) The type of hypotheses and the frequency with which
they are formed is a function of both the level of
motivation and the type of motivation.
Three major characteristics of hypothesis behavior could
vary as a function of motivational level; these may be

17
stated in question form:
a) Do animals at a low motivational level form
fewer, or more, hypotheses than animals more
highly motivated?
b) Do animals characteristically form visual
hypotheses at one level of motivation, and
spatial hypotheses at another?
c) Is motivational level related to shifts between
types of hypotheses?
Petrinovitch and Bolles had suggested that food deprivation
might lead to more variation in hypothesis behavior than
water deprivation. Accordingly, when level of deprivation
is held constant, the following statement will be tested:
2) Do food deprived animals show more shifts between
hypotheses than water deprived animals.
The establishment of equivalent levels of motivation in two
different modalities, food and water, presents difficulties
in both laboratory technique and research design. True
equivalence is impossible to attain since hunger and thirst
are different experiences. An effective equivalence may
possibly be approached by using equivalent operational
definitions. The more usual approach of using hours of
deprivation would not prove to be a satisfactory procedure;
Stellar and Hill (194-5) observed that rats may adjust their
water consumption, within a limited time of access, to fit

18
their deprivational state. Brown (1961, pp. 71-74)
presented data which leads to the same conclusion. The
method which was used in the present study determined the
ad libitum food or water intake for a pilot group of animals
and then provided that fixed percentage of such an amount
which would result in equivalent running speeds in a
straight runway. The experimental animals were also weighed
just before this limited amount deprivation schedule was
started, and, immediately after it ended; this measure
provided an additional means of defining motivational level.
3) The adaptability of animals is related to the type and
frequency of the hypotheses which they display.
Rosenzweig, Krech, and Bennett (1958) have argued that
spatial preferences and hypotheses, and shifts of preference
and hypotheses, represent more adaptable behavior since the
animal has given up responding to an initial (apparently
more dominant) aspect of the situation. The measure of
adaptability used in the present study involved a similar
logic; the animal was presented with the doors and lights
of the hypotheses maze arranged so that one of the
hypothesis alternatives was solvable. When criterion was
reached, the maze was adjusted for solution of a second
hypothesis alternative. Adaptability was thus equated
with the number of trials to adopt new modes of response as
they became more efficient. The appropriate questions are*

19
a) Do animals that demonstrate spatial hypotheses
and preferences take fewer trials to solve the
problems than animals with visual hypotheses and
preferences?
b) Do animals that have many hypotheses take fewer
trials to solve the problems than animals with
few or no hypotheses?
c) Do animals that shift hypotheses and preferences
take fewer trials to solve the problems than
animals that do not shift?
4) The adaptability of the animals is related to the type
and level of motivation.
The Yerkes-Dodson (1908) law holds that optimal motivational
level is inversely related to problem difficulty; in the
present experiment problem difficulty was not effectively
manipulated, therefore, according to this law an optimal
level of motivation should be apparent. The positions of
Duffy (1951 1957) and Malmo (1957)* are also consistent
with this. Brown (1961, p. 9*0* however, found equivocal
results in the area of discrimination learning and concluded
that further studies were "urgently needed." The relation
ship can be posed as the following question*
a) Does a linear relation exist between motivational
level and number of trials to solve the problems
such that increased motivational level results in
fewer trials?

CHAPTER II
PROCEDURE
The general experimental procedure as described by
Krechevsky (1932b)# and Rosenzweig, Krech, and Bennett
(1958) was followed with certain exceptions. These
exceptions were a difference in kind and amount of incentive,
a slight modification of the maze, and a slightly higher
criterion for determining hypotheses. The criterion was
placed at 36 choices out of 48 for the following reasons
Since each animal had a number of days on which to be
scored for hypotheses, the level of significance should be
adjusted accordingly. Similarly, the scoring of each
animal's daily responses for hypotheses increased the
likelihood of finding hypotheses in direct proportion to
the number of alternative hypotheses. The four alternating
and perseverating hypotheses were not scored so the level
of significance was not adjusted to include these data in
the determination of hypotheses. Significance levels
were adjusted to remain for each animal at P < .01 (see
Ryan, 1959) in consideration of nine days opportunity to
form an hypothesis and four scorings of each day's behavior
(spatial -right, -left, and visual -light, -dark).
20

21
Subjects> Subjects were 32 naive male hooded rats,
varying in age from 92 to 127 days at the start of the
experiment* Of this group, one animal was lost from the
experiment due to escape from the cage and an undeterminable
amount of eating. All animals were housed individually.
Water-deprived animals had ad libitum Purina rat pellets;
food-deprived animals had ad libitum water. They were
maintained on a 12 hour light-dark cycle in an air
conditioned room which had a mean temperature of approxi
mately 78 F.
Apparatus. The standard Krechevsky hypothesis maze
was used, modified with five guillotine doors to prevent
retracing among the start box, goal box, and discrimination
units. The top was covered with window screening painted
with aluminum paint to reflect the interior light of the
maze and reduce room cues. Determining each animal's
choices by mirrors proved unsatisfactory, and direct
observation through the screen was used. The maze was
placed on a table at the side of a small room. The only
light came from the apparatus itself and a 7# watt bulb
suspended over the goal box. This light permitted the
experimenter to observe when the animal had reached the
goal box and to record the animals' responses. The straight

22
runway which was used in the pretraining period was similar
to the hypothesis maze but with the center sections removed
so that no alleys were formed. The guillotine doors were
retained and curtains were hung at the three center doors
to provide habituation for the animals.
Procedure. Animals were treated in groups of about
eight, with random assignment to each condition. The
animals assigned to the food-deprivation category (PD) were
placed on ad libitum powdered Purina rat chow and water
schedules; animals in the water-deprivation category (WD)
received the same treatment except they were given pellets
rather than powdered food. After a four-day adjustment
period, measurements of each animal1s food or water (as
appropriate) consumption was taken for 14- days. At the end
of this time the average daily consumption was computed and
the animals were weighed.
A pilot study had indicated that 65 per cent of the
ad libitum food intake and 70 per cent of the ad libitum
water intake vjould result in approximately equivalent
running times in a straight runway; similar results were
found for a 45 per cent food and a 50 per cent water level.
Therefore those four levels were adopted. The animals
were placed on the appropriate schedule one day prior to
the start of the pretraining trials and remained on these
diets until the conclusion of the experiment.

23
Pretraining. The animals were given 5 trials the
first day, 5 trials the second day, and then 10 trials each
day for nine successive days; the animals were run every
24- hours. The incentive was approximately 0.1 gram of
powdered chow or 0.1 c.c. of water, as appropriate. The
incentive was taken from each animal's daily ration.
Hypothesis trials. The animals were given 6 trials
on both the first and second days, and 12 trials each of
the successive eight days. The Krechevsky order (see Table
1) was followed. After the final day's trials within the
hypothesis maze were completed, the hypotheses for each
animal were computed. Weights were obtained for each animal.
Adaptability trials. On the basis of each animal's
hypotheses and membership in the four experimental groups,
a decision was made as to which of two visual problems and
two spatial problems were to be solved. The criterion for
solution was the same as for hypotheses; however, whereas
hypotheses were computed on the basis of an animal's daily
performance, these trials were counted serially without
respect to the termination of each days trials; such a
procedure is necessary to prevent grossly unequivalent
amounts of practice. An animal was considered to have
reached criterion when it first attained 36 out of any 46
consecutive choices which were correct; animals that reached
criterion on the first, second, or third choice-points of a

24
trial were allowed to finish that trial and attain the
incentive.
Roughly half of the animals which had made visual
hypotheses were assigned spatial problems firstj spatial
hypothesis animals were treated similarly. When animals
had made both types of hypotheses, or none, the assignment
was made on the basis of the dominant preference. In all
cases, the assignment of any problem within either stimulus
modality was opposite to that animal s dominant preference.
The number of daily trials and the incentives
remained the same as in the latter part of the hypothesis
maze experiment. At the conclusion of the first problem,
each animal was immediately shifted to the second problem
on the next trial. The scores were each animals total
number of trials to solve both problems.

CHAPTER III
RESULTS
This study was concerned with certain relationships
between motivation* hypothesis behavior* and adaptability.
Two semi-independent measures of motivation were used.
First there was a deprivation schedule providing for
restricted amounts of food (FD) or water (WD) at two
different levels, moderate and severe. Second, a measure
ment of body weight change (BWC) over the course of the
experiment was obtained.
Since the experimental design required equivalent
levels of motivation in the FD and VD groups for testing
certain hypotheses, evidence concerning this equivalence
will be presented first. Percentages of ad libitum food
and water had been chosen to obtain equivalent running
times. Speed of manning was thus used as a criterion of
equivalence. The concept of equivalence can be assumed to
have been validly met if it can be shown that the
deprivation schedules establish equivalent running times
within each level of deprivation.
There are two sources of running time data in the
experiment, one from the training trials and the other from
25

26
tlie trials in the Krech apparatus. Response time data were
available from 19 animals during the training trials to use
as a check against this method. These data, presented in
Table 2, showed that only a slight degree of equivalence
existed between each of the moderate and each of the severe
groups. These data were obtained in a straight runway and
were therefore similar to the conditions of the pilot group.
However, it is of more pertinence to determine whether
equivalent running speeds existed at the time of the
hypothesis behavior trials.
TABLE 2
MEAN RUNNING TIME IN SECONDS IN THE
TRAINING TRIALS
(Days 3 through 10, N 19)
Deprivation
Level
Pood
Water
Moderate
23.9
17.7
Severe
7.6
15.6
Response time data for a sample of 20 animals in
the hypothesis maze are presented in Table 3. It will be
noted that the moderate PD and severe WD groups were
approximately equivalent, but the other groups occupy
extreme positions. This indicated that type of deprivation,
food or water, was possibly confounded with level of
motivation. Analysis of variance established that such

27
TABLE 3
MEAN RUNNING TIME
IN
SECONDS
IN THE
HYPOTHESIS
MAZE
(Days 1-9,
N
- 20)
Level
Pood
Water
Moderate
32.1
49.2
Severe
20.4
35.2
confounding did take place. According to the original
experimental procedure, a significant difference in running
time should be found for the level of motivation variable
(since the restricted diets were deliberately chosen to
produce different running times). As the summary in Table
4 indicates, animals differed significantly (P < .005) in
level of motivation as measured by running time level.
However, these restricted diets were also deliberately
chosen with the intent of producing no significant differ
ence between running times on the type of deprivation
variable. Here, unfortunately, they also differed and this
difference is significant (P < .001). Clearly, testing the
experimental hypotheses relating to the food-water variable
would introduce motivational level as a contaminating
factor.

28
TABLE 4
ANALYSIS OF VARIANCE ON THE EFFECTS OF TYPE AND
LEVEL OF DEPRIVATION ON RUNNING TIMES IN THE
HYPOTHESIS MAZE
(X* log X transformation)
Source
d.f.
M.S.
F
P<
Type (Food-Water)
1
.230
20.35
.001
Level (Moderate-Severe)
1
.138
12.21
.005
Type x Level
1
.004
55
Error
16
.011
Total
19
Another indication that equivalence did not exist
was found when the mean BVC was computed for each of the
cells. Table 3 with running time data and Table 5 with
mean body weight changes reflect the same general trend.
This suggests that the two measures are correlated. This
turned out to be true, with a rank order correlation of
.680 (P < .01). Because of this correlation, per cent
of BVC, rather than the dichotomous Moderate-Severe
categories, was used as the indicator of motivational level
in further analysis of the data.

29
TABLE 5
MEAN PEH CENT OP BODY WEIGHT CHANGE OVER
THE COURSE OP
THE EXPERIMENT.
(H 31)
Deprivation
Level
Pood
Water
Moderate
-13.1
1.8
Severe
-28.9
8.8
To summarize the results obtained so far,
1. The desired two equivalent levels of motivation were not
attained, since type of motivation was contaminated by level
of motivation.
2. Per cent of BWC was significantly correlated with
running time and was used as the independent variable
representing motivational level.
Tests of the Hypotheses
* \
A. The effect of varying motivational level.
The hypothesis was advanced in the introduction that
the hypothesis-behavior of the animals would be a function
of level of motivation. It is apparent from Figure 2 that
varying motivational level induces major changes in both
the gross amount and type of hypothesis behavior, spatial
(S) or visual (V). The more negative the BWC, the more S
hypotheses the animal forms.

30
Pig.2.-Number and type of hypotheses formed as a function
of body weight (totals of all hypotheses within a
class interval, divided by the number of subjects,
N, of the class.)

31
The form of the data and the nature of the inquiry
lent itself in this case and in many of the tests to
p
follow to the use of the*X test of independence. The
data are usually in frequency form and also in dichotomous
categories such as S or V hypotheses. Even where dichotomous
categories do not exist, it was meaningful to create at times
such classes as above or below the median. The Tates
correction of the formula was used (McNemar, 19551 formula
86a). The null hypothesis (Ho) was rejected when*X 2.05,
1 d.f. 5*84 was surpassed.
Ho; Animals do not differ in the number of hypotheses
formed as a function of being above or below the
median in BWC. (N 31).
p
Result: *X 5.59 and the Ho is rejected. Another indi
cation of this relationship is given by which has a value
of .663 (P < .001). (The F ratio between and r 1.057
and linear regression may be assumed). It is concluded that
the number of hypotheses formed is a function of BWC; more
hypotheses are formed with severe losses of body weight and
fewer hypotheses are formed with moderate increases of
weight.
Ho; That animals do not differ in forming S or V
hypotheses predominantly as a function of being
above or below the median in BWC.

52
In this test, N > 28 and animals forming both S and V
hypotheses were included providing one type occurred with
greater frequency. Bight animals were included from this
mixed category and one was excluded.
Result: 5.89 and the Ho is rejected. It is concluded
that S hypotheses predominate with major negative BWC, and
V hypotheses predominate with slight increases or decreases
in body weight.
Ho: That animals do not differ in shifting or not
shifting hypotheses as a function of being
above or below the median in BWC.
Since 29 of the 31 animals formed hypotheses, N 29.
However, of these, only nine animals shifted hypotheses.
p
Results: Y a 0.46 and the Ho cannot be rejected. It
should be noted from Figure 2 and from the results pre
sented above that S hypotheses increase while V hypotheses
decrease as the animals lose weight. Since shifting
hypotheses involves by definition both V and S hypotheses,
some restriction of this shifting within the more moderate
levels of BWC might be expected. Inspection of the data
in the Appendix, Table 9* reveals that six of these nine
animals are above the median BWC. There are thus empirical
grounds cautioning against acceptance of this Ho.

33
B. The effect of varying food and water.
The contamination of the type of deprivation by the
level of deprivation prevented an adequate test of the
effect of food and water on hypothesis behavior. Some
indication of the effect may be gained by constructing
roughly equivalent groups. Two such groups are the moderate
FD and severe WD from the original design (note their
similarity in Tables 3 and 5). Table 6 presents a comparison
of these two groups and it will be seen that there is little
difference in the frequencies.
TABLE 6
COMPARISON OF THE MODERATE FOOD DEPRIVED
AND SEVERE WATER DEPRIVED GROUPS FOR
FREQUENCIES OF DOMINANT HYPOTHESIS TYPE
Preference
Food
Water
Spatial
5
5
Visual
2
3
Matched groups can
also be
composed by choosing
animals with nearly equivalent BWC. Unfortunately, these are
too few to warrant statistical test but the data are pre
sented in Table 7 for inspection. There is a slight
indication that food deprivation may more likely be
associated with S hypothesis formation than water deprivation.

TABLE 7
POOD AND WATER DEPRIVED ANIMALS MATCHED FOR
EQUIVALENT BODY WEIGHT CHANGE
34
BWC %
Number and Type Hypothesis
Trials
to Solve
AP
Pood
Water
Todd
Water
T
V
s
V
Pood
Water
- 5.7
+
vn

CD
3
0
0
1
38
36
- 5.5
- 4.7
3
1
0
2
21
35
- 7.9
- 7.7
1
3
2
1
33
27
-18.5
-16.8
2
0
3
0
25
23
-31.2
-31.4
6
0
5
0
33
36
It will
be shown in
section C
that
BWC was
not
significantly related to the shifting of hypotheses. Since
it therefore seems probable that motivation is not a
contaminating factor, we may test the food and water
deprivation variable on the shifting of hypotheses. The
resulting .016 and the Ho is not rejected. The effect
of this variable on the AP trials was also tested and
'X .0039 and the Ho is not rejected. It is concluded
that type of deprivation was related neither to the shifting
of hypotheses nor to the ease of solving the adaptability
problems. It was not possible to test the influence of
deprivation type on the kind of hypotheses formed.

35
C. The relation of hypothesis behavior to adaptability.
Adaptive behavior was measured by the total number
of trials to solve two problems. These problems (AP) were
arrangements of the doors of the Krech apparatus such that
one of the four hypotheses S right (SB) or left (SL), V
light (TO) or dark (TO), was a correct solution. Bach
animal was presented with one of the S problems and one of
the V problems, the order of presentation (S or V first)
was determined by the experimenter who assigned roughly
equal amounts of the orders to the various categories of
hypothesis-forming animals.
It is possible that order effects exist in the
presentation of these problems. If an animal had shown an
SB preference or hypotheses the use of problems SL and TO
in that order may have been more, or less, intrinsically
difficult than the order TO, SL.
A second source of order effect might be found in the
choice of the particular cue of each modality; that is, an
SB (or TO) problem also might be more or less -
intrinsically difficult than SL (or TO) problems. However,
this second order effect was controlled by giving all
animals problems which differed from their last demonstrated
preference (as revealed by the largest number of choices
within the SB, SL, TO, TO categories). If an animal had
shown SB and TO preferences during the hypothesis trials,

36
the problems presented would be SL and VL, although not
necessarily in that order, Thus, the first type of order
effect varied while the second type was partially controlled.
An indication of the effect of this first type of
order can be obtained from the point-biserial correlation
between order (first problem the same, first problem
opposite demonstrated preference) and the distribution of
number of trials on the AP. This correlation has the value
r ^ .013 which of course is not significant. When tested
by A (Same-Opposite x More-Less median trials on the AP),
2
"X .028 and is not significant. Therefore it will be
assumed that an order effect does not exist in the analysis
of the data to follow.
Some statistics of the animals* gross behavior are
presented in Table 8, It will be noted that more animals
formed S hypotheses and more frequently, than any other
TABLE 8
TYPE AND FREQUENCY OF HYPOTHESIS BEHAVIOR
Number of
Animals
Type of Hypothesis
Formed
Meqn Number of
Days Persisted
14
S only
4.57
6
V only
2.67
9
Both S and V
4.11
2
none


37
category. This differs from Kreoh*s finding and will he
discussed in the next chapter. However, on the basis of
his argument for adaptability there appears to be a logical
basis for inferring that animals which form V hypotheses
are more adaptive since S hypotheses are more "popular."
The California group had stated that on the first five or
six trials their animals displayed a decided preference for
the lighted alleys (Rosenzweig, et al.. 1958). Accordingly,
the data from these trials t/ere analyzed. The Spatial minus
Visual Preference score was computed for the first six
trials and found to average -2.4 per cent. This indicates
that there may be a very slight general orientation of the
animals to the light cues, light or dark. When the
individual visual preferences are examined it is found that
13 preferred the lighted alleys to some degree. Thirteen
of the animals also exhibited a preference for the darkened
alleys, and for five of the animals the preferences were
equal. When the data were examined for the most marked
preference, S or V, it was found that 16 animals preferred
S, 14 preferred V, and with one animal the choices were
tied. Thus, it is not clear which category of hypotheses
Krech would identify as the more adaptive on the basis of
his line of argument. All alternatives will be tested.
Hot Animals forming only S hypotheses and all
animals forming V hypotheses do not differ in

38
respect to being above or below the median in
AP trials.
A
Result: .281 and the Ho is not rejected. A test of
the same type on V vs. S plus S and V frequencies yields
a*X2 .0023, and it is concluded that there is no evidence
to identify either a Y or S hypothesis factor with
adaptability. The test may be applied to animals that
shift hypotheses. (In this case, all animals that shifted
hypotheses did so from one modality to another. Hone of
A
the animals shifted within modalities.) The'X^ .016 and
the Ho is not rejected. It is concluded that there is no
evidence to identify the shifting or non-shifting of
hypotheses with adaptive behavior.
It may be argued that the more hypotheses an animal
forms, the more adaptive that animal is. This proposition
was tested by dividing the animals into categories above
or below the median total number of hypotheses formed and
above or below median number of trials on the AP. 2 .152
and the hypothesis is not rejected. It is concluded that
there was no experimental basis for identifying frequency
of hypothesis formation with adaptability.
In the recent publications of the California group,
the implications of hypothesis behavior have been retained
but the measure more usually relied on for the interpretation
of the data has been the Spatial-Visual Preference Score
(PS). The authors have equated a S preference as being more

59
innately adaptive. This may be tested by a correlation
between each animal's PS and the trials on the AP. A
significant negative correlation would indicate that V
preferences are associated with few trials on the AP;
positive correlations would associate S preferences with
ease of solution. The correlation (r ) between these two
measures was -.013; this is taken as indicating no relation
existed between an animal's PS and its adaptability. When
the number of shifts in algebraic sign of the PS are cor
related with the AP trials, r -.229 which is not signifi
cant (P > .10).
To summarize the results of this section, no evidence
was found in the present study to identify either type or
amount or method of measuring hypothesis behavior with the
measure of adaptive behavior (AP).
D. The effect of level of motivation on adaptability.
As will be noted from Figure 3* there is a general
U-shaped relation between motivational level and AP trials;
an optimal level appears to exist between a -10 per cent
and -20 per cent BWO. The extent of this relation is given
by the coefficient which is .486 (P > .10) and by
I* ** -275 (P > .10). This also results in an insignificant
It is concluded that there is little statistical
evidence for an optimal level of motivation for solving the
AP, although a slight U-shaped relation is present.

Average Number of Trials
40
% change in body weight
Fig.3.-The relation between body weight change and
average number of trials to solve both
adaptability problems.

CHAPTER IV
DISCUSSION
Consideration of tlie Experimental Hypotheses
A. Motivation and hypothesis behavior. One research
hypothesis advanced earlier in this paper stated that the
frequency, types, and shifts of hypothesis behavior would
be some function of motivational level. The results have
shown: (1) that an increase in motivational level, as
measured by body weight change, was accompanied by an
increase in the total number of days a hypothesis was
exhibited; (2) that, while S and V hypotheses occurred with
approximately equal frequency at the lower levels of
motivation, V types declined but S types increased as
motivational level increased; (3) that no statistically
significant relation between motivational level and shifting
of hypotheses was present, although six of the nine animals
that did exhibit shifts were below the median in amount of
body weight lost.
The effect of motivational level on hypothesis
behavior had not been investigated prior to this study, and
in neither the prior work of Krechevsky, nor in his more
current work with the California group, has there been any
41

indication that these effects might exist* Particular
forms of hypothesis behavior (S preferences and shifts in
preference) were assumed by the California group to represent
more adaptive behavior than V preferences and lack of shift
in preference. Since shifts predominate at lower levels
while S hypotheses predominate at higher levels of moti
vation, the results are apparently inconsistent with these
assumptions. Rather, the present results indicate that
such shifts in hypothesis behavior should more parsimoniously
be related to motivational levels rather than to the relative
adaptive level of some particular form of hypothesis behavior
or the "adaptability" of the animal.
A second hypothesis stated that food and water
deprivation would lead to different kinds of hypothesis
behavior. This hypothesis proved to be untestable since
the attempt to establish the required operationally
equivalent motivational levels (measured by running speed
and weight change) failed. This failure may be attributed
to three causes. First, too small a sample of animals was
used in the pilot study. Second, although it was not
apparent from the pilot study, it now seems that the use of
a percentage of ad libitum food or water intake does not
produce sufficiently homogeneous running times or weight
changes. Third, the relatively long period (about a month)
of the deprivation schedule may have led to different
effects within the food and water conditions.

43
The method which was utilised (reduced percentage
of ad libitum intake) was an attempt to overcome certain
difficulties of the traditional limited time schedules
(water deprived rats may adjust their water intake to the
amount of time available) and limited amount schedules
(intakes and nutritional requirements vary partly as a
function of age, body size, and metabolism). It now appears
that sufficiently precise control would be obtained by
reducing the body weight of each animal to some predetermined
level before the start of the experimental program; this
body weight level could be maintained by making daily
adjustments in each animal*s daily food or water ration.
B. Hypothesis behavior and adaptability. A third
experimental hypothesis asserted a positive relationship
between type and frequency of hypothesis behavior and
adaptability (as measured by the trials to solve the
adaptability problems, A?). In the present study no
relation was found between the AP measure and total number
of hypotheses formed, shifts in hypotheses, or types of
hypotheses. This is a result which is in apparent oppo
sition to the assumptions of Krech and his associates since
a particular type of hypothesis behavior (spatial) and shifts
in hypothesis behavior had been identified as "adaptive."
There has not been any previous empirical evidence comparing
hypothesis behavior with some other measure of adaptability

44
These results may be interpreted to mean that
hypothesis behavior is not a valid measure of adaptability.
However, it is possible that the measure of adaptability,
AP, used in the present study, is invalid. This is a
crucial question, but since there is little precedent and
no consensually agreed upon definition of adaptive behavior
in the psychological literature, this point must at present
be settled on logical grounds. Hypothesis behavior has
been the major, but seldom used, convention for assessing
this variable. While other measures such as the Hebb-
Williams apparatus might have been used, the AP measure
was chosen because it uses the same apparatus, procedure,
criteria, and perceptual-motor capacitiesj equally
important, its logic is congruent with Krech's reasoning
that hypotheses are adaptive. It differs essentially in
permitting a solution to take place. There exists such a
high degree of similarity between hypothesis behavior and
the AP measure, that the lack of any apparent relation
between the two raises serious doubts as to the "adaptive
nature of hypothesis behavior. It also raises the question
as to what hypothesis behavior is.
C. Motivation and trials to solve the two adaptability
problems. A fourth hypothesis was advanced that adaptability
would be related to motivational level. The results were

4-5
inconclusive; a U-shaped relation indicating an optimal
motivational level (between -10 and -20 per cent BWC) was
found but statistically significant correlations were not
obtained. The concept of an optimal motivational level
dates back to an original formulation by Yerkes and Dodson
(1908). It has been found to be generally true in such
studies as the Columbia obstruction box experiments (Warden,
1951), and more recently arousal level theorists have
postulated a similar relation (Freeman, 194-0; Duffy, 1957).
* *
D. Conclusions. The relation of various hypothesis
behavior measures to the solution of two problems has been
examined and no support was found for the assumption that a
particular type of hypothesis behavior can be considered
intrinsically more adaptive than some other type. It has
also been shown that the hypothesis behavior exhibited by
an animal is a function of the animal's motivational level.
It was therefore considered more parsimonious to relate
these hypothesis behavior characteristics to motivational
level rather than to assumed levels of adaptability. It
was not possible to test the effects of food and water
deprivation on hypothesis behavior since the attempt to
establish operationally equivalent motivational levels in
these categories was not successful. Finally, some evidence
of an optimal level of motivation for performance was
obtained but the evidence was not statistically significant.

CHAPTER V
SUMMARY
This study investigated relationships "between:
Krech's various measures of hypothesis behavior in the rat
and adaptability, motivational level and hypothesis
behavior, and motivational level and adaptability.
Thirty-one hooded rats were placed on limited food-
or water-diets. After receiving trials to determine
hypothesis behavior in the Krech insoluble maze, the
animals were tested for adaptability by number of trials
to solve two problems in the same apparatus.
The results indicated that:
1. There was no apparent relation between type,
duration, or shifts in Krechs hypothesis behavior and
the measure of adaptability.
2. The type of hypothesis behavior and its duration
was a function of motivational level; hypothesis behavior
which utilized light cues gradually diminished with
increased body weight loss while hypothesis behavior based
on spatial cues rapidly increased.
3. There was slight evidence of a curvilinear
relation of adaptability to motivational level but it was
not statistically significant.
46

47
4. It was not possible to determine differences in
hypothesis behavior as a function of type of deprivation
because of an inability to establish equivalent motivational
levels*
The results were interpreted as evidence against
Krech's assumptions that hypothesis behavior is adaptive.
A motivational interpretation was considered to be the most
parsimonious organization of the data.

BIBLIOGRAPHY
Boring, E. G, A history of experimental psychology.
(2nd od.) Hew York* Applepon-Century-Crofts, 1950.
Brown, J. S. Motivation. New York: McGraw-Hill, 1961.
Duffy. E. The concept of energy mobilization. Psychol,
Rev., 1951, 8, 50-40.
The psychological significance of the concept
of harousal" or "activation." Psychol. Rev.. 1957.
64, 265-275.
Freeman, G. L. The relationship between performance
level and bodily activity level, J. exp. Psychol..
1940, 26, 602-608.
Hamilton, G. V. A study of trial and error reactions in
mammals. J. Anim. Behav.. 1911, 2 53-66.
Hilgard, E. R. Theories of learning. New York: Appleton-
Century-Crofts,
Krech, D. Cognition and motivation in psychological
theory. In W. Dennis (Ed.), Current trends in
psychological theory. Pittsburgh: Univ. Pittsburgh
Press, 1951. £p. lll-l39.
Rosenzweig, M. R., Bennett, E. L., & Krueckel, B.
Enzyme concentrations in the brain and adjustive
behavior patterns. Science. 1954, 120. 994-996.
Krechevsky, I. "Hypotheses" versus "chance" in the pre
solution period in sensory discrimination learning.
Univ. Calif. Publ. Psychol.. 1952, 6, 27-44.
The genesis of "hypotheses" in rats. Univ.
Calif. Publ. Psychol.. 1932, 6, 45-64.
. The docile nature of "hypotheses." J. comp.
Psychol.. 1933, 1, 429-443.
. Hereditary nature of "hypotheses." J. conro.
Piychol.. 1933, 16, 99-116.

49
Krechevsky, I. 3rain mechanisms and ^hypotheses.M J.
comp. Psychol., 1935 12 425-468.
. Brain mechanisms and variability: I. Variability
wITEin a means-end-readiness. J. comp. Psychol.. 1937.
2t 121-138.
. Brain mechanisms and variability: II. Variability
where no learning is involved. J. comp. Psychol., 1397*
22, 139-163.
Brain mechanisms and variability: III. Limita-
tions of the effect of cortical injury upon variability,
j. comp. Psychol.. 1937, 22 351-364.
Lashley, Karl S. Brain mechanisms and intelligence.
Chicago: Univ. Chicago Press, 19-9.
Malmo. R. B. Anxiety and behavior arousal. Psychol. Rev..
1957, 64, 276-287.
McNemar, Q. Psychological statistics. (2nd ed.) New York:
Wiley, 1955.
nmmiolion, w th9 ra
Osgood, C. E. Method and theory in experimental psychology.
New York: Oxford University Press, 1953.
Petrinovitch, L., & Bolles, R. Deprivation states and
behavioral attributes. J. comp, physiol. Psychol,.
1954 42, 450-453.
Rosensweig, M. R., Krech, D.,J& Bennett, E. L. Brain
chemistry and adaptive behavior. In H. P. Harlow &
C. N. Woolsey. Biological and biochemical bases of
behavior. Madison:University of Wisconsin Press,
567-400.
Ryan, T. A. Multiple comparisons in psychological research.
Psychol. Bull.. 1959, 6, 26-47.
Spence, K. W. The nature of discrimination learning in
animals. Psychol. Rev.. 1936, 42, 427-449.
Continuous versus non-continuous interpretations
of discrimination learning. Psychol. Rev.. 1940, 47,
271288.

50
Spence, K. W. An experimental test of the continuity and
non-continuity theories of discrimination learning.
J. exE- Psychol.. 1945* 255-266.
Stellar, E., & Hill, J. H. (She rat's rate of drinking as
a function of water deprivation. J. comp, physiol.
Psychol.. 1952, 4, 96-102.
Stevens, S. S. (Ed.), Handbook of experimental psychology.
Hew York Wiley, l9i?l. irTnrT'r,nnnTrl,'IM Tlnrn Tr'mrrr.mmT
Tolman, E. C. There is more than one kind of learning.
Psychol. Rev.. 1949, 6, 144-155.
Warden, 0. J. Animal motivation studies. The albino rat.
New York: Columbia tfniv. Press, l93l*
Witkin, H. A. 'Hypotheses' in rats: An experimental
critique. III. Summary evaluation of the hypotheses
concept. Psychol. Rev., 1942, 4£l, 541-568.
Yerkes, R. M., & Dodson, J. D. The relation of strength
of stimulus to rapidity of habit formation. J. comp.
Neurol. Psychol.. 1908, 18, 459-482.

APPENDIX

52
TABLE 9
PER CENT BODY WEIGHT CHANGE, TYPE AND OCCURRENCE
OF HYPOTHESES, AND TRIALS TO SOLVE THE
ADAPTABILITY PROBLEMS
Per Cent
Body-
Weight
Change
Deprivation
Condition
Food (F) and
Water (W)
Hypothesis Formed
on Day Number
Order of Presentation
and Trials to Solve
Adaptability Problems
V
S
V
+ 9.7
70 W
VD 3; SR 7
19
23
+ 5.8
70 W
VD 8
21
15
+ 5.7
65 F
SL 1, 2, 3
18
20
+ 3.1
70 W
VD 5, 6, 7. 8. 9
12
12
+ 2.4
70 W
none
13
12
+ 2.2
70 W
SL 6, 7. 8, 9
30
18
+ 1.7
70 W
SR 5, 6} VD 7, 8, 9
18
21
+ 0.9
50 W
none
12
15
- 1.6
50 W
SR 2, 7
13
13
-2.4
50 W
SL 5,6
13
14
-2.7
70 W
VD 4, 5; SR 9
25
11
-4.7
50 W
VD 5, 6
22
13
- 5.5
65 F
SR 5, 8, 9; VD 7
11
10
- 6. 8
50 W
VD 7, 8, 9
11
13
-7.5
50 W
VD 4, 5, 8, 9
24
15
-7.7
70 W
VD 2; SR 6, 9
14
13
- 7.9
65 F
VD 3, 7, 9; SL 4
20
13
-11.7
65 F
SR 3, 5, 6, 7; VD 9
12
15

53
Table 9 Cont.
Per Cent
Body
Weight
Change
Deprivaition
Condition
Food (F) and
Water (W)
Hypothesis Formed
on Day Number
Order of Presentation
and Trials to Solve
Adaptability Problems
V
S
V
16.8
50 W
SL 3, 4, 5
11
12
-18.5
65 F
SR 6,8
13
12
-20.5
65 F
YD 8
12
16
-21.7
65 F
SL 4, 5, 6, 7. 8, 9
12
13
-23. 1
45 F
SR 5, 6, 7, 8, 9
18
18
-23.5
65 F
VD 3; SR 1, 7. 8, 9
16
17
-25.3
45 F
SL 2, 3, 4, 6. 7, 8
10
14
-26. 6
45 F
SL 4, 5, 6, 7, 8, 9
22
11
-28.9
45 F
SR 3,4, 5, 6; VL 7
18
13
-31.2
45 F
SL 4, 5, 6, 7, 8, 9
16
17
-31.4
50 W
SR 3, 5, 7, 8, 9
18
18
-33.6
45 F
SL 4, 5, 6, 7, 8, 9
12
13
-33.7
45 F
SL 2, 3, 4, 5, 6, 7, 8, 9
12
12

BIOGRAPHICAL SKETCH
Robert Louis Procter was born March 26, 1928 in
Washington, D. C. The son of a Naval Officer, he lived
in Virginia, New York, California, and Massachusetts until
his father's retirement in Vermont in 1937. He graduated
from Springfield (Vt.) High School in 194-5 and was accepted
into the Army Specialized Reserve Training Program and
assigned to Massachusetts State College. He was later
transferred to Korea and then discharged from the Army in
194-7 He was admitted to the University of Missouri in
194-8 and received his B.A. degree in 1951 and his M.A.
degree in psychology in 1953* He returned to the Army in
that year and was discharged in 1955. He entered the
University of Florida in 1955 and held a series of graduate
assistantships, primarily in clinical and comparative
psychology. He was also employed at Sunland Training
Center, Gainesville, Florida, and Rome (New York) State
School, where he was Senior Clinical Psychologist. His
internship was at the Veterans Administration Hospitals in
Gulfport and Biloxi, Mississippi. He is a member of
Psi Chi, psychology honorary, and the Florida Psychological
Association.
54

This dissertaion was prepared under the direction
of the chairman of the candidate's supervisory committee
and has been approved by all members of that committee.
It was submitted to the Dean of the College of Arts and
Sciences and to the Graduate Council, and was approved as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
June 11, 1962
Dean, Graduate School
Supervisory Committee*



22
runway which was used in the pretraining period was similar
to the hypothesis maze but with the center sections removed
so that no alleys were formed. The guillotine doors were
retained and curtains were hung at the three center doors
to provide habituation for the animals.
Procedure. Animals were treated in groups of about
eight, with random assignment to each condition. The
animals assigned to the food-deprivation category (PD) were
placed on ad libitum powdered Purina rat chow and water
schedules; animals in the water-deprivation category (WD)
received the same treatment except they were given pellets
rather than powdered food. After a four-day adjustment
period, measurements of each animal1s food or water (as
appropriate) consumption was taken for 14- days. At the end
of this time the average daily consumption was computed and
the animals were weighed.
A pilot study had indicated that 65 per cent of the
ad libitum food intake and 70 per cent of the ad libitum
water intake vjould result in approximately equivalent
running times in a straight runway; similar results were
found for a 45 per cent food and a 50 per cent water level.
Therefore those four levels were adopted. The animals
were placed on the appropriate schedule one day prior to
the start of the pretraining trials and remained on these
diets until the conclusion of the experiment.


23
Pretraining. The animals were given 5 trials the
first day, 5 trials the second day, and then 10 trials each
day for nine successive days; the animals were run every
24- hours. The incentive was approximately 0.1 gram of
powdered chow or 0.1 c.c. of water, as appropriate. The
incentive was taken from each animal's daily ration.
Hypothesis trials. The animals were given 6 trials
on both the first and second days, and 12 trials each of
the successive eight days. The Krechevsky order (see Table
1) was followed. After the final day's trials within the
hypothesis maze were completed, the hypotheses for each
animal were computed. Weights were obtained for each animal.
Adaptability trials. On the basis of each animal's
hypotheses and membership in the four experimental groups,
a decision was made as to which of two visual problems and
two spatial problems were to be solved. The criterion for
solution was the same as for hypotheses; however, whereas
hypotheses were computed on the basis of an animal's daily
performance, these trials were counted serially without
respect to the termination of each days trials; such a
procedure is necessary to prevent grossly unequivalent
amounts of practice. An animal was considered to have
reached criterion when it first attained 36 out of any 46
consecutive choices which were correct; animals that reached
criterion on the first, second, or third choice-points of a


12
tlie development of hypotheses (or the complete nonoccur-
rence of them), shifts between hypotheses, and the
persistence of hypotheses, are all theoretically explainable
post hoc providing certain assumptions regarding initial
response strengths and reinforcements can be made.
Unfortunately, the Krechevsky hypothesis maze does not lend
itself to an empirical test of the Spence position.
There was no further work on hypothesis behavior
until 1954 when Krech1, Eosenzweig, and Bennett began
publishing the results of a series of experiments which
correlated brain chemistry with adaptability. Despite the
earlier criticisms of hypothesis behavior, this technique
was used as the measure of adaptability. Two changes were
made in this measure. First, a reduced level of signifi
cance was used as a criterion of hypotheses (33 rather than
35 choices of the 48 occurring each day), and second,
Krechevsky*s (1933) spatial-visual Preference Score was
also used. This Preference Score provided a continuous
scale of relative preference for either visual or spatial
modalities \d.thin each animal. It is computed in this
fashion1 1) the number of choices in each of the eight
hypothesis alternatives is determined for an animal for a
T
Krechevsky changed his name to Krech.


4
Fig. 1.-Floor Plan of the Krechevsky Hypothesis Maze.
This diagram shows the maze as it appears on trial 7
An animal following a visual-dark hypothesis would
choose alleys in the order left, left, right, right.
correction method was used at each choice-point and the
reward was available in the goal box at the completion of
every trial regardless of the correctness of the animal's
four choices. Krechevsky used a 23-hour food deprivation
schedule; a cube of bread soaked in milk was used as the
incentive. At each choice-point there were two alleys, a
right and a left, between which the animal could choose.
Over each alley was an electric light, but only one side
was illuminated. The animal might choose on the basis of
spatial cues (right, left) or visual cues (light, dark).
A curtain at the end of each alley prevented the animal
from seeing a swinging door which could be locked to either
side, thus preventing the animal from using one or the
other alleys. One alley of each of the four units was
always locked. When the animal chose an alley which had
an open door, it went through to the next unit. When the


TABLE 7
POOD AND WATER DEPRIVED ANIMALS MATCHED FOR
EQUIVALENT BODY WEIGHT CHANGE
34
BWC %
Number and Type Hypothesis
Trials
to Solve
AP
Pood
Water
Todd
Water
T
V
s
V
Pood
Water
- 5.7
+
vn

CD
3
0
0
1
38
36
- 5.5
- 4.7
3
1
0
2
21
35
- 7.9
- 7.7
1
3
2
1
33
27
-18.5
-16.8
2
0
3
0
25
23
-31.2
-31.4
6
0
5
0
33
36
It will
be shown in
section C
that
BWC was
not
significantly related to the shifting of hypotheses. Since
it therefore seems probable that motivation is not a
contaminating factor, we may test the food and water
deprivation variable on the shifting of hypotheses. The
resulting .016 and the Ho is not rejected. The effect
of this variable on the AP trials was also tested and
'X .0039 and the Ho is not rejected. It is concluded
that type of deprivation was related neither to the shifting
of hypotheses nor to the ease of solving the adaptability
problems. It was not possible to test the influence of
deprivation type on the kind of hypotheses formed.


2
level by a certain amount, hypotheses are said to be formed.
Hypothesis behavior and hypotheses are terms which have
meaning only when the apparatus, the cues, the pattern of
choices, and the statistical level are understood.
Hamilton (1911) noticed in the course of a trial and
error learning experiment that rats and other animals make
apparently systematic attempts at solving problems that
cannot be solved; they may go consistently to the right, or
left, or to the last place found successful, and so on.
Lashley (1929) also noted this feature of behavior in
discrimination problems. He indicated that while there was
no experimental technique available at that time to
investigate this behavior, these 'attempted solutions"
probably represented a significant aspect of the learning
process. Lashley's observations served as at least a
partial basis for Krechevsky's investigations; Krechevsky's
experiments were designed to provide the lacking experi
mental technique. In one of his earliest published works
he states, "The data from the present experiment have been
examined in the light of Lashley's suggestion and an attempt
has been made to devise a method for the objective
determination of the validity of that suggestion."
Krechevsky concluded from this study that "...it is shown
quite definitely in the presolution period that...the
animal is engaged in bringing to perfection various


Average Number of Trials
40
% change in body weight
Fig.3.-The relation between body weight change and
average number of trials to solve both
adaptability problems.


30
Pig.2.-Number and type of hypotheses formed as a function
of body weight (totals of all hypotheses within a
class interval, divided by the number of subjects,
N, of the class.)


35
C. The relation of hypothesis behavior to adaptability.
Adaptive behavior was measured by the total number
of trials to solve two problems. These problems (AP) were
arrangements of the doors of the Krech apparatus such that
one of the four hypotheses S right (SB) or left (SL), V
light (TO) or dark (TO), was a correct solution. Bach
animal was presented with one of the S problems and one of
the V problems, the order of presentation (S or V first)
was determined by the experimenter who assigned roughly
equal amounts of the orders to the various categories of
hypothesis-forming animals.
It is possible that order effects exist in the
presentation of these problems. If an animal had shown an
SB preference or hypotheses the use of problems SL and TO
in that order may have been more, or less, intrinsically
difficult than the order TO, SL.
A second source of order effect might be found in the
choice of the particular cue of each modality; that is, an
SB (or TO) problem also might be more or less -
intrinsically difficult than SL (or TO) problems. However,
this second order effect was controlled by giving all
animals problems which differed from their last demonstrated
preference (as revealed by the largest number of choices
within the SB, SL, TO, TO categories). If an animal had
shown SB and TO preferences during the hypothesis trials,


24
trial were allowed to finish that trial and attain the
incentive.
Roughly half of the animals which had made visual
hypotheses were assigned spatial problems firstj spatial
hypothesis animals were treated similarly. When animals
had made both types of hypotheses, or none, the assignment
was made on the basis of the dominant preference. In all
cases, the assignment of any problem within either stimulus
modality was opposite to that animal s dominant preference.
The number of daily trials and the incentives
remained the same as in the latter part of the hypothesis
maze experiment. At the conclusion of the first problem,
each animal was immediately shifted to the second problem
on the next trial. The scores were each animals total
number of trials to solve both problems.


indication that these effects might exist* Particular
forms of hypothesis behavior (S preferences and shifts in
preference) were assumed by the California group to represent
more adaptive behavior than V preferences and lack of shift
in preference. Since shifts predominate at lower levels
while S hypotheses predominate at higher levels of moti
vation, the results are apparently inconsistent with these
assumptions. Rather, the present results indicate that
such shifts in hypothesis behavior should more parsimoniously
be related to motivational levels rather than to the relative
adaptive level of some particular form of hypothesis behavior
or the "adaptability" of the animal.
A second hypothesis stated that food and water
deprivation would lead to different kinds of hypothesis
behavior. This hypothesis proved to be untestable since
the attempt to establish the required operationally
equivalent motivational levels (measured by running speed
and weight change) failed. This failure may be attributed
to three causes. First, too small a sample of animals was
used in the pilot study. Second, although it was not
apparent from the pilot study, it now seems that the use of
a percentage of ad libitum food or water intake does not
produce sufficiently homogeneous running times or weight
changes. Third, the relatively long period (about a month)
of the deprivation schedule may have led to different
effects within the food and water conditions.


33
B. The effect of varying food and water.
The contamination of the type of deprivation by the
level of deprivation prevented an adequate test of the
effect of food and water on hypothesis behavior. Some
indication of the effect may be gained by constructing
roughly equivalent groups. Two such groups are the moderate
FD and severe WD from the original design (note their
similarity in Tables 3 and 5). Table 6 presents a comparison
of these two groups and it will be seen that there is little
difference in the frequencies.
TABLE 6
COMPARISON OF THE MODERATE FOOD DEPRIVED
AND SEVERE WATER DEPRIVED GROUPS FOR
FREQUENCIES OF DOMINANT HYPOTHESIS TYPE
Preference
Food
Water
Spatial
5
5
Visual
2
3
Matched groups can
also be
composed by choosing
animals with nearly equivalent BWC. Unfortunately, these are
too few to warrant statistical test but the data are pre
sented in Table 7 for inspection. There is a slight
indication that food deprivation may more likely be
associated with S hypothesis formation than water deprivation.


CHAPTER IV
DISCUSSION
Consideration of tlie Experimental Hypotheses
A. Motivation and hypothesis behavior. One research
hypothesis advanced earlier in this paper stated that the
frequency, types, and shifts of hypothesis behavior would
be some function of motivational level. The results have
shown: (1) that an increase in motivational level, as
measured by body weight change, was accompanied by an
increase in the total number of days a hypothesis was
exhibited; (2) that, while S and V hypotheses occurred with
approximately equal frequency at the lower levels of
motivation, V types declined but S types increased as
motivational level increased; (3) that no statistically
significant relation between motivational level and shifting
of hypotheses was present, although six of the nine animals
that did exhibit shifts were below the median in amount of
body weight lost.
The effect of motivational level on hypothesis
behavior had not been investigated prior to this study, and
in neither the prior work of Krechevsky, nor in his more
current work with the California group, has there been any
41