ENVIRONMENTAL PREFERENCE: A MULTIDIMENSIONAL ANALYSIS OF THE RAT’S RESPONSE TO THE COMPLEXM 0F ETS SURROEENDIW Thesis for the Degree of M. A. PMCHlGAN STATE UNIVERSEW MICHAEL BENNY 1974 N W“ \\ \\ WU\\\\\\\\\\\\\\\\\\\\\\\\L\\\\1\\\\l w. _ _.-3 1293 10382 910 7"? .. e , L .I {:5 .P I f a]: .. . hi. -’-- .. -- F A l I.” PIE-\jlfl‘ji ;. y t ’3: k: U 111‘] C 1 SH" I i; I ”$.31“ I"?! fl ABSTRACT ENVIRONMENTAL PREFERENCE: A MULTIDIMENSIONAL ANALYSIS OF THE RAT'S RESPONSE TO THE COMPLEXITY OF ITS SURROUNDINGS BY Michael Denny The complexity of an animal's environment can be defined on numerous variables. The significance of the level or the presence or absence of these variables varies among species. Further, the Optimal level of environmental complexity which an animal may seek can change as the organismeenvironment interaction changes. Environmental preference has usually been determined by examining a few specific responses to an operationally defined complexity dimension. The common dependent measure has been approach behavior in thaform of a choice response measured under the assumption of a fixed organism-environment interaction. The present study used 20 albino rats to examine the multiplicity of responses made during a 48 hour period of confinement to a large (32 square feet) four compartment cage. The four compartments represented four levels of three-dimensional complexity. Each compartment, containing its own food, water and resting box, had a monitoring capability which allowed continuous recording of movement in and out of Michael Denny the compartment as well as general locomotor activity. An artificially controlled 12 hour light and dark cycle was maintained throughout the experiment and was treated as a two level factor in an ANOVA along with the four level complexity factor. The ANOVA was applied independently to a number scores in order to assess the differential responses to complexity and illumination. A total of 15 different scores derived from general activity including locomotion, feeding, resting and defecation were analyzed under ANOVA and appropriate nonparametric tests. In addition, a cor- relational review of the scores was made. Results indicated that the highest complexity level was preferred over the others on a number of scores reflecting appetitive behavior and one score reflecting consummatory behavior - resting. In general, preference in terms of. these criterion scores was a positive monotonic function of complexity. It was also found that the complexity effect was considerably reduced during the night periods. In fact, the overall behavior patterns were markedly different between day and night with most activity and feeding occurring in the night period. It was suggested that high complexity is preferred because of its association with shelter and relaxation. Movement between compartments was discussed in terms of stimulus discrepancy from an adaptation based standard which was found to account best for short-term patterns. Arousal explanations of complexity preference (e.g. Fiske and Maddi, Michael Denny 1961) were not confirmed. The study also included a discrete trial phase incor—K porating a more traditional paradigm for preference testing. The test tended to reinforce the first phase findings. Based on a correlational comparison of the two phases the complex- ity effect was interpreted as a shelter seeking response reflecting an escape tendency induced by the mildly aversive character of the procedure. ENVIRONMENTAL PREFERENCE: A MULTIDIMENSIONAL ANALYSIS OF THE RAT'S RESPONSE TO THE COMPLEXITY OF ITS SURROUNDINGS BY (3 ‘ Michael“Denny A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1974 \5‘ ACKNOWLEDGEMENTS I would like to express my appreciation to Ralph Levine for his guidance, and Florence Denny for her assistance in preparing this thesis. ii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION- The Significance of Environmental Complexity Stimulation Seeking State Variables Behavior Classes Experimental Design and Hypotheses EXPERIMENTAL DESIGN Subjects Apparatus Enclosures Complexity Inserts Locomotor Activity Sensors Recording Devices Experimental Environment Procedure Preliminary Ad.LEb_Phase EIscrete Trial Testing RESULTS Ad Lib Phase The Multivariate Design Basic Scores Ratio Scores General Behavior - Effects of Light Period Effects of Complexity Daytime Nighttime Light Period X Complexity Interaction X2 Test of "Exit Patterns" iii vii-viii Table of contents (continued) Discrete Trial Phase Design and Scores Effects of Complexity Correlation with Ad Lib Phase Summary DISCUSSION Approach Responses Staying Response Feeding and General Activity Dimensionality of Complexity Stimulus Discrepancy Hypotheses State Considerations Conclusion APPENDIX Correlation Among the Dependent Measures Within Treatment Conditions Basic Scores Ratio Scores Correlations Independent of Treatment LIST OF REFERENCES iv Page 55 55 57 59 64 67 68 69 72 74 75 81 83 85 85 85 90 91 93 10. ll. 12. 13. 14. LIST OF TABLES Page Procedural timetable of 15 day experimental session. 23 Overall 12 hour means of basic scores. 29 ANOVA on the effects of light period. 31 ANOVA on the effects of complexity under day conditions: basic scores. 34 Planned comparisons on the effects of complexity under day conditions: E tests on basic scores. 35-36 Planned comparisons on the effects of complexity: Wilcoxon rank difference test on occupancy time. 38 ANOVA on the effects of complexity under night conditions: basic scores. 43 Planned comparisons on the effects of complexity under night conditions: E tests on basic scores. 44-45 ANOVA on the effects of complexity under night conditions: ratio scores. 43 Planned comparisons on the effects of complexity under night conditions: E tests on ratio scores. 49-50 ANOVA on the interactionrbetween light period (LP) and complexity (C). 54 Chi-squared test of "exit pattern" distributions. The observed number of movements from one compartment to another compartment appears as the upper entry in each cell. The lower value is the expected number as explained in the text. 56 Planned comparisons on the results of the discrete trial phase: Wilcoxon rank difference test on choice score. ‘ 58 ANOVA on the effect of complexity: discrete trial scores. 60 List of Tables (continued) Page 15. Planned comparisons on the effects of complexity; E tests on discrete trial scores. ‘61 16. Correlations among discrete trial scores and ' the following basic scores: occupancy time (OT), locomotor activity (LA), feeding time (FT), resting time (RT), entry frequenCy (EF), food consumption (FC), water consumption (WC), and defecation (D). 7 62 17. Intercorrelations among discrete trial scores including latency (L) and locomotor activity (LA), and correlations with the following ratio scores: mean occupancy time (MO), mean active occupancy time (MAO), locomotor activity rate (AR), resting~rate (RT), feeding speed (FS), and defecation rate (DR). 63 A-l. Similarity scores of the correlational patterns of responding (basic scores) within each condition. The value of the table entries represents the degree of similarity found in the response patterns of each contrast between conditions. 88 A-2. Intercorrelations among basic scores under day (left-hand matrix) and night (right-hand matrix) conditions. The scores include occupancy time (OT), locomotor activity (LA), feeding time (FT), resting time (RT), entry frequency (EF), food consumption (FC), water consumption (WC), and defecation (D). 92 vi LIST OF FIGURES Page Four views of the experimental cage: A)top view showing center section and four compartments, B) end view of high and open complexity compartments, C) view of a center section door, and D) side view of one compartment showing resting box, foot plates and litter tray. 17 The distribution of activity in terms of entry frequency under night conditions. 32 Comparison of the effects of complexity under day conditions on four basic scores; resting time (R), occupancy time (0), locomotor activity (A), and entry frequency (E). 39 Comparison of the effect of complexity under day conditions on four scores; food consumption (F), water consumption (W), feeding time (T), and defecation (D). 40 Comparison of the effects of complexity under night conditions on four basic scores; resting time (R), occupancy time (0), locomotor activity (A), and entry frequency (E). 41 Comparison of the effects of complexity under night conditions on four basic scores; food consumption (F), water consumption (W), feeding time (T), and defecation (D). 46 Comparison of the effects of complexity under night conditions on three ratio scores; resting rate (R), locomotor activity rate (A), and defecation rate (D). 47 Comparison of the effects of complexity under night conditions on two ratio scores; water consumption rate (W), and food consumption rate (F). 52 Comparison of the effects of complexity on three discrete trial scores; choice (C), latency (L), and locomotor activity (A). 53 vii List of Figures (Continued) Page A-l. Response patterns for each complexity level under day and night conditions. The intercorrelations of the basic scores have been reduced to positive or negative relations. A "0" indicates the lack of a significant relationship. 87 viii INTRODUCTION The experimental study of environmental preferences has historically been limited to two rather distinct areas of investigation. On the ethological side, the habitat pref- erences of a variety of organisms have been tested along specified environmental dimensions (e.g. Harris, 1952; Sexton EE. il’! 1964; Kolpfer, 1963; Sale, 1969; Reese, 1963; Kolpfer and Hailman, 1965). The dimensions of interest were of a spe- cies-specific nature, simulating natural conditions (e.g. foliage or cover type, water depth, shell configuration, etc.) For example, in the field, Wecker (1963) experimentally tested deer mice of different environmental-genetic backgrounds for selection of natural woods or grassland habitats using a number of behavioral measures. He found that both heredity and early experience contributed to the definition of a habitat preference. The other line of investigation has dealt with preference for a specific, biologically weak, aspect of the environment. CNnis means any factor of an environment, which is not essen- ‘tial to the maintenance of the organism as opposed to such faxztors as food, shelter and temperature. Stimulus complexity, ha‘ving been treated as an elicitor of exploratory or approach befnavior (Dember 2E EE., 1957; Berlyne, 1950; Welker, 1957) and as an Operant reinforcer (Butler, 1953; Barnes and Baron, 1961), has been experimentally the most productive of these environmental variables. Although some of these early studies failed to isolate complexity from novelty they did show that the preference for complexity increased over exposure time suggesting an interaction between novelty and complexity. A major reason responsible for the confounding of com- plexity with novelty was the failure to define adequately the concept of stimulus complexity. This situation has since been rectified primarily by the distinction made between two operationally defined forms of complexity. Berlyne and Slater (1957) produced complexity by either changing the stimulus on each trial ('Y' maze problem) or providing a well differentiated complex stimulus on all trials. They con- sidered these procedures to yield 'successive' and 'simulta- neous' complexity, respectively, where only successive com- plexity involved novelty. Persistent exploration was found only in the simultaneous complexity condition. Another way of viewing this distinction separates complexity into dynamic and static stimulus states. Thus, if an environment is highly complex in physical structure but is unchanging it is reflecting static complexity. On the other hand, a stimulus can be considered complex because short-term or frequent changes in its quality or magnitude occur. This would be an example of dynamic complexity. The concern of the present study is with static com- plexity which can be defined by the number and variety of three-dimensional elements fixed in a given space or in other words, the density and heterogeneity of elements in the stimulus field (Berlyne, 1960). These independent parameters of complexity were linked together such that progressively. higher operational levels of complexity always represented increases in both the density and heterogeneity of elements in the environment. In all other ways environmental stimuli were constant across levels and over time (excluding illumi- nation cycles). This means that, essentially, after an initial exposure to all levels, the animal becomes familiar with the environment and no stimuli can be considered novel except in the restrictive sense of short-term novelty (Berlyne, 1960) which is present any time a change occurs in the animal's perception of the proximal stimulus complex. The Significance of Environmental Complexity. The relevance of static environmental complexity to the individual is primarily a subject of conjecture, however, a number of possibilities exist. Ecological studies have only considered environmental complexity in terms of species diversity (e.g. Kohn, 1967; Rosenzweig and Winakur, 1969) but the higher abundance as well as diversity of fauna found to be associated with more complex environments suggests an environmental contribution to the success or viability of the individual. The contribution may stem from the organism's use of complexity as a stimulus dimension to pattern its physical world. For example, its cue value (association with or information about the environment) may be instrumental in locating appropriate shelter, food or conspecifics. At this juncture, any biological effect of environmental complexity is a mediated one not necessarily crucial to the maintenance of the organism. In this vein, Hilden (1965) suggests that: The process of habitat selection is not likely to be a response to ultimate factors, but to series of proximate factors. These may, in themselves, lack any direct biological meaning for the organism, but will collectively define habitats likely to possess the necessary ultimate factors. Conversely, environmental complexity may contribute directly to the develOpment and maintenance of the biology, including nervous functioning of the animal. Investigations have shown that environmental complexity can affect brain development (Rosenzweig, EE il‘r 1962; Riege, 1971), learning behavior (Bingham and Griffiths, 1952; Hynovitch, 1952) and emotional reactivity (Denenberg, 1964, 1967). The adaptive value of approaching complex stimulus situations, considering these results, is inherent in the consequences of experiencing the complexity. Hebb (1949) has theorized that Optimal perceptual and intellectual functioning requires a background of some sufficient level of sensory stimulation. The idea that organisms tend to actively seek this level of stimulation has been forwarded rather independently by a number of re- searchers including Hebb (1955), Leuba (1955), Dember and Earl (1957), McClelland, Atkinson, Clark and Lowell (1953), Schneirla (1959) and Butler and Harlow (1954). Stimulation Seeking Excluding Schneirla and Harlow, most discussions of stimulation seeking have included the hypothesis that approach or orientation behavior is afN-shaped function of the dis- crepancy between a standard and a new level of stimulus complexity or intensity. While these interpretations rest at the stimulus level some authors have extended their theoretics to focus on the concomittant effects of the stimulus. More precisely, they posit that higher animals tend to select stimulus levels which result in moderate changes in their "arousal" level (Berlyne, 1960; Fiske and Maddi, 1961). Thus we find both stimulus-based and response- based forms of the discrepancy hypothesis. The response-based form implies that the attractiveness of a discrepant stimulus rests on the character of the resultant responses induced by the stimulus. The stimulus—based form implies that the attractiveness is simply the approach eliciting strength of the stimulus itself. If the level of induced arousal is positively and monotonically related to the degree of change in a stimulus, then both forms yield identical predictions of the stimulus' attractiveness. Attractiveness is used here as if it were Operationally defined in terms of approach or choice strength. To a large extent, for either form, the prediction depends on what dynamics are imparted to the internal stand— ard used to compare a new stimulus level. That is, whether the standard is intrinsically fixed or a product of adapta- tion (i.e. transient). If, when dealing with static complexity, the standard is fixed, exploratory behavior in the form of stimulus change seeking would be expected to decrease and remain at low levels when the optimum discrepancy from the standard is attained. Conversely, when the standard is transient, the reduction in exploratory behavior would be temporary. This is because a trend to actively seek a moderate change in the stimulus level develOps after some passage of exposure time to the new stimulus situation. At this point, the new stimulus situation has replaced the original standard as the reference point for determining stimulus discrepancy. As a potential elicitor of approach behavior, recent studies have demonstrated that, intermediate levels of stimulus complexity are often the most effective (e.g. Sales, 1968; Dutch and Brown, 1971). These results question the assumption that approach tendency is positively and mono- tonically related to stimulus complexity and lend support to the discrepancy hypothesis with an implied fixed internal standard. In similar studies, approach preference for moderate complexity is found to shift to higher complexity levels after sufficient exposure to the environment (Walker and Walker, 1964; May, 1968; Haben, 1969), supporting the notion of an adaptation based standard. The recurring problem Of novelty, however, may be responsible for the Observed shift. For example, the approach eliciting strength of high stimulus complexity may be suppressed by the commensurate high novelty which can elicit fear or avoidance responses. Following commerce with the environment the animal habituates to the novelty and in this situation tends to show an affinity toward higher complexity. Evidence supporting this interpretation has been reported by Montgomery (1955) and Welker (1957). The behavioral time course of approach behavior deserves scrutiny because as a component of preference (preference as used here implies seeking and maintaining an Optimal level of stimulus quality; after Dember, 1965) the importance of approach behavior is directly related to the persistence of the behavior. The present study circumvents this problem by allowing the animals ample time to habituate to the novelty of the environment before any behavior is sampled. State Variables The discussion so far has considered the way parameters of a stimulus situation, (i.e. environmental complexity) are processed by the animal in terms of a preference response and the possible consequences of this preference. The discussion has been intentionally general in nature but it is important that a few factors be introduced which may conditionalize preference behavior. Obviously, in contrast to the mechanisms for selection of habitats along biologically imperative dimensions, the motivational base for those behaviors which temporarily put an animal in its preferred or psychologically optimal environment, and the consequences of this placement, are Open to substantial variation. Any organism is usually characterized by having the capacity to Operate under a variety of shOrt-term internal or motivational states, (e.g. fear, hunger, and fatigue or sleepiness). Approach toward or non-withdrawal from stimulus complexity must be viewed in terms of the animals current state. Previous research has been primarily concerned with the effects of stimulus complexity on arousal (Berlyne, 1960) rather than the effects of arousal state on the response to stimulus complexity. However, there is evidence that hunger heightens the exploratory response to novelty (Fehrer, 1956; Hughes, 1965) and maze complexity (Adlerstein and Fehrer, 1955). While fear can be induced by high levels of stimulus complexity, as mentioned before, there is no current evidence that fear influences complexity preference. In fact, Montgomery and Monkman (1955) found that electric shock and loud auditory stimulation preceding a maze experience had no affect on subsequent exploratory behavior. The present experiment was designed to look at the behavior of rats within an environment over an extended time period (48 hours) during which the animal's motivational state is likely to change several times. One way of identi- fying the states is by their behavioral manifestations. Thus behaviors such as sleeping, locomotion and eating can serve as state indicators as well as criterion measures of environ- mental preference. Obviously the functional properties of these behaviors are basically different and, likewise, the extent to which each is vulnerable to the stimulus control of environmental complexity would be expected to differ. The mechanics of how the state of the animal might modify stimulus control is not within the scope of this study. However a plausible and parsimonious explanation is that it alters stimulus sensi- tivity or attentional threashold levels (Campbell and Sheffield, 1953). Behavior Classes Another consideration which may be related to the concept of state is the nature of behavior classes. A widely accepted scheme for behavior classification, especially among etholo- gists, distinguishes between appetitive and consummatory responses. In some cases a third class, post-consummatory, is also distinguished (see Denny and Ratner, 1970). A classificatory scheme is used as a device to identify and group the distinct purposive behaviors in an animal's repetoire. It‘s functional value comes primarily from the ability to impart "motives" or "intentions" to these behaviors based on how they are grouped. Appetitive behavior is viewed as those responses instrumental in achieving contact (percep- tual or physical) with a particular stimulus. This generally includes seeking, orientation, and approach responses. Con- summatory behavior is considered to be a response which regularly consummates or terminates a recurring behavior sequence, usually of a species-specific stereotyped nature (e.g. resting, contacting, eliminating, drinking and feeding). Finally post-consummatory acts are defined as coordinated 10 responses which disengage the animal from a consummatory act and serve as a transition into the appetitive components of subsequent behavior. An application of this classificatory model to explo- ratory behavior by Fowler (1967) treats orienting toward, attending to, locomoting toward, and manipulation of the stimulus Object as consummatory action. Appetitive acts are those which involve the animal in any response which alters or changes those stimuli currently impinging upon it. These stimulus seeking behaviors increase the probability of new or different (e.g. more complex) stimuli being per- ceived. It is worth noting that this application implies a Spencian concept of drive (Spence, 1956) and is consistent with exploratory drive concepts forwarded by Harlow 23.2l (1950) and Berlyne (1960). Obviously, the usefulness of this schema is relative to the preciseness with which the observer can infer the animal's state from its ongoing behavior or the existing environmental conditions. Approaching a specific stimulus complex may be a "consumption" of the perceptual qualities of the complex; but, on the other hand, the animal may be approaching the complex in search of food or as a place to defecate, etc.f If so, the behavior would be classified as appetitive not con- summatory. In a manner of speaking it is the "intention" of the animal that has to be known or determinable before a behavior can be classified. If a complete behavior sequence can be identified then 11 the "intention" can be inferred from the terminal element. In addition, knowing the state of the organism usually increases the validity of a classificatory conclusion. For example, if the animal has been deprived of food and after. approaching a stimulus complex it immediately engages in feeding behavior, the approach behavior is obviously appe~ titive. If a well fed animal approaches a specific stimulus complex and subsequently eats there, the interpretation is more difficult. The animal may have been drawn to the stimulus complex because it elicited exploratory reSponses and only incidentally did the animal find and eat the food. If the animal does not eat but continues to stay in the stimulus complex it would be possible to label the approach behavior as consummatory (after Fowler) assuming that the "intention" of the behavior is to maintain an orientation to, and the perceptual impact Of, the stimulus complex. A more convincing demonstration of the animal's intention, and thus, the reasonableness of such a classification, would be for the animal to approach and remain within the stimulus complex without eating when it was known to be hungry, that is under food deprivation. Without complete information on the sequential organiza- tion of ongoing behavior, adopting Fowler's classificatory cOncept is riggy. For this reason the present study will rely on the more restrictive traditional conception of consummatory behavior, that is feeding, drinking, resting/sleeping, and defecation. All other behaviors which are considered in this study will be classed as appetitive. 12 Experimental Design and Hypotheses The stimulus complexity studies mentioned have relied on using one or two measures over short sessions of responding to aSsess the animal's apparent preference for a more complex (or more novel) environment. Because Of the ubiquitous nature of the stimulus condition of complexity and the vaguely defined nature Of the responses (exploratory or approach activity), a multivariate analysis of various behaviors could lead to a1 clearer picture of the multiplicity of specific responses differentially elicited by the complexity dimension. Speci- fically, determining the nature of the relationship between these behaviors--the response pattern--is needed for an understanding of complexity preference. Few multidimensional investigations of free responding to complexity have been made. Multidimensional studies of the rat have suggested relation- ships of feeding, grooming and exploration to complexity (Pereboom, 1968; Hughes, 1968; Bindra and Spinner, 1958) but these have not attempted to formulate a response pattern directly by correlating the behavioral measures. Some correlational work has been done on the free responding situation of the home cage (e.g. Jennings, 1971) but no manipulations of complexity were made. In an attempt to assess the complexity preference of laboratory rats in broader terms than has been previously achieved the present study attends to a number of different behaviors. Using both a longterm free-choice or 29.112. situation and a more traditional discrete trial situation, 13 response patterns within environments of specific complexity were compared. An environment, in this study, is viewed as a place (a set of conditions) to feed, a place to locomote, a place to rest, a place to defecate, etc.: In the 3g liE_phase of the experiment a rat had free access to four large compartments (32 square feet of area) differing in the density and heterogeneity of chains hanging down to the floor from a false-ceiling insert. The large area was intended to minimize the extent to which exploratory behavior can draw the animal into compartments of less pre- ferred complexity levels. That is, each compartment was sufficiently large to provide considerable area for explora- tion and general exercise. As well, each compartment provided sufficient environmental support in terms of food and water resources and apprOpriate resting sites. Assuming that the propensity displayed by rats to approach environments of higher complexity reflects a general preference rather than simply a transient attraction, it would be expected that rats would tend to favor high complex- ity as a place to perform a wide variety Of behaviors. The strength of this preference, however, was expected to vary according to the behavior class examined and the internal state of the animal. Adopting the belief that, after the effects of any fear inducing stimuli are partialled out, the incentive value of an environment increases with its complex- ity (up to some extreme limit) it was expected that the appetitive class of behaviors would be most highly associated 14 with high complexity. This assumed that exploratory responses would be a major component of appetitive behavior. For the same reason and because it was assumed that satiation rates would decrease with increasing complexity (a notion proposed by Glanzer, 1953, and applied to stimulus-seeking by Myers and Miller, 1954), it was expected that the cOnsummatory class of behaviors would be most highly associated with high complexity as well. This assumed that once attracted to a high complexity area the animal would tend to stay in the area to perform most consummatory acts. State variables were expected to exaggerate or diminish the relative differences in response strength between the different complexity levels depending on whether the response probability of interest was raised or lowered by the specific state condition. Because identification Of state conditions suffers from the same shortcoming as behavior classification, namely a requirement for complete knowledge of the sequence of behavior, a systematic examination of state effects in the EE_EEE phase of the present experiment was limited to active versus inactive (resting) states determined by illumination level. In this context if inactivity was found to be con- trolled by complexity the effect would be most striking under day illumination. Obviously this is only relevant to con- summatory behavior as appetitive responses are characterized by locomotor activity. If, on the other hand, some active behaviors were found to be under the control of complexity A the effect would be exaggerated during night conditions. 15 This expectation applied to both appetitive and consummatory behaviors. . A discrete trial phase of the experiment was included to compare the results of the 2E EiE_phase to those of a procedure typical of previous research on stimulus complexity preferences. Here, rather than being left to roam through four complexity compartments for 48 hours, the rat was given a choice of compartments which once made terminated the trial. Overall, the strength or frequences of all measured responses including choice, feeding, locomoting and resting were hypothesized to increase as the complexity level in- creased across the four environments available to the animal. EXPERIMENTAL DESI GN I. SUBJECTS. Twenty male albino rats, 70-80 days old when received from the suppliers, were used. Sixteen of the rats were obtained from Sprague-Dawley in Madison, Wisconsin and the remaining four were of Sprague-Dawley descent obtained from a local supplier. II. APPARATUS. Four 4-compartment cages with removable complexity inserts and recording cabability were used. A. ENCLOSURES. Each cage was a 4'x8'x2' wooden frame constructed of 1%"x18" pine and covered with 8" hardware cloth forming the floor, ceiling and outside walls. Each cage was subdivided into four equal sized compartments (approximately 2'x4'x2') by 1/8" masonite panels bisecting each wall and extending to within 4 3/4" of the center. These panels were joined at the cage center by four 8"x2' strips Of %" hardware cloth which formed a small (95" square) section at the center of the enclosure. Centered at the base of each hardware cloth strip was a 2k"x3" clear plexiglass door, hinged at the top. This allowed access, through the center section, from one compartment to any of the other three compartments (see Figure 1). The floor of the center section was 24 gauge galvanized sheet metal. The hardware cloth ceilings of each compartment were 16 Figure l. 17 D. Four views of the experimental cage: A) tOp view showing center section and four compartments, B) end view of high and open complexity compartments, C) view of a center section door, and D) side view of one compartment showing resting box, foot plates and litter tray. 18 hinged and a removable cover was separately fixed over the center section. The latter was provided for introduction of the animal to the enclosure. The cages were raised 4" above the floor so that a galvanized sheet metal tray filled with crushed corn husks and topped by paper could be positioned under each compart- ment and easily removed for Observation and cleaning. B. COMPLEXITY INSERTS. Complexity was created by fixing one of three inserts in a compartment. The inserts were approximately 20"x44" and consisted of three or five longitudinal pieces connected to 3 crossmembers all of 3/4" square pine. Eighteen inch lengths of chain of three types were hung from these frames. The attached chain lengths were approximately equally spaced and intermixed along the longitudinal members of a frame in the following fashion: Open complexity: no insert Low complexity insert: 9 lengths of bead chain 9 lengths of bead chain and Medlum C°mpleX1tY lnsert‘ 13 lengths of furnace chain High complexity insert: 9 lengths of bead chain, 13 lengths of furnace chain, 13 lengths of tensor link chain Aluminum brackets mounted along the top of the cage walls allowed easy attachment and removal of the insert frame at a fixed height 18" above the floor of the cage. In this position the chains extended down to within 1/8" of the floor. A view of the cage from one end showing the insert as well as footplates, rest box and food hopper is included in Figure l. 19 C. LOCOMOTOR ACTIVITY SENSORS. Each compartment contained eight 4"x5" sensor plates distributed across the floor. The plates, constructed of 28 gauge galvanized sheet metal, were attached to the hardware cloth floor at one end of their long axis by two maChine screws with a narrow phenolic spacer (1/16" thick) between the plate and the hardware cloth at the point of mounting. A wire terminal was connected under- neath the cage to one of the mounting screws of each plate. The upper surface of the plates were painted with a grey enamel. The plates mounted in the above fashion formed SPST normally Open momentary contact switches with the underside of the plate as one contact and the cage floor as the other contact. These switches closed when a rat stepped upon any part of the plate except the edge used for mounting. Of the eight plates located in a compartment (See Figure 1) one was positioned in front of the food hopper and water bottle which were attached to the compartment wall. To another plate a four-sided aluminum box was attached creating a small enclosure, 4"x5"x4", open at one end. This will be referred to as the resting box. In addition to the front plate switches,a switch was integrated with each center section door. The hinges (2 per door) were 'U' shaped brass strips attached to the top of each plexiglass door and lOOped over a strand of wire in a cutout made in the hardware cloth sides of the center section. Attached to the brass strips, were two loops of stainless steel wire, one on either side of the hardware cloth wall. 20 These loops projected upwards and about %" out from the wall. Attached to either side of the wall, directly in line with the wire lOOps, was a 29 length of stainless steel tape insulated from the wall itself by a phenolic washer. A wire terminal was attached to the 2 strips (see Figure l). The above arrangement resulted in a normally Open SPDT momentary contact switch with the wire lOOps as one contact and the steel tape as the other contact. These switches closed when a rat passed through the door from either directiOn, as the wire lOOp moved with the door and was brought in contact with the steel tape on the wall above the door. Because electrical continuity existed between the wire loops and the cage floor through the brass hinges and center section wall, all switches within a cage had a common ground. D. RECORDING DEVICES. Two 20 channel Esterline Angus event recorders Operated in tandem were used to record activation of the plates and door switches. An 18 volt power supply was connected in series between the common ground of the recorders and the cage bottoms. In addition each pen terminal was jumped to ground through a reverse polarity silicon diode to suppress electrical arching across the foot plate switches. For each cage eight pens were used, one for each door and four for the foot plates. The nest box and food-water plates were each connected to a separate pen while the remaining 3 plates in the rear of the compartment (away from the center section) were connected in parallel to a single pen. In like manner the 3 foot plates in the forward half of the 21 compartment were wired to a common pen. The recorders were run at 12" per hour which allowed discrimination between repeated events occurring at inter- vals as short as 5 seconds. E. EXPERIMENTAL ENVIRONMENT. The experiment was con- ducted in two adjacent rooms (approximately 15'x17'). The rooms were cinder block with concrete floors and heated by steam with ambient temperatures ranging between 70°F and 800F. The continuous sound of exhaust fans tended to mask most outside noices. All Outside sourCes of light including the windows were covered with black plastic sheeting. The Esterline Angus recorders and associated electronics were located in one room and were surrounded with 3" thick sound insulation. An upright cage rack containing 16 small cages was located in the other room. These cages were used for holding animals between testing and for initial habituation to the room environment. III. PROCEDURE. The 3E EEE phase, consisting of 7 days within the experimental cages (with high, medium, low and open complexity), was followed by 3 days of discrete trial testing in the same cage. The first five days of ES.££B occupancy served as a habituation period intended to stabilize responding. Table l is a timetable of procedures for a complete experimental session. The SS LEE procedure guarantees that stimulus satiation is develOped enough to exclude the possibility of stimulus 22 novelty Operating at any meaningful level. Indeed habitua- tion to novelty has been shown to be rapid (Montgomery, 1951; Berlyne, 1955; Glanzer, 1961) and long lasting (Blanchard EE El! 1970). The novelty satiation effects as well as the animals' general familiarity with the cage and its layout are expected to carry over into the discrete trial testing. Thus the order of implementation of the two phases is import- ant in eliminating novelty effects and in assuming stability of responding. Meeting one objective of the study, comparison of the two testing modes, requires maximizing the consistency of correlations between the two phases. . A. PRELIMINARY. The foot plates including the nest box and the doors were cleaned with "Lime-AwaY" before every experi- mental session. Three complexity inserts were fixed in each cage leaving one compartment empty. The pattern of placement was semi-random such that 1) each cage contained three different inserts ("high", "medium", and "low"), 2) no com- partment had the same kind of insert for two consecutively run subjects, 3) during a session no two cages had the same pattern of insert placement. Subjects were obtained from the supplier in lots of four and were held in individual cages for 5 days with free access to food and water. During this holding time the animals were handled once daily for about one minute. B. 52 Egg PHASE. At 8 a.m. on the morning following the fifth day of habituation to the room environment one rat was placed into the center section of each cage. After the 23 Table 1. Procedural Timetable of 15 day Experimental Session Phase Day Time Procedure preliminary -5 a.m. -4,-3,-2,-l varied 3g lib 0 8 a.m. 4 8 a.m. 4, 5 8 a.m. + 8 p.m. discrete 6 8 a.m. trial 6 8 p.m 7, 8 8 a.m. + 8 p.m. 9 8 a.m. Animal placed in holding cage with free access to food and water. Animal removed from holding cage and handled for approximately one minute. Animal placed in center section of experimental cage with free access to the four complexity com- partments (habituation). Chart recorders turned on. Food and water supplies weighed and replenished, fecal boli counted and removed. Animal removed from experimental cage and returned to holding cage. Animal removed from holding cage and given 3 choice trials (30 minute ITI) in experimental cage and then returned to holding cage. As above As above, animal is terminated. Note: the lights were automatically switched on at 8 a.m. and off at 8 p.m. except during the evening testing sessions on days 6 through 8 when the light period was extended to about 9 p.m. 24 animals had spent 120 hours in the cage, the recorders were turned on and measured amounts of food and water were put into their respective containers. In addition the litter trays were cleaned. Subsequently, food and water consumption and defecation were measured every 12 hours. Food and water consumption was determined by weighing the remaining food pellets ("Purina Lab Chow") and the water bottle, then replenishing the supplies to a standard weight. Defecation was measured by counting and removing the fecal boli from the litter trays. Following 48 hours Of this regime the animals were remOved from their cages and returned to their respective holding cages. C.. DISCRETE TRIAL TESTING. At 8 p.m. on the day of removal from the experimental cages subjects were given 3 discrete testing trials. The procedure for any of the subjects was as follows, for each of six sessions which were given every 12 hours.‘ The animal was transported in his holding cage to the experimental cage it previously occupied for one week. It was then introduced into the center section of the cage and once it entered into one of the four compartments the door was blocked to prevent the rat from leaving the compart- ment. After 3 minutes the subject was removed from the compartment and returned to the holding cage. This procedure was repeated twice more with an intertrial interval.of 30 minutes. During the testing, the recorders were operative and 25 activity was recorded. In addition the latency to enter a compartment after placement into the center section was noted. Each animal received a total of 18 trials over 6 sessions. The side of the cage from which the rat was intro- duced into the center was alternated from one trial to the next. RESULTS I. fig Lib Phase. A. The multivariate design. The experimental design was essentially a 4 x 2 facto- rial with repeated measures. Viewed this way there were four complexity levels, one for each compartment, and two. illumination levels. The latter, although artificial light on a 12 hour schedule (light on from 8 a.m. to 8 p.m.), approximately coincided with the true daylight period and will be referred to as "day" and "night". The design implies that each subject served in each of the eight conditions. Contrary to procedures typifying this factorial design, the complexity factor was subject selected. That is, the rat and not the experimenter determined when and for how long 'the rat served in each level, or in other words, when and for how long it occupied each compartment. 1. Basic Scores. From the recordings on the 6th and 7th days of experimental cage occupancy a number of scores were derived. In doing so, the two day sessions of the 48 hours were combined as were the two night sessions. Thus a day and night score was derived for the following measures for each of the four compartments representing high, medium, low and open complexity levels. 26 5. 27 Occupancy time (total time in minutes within compartment). Locomotor activity (total number of discrete depressions of all foot plates). Feeding time (total time in minutes foot plate in front of food and water containers was depressed). Resting time (total time in minutes foot plate at bottom of rest box was depressed). Entry frequency (number of entries made into compartment). In addition to scores based on data obtained from continuous recording, the following scores were derived from observations made every 12 hours. 6. Food consumption in grams. 7. Water consumption in grams. 8. Defecation (number of fecal boli). 2. Ratio Scores. Six ratio scores, generated from the eight basic measures, were also used. Because of low general activity during day sessions which created some scores with zero as a denominator the scores given below were considered only for the night sessions. 1. Mean occupancy (occupancy time divided by number of entries). Resting rate (rest box occupancy time divided by occupancy time). Locomotor activity rate (locomotor activity count divided by the sum of occupancy time minus resting time). Defecation rate (feces count divided by the sum of total time minus resting time). Mean active occupation (the sum of occupancy time minus resting time divided by number of entries). 28 6. Feeding speed (the sum of food and water consumption divided by feeding time). A final score was derived from the continuous recordings and is the only one which reflects a temporal trend, El beit short term. The composite "exit pattern" score is the frequencies of entry into each of the other compartments made directly from the compartment in question. As mentioned before, general activity was particularly low during the day. For example the mean number of entries per day was 6.00 as compared to 74.72 per night, however, two animals spent the total 48 hours in one compartment (high complexity for one and low for the other). These subjects were not included in any data analyses. Analysis of data from the remaining 18 animals included a test of the overall design (4 x 2 ANOVA), separate tests of day and night behavior (4 x 1 ANOVA's) including planned comparisons, and a correlational review of the dependent measures. NOn- parametric tests of a few special cases were also conducted. B. General Behavior - Effects of Light Period. The following results represent distinctions between day and night behavior without regard to which complexity compartment it occurred in. That is, data from the four compartments are combined within an illumination level. The overall 12 hour means of the basic scores for day and night are presented in Table 2. The mean basic scores all showed a marked depression of active behavior in daylight hours. All scores were significantly greater for the night period (except resting 29 Table 2. Overall 12 hour Means of Basic Scores Score Day Night Locomotor activity 26.20 342.80 Feeding time i 2.20 25.78 Nesting time 511.06 143.02 Entry frequency 6.00 74.72 Food consumption 3.20 18.80 Water consumption 8.88 31.50 Defecation 5.22 23.86 30 time which was significantly less) at 0(<.0005 under an ANOVA with l and 17 df (see Table 3). A look at the dis— tribution of activity showed most daylight behavior occurring in the initial hour which appears to be, in part, a contin- uation of a build up in activity during the terminal hours' of darkness. The sudden change of environmental state from dark to light could also contribute to this phenomenon. The nighttime distribution, using the number of entries as a representative activity score, was a relatively smooth trimodal curve peaking during the first, last and 5th hours as shown in Figure 2. The highest peak was during the first hour of darkness suggesting, again, an excitatory effect of state change. C. Effects of Complexity. The following results represent the incidences of various behaviors compared across complexity levels within day and night periods (basic scores) or only night periods (ratio scores). Planned comparisons on each score were made in addition to the overall ANOVA. These were based on the original hypothesis, applied to all basic scores, that the direction of differences in magnitude between complexity levels would conform to high medium low open. Three pair- comparisons of cell means, high-medium>0, medium-low) 0, and low-open>0, served as a critical test of the hypothesis. This was accomplished statistically using correlated E tests or, in the case of occupancy time, the Wilcoxon rank differ- ence test. 31 Table 3. ANOVA on the Effects of Light Period Score Source df Mean F 0< square Locomotor Lt. period 1 9082183 115.102 ‘<.0005 activity Error 17 7838.1 Feeding Lt. period 1 4999.7 36.734 (.0005 time Error 17 136.11 Resting Lt. period 1 646617 72.908 (.0005 time Error 17 8868.9 - Entry Lt. period 1 42504 257.921 «(.0005 frequency Error 17 164.80 Food Lt. period 1 2185.6 189.262 (.0005 consumption Lt. period 17 11.548 Water Lt. period 1 4601.4 127.630 ‘<.0005 consumption Error 17 36.052 Defecation Lt. period 1 3124.8 369.877 <:.0005 Error 17 8.4482 32 .msofluflpcoo uzmfls “Opus wososwmnm muucm mo mason as mufl>fluom mo COHDSQHHDNHU one Q 0 H m m m B m U H Z m 0 m D O m nvdd . nub flan 0km UOA NH .m musmflm SIIUJNI JO HIEHDN IV!" 33 For some contrasts two alpha values may be given. The first, which is always given, indicates the alpha value consistent with the one-tailed nature of the original hypothesis. The second value is included when the direction of the difference is Opposite to that predicted. In this case the alpha is two-tailed. The use of E test based planned comparisons is not particularly conservative, statistically, but, based on the remarkable consistency in the trend re- vealed by the contrasts, the procedure is justifiable. 1. Daytime. For daytime behavior complexity level was found to significantly affect locomotor activity, resting time, and entry frequency all at an alpha level less than .0005 (ANOVA with 3 and 51 degrees of freedom). All other basic scores were insignificant (see Table 4). The last column of Table 4 (and comparable tables that follow) contains the 2E3 squared statistic which indicates the proportion of the total variance of the score explained by the complexity variable. The means of these scores are presented in Table 5. The analysis of the occupancy time data was treated somewhat differently because of certain restrictions imposed by the experimental design. Because the sum of occupancy time across the four complexity levels was constant for all subjects (i.e. total occupancy time in a 12 hour period has to equal 12 hours) a Friedman test of the ranked data was used in place of an ANOVA. This procedure was applied to the nighttime results as well. For daytime periods occupancy 34 Table 4. ANOVA on the Effects of Complexity under Day Conditions: Basic Scores Mean 2 Score Source df square o( eta Locomotor Complexity 3 696.6800 11.708 (.0005 .41 activity Error 51 59.5000 Feeding Complexity 3 7.7442 2.279 .0910 .12 time Error 51 3.3986 Resting Complexity 3 906908.0000 9.379 .800 Low 83.2 2.70 23 <,.005 Open 12.7 1.64 -NIGHT— High 233.4 3.05 47 < .050 Medium 138.0 2.61 76 ) .300 Low 146.2 2.61 32 < .010 Open 71.8 1.72 39 o7fi a R— — .5. E ..J 4 =3 8 B O —— — B = [u .2 I uT—I -— ——————— —o—o—-—- —._-—o— ----- z A £- 0 = H — I?" m .A E ° F a. — o _ — m I! O - 0. HIGH MEDIUM LOW OPEN C O M P L E X I T Y L E V E L Figure 3. Comparison of the effects of complexity under day conditions on four basic scores; resting time (R), occupancy time (0), locomotor activity (A), and entry time (E). - 40 .50 .4 4 B o T 9 I _ D 8: fl 0 z.2 = —--I—a- 0 — — H —- - E. — _ M _ F -T o a 9- =— 0 _— D a — m — HIGH MEDIUM LOW OPEN C O M P L E X I T Y L E V E L Figure 4. Comparison of the effects of complexity under day conditions on four basic scores; food consumption (F), water consumption (W), feeding time (T), and defecation (D). o 0 v .1 ‘ . . ‘ a . ' . ._ . 41 P R O P O R T I O N O P T 0 T A L _— —— Low HIGH MEDIUM C O M P L E X I T Y L E V E L Figure 5. Comparison of the effects of complexity under night conditions on four basic scores; resting time (R), occupancy time (0), locomotor activity (A), and entry frequency (E). 42 2. Nighttime. The results from night periods were notas differen- tiated as those of daylight behavior. Occupancy time and rest time showed significant effects of complexity at alpha levels of .020 and .037, respectively, while activity and entry frequency approached significance (°<¢¥u069 and .060, respectively). As Table 7 indicates, no other results were significant based on ANOVA with 3 and 51 degrees of freedom. Occupancy time was again analized using Friedman's procedure (I? = 10.1 with 3 degrees of freedom). Tables 6 and 8 show that the hypothesis based contrasts between high and medium were significant for occupancy time (“4.05), activity (“8.006) and rest time (“3.009) . Again, the general pattern was sustained where, except for feeding and defecation scores, the low-open comparisons showed relatively large and positive differences while the medium-low differences were small and negative. The distributions across complexity of these scores are compared in Figures 6 and 7. When nighttime beHavior is viewed in terms of rates the differences among complexity levels are more striking. Analysis of the ratio scores (ANOVA with 3 and 51 degrees of freedom as summarized in Table 9) revealed Significant effects on mean occupancy time (O<¢¥.028), mean active occupancy (°<29.023), locomotor activity rate (c£‘<.0005), and defecation rate (“$.001). Resting rate approached significance (“8.073) while feeding speed was not significant. The hypothesis based contrasts (Table 10) revealed a Table 7. ANOVA on the Effects of Complexity Under Night Conditions: Basic Scores. Mean Score Source df square (y: eta Locomotor Complexity 3 13098.000 2.517 .069 .13 activity Error 51 5204.600 Feeding Complexity 3 25.981 .132 .941 .01 time Error 51 197.500 Resting Complexity 3 '44659.000 3.052 .037 .15 time Error 51 14634.000 Entry Complexity 3 330.980 2.637 .060 .13 frequency Error 51 125.510 Food Complexity 3 48.476 .630 .599 .04 consumption Error 51 77.003 Water Complexity 3 182.080 .930 .433 .05 consumption Error 51 195.740 Defecation Complexity 3 53.124 .426 .735 .02 Error 51 124.760 44 Table 8. Planned Comparisons on the Effects of Complexity Under Night Conditions: E Tests on Basic Scores. Score Complexity Mean E value of o(under dunder difference original two-tailed hypothesis test Locomotor High 100.1 activity 2.82 .006 Medium 83.2 .23 .410 Low 80.5 .55 .296 Open 76.0 Feeding time High 6.32 . -.12 .546 908 Medium 6.59 —.24 598 804 Low 7.15 .62 .272 Open 5.71 Resting time High 71.34 2.62 009 Medium 18 43 -.80 783 .434 Low 34.58 .79 218 Open 18.65 Entry frequency High 21.11 ‘ 1.15 .134 Medium 18.97 .12 .454 Low 18 75 1.56 .070 Open 15.89 Food consumption High 5.65 .37 .359 Medium 5.11 .77 .223 Low 3.99 -.04 .516 968 Open 4.04 45 Table 8 (Continued) Water _ consumption High 7.81 -.02 .508 .982 Medium 9.18 -.05 .520 .960 Low 9 29 1.32 .103 Open 6 22 Defecation High 5.69 . -.75 .767 .466 Medium 7 08 1.10 .143 Low 5 03 -.55 .704 .592 P R O P O R T I O N O F T O T A L Figure 6. "i: ll l|||| L HIGH MEDIUM LOW C O M P L E X I T Y L E V E L Comparison of the effects of complexity under night conditions on four basic scores; food consumption (F), water consumption (W), feeding time (T), and defecation (D). 47 .5 o-J a: E" O E" ha 0 D .2 o o—o—n—u—u—u-nu—u—o—o—u—u-g—u—u z - o A— R A H E" a: O n. O m a. - HIGH MEDIUM LOW OPEN COMPLEXITY LEVEL Figure 7. Comparison of the effects of complexity under night conditions on three ratio scores; resting rate (R), locomotor activity rate (A) and defecation rate (D). Table 9. ANOVA on the Effects of Complexity Under Night Conditions: Ratio Scores. ‘ Mean Score Source. df square o< eta Mean Complexity 3 117.24 .287 .028 .16 occupancy Error 51 35.666 Mean active Complexity 3 59.001 .465 .023 .17 occupancy Error 51 17.029 ’ Activity Complexity 3 3.6107 .712 (.0005 .34 rate Error 51 .41445 Resting Complexity 3 .087219 .458 .071 .13 rate Error - 51 .035482 Feeding Complexity 3 2.6005 .105 .957 .01 speed Error 51 24.849 Defecation Complexity 3 .031201 .87 .001 .28 rate Error 51 .0045390 Table 10. 49 Planned Comparisons on the Effects of Complexity Under Night Conditions: E Tests on Ratio Scores. Score Complexity Mean t value of «under oLunder Hifference original two-tailed hypothesis test Mean occupancy High 10.817 2.94 .005 Medium 6.956 - 35 .663 .734 Low 7.661 - - 1.52 .074 Open 4.633 Mean active occupancy High 7.742 2.26 .019 Medium 5.994 .29 .388 Low 5.581 1.82 .044 Open 3 341 Activity rate High .952 - 59 .718 .564 Medium 1 079 80 .217 Low .907 -4 46 999 .0004 Open 1.863 Resting rate High 2528 2.61 .009 Medium .0899 - 61 .723 552 Low 1281 -.38 .644 .712 Open 1516 Feeding speed High 4.849 26 404 Medium 4 421 30 .384 Low 3.919 ' -.28 .608 .784 Open 4.387 50 Table 10 (Continued) Defecation rate High .03689 -l.50 .924 .152_ Medium .07056 ' ' ' 1.28 .112 Low ' .04183 -3.83 .999 .0014 Open .1275 51 pattern similar to those based on other scores except that in some cases the relationship was reversed. Specifically, activity rate and defectation rate showed a negative relation- ship to complexity level. The largest difference was between low and open with two-tailed alphas of approximately .0004 for activity and .0014 for defecation. Confirming the general hypothesis were contrasts between high and medium for mean occupancy time (cx 25.005) , mean active occupancy “2.019) and resting rate (o.70 Low 4.94 2.75 20.5 (.005 Open 2.89 1.70 First trial of each session High 2.44 3.00 47.5 (.05 Medium 1.28 2.50 89.0 ).50 Low 1.56 2.61 38.0 ).025 Open .72 1.89 59 exposure to other complexity levels. For this reason a score was devised which weighted the total trial score by the number of first trial instances it contained. The weighted choice score was obtained by multiplying the total trial score by the first trial score plus one. ANOVA with 3 and 17 degrees of freedom on this new score showed a significant difference among complexity levels (O<6¥.004). Table 15 shows that both.the high-medium and low-Open hypothesis-based comparisons were different from zero («fiaOll‘and .035, respectively). No overall differences for activity or latency were found with ANOVA (see Table 14), however, under a two-tailed t test the low-Open latency contrast was nearly significant but negative (OCQ:084). The weighted choice score is com— pared to the other scores in Figure 9. B. Correlation with Ad Lib Phase. One objective of the study was to compare the results of the two experimental phases in order to determine whether "preference" as tested in the more traditional discrete trial manner was consistent with "preference" derived from a more extensive long term analysis of behavior. For this reason Table 16 and Table 17 are included to provide the correlations between three discrete trial scores and the basic and ratio scores of the first phase. As can be seen from the Tables, choice correlates relatively well with occupancy time, resting time, and entry frequency in that order and more strongly for the night version 60 Table 14. ANOVA on the Effects of Complexity: Discrete Trial Scores Score Source df Mean F (ii 'eta square Weighted Complexity 3 862.19 4.966 .004 .23 choice Error 51 173.61 Locomotor Complexity 3 5.2761 .402 .752 .02 activity Error 51 13.129 Latency Complexity 3 35.697 1.700 .179 .09 Error 51 20.993 61 Table 15. Planned Comparisons on Effects of Complexity: E Tests on Discrete Trial Scores. Score Complexity Mean 13 value of «under o‘under difference original two-tailed hypothesis test Weighted choice High 22.39 2.51 .011 Medium 11.39 - .65 .736' .528 Low 14.28 ~ 1.94 .035 Open 5 78 Locomotor activity High 6.944 ‘ - .97 .826 .348 Medium 8.111 68 .258 Low 7.289 - .51 .691 618 Open 7 911 Latency High 7.472 .45 .331 Medium 6.794 - .33 .626 .748 Low 7.294 -l.74 .958 .084 Open 9.944 62 Table 16. Correlations among Discrete Trial Scores and the Following Basic Scores: Occupancy Time (OT), Locomotor Activity (LA), Feeding Time (FT), Resting Time (RT), Entry Frequency (EF), Food Consumption (FC), Water Consumption (WC), and Defecation (D). -DAY- OT LA FT RT EF FC WC D Weighted . choice .445 .204 .098 .454 .296 .007 .256 .180 Locomotor activity -.077 .124 .141 —.070 -.004 .039 -.109 -.l98 Latency -.O38 -.026 -.138 -.063 -.153 -.086 .032 -.162 -NIGHT- Weighted choice .653 .246 -.058 .578 .395 .133 .211 .356 Locomotor ' activity -.099 .242 .010 -.032 -.101 -.O93 .007 -.032 Latency .051 -.026 -.007 .085 -.042 -.018 -.038 .067 63 Table 17. Intercorrelations among Discrete Trial Scores including Latency (L) and Locomotor Activity (LA), and Correlations with the following Ratio Scores: Mean Occupancy Time (MO), Mean Active Occupancy Time (MAO), Locomotor Activity Rate (AR), Resting Rate (RR), Feeding Speed (FS), and Defecation Rate (DR). MO MAO AR RR FS DR L LA Weighted choice .488 .110 -.242 .301 .006 .026 .516 .125 Locomotor activity .146 .106 .497 -.081 .127 -.027 -.362 Latency .310 -.o77 .087 .268 -.045 .062 ' 64 of these scores. The discrete trial activity and latency scores appear essentially unrelated to the dd lib scores with the exception of locomotor activity which correlates .497 with nighttime dd_1db_1ocomotor activity rate. Table 17 also includes discrete trial score intercorre- lations demonstrating a relatively strong relationship between latency and choice. Latency also shows a negative relation- ship to activity. III. Summary. The idea that rats "prefer" higher complexity was con- firmed, in part, by a number of different measures. Occupancy time, locomotor activity, feeding time, resting time, entry frequency, food consumption and defecation were all greater for higher complexity levels in at least one comparison. In general these scores were ordered across complexity levels as follows, with greatest magnitude first: high-low-medium- open. The smallest difference occurring between low and I medium levels. The graphs in Figures 3, 4, 5, and 6 show the relationships of these measures to complexity. The graphs indicate proportion of the total contributed by each level for day and night behavior. The differences due to com- plexity were greater during the day when general activity was lower. Figures 7 and 8 include similar graphs for the ratio- scores derived from night behavior and the discrete trial results. The graphs reveal that resting rate and the weighted discrete choice score take on patterns similar to 65 the basic_scores. In contrast, locomotor activity rate, *defecation rate, and food and water consumption rates tend to decrease with increasing complexity. The comparable results from the discrete trial and dd lib phases, as suggested graphically, are also indicated by the relatively high correlations between the choice score and occupancy time and resting time, particularly at night (.653 and .578,respectivelx,as given in Table 16). The preference for a particular compartment seemed to be partially conditioned by the rats immediate history of complexity exposure. A strong preference for high complexity was evident when exiting from the low compartment and a preference for medium complexity existed after having experienced high complexity (during the night). Finally, the data was also analyzed correlationally Iusing Pearson's product moment coefficients. The Appendix includes comparisons of separate response patterns for each complexity level. It also presents the overall correlations among the basic scores for day and night behavior. A short summary of the response pattern findings is given below. In respect to behavioral organization in the $9,112 phase, medium and low complexity conditions were similar during both day and night periods, although the patterns underwent substantial change from day to night. During the day these complexity levels shared little similarity with high and Open behavior patterns which were quite similar to 66 one another. Under night conditions behavior in the high complexity compartment tended to change while Open complexity showed no such shift. The high complexity shift brought the behavioral pattern into substantial agreement with those displayed in the low and medium compartments under Opposite illumination conditions (day). DISCUSSION Interpreting the value of environmental complexity, as this study has operationally defined it, can be approached in a number of ways. The simplest and most direct method is to assess the approach eliciting power of the stimulus complex. This requires examination of those appetitive behaviors which bring the rat into proximity with the different complexity areas. Approach, however, only repre- sents part of the process of preference (e.g. when occupancy time is considered). The other important component is what this author will call the "staying response". This includes all behaviors which have the effect of keeping the animal in its proximate environment or more specifically in the same' compartment. For example, feeding, grooming, Sleeping, and even exploration can contribute to this staying response. It is apparent that, on the whole, these behaviors are consummatory. On the other hand, consummatory behaviors like feeding can be considered separately as complexity preference criteria. Determining in which complexity situation a rat is most likely to eat is such an attempt to define complexity preference in terms of a specific biologically relevant variable. At a higher level of organization, these behaviors can be interpreted as contributors to a response complex 67 68 representing a dynamic preference which is conditionalized by the state of the organism. The approach at this level is more amenable to consideration of the value of complexity in terms of habitat use and selection. Approach Responses The power of complexity to elicit approach responses is reflected by two separate dependent measures. Both entry frequencies Of the dd lid phase and choice frequencies of the discrete trial phase each state the relative instances of approach to the four levels of complexity. Examining these scores indicates that complexity does serve as a differential approach eliciting stimulus, however, the marginal differences under dd lib night conditions suggest that the comparative efficiency Of the stimulus dimension is subject to attenuation. It will be useful at this point to distinguish between two antecedents of approach. Once the animal is in the com- partment it may engage in consummatory behavior, tending to stay in the compartment for long durations, or it may engage in nonspecific exploration and tend to exit after a short time. During the initial five day habituation period the situation may have been quite different in that lengthy bouts of specific exploration probably occurred. These bouts being motivated by the unfamiliarity of the cage. Although complexity has been shown to elicit differential levels of appetitive behavior it is suggested that during the early stages of habituation the exploratory incentive offered by higher complexity was greater than is indicated. The basis 69 for specific exploration such as novelty and the necessity to find food and water are greatly diminished as the animal spends more time in the environment. Once the rat is com- pletely familiarized with the environment and the variety of surroundings it offers, appetitive kinds of behavior are likely to be very efficient with no excessive incidence of exploratory approach behavior which does not lead to a consummatory act. Under this assumption frequency of consummatory responses becomes the critical measures of complexity effects. Stayinngegponse. Occupancy time, representing the undefined collection of behaviors which keep the rat where it is, provides a better differentiation of complexity effects even though a night- time leveling of the distribution across complexity levels still occurs. Occupancy time appears to be primarily a result of resting behavior (intercorrelations of .890 and .730 under day and night conditions, respectively)(see Appendix) with little relationship to feeding behaviors or locOmotor activity. Consequently, resting can be assumed to be the predominate staying response responsible for keeping the rat in its immediate environment. Particularly during the day, when the resting rate is greatest, occupancy time in the high complexity compartment is extreme. In nature it is quite rare to see animals sleeping or resting in the Open. A rather obvious interpretation is that complexity is associated with shelter or hiding places 70 for the rat. In this sense the high complexity compartment or environment becomes "home base" for the animal from which it ventures out into surrounding territory (other compart- ments) to explore, find food, or in other ways interact with the more remote surroundings. Shelter has previously been shown by Sale (1969) to be an important parameter of suitable habitat for fish. In both field and laboratory studies Sale found that plant and rock cover were major factors in habitat selection to the exclusion of many presumably salient variables. The importance of shelter as the salient parameter of complexity preference may be seen in the loss of a strong differentiated response under night conditions. While night behavior was concerned with feeding and general locomotion; day behavior was characterized by resting or sleeping. Thus, one may conclude that the reason for the apparent greater approach value associated with higher levels of complexity is primarily founded on its attractiveness for resting (or shelter) rather than for feeding or locomotor activity. Indeed, feeding and activity rates are highest in the least complex environments. Results of the discrete trial testing are also compatible with this interpretation if the experience for the animal is considered slightly traumatic and remembering that day illumination was present. The consequences of the approach response may well be escape (to an environment associated with shelter and relaxation). The consistently short 71 latencies to enter a compartment supports the escape notion. If it can be assumed that the animal knows where it is going when it leaves the center section, it is possible to conclude that the discrete trial procedures primarily test the "home base" qualities of the stimulus complex. Considering the previous dd_ldb exposure, familiarity with all levels of the complexity dimension seems certain. It is the 32.112 results, in fact, which have allowed a much better specifi- cation Of the parameters of discrete trial choice. The discrete trial phase of this experiment yielded a choice score based on a response which would conventionally be thought of as appetitive in nature. The power of com- plexity manifested in the diScrete trial procedure by high approach frequencies is interesting in its unique relation- ship to complexity effects of the dd lib phase. The choice score, rather than reflecting the entry frequencies of the first phase, best reflects occupancy and resting time. Its correlation with nighttime occupancy is .653 as compared to a .296 correlation with daytime entry frequency (see Appendix). In other words, there is a strong relationship between the appetitive behavior in one situation and the consummatory behavior in another. Thus, the choice score takes on a ~ predictive value associated most strongly with the most powerful effect of environmental complexity under freer conditions. The discrete trial technique for testing com- plexity preference, then, seems to be justified in terms of its validity, once the motivational base of the response is 72 understood. Feeding and General Activity. While rats tend to use the high complexity compartments for daytimeinactivity,their bouts of locomotor activity are not so reStricted to high complexity. Likewise, food and water are consumed equally in all compartments (except Open where eating is uncommon during the day). In general the rat's attraction to high complexity is subordinated by behaviors which reflect an internal state consistent with locomotor activity and feeding. This internal state can be interpreted further as one consistent with non-specific exploration. Exploration is assumed, here, from high loco- motor activity, frequent entries into all compartments and short mean occupancy intervals. During these periods of activity is when most feeding Occurs and feeding may well be a prime incentive for the locomotion displayed (locomotor activity correlating .545 with feeding time) (see Appendix). Another incentive for the locomotion could be explora- tion for its own sake as Butler and Harlow (1954) and Berlyne (1960) have treated such behavior. Experiments where food and stimulus complexity have been kept orthogonal (e.g. Timberlake and Birch, 1967; Taylor, 1971) usually indicate that high levels of complexity can substantially reduce the probability of a deprived animal choosing to approach the food area in deference to the complexity area. In terms Of approach behavior it is reasonable to assume that locomotor activity can reflect separable elements of both exploration 73 and feeding. A somewhat surprising characteristic Of the rats' behavior was that no particular compartment became a favored place to feed. This not only demonstrates that complexity is an irrelevant factor in the elicitation of food searching or feeding itself, but also that rats tend not to form place habits for feeding sites. Whether this is an exclusive trait of confined animals is not known and would be hard to deter- mine because studies in the field, where food and water are not homogeneously distributed, would be unable to detect feeding site habits that were not biased by the distribution of food resources. 3 Because feeding behavior is nearly independent of environmental complexity in the context Of this study, the difference in locomotion between that highest complexity level and lower levels is probably a result of a heightened exploratory response. Exploration, though it may occur most in high complexity areas, which have the greatest occupancy times, is also a function of exposure time or familiarity. The lower the complexity level the lower the total occupancy time and the greater the locomotor activity_rate. It appears that regardless Of time spent within a compartment the organism is inclined to equalize the total expense of activity across all levels of complexity. This phenomenon applies to feeding behavior as well. Thus, the primary factor in complexity preference appears to be manifested in the resting/sleeping response. This is 74 in contrast to other consummatory responses such asfeeding, drinking and defecation, the frequencies of which are relatively homogenously distributed across the four com- plexity levels. Obviously, the original expectation of a pervasive complexity effect has not been born out. Even the expectation that the strength of appetitive behaviors would be positively related to environmental complexity is not fully confirmed. Dimensionality of Complexity. An important aspect of the experimental design is the assertion that the complexity variable is essentially uni- dimensional or monotonically organized across the four com- partments. This can only be ascertained indirectly by looking at the unidimensionality of the variable's effect. The rather consistent ranking from high to Open on most dependent measures supports the assumption. The exception is evidenced by the medium-low comparisons where no significant differences were found. In addition, the intercorrelational patterns Of these two levels were similar during both day and night periods. The similarity is particularly signifi- cant in view of the substantial shift that occurred in the patterns from day to night (see Appendix). Considering these facts, condensing the complexity variable into three levels by treating medium and low as one level would be a logical step to ensure monotonicity. Although this was not done for any analysis, it would be a good way to look at the results post hoc. 75 Another question relating to unidimensionality concerns behavior in the Open compartment. Only in this compartment 'was the behavioral pattern constant across illumination levels. Rates of locomotor activity and defecation were higher than in any other compartment suggesting arousal effects similar to those found in standardized openfield situations. In addition, water was consumed at the fastest rate in Open complexity. Drinking in this situation could be an emotional response to the arousal elicited by the Open environment. The arousing prOperties of the open compartment may explain the lack of a shift in day-night behavioral patterns. That is, the arousing prOperties of night may have little effect because the Open environment is already arousing even under day conditions. Stimulus Discrepancy Hypothesis. A limited arousal explanation of behavior has been prOposed to handle the apparently unique response to the Open compartment. A much broader application of the arousal Concept has been adOpted by some theorists to explain stimulus seeking behavior in general. Fiske and Maddi (1961) contend that stimulus deviations in either direction from a familiar standard serve to arouse the animal and can elicit approach responses toward the moderately dissimilar stimulus. If we assume that on the average the familiar standard in the dd Add situation is the mean level of stimulation pro- vided by all four compartments then the Fiske and Maddi interpretation would predict that both extremes of the 76 complexity dimension would be arousing and good approach elicitors. This interpretation might be applicable to the extent that it concerns general arousal, however, in respect to the approach value associated with these moderately' discrepant stimulus levels, the hypothesis clearly has some problems when the frequency Of visiting the open compartment is less than chance. Assuming that an animal is Operating with a middle- valued standard it can be argued that stimulus discrepancy does produce arousal. When arousal levels are inferred from appetitive response rates, the high and low ends of the . complexity dimension appear to be the most arousing. The similarity in the daytime response patterns for these two levels (see Appendix) provides further justification for the claim that high and Open levels can produce the same effects. Beyond this claim, however, it is clear that arousal cannot work as an approach incentive as Fiske and Maddi suggest. If it did, Open complexity would be approached equally as Often as high complexity. Yet, the preference for high complexity is much stronger in terms of at least one appetitive response (approach frequency) and at least one consummatory response (resting). In this example, the response consequences of the stimulus complex apparently have no effect on the complex's attractiveness. I Forgetting about this contradiction for a moment, suppose the high complexity level attracts the animal because it is arousing. Why, then,does the animal select this environment 77 for resting and sleeping? Obviously, high complexity can not be arousing the animal very consistently. In general, the application of an arousal process to explain the ‘ mechanics of environmental preference is unsatisfactory. The simple alternative to Fiske and Maddi's reSponse- based explanation is to exclude arousal as an intervening variable.) Instead of assuming that the effect of environ- mental stimuli are response produced it is necessary only to consider them as elicited. Particularly, there is no need to require that a preference response be reinforced by a change in arousal state. Appealing simplicity is achieved when the mechanics of a process are stimulus bound and response pro- duced effects, although they may have beneficial consequences, are not construed as important motivational variables. The discrepancy hypothesis, in its stimulus-based form, probably works best in situations where the standard can be viewed as a transient internal representation reflecting short-term habituation to the immediate stimulus surround. Under this assumption the standard becomes whatever complexity level the animal has been exposed to most recently. If the relationship between attractiveness and stimulus differences is formulated as all-shaped function, complexity levels Of moderate rather than extreme or minimal differences from the currently occupied compartment would be more attractive. Some support for stimulus discrepancy in this context is found in the exit patterns from the high and low complexity compartments. When exiting from high complexitylrats showed 78 a heightened preference for medium complexity while the attraction of the Open compartment was suppressed. In other words, under the short term view of immediate change, preference for a moderate but not an extreme change in environmental complexity was suggested. Exits from the low complexity compartment indicated a heightened attrac- tion toward high complexity and a reduced preference for medium complexity. In other words, preference for a moderate but not a minimal immediate change in environmental complexity was suggested. In dealing with the relative differences between complexity levels it is important to remember that the low and medium levels, because of the unified way in which the rats responded to them, should be considered as nearly identical. Another possibility is that the standard on the average is better represented by high complexity as a consequence of the animal's disproportionately greater exposure to this level. Previously described discrepancy hypotheses would be unable to predict the results Obtained in the present study under this definition of a standard. Further, while the above standard is conceived as a product of adaptation, emphasis on the current state of habituation to a specific complexity level is not necessary. In the long run, an overall relatively intransient standard may develOp. Again this alternative could not be handled by stimulus discrepancy hypotheses. There is, however, a unilateral discrepancy hypothesis capable of dealing with the proposition that high complexity serves as the discrepancy standard. 79 Dember and Earl (1957) as proponents Of the adaptation vieWpoint, have modified the idea of the/l-shaped approach function. They have suggested that when looking at long-term shifts in complexity preference the course of stimulus seeking is one-way. That is, only stimuli of greater complexity than the adaptation level are approached. The approach curve is thus reduced to alfl-shaped function of positive discrepancy. An animal operating with high complexity as the discrepancy standard would be limited in its approach prefer- ences as no higher level of complexity is available. In this case, the approach tendency would be expected to decrease monotonically as the stimulus discrepancy deviates away from the ideal of moderate positive discrepancy. That is, as it becomes more negative. This is exactly what was found in the present study. During the habituation phase of the experiment, it is. possible that the rats initially were Operating with low complexity standards derived from the starkness of their home cages. Under Dember and Earl's adaptation hypothesis the standard would have shifted upwards finally reaching and settling at the high complexity level. Subsequently, the long-term.aspects of the standard would be fixed. Alterna- tively, the animals may have entered with a fixed standard already at a high level. The latter seems unlikely but discrimination between the tWo conditions is impossible based on the limited data on early behavior. The preceding explanation is primarily intended for 80 long-term trends and not the minute to minute behavior of the rats. Obviously, as previously discussed, short-term patterns in movement from one compartment to another exist. Indeed, the concept of stimulus discrepancy, in its various applications, should be taken only as a reflection of general trends within short-term or long-term preference behavior. H When thinking of Dember and Earl's construction it h‘fiw should be remembered that short-term behavior constantly intervenes. Excursions from the high complexity compartment H may be motivated by a multitude of factors including a search h; for variety in environmental stimulation. Thus, achieving a reduction in the complexity of its surroundings could be con- sidered a desired consequence of the rat's motility. The determination of the role of environmental complexity in the selection of suitable or Optimal habitats and its role in guiding the animal within its adopted habitat await exten— sive field investigations. However, the present study has found that the more complex an environment is, the more likely it is that the rat will use the region as a home base. ‘Whether the home base quality is attractive because it (provides exploratory incentives, stimulation for general arousal, stimuli compatible with seclusion and relaxation, or (a.combination of these factors is not completely clear. It appears that both relaxation stimuli and exploratory estimuli are important for the formation of a relatively sstable home base, while arousal stimuli are of tertiary C20ncern. Especially, in view of the simple elicited nature 81 of preference behavior, arousal is a burdensome hypothetical construct. The parsimony of a stimulus-based determination of both short and long-term reactions to environmental complexity is entirely adequate to explain previous findings as well as the present results. The stimulus bound formulation also works well in ethological terms being consistent with Hilden's (1965) notion that animals tend to select habitats on stimulus variables which are often irrelevant to the animals in a biological sense. The cue value of the stimulus or stimulus dimension is presumably determined on a genetic/evolutionary basis, although early experience could modify this (e.g. Wecker, 1963). State Considerations. The fact that the behavior most affected by complexity level is also strongly influenced by light period demonstrates the importance of accepting environmental preference as a dynamic process. This author has suggested that the dimi- nuation of complexity effects associated with a change of illumination to night levels reflects a change in the rat's internal state. The dramatic increase in locomotor activity lends obvious support to this interpretation. Yet, this shift in behavior could be explained by the loss of visual cues under the no-light condition of night. For example, it could be argued that with the loss of visual cues the rat is either unable to discriminate between different complexity levels or that without visual support the discrimination is meaningless in terms of any differential response eliciting 5' Aims»; .1!— =1 _L 82 potential. It is easily assumed, however, that the rat is Operating with both visual and tactual modalities and that the complexity dimension is just as meaningful tactually as it is visually. Considering the extensive use of vibrissae by albino rats in unfamiliar situations this assumption seems reasonable. Even if the visual mode predominates it seems that the animal has had sufficient experience with the environment to effectively integrate the associative aspects of the different stimulus dimensions (visual and tactual) to the extent that the "internal representation" of the environ- ment is the same day or night. A reliable test of the extent to which any behavior shifts reflect a reliance on visual information would require the nighttime illumination to be raised to a level which allows adequate visual acuityand still represents a substan- tial drOp from the daytime illumination level. If the lack of a strong complexity preference continued, then the effect would appear to be caused by something other than loss of visual information. A slightly different approach would have to be taken if the effects of illumination cycle BEE dd versus the effects of a circadian rhythm, which in the present experiment are confounded, are to be determined. For example, illumination levels could be changed every few hours and if a diminuation of the differential response to complexity occurred synchro- nously it could be concluded that the effect is essentially 83 independent of a circadian rhythm. Accepting environmental preference as a dynamic process, Of course, suggests the necessity of plotting the temporal course of behavior much more closely than the present study has attempted. This does not require a change in experi- mental design or a major change in recording procedure but rather a more sophisticated analysis of behavior. That is, a behavior sampling procedure yielding a fairly continuous flow of data is needed. While this existed in the present experiment for locomotor activity, compartment entry, and resting box occupancy; because of the complex and voluminous nature of the data, a sequential analysis was not attempted. Further, the sampling of food and water consumption and defecation was limited to twelve hour intervals. Continuous recording of at least feeding and drinking would be essential to an accurate assessment of the short-term temporal course of preference behavior. Continuous monitoring of feeding and drinking is particularly important in reference to behavior classification and state identification. As previously discussed, this kind of information is critical to the identification of appetitive behavior and short-term state conditions. Obviously, the day-night illumination variable is a very gross and restrictive division of state levels. Conclusion. To summarize, it has been shown that rats select one environment for most of their daytime resting. This is 84 usually the high complexity and never the open complexity compartment. This preference appears stable over the two days and is also reflected, to a lesser degree, under night conditions. The nighttime preference is in terms Of occupancy and resting times and entry frequency and not in terms of more active behaviors such as feeding, drinking and nonspecific locomotion. It was suggested that high complexity evolves as a "home base" from which the animal initiates exploration and feeding activity. These behaviors were found to be controlled by illumination level. The short-term bouts of stimulus seeking showed evidence of being all-shaped function of sti- mulus discrepancy. The long-term evolution of a standard environment (high complexity) was proposed dld_Dember and Earl (1957). The existence of a relatively fixed innately determined standard is the alternative explanation. The inclusion of arousal in a motivational theoretic was found to be unworkable or unnecessary. Particularly in relation to discrepancy hypotheses the concept was problematic. It may be useful, however, in explaining the unique response to the Open compartment. The general avoidance of this environment and the high rates of appetitive and consummatory responding, once the animal was in it, can be viewed as derivitives of high induced emotionality. Finally it was found that the more traditional discrete trial procedure seemed to tap environmental preference founded on the home base qualities of complexity as defined by the dd lib results. APPENDIX APPENDIX A. Correlation Among the Dependent Measures Within Treatment Conditions. For each of the complexity level x illumination level (4 x 2) conditions the intercorrelations Of the set of dependent measures under consideration was computed. Thus, when the basic scores were used eight 8 x 8 matrices were constructed. Similarly, eight 6 x 6 matrices were compared when the ratio scores were examined. 1. Basic Scores. The lack of consistency between the correlational patterns of the different conditions was striking. Even under the same illumination level the behavior in compart- ments of adjacent complexity level displayed little simi- larity. Because of high variability and an insufficient number of subjects no comprehensive review of this data such as cluster analysis was attempted. Two rather crude surveys of the data, however, were conducted. One looked for measures which tended to show similar correlations with the other measures across conditions. The other compared the overall correlational patterns among conditions. The latter will be treated first. The eight matrices based on treatment condition, as described earlier, were divided along their main diagonals. 85 86 Using these matrix halves all correlations were reduced to '+', '-', or '0' using the inverse of the square root Of the sample size as the error term. Thus all matrix entries greater than .235 were coded as '+', less than -.235 as '—' and all others received a '0' code. The transformed matrices are given in Figure A-l. All possible pair-comparisons between the matrices were made. Each comparison yielded a congruency or "similarity score" when one matrix was laid atop another and the number and extremeness of deviations in the correSponding correla- tional entries were tallied. If two corresponding entries were the same, a zero was registered, if one was '0' and the other a '+' or '-' then the value 1 was registered and if one was a '+' and the other a '-' the value 2 was registered. The resulting sum of these deviation values was subtracted from 56 (the maximum possible deviation score) and then divided by 56 giving a percentage similarity score ranging from 1.0 (prefect congruency) to 0.0 (complete dissimilarity). These values for the 28 pair-comparisons are given in Table A-1. By chance alone a similarity score would have the expected value of .33. The table shows that all scores were above this value suggesting some thread of relationship across all conditions. This is the least that would be expected. Further examination of the condition pairs yielding the more extreme similarity scores led to some tentative conclusions. Reemphasis of the procedural crudeness OPEN M E D I U M H I G H i O +-° O + + o + + + + + O O O O O + + + + + O + + + + O + - - O - + + o W O + o + - O D O O + + (Emmi: DAY OOOOOO OOOOO‘P OO+OO + + + + + + + + + + + + + '0 O + + + O + O -0 87 O A T R E F W O A T R E F W O A T R E F W O A T R E F W + O +1+ o+-+ 49m *000 0.00 + O O - NIGHT E o - — o -000- .000- O+OO- FOOOO+ ~0+O¢O ’O+-O 0000- OO++-O WO++OO+ D -O O+O+O OOO+OO- + + + + O + + Response patterns for each complexity level under day and night Figure A—l. The intercorrelations of the basic scores have been indicates the to! A reduced to positive or negative relations. lack of a significant relationship. conditions. 88 Table A-1. Similarity Scores of the Correlational Patterns of Responding (Basic Scores) within each Condition. The Value of the Table Entries Represents the Degree of Similarity found in the Response Patterns Of each Contrast between Conditions. DAY NIGHT COMPLEXITY High Medium Low Open High Medium Low Medium .48 DAY Low .61 .74 Open .88 .46 .67 High .59 .76 .78 .65 Medium .74 .48 .69 .70 .61 NIGHT Low .74 .43 .67 .57 .57 .85 Open .79 .48 .72 .81 .67 .69 .79 89 is necessarily a caution toward accepting these conclusions. During the day, behavior in the high and Open compart- ments showed similar patterns (similarity score = .88) as did the low and medium compartment behavior (.74). The pre- dominate source of disagreement for both pairs was in the food and water scores where some positive correlations were lacking in one of the compartments (open in the first case and low in the second). In addition substantial dissimilarity existed between the medium and high and the medium and low complexity levels (.48 and .46 respectively). Nighttime behavior generally showed a different relation- ship between the compartments and different correlational patterns in the same compartment as compared to the day results. Low and medium maintained their relative similarity (.85), although the relationship was different from that found for daytime behavior. While the food and water corre- lations tended to match, the intercorrelations of occupancy time, activity, and resting time lost their similarity. The previous high and low congruency was not evident during night periods. Further;open was most similar to low complexity (.74). Behavior organization at night in high complexity resembled that displayed during the day under medium and low complexity conditions (.76 and .78 respectively). On the other hand the low and medium night patterns were similar to day behavior in the high compartment (.74 and .74) with the major exception of low correspondence for correlations 90 associated with the activity score. In general, behavioral organization seems to reflect an interaction between light period and complexity level. That is, similar patterns were associated with daylight behavior in high and nighttime behavior in low and medium compartments. Conversely, nighttime behavior in high was most like daytime behavior in low and medium. Behavior in the Open complexity compartment did not appear to be affected by this behavioral shift and remained relatively constant. The other approach to the correlational data was to look for scores which displayed consistent relationships with the other scores across complexity and illumination conditions. Using the same coding (+, -, and O) the matrices were re- arranged so that a separate matrix was obtained for each dependent measure. Dimensions were conditions (8) X remain- ing scores (7). Only two meaningful intercorrelations were found to be consistent. Resting time correlated positively with defecation with the single exception of zero correlation for Open complexity during the day. Nesting time was also found to have no relationship to food consumption except for medium complexity which showed a positive correlation during the day. 2. Ratio Scores. Because of the intrinsic statistical problems of inter- preting correlations among ratios based on common scores and because a scan of the matrices reveled no discernable patterns these results were not pursued. 91 B. Correlations Independent of Treatment. The correlations among the various basic scores inde- pendent of complexity level are given in Table A-2. Separate correlations are given for day and for night results. 92 Table A—2. Intercorrelations among Basic Scores under Day (Left-hand Matrix) and Night (Right-hand Matrix) Conditions. The Basic Scores include Occupancy Time (OT), Locomotor Activity (LA), Feeding Time' (FT), Resting Time (RT), Entry Frequency (EF), Food Consumption (FC), Water Consumption (WC), and Defecation (D). OT LA FT RT EF FC WC D OT .377 .115 .730 .587 .284 .160 .650 LA .402 .545 .319 .420 .294 .063 .253 FT .245 .448 -.057' .339 .618 .411 -.001 RT .900 .406 .041 .413 .080 .116 .390 EF .534 .808 .502 .514 .308 .156 .502 PC .218 .223 .241 .194 .243 .328 .209 WC .094 .224 .037 .070 .203 .101 .155 D .482 .172 .118 .500 .382 .340 .076 LIST OF REFERENCES LIST OF REFERENCES Adlerstein, A., and Fehrer, E., 1955, The effect of food deprivation on exploratory behavior in a complex maze. d. comp. physiol. Psychol., 48, 250-253. Barnes, G. W., and Baron, A. 1961. Stimulus complexity and sensory reinforcement. d. comp. physiol. Psychol., 50, 228-232. Berlyne, D. E. 1950. Novelty and curiosity as determinants of exploratory behavior. Brit. d. Psychol., 41, 68-80. Berlyne, D. E. 1955. The arousal and satiation of perceptual'/ curiosity in the rat. d. comp. physiol. Psychol., 48, 238-246. Berlyne, D. E., and Slater, J. 1957. Perceptual curiosity, exploratory behavior and maze learning. d. comp. physiol. Psychol., 50, 228-232. Berlyne, D. E. 1960. Conflict, arousal and curiosity. New York: McGraw—Hill. Bindra, D., and Spinner, N. 1958. Response to different degrees of novelty: The incidence of various activities. d. exp. anal. Behav., 1, 341-350. Bingham, W. E., and Griffith, W. J. 1952. The effect of different environments during infancy on adult behavior in the rat. d. comp. physiol. Psychol., 45, 307-312. Blanchard, R. J., Shelton, V. F., and Blanchard, D. C. 1970. Historical effects of stimulus exposure: readiness to eat and object exploration. Learning and Motivation, 1, 432-444. Butler, R. A. 1953. Discrimination learning by rhesus monkeys to visual-exploration motivation. d. comp. physiol. Psychol., 46, 95-98. Butler, R. A., and Harlow, H. F. 1954. Persistence of visual exploration in monkeys. d. comp. physiol. Psychol., 47, 258-263. 93 94 Campbell, B. A., and Sheffield, F. D. 1953. Relation of random activity to food deprivation. d. comp. physiol. Psychol., 46, 320-322. Dember, W. N. 1965. The new look in motivation. Amer. Scientist, 53, 409-427. Dember, W. N., and Earl, R. W. 1957. Analysis of exploratory, manipulative and curiosity behavior. Psychol. Rev. 64, 91-96. Dember, W. N., Earl, R. W., and Paradise, N. 1957. Response by rats to differential stimulus complexity. d. comp. physiol. Psychol., 50, 514-518. Denenberg, V. H. 1964. Critical periods, stimulus input and emotional reactivity: a theory of infantile stimulation. Psychol. Rev., 71, 335-351. Denenberg, V. H. 1967. Stimulation in infancy, emotional reactivity and exploratory behavior. £2 D. C. Glass (ed.) Neurophysiology and Emotion, New York: Rockefeller Univ. Press for Russell Sage Foundation. Denny, M. R., and Ratner, S. C. 1970. Comparative Psychology, Homewood, Ill: Dorsey Press. Dutch, J., and Brown, L. B. 1971. Visual stimulus complexity and approach behavior in rats. Psychon. Sci., 22, 162-163. Fehrer, E. 1965. The effects Of hunger and familiarity of locale on exploration. d. comp. physiol. Psychol., 49, 549-552. Fiske, D. W., and Maddi, S. R. 1961. Functions 9: Varied Experience. Homewood, Ill: Dorsey Press. Fowler, H. 1967. Satiation and curiosity. £2? The Psychology dd Learning and Motivation. Advances in research and theory, Spence, K. W., and Spence, J. T. (eds).: New York, Academic Press, vol. 1, 157-227. Glanzer, M. 1953. Stimulus satiation: an exploration of spontaneous alteration and related phenomena. Psychol. Rev., 60, 257-268. Glanzer, M. 1961. Changes and interrelations in exploratory behavior. d. comp. physiol. Psychol., 54, 433-438. Haben, T. 1969. Unidirectional changes in preference for increasing visual complexity in the cat. d. comp.physiol. Psychol., 68, 296-302. 95 Harlow, H. F., Harlow, M. K., and Meyer, D. R. 1950. Learning motivated by a manipulation drive. d. exp. Psychol., 40, 228-234. Harris, V. T. 1952. An experimental study of habitat selection by prairie and forest raCes of the deer mouse, Peromyscus maniculatus. Contrib. Lab. Vert. Biol., Univ. dd Michigan, 56, 1-53. Hebb, D. O. 1949. The grganization of Behavior. New York: Wiley. '— Hebb, D. O. 1955. Drives and the C.N.S. (conceptual nervous system). Psychol. Rev., 62, 243-254. Hilden, O. 1965. Habitat selection in birds, a review. Ann. 2001. Fenn., 2, 53-75. Hughes, R. N. 1965. Food deprivation and locomotor exploration in the white rat. Anim. Behav., 13, 30-32. Hughes, R. N. 1968. Behavior of male and female rats with free choice of two environments differing in novelty. Anim. Behav., 16, 92-96. Hymovitch, B. 1952. The effects of experimental variations on problem solving in the rat. d. comp. physiol. Psychol., 45, 313-321. Jennings, W. A. 1971. Total home-cage activity as a function of the estrous cycle and wheel running in the rat. Psychon. SCio, 22’ 154-1650 Klopfer, P. H. 1963. Behavioral aspects of habitat selection: the role of early experience. Wilson Bull., 75, 15-22. Klopfer, P. H., and Hailman, J. P. 1965. Habitat selection in birds. Id; Lehrmen, S., Hinde, R. A. and Shaw, E. (eds.) Adx. $3 the Study dd Behavior, New York: Academic Press, vol. 1, 279-303. Kohn, A. J. 1967. Environmental complexity and species diversity in gastropod genus Conus on Indo-West Pacific reef platforms. Am. Nat., 101, 251-259. Leuba, C. 1955. Toward some integration of learning theories: the concept of optimal stimulation. Psychol. Rev., 1, 27-33. May, R. B. 1968. Pretest exposure, changes in pattern complexity and choice. d. comp. physiol. Psychol., 66, 139-143. 96 McClelland, D. C., Atkinson, J. W., Clark, R. W., and Lowell, E. L. 1953. The Achievement Motive. New York: Appleton- Century. Montgomery, K. C. 1951. The relation between exploratory behavior and spontaneous alternation in the white rat. d. comp. physiol. Psychol., 44, 582-589. Montgomery, K. C. 1955. The relations between fear induced by novel stimulation and exploratory behavior. d. comp. physiol. Psychol., 48, 132-136. Montgomery, K. C, and Monkman, J. A. 1955. The relation between fear and exploratory behavior. d. comp. physiol. Psychol., 48, 132-136. Myers, A. K., and Miller, N. E. 1954. Failure to find a learned drive based on hunger: evidence for learning motivated by "exploration". J. comp. physiol. Psychol., 47, 428-436. _ Pereboom, A. 1968. Systematic-representative study of spontaneous activity in the rat. Psychol. Rep., 22, 717- 732. Reese, E. S. 1963. The behavioral mechanisms underlying shell selection by hermit crabs. Behavior, 21, 78-126. Riege, W. H. 1971. Environmental influences on brain and behavior of year-old rats. Dev. Psychobiol., 4, 157-167. Rosenzweig, M. R., Krech, D., Bennett, E. L., and Diamond, M. C. 1962. Effects of environmental complexity and training on brain chemistry and anatomy: a replication and extension. d. comp. physiol. Psychd£., 55, 429- 437. Rosenzweig, M. L., and Winakur, J. 1969. Population ecology of desert rodent communities: habitats and environmental complexity. Ecology, 50, 558-572. Sale, P. F. 1969. A suggested mechanism for habitat selection by the juvenile manini acanthrus truostegus sandvicensis WStreets. Behavior, 35, 27-44. Sales, S. M. 1968. Stimulus complexity as a determinant Of approach behavior and inspection time in hooded rats. Can. d. Psychol., 22, 11-17. Schneirla, T. C. 1959. An evolutionary and developmental theory of biphasic processes underlying approach and withdrawal. Id: M. R. Jones (ed.), Nebraska Symposium dd Motivation. Lincoln, Nebraska: UniverSIty of Nebraska Press. 97 Sexton, O. J., Heatwale, H., and Knight, D. 1961. Correlation of microdistribution of some Pannamanian reptiles and amphibians with structural organization of the habitat. Carib. d. §E£°r 4, 261-295. Spence, K. W. 1956. ,Behavior Theory and Conditioning. New Haven, Yale Univ. Press. Taylor, G. T. 1971. The incentive value of complexity. Psychon. Sci., 22, 143-144. Timberlake, W., and Birch, D. 1967. Complexity, novelty, and food deprivation as determinants of the speedvof a shift from one behavior to another. d. comp. physiol. Psychol., 63, 545-548. - Walker, E. L., and Walker, B. E. 1964. Response to stimulus complexity in the rat. Psychol. Rev., 14, 484-497. Wecker, S. C. 1963. The role of early experience in habitat selection by the prairie deer mouse, Peromysous maniculatus bairdi. Ecol. Monographs, 33, 307-325. Welker, W. I. 1957. "Free" versus "forced" exploration of a novel situation by rats. Psychol. Rep., 3, 95-108. MICHIGAN STQTE UNIV. LIBRQRIES 11111 31 111 1111" "ll llll 11111111 9 01 2 31038291