ABSTRACT THE INTENSITY, HABITUATION, AND RETENTION OF HABITUATION OF THE THREAT DISPLAY IN MALE BETTA SPLENDENS (REGAN) AS A FUNCTION OF ELICITING STIMULI BY Michael H. Figler The threat display of the male Siamese fighting fish (EEEEE splendens) is a highly integrated phenomenon, involving approach, extension of the gill covers (opercu- 1ae) and brachiostegal membranes, erection of the medial fins, intense deepening of body color, orientation move- ments, and undulatory movements. Many qualitatively different stimuli are capable of eliciting the complete threat display (e.g. live male £3522, mirror image, two dimensional cut-out). However, therelationsbetween the initial eliciting strength of the threat stimulus and the degree of habituation, retention, and recovery of habituation have not been systematically investigated. These relations are important in an elicit- ing stimulus and competing response theory of habituation like that of Ratner (1970). To investigate the above relations, each g in four independent groups was exposed to one of four different Michael H. Figler eliciting stimuli for two 40 minute sessions separated by a 24 hour rest period. The four threat elicitors were: a live unhabituated, highly responsive male getta (Unhab. male); a mirror image (Mirror); a live habituated, relatively un— responsive male EEEEE (Hab. male); and a two dimensional cut-out of a male EEEEE in lateral display (Cut-out). Dur- ing the observation periods, five different reliable com- ponents of the display were simultaneously recorded: air gulping frequency (AG), gill cover erection frequency (GF), gill cover erection duration (GD), fin erection frequency (FF), and fin erection duration (FD). For all five dependent measures, the Unhab. male stimulus was the strongest elicitor followed by the Mirror, Hab. male and Cut-out stimuli, respectively. Only two of the dependent measures (GD, FD) showed a systematic decre- ment for all stimulus groups. With respect to both of these measures, initial eliciting stimulus strength was positively related to absolute response decrement, and pro- portion of recovery of habituation. Initial eliciting stimulus strength was inversely related to the proportion of decrement across sessions, and the proportion of reten- tion of habituation over a 24 hour period. Four of the threat components (GF, GD, FF, FD) were highly positively related. AG was shown not to be an integral component of the display,itself. The data provided support for portions of an eliciting stimulus and competing response theory of habituation. THE INTENSITY, HABITUATION, AND RETENTION OF HABITUATION OF THE THREAT DISPLAY IN MALE BETTA SPLENDENS (REGAN) AS A FUNCTION OF ELICITING STIMULI BY Michael H. Figler A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1970 To Sonia, my loving wife ACKNOWLEDGMENTS I would like to thank my major professor, Dr. Stanley C. Ratner, for his advice, encouragement and friendship throughout my entire doctoral program. Special thanks also goes to Dr. James C. Braddock of the Department of Zoology for his personal assistance, and the generous use of his laboratory facilities. To Drs. M. Ray Denny and Ralph Levine goes sincere appreciation for their constructive criticism during the initial stages of the research and evaluation of the manuscript. I am also indebted to Dr. Andrew M. Barclay for his assistance in the computer analysis of the data. And to my loving wife Sonia, thank you for your love, patience and support throughout my entire graduate career. LIST OF TABLES . LIST OF APPENDIX LIST OF FIGURES INTRODUCTION . . METHOD . . . . . RESULTS . . . . DISCUSSION . . . REFERENCES . . . APPENDIX . . . . TABLE OF TABLES CONTENTS Page iv Table l. LIST OF TABLES Page Summary Table for Multivariate Analysis and Univariate Analyses of Variance for Each Dependent Variable: Air Gulping Frequency (AG), Gill Cover Erection Frequency (GF), Gill Cover Erection Duration (GD), Fin Erection Frequency (FF), and Fin Erection Duration (FD) . . . . . . . . . . . . . . . 38 Pearson Es for Dependent Variables in Session 1 and Session 2; Stimulus Groups are Con‘bined (E = 40) a o a o o a o I o o o 55 Kruskal-Wallis H Test Summary Table for Four GD Proportional Variables (Proportion of Decrement in Session 1, Proportion of Across Sessions Decrement, Proportion of Retention, and Proportion of Recovery): Sum of Ranks and Mean Ranks for Each Stimulus Group . . . . . . . . . . . . . . 61 Kruskal-Wallis H Test Summary Table for Four FD Proportional Variables (Proportion of Decrement in Session 1, Proportion of Across Sessions Decrement, Proportion of Retention, and Proportion of Recovery): Sum of Ranks and Mean Ranks for Each Stimulus Group . . . . . . . . . . . . . . 65 LIST OF APPENDIX TABLES Table Page A. Pearson £5 for Dependent Variables in Session 1 and Session 2; Unhabituated Male Stimulus Group (E = 10) . . . . . . . . 86 B. Pearson Es for Dependent Variables in Session 1 and Session 2; Mirror Stimulus Group (E = 10) . . . . . . . . . . . . . . . 87 C. Pearson £5 for Dependent Variables in Session 1 and Session 2; Habituated Male Stimulus Group (E = 10) . . . . . . . . 88 D. Pearson rs for Dependent Variables in Session 1 and Session 2; Cut—out Stimulus Group (E = 10) . . . . . . . . . . 89 Vi Figure 1. LIST OF FIGURES Test aquarium . . . . . . . . . . Mean air gulping frequency per ten minute observation period . . . . . . Mean gill cover erection frequency per ten minute observation period . . Mean total gill cover erection duration (seconds) per ten minute observation period a o a o o u o o I o I I Mean fin erection frequency per ten minute observation period . . . . . . Mean total fin erection duration (seconds) per ten minute observation period Page 26 40 42 46 49 52 INTRODUCTION When a stimulus is repeatedly presented to an organ- ism in a well-controlled situation, a decrement in the re- sponse to the stimulus usually occurs. This decrement is termed habituation. The process of habituation does not depend upon effector fatigue since a response which has been completely habituated to a particular stimulus may be immediately and fully elicited by a novel stimulus of equal intensity (Harris, 1943; Thompson and Spencer, 1966). It is also clear that habituation may be separated from sensory adaptation or fatigue since it persists even when long ‘ periods of rest intervene between presentations of the stimulus. In other words, there is retention of the re- sponse decrement which exceeds that which can be attributed solely to peripheral fatigue (Thompson and Spencer, 1966). It is generally accepted that the process of habituation is a form of learning and, therefore, must involve some sort of changes in the central nervous system and/or some type of long lasting changes (Harris, 1943; Thorpe, 1963; Thomp- son and Spencer, 1966; Denny and Ratner, 1970). Habituation occurs for species in all phyla (Harris, 1943; Denny and Ratner, 1970), and is considered to have very important ecological significance for the individual organism. If an organism ceases responding to a stimulus Of little significance to the organism, considerable energy is conserved (Thorpe, 1963). Many different stimulus— response systems are subject to habituation, a characteris— tic which, combined with wide phylogenetic occurrence, provides the opportunity for a comparative analysis across both stimulus-response systems within a single animal, and cross-species comparisons. A further advantage of habitua- tion as a comparative tool is that it is not restricted to any particular apparatus as are many other forms of behav— ioral change. A final advantage in investigating the process of habituation is that the behavioral characteris- tics of this process are parallel to the underlying neural correlates, as postulated by Thompson and Spencer (1966). Although there are many parametric characteristics involved in habituation (Thompson and Spencer, 1966) and ways of measuring the event (Askew, 1969), the present study focuses on the influence of the initial eliciting stimulus strength on the consequent habituation and reten- tion of habituation of the response. The organism which is to be used is the Siamese fighting fish, Betta splendens (Regan) and the behavior looked at is its species specific threat display. In the succeeding sections of the intro- duction, a review of the literature on stimulus intensity and its role in habituation will be presented, followed by a description of Betta splendens and some of the previous work done with that species. Finally, some of the past research on habituation of stimulus-response systems in fish is presented, ending with the statement of the prob- lem for the present study. Stimulus Intensity This section summarizes some of the past research done on the effect of stimulus intensity on reSponse dimen— sions such as initial response level, absolute amount of habituation, degree of habituation, and retention of habit- uation. Relevant data using fish as subjects is reserved for a later section. Miller and Murray (1966) reported on the stimulus conditions necessary to elicit an "immobility" response (pause in ongoing drinking behavior) in guinea pigs. In- dependent groups of animals were eXposed to either a 47 or 77 db. tone. It was found that the 77 db. group had a significantly higher initial response level and a greater absolute amount of decrement (initial minus terminal re- sponse level) when compared to the 47 db. group. However, the rate of habituation to a zero response level was not significantly different for the two groups. In a study by Martin and Melvin (1964), the habit- uation of the freezing response in bobwhite quail was in- vestigated using either a live red-tailed hawk or a life- sized silhouette as the eliciting stimulus. One stimulus PasSed over §_ twice a day for five days. Then the other stimulus was used in a similar manner for six more days. Half of the §s saw the silhouette first and the other half saw the live stimulus first. The results revealed a much higher duration of the freezing response when gs were ex- posed to the live stimulus. There was a larger absolute decrement in response for the "live" group as compared to the "silhouette" group. This particular measure of habit— uation, however, is complicated by the differences in the initial elicitation values of the two stimuli (i.e. the "live" group had a much greater potential decrement). Some measure taking into consideration the initial differences in response levels would have been more meaningful (propor- tion or percentage of decrement). Bobwhite quail were used in another study of habit- uation of the freezing response, this time a response suppression technique was used (Melvin and Cloar, 1969). The gs were first trained to key peck on a variable inter- val reinforcement schedule. When a stable response rate llad been reached, the gs were exposed to either a live vaainson's hawk or a live silver king pigeon, with the stimulus being held behind a glass partition. Each S was eXposed to one stimulus twice daily for four days, and tilen shifted to twice daily exposures of the other stimulus f<>r three more days. Using the duration of freezing and dilration of suppression of key pecking as the dependent variables, it was found that the live hawk stimulus elicited an initial freezing response of much longer duration and thus suppressed key pecking longer. In terms of the habit- uation of the freezing re3ponse a greater absolute amount of decrement was associated with the hawk stimulus. An in- teresting auxiliary finding was the lack of response re- covery (retention of habituation) to the presentation of the hawk, 18 days after the last experimental exposure. Kuenzer (1958) investigated the habituation of the withdrawal response in earthworms to different intensities of electric shock. He found that a greater initial response was associated with the stronger stimuli. Also, the stronger the stimulus intensity, the greater the absolute amount of habituation, and the slower the rate of decrement to a zero asymptotic response level. In addition, this study seems to be the only one that attempted to look at the retention of habituation. He found that after a 24 .hour rest period, recovery was complete for the weaker stimmli but recovery for the strong stimulation was only about 75 percent. Askew (1969) in a systematic study of the "head- Sllake" response of male rats (rapid rotation of the head about the front-to-rear axis) to various intensities of air that was blasted into the ear, provided some clear data on the influence of stimulus intensity on the habit- uation process. He found that the higher the intensity of the stimulus (air pressure), the greater the initial head— shake response level, the greater the absolute amount of habituation, the higher the asymptotic terminal response level, and the slower the rate of decrement. Stimulus in- tensity did not have a measurable effect upon retention of habituation after a 30-minute rest interval between sessions. Dunlop, Webster, and Rodger (1966), in a neurologi- cal study of habituation, examined amplitude changes in evoked potentials to auditory stimuli at the cochlear nu- cleus, inferior colliculus, and medial geniculate body of unanesthetized cats. At all three locations initial and terminal response levels were a direct function of stimulus intensity. A significant decrement in response resulted for all stimulus intensities. In the cochlear nucleus and inferior colliculus no significant relation was found be- tween stimulus intensity and amount and rate of habituation. .At the medial geniculate, however, the stronger auditory .stimuli produced a greater absolute amount of habituation Eilthough there did not appear to be any difference in the time taken for the response to reach asymptote. In a study measuring some components of the orient- illg'reflex, Uno and Grings (1965) used five intensities of 311 auditory stimulus to investigate the effects of stimulus irrtensity on habituation of skin conductance, skin poten- tixal, heart rate, digital blood volume and digital pulse - -.-.- F— "' -¢_—-—_-—.—e .h... gun-.— ._.. _..-- ‘ amplitude in human subjects. A within-subjects design was used, with each intensity value being given five times dur- ing the session. All of the dependent variables showed a different initial re3ponse level as a function of the dif- ferent intensities: the stronger the stimulus the greater the response. There was an indication of a ceiling effect at the higher intensities, however. Although response magnitude generally increased with stimulus intensity and decreased with repetition, the relation between stimulus intensity and response habituation was different for the various dependent measures. Some measures showed no in- tensity by habituation interactions, while for those that did, the relations varied. One general finding was that the higher intensity stimuli were associated with higher terminal response levels. However, there were exceptions to this relationship, too. From the above survey of studies on the effects of stimulus intensity (i.e. "adequacy" of the stimulus on lhabituation, it appears that the simple rules of Thompson and Spencer (1966) meet with many exceptions. It is quite Cllear that the initial response level is almost always greater for the higher stimulus intensities. Also, for lower asymptotic response levels above zero, higher in- tensity stimuli are usually associated with higher terminal asymptotes. As far as the absolute amount of habituation iJB concerned, one usually finds a positive relation with stinuxtus intensity, although instances of negative or no relation can be found in the current literature. Few, if any, systematic studies of the relations between stimulus intensity and the degree of retention and/or recovery of habituation have been attempted. Overall, it seems that different stimulus-response systems may have characteris— tic stimulus intensity-decrement relationships. More re- search with different species and stimulus-response systems is obviously needed. Experimental Animal The Siamese fighting fish, Betta splendens (Regan), is a member of the order Labyrinthici, family anabantidae. Each individual of this family possesses a labyrinth which consists of a pair of vascularized cavities located in the sides of the head above the gills. The fish supplements breathing through its gills, by gulping air from the water surface into these cavities (Smith, 1937; Forselius, 1957). Since Bettas normally inhabit shallow, stagnant ponds and <3anals with low oxygen content, the labyrinth serves as an eruxiliary organ for obtaining sufficient oxygen. The fish Huakes frequent excursions to the water surface, expels a 1bubble of vitiated air, takes in a new supply, and rapidly rsatreats toward the bottom. The bubble blowing behavior 1&3 particularly well evolved in the fish, since the species are bubble nest builders. Normally, this facet of reproductive behavior is performed by the male, although it has been reported that the female will also perform these functions (Forselius, 1957). After an elaborate courting sequence, the male and female squeeze each other under the bubble nest which has been constructed by the male. As the female drops her eggs and the male fertilizes them, he picks them up and blows them into the nest. At this point, the female is usually driven away from the nest area, since the male almost always cares for the young (see Forselius, 1957, for a complete elucidation of the reproductive cycle). In the limited space of the aquarium, the female may even be killed, if allowed to remain for too long a time (Gor- don and Axelrod, 1968). During the incubation period (30 to 40 hours), the male continually repairs the nest by replacing burst bub- bles. This replacement has two major functions: to increase the solidity of the mass of bubbles and to raise the eggs in the nest slightly closer to the surface where oxygen is Inore plentiful (Forselius, 1957). After hatching, the male <2an be seen incessantly retrieving and replacing any young that may have fallen out of the nest. After the free swim- nning.stage is reached (three or four days), the nest is abandoned . Eflbreat and Attack Behavior The characteristic that makes this species so well lenown, and an ideal preparation for the present investigation, 10 is its threat display and attack behavior (which combined will be termed fighting behavior). A conspecific male (Lissman, 1932; Smith, 1937; Adler and Hogan, 1963), a mirror-image (Baenninger, 1966; Simpson, 1968), a model of a male in display (Lissman, 1932; Thompson and Sturm, 1963), a two-dimensional cut-out of a male (Hess, 1952), or a live female Bgttg (Simpson, 1968) will all elicit the fighting sequence in a male Bgttg. Stimulus configura- tion is an important factor in the organization, frequency, and intensity of these behaviors. The behavioral components of the threat display, itself, include approach, extension of the gill covers (operculae) and brachiostegal membranes, erection of the medial fins, intense deepening of body and fin color, orientation movements, and undulatory movements. Attack behavior includes biting and nipping, jaw locking, pectoral fin beating, pelvic fin flickering, tail beating, and tail flashing (Simpson, 1968). Although most research has concerned itself with the fighting behavior of males, Braddock and Braddock (1955) have reported that aggressive behavior between females is common and very similar to that of males. Although fights between males have been reported 'to last for as long as 24 hours, they usually last less than an hour (Smith, 1937). Contrary to popular belief, Eflgttas rarely if ever fight to the death under natural circumstances. Sooner or later one contestant discontinues 11 the fight and swims away. The loser's color becomes pale as the retreat takes place, and the victor usually manages a final few nips at him (Forselius, 1957). In the process of fighting, the drape-like fins of both victor and van- quished may be seriously shredded and scales lost. How- ever, under prOper conditions the damaged fins and lost scales will regenerate in about 48 hours (Gordon and Axel- rod, 1968). As one can see, the richness of the fighting repertoire of ggttg splendens provides an excellent prep- aration for the investigation of aggressive behavior. Many researchers have already capitalized upon the opportunity. Thompson (1963) reported that male Bgttgs will learn an Operant swimming response if reinforced by the visual image of another male of the species. He found that the three types of visual stimuli he used (mirror-image, moving male model, stationary model) were of differential reinforcing value. The mirror image was the strongest reinforcer, in terms of acquisition rate, number of responses, and resis- tance to extinction. Compared on the same measures, the Inoving model was superior to the stationary model. In a £3imilar study of fighting cocks, mirror-image reinforce- Inent maintained a lower key-pecking response output, than another live rooster presented behind a window (Thompson, 3L954). Another variable important in the operant behavior (bf the male Betta is the color of the reinforcing stimulus. 12 Operant reSponse rate varies as a function of the color of the model with respect to the color of the subject, the response rate becoming higher as their colors become more dissimilar (Thompson and Sturm, 1965b; Thompson, 1966). Using additional controls in an attempt to replicate an earlier study of Thompson (1963), Goldstein (1967) taught male Bgttgs to swim through a ring in order to have an opportunity to display to their mirror-images. Hogan (1967) obtained similar results for the Operant swimming response, and also demonstrated that males learn to swim through a runway when a presentation of the mirror was made contingent upon the correct response. A comparison in the runway of performance for mirror display and for food revealed that food reinforcement produced significantly better performance. In an experiment by Thompson and Sturm (1965a), unconditioned threat display behavior elicited by the mir- ror image of a male Bgttg was brought under control of a previously, ineffective stimulus by classical conditioning. Relative rates of acquisition of four components of the display were compared. Fin erection and undulating move- Inents were acquired most rapidly, while gill cover erection iand frontal approach were acquired most slowly. A dis- <2riminative conditioning procedure, using two different <:olors of light as the CS, revealed that the responses were that artifacts of sensitization. Using electric shock as 13 the CS.and a mirror image as the UCS, Adler and Hogan (1963) were successful in conditioning the gill cover (operculae) erection. In further experimentation, the same investigators found that gill cover erection in the presence of a mirror image or another live male in an outside compartment could be suppressed by an electric shock contingent upon the elicitation of the response. Mild punishment of an operant swimming response reinforced by a mirror presentation has been shown to temporarily increase the rate of responding and the vigor of the con- sequent threat display in male Egttgs (Melvin and Anson, 1969). This effect may be explained within the pain- elicited aggression paradigm described by Ulrich (1966) for a wide variety of species. Many other methods have been utilized for increas- ing or decreasing the organization, frequency, and inten— sity of the components of the fighting sequence in Bgtta _§p1endens. Various drugs alter the aggressive patterns in 'this species. Evans §t_gl, (1956) reported changes in the ‘regetative, motor, and behavioral characteristics of Bgttas following the administration of LSD-25, eight other ergot. éierivatives, mescaline, and demerol. Although LSD-25 had ‘tflne most numerous effects, the other drugs had distinct effects on M's behavior. In a later study focusing on ‘tflle change of fighting behavior following LSD-25 administra- tion, Evans et a1. (1958) reported a marked increase in the 14 aggressiveness of the subjects. Whether the drug was given to all the individuals in an established social hierarchy or only to one of the lower.ranking members of the group, the results indicated that narcotized female Bgttas were more aggressive than controls. After LSD-25 treatment, low ranking individuals rose in the social hierarchy because of their increased aggressiveness. These reversals in social dominance seemed to be relatively long lasting. McDonald and Heimstra (1964) were not able to replicate this data since they could not even get the Bgttgs to form clearly defined dominance hierarchies. In nearly all instances, the fish that were placed in groups fought to the death before any stable hierarchies developed. The same investi- gators did succeed with another species (green sunfish). However, it was found that in four subject hierarchies and pairs of fish, LSD-25 increased the frequency of attack behavior by the submissive fish which were given the treat- lnent. In both hierarchical formations and pairs, the drug teffects were temporary and no changes in the dominance llierarchies were noticed. Rather, the overall aggressive- 11688 of the four member groups and the pairs increased. Marrone (1966) found spontaneous performance of £1111 cover erection, fin erection, and increased coloration liri isolated male ggttas when they were placed in solutions Chantaining norepinephrine bitartrate. Subjects placed in ear; epinephrine bitartrate solution resembled controls 15 (plain.water) except for color change; the epinephrine group bleached during observation periods. In another study concerned with the effect of externally admini- strated amines on threat behavior in the male Bgttg, Baenninger (1968) studied mirror elicited threat behavior instead of tg.yggtg_responding. Epinephrine and norepine- phrine suppressed display behavior, compared with controls in sodium chloride solution or plain water. Since these two investigators were interested in different conditions of display elicitation, it is difficult to interpret their contradictory data. In addition to the above mentioned means of manipulating fighting behavior in the ggttg, re- sponse changes occur with repeated presentation of the eliciting stimulus. In other words, the components of display are subject to habituation, as will be discussed in the following section. Habituation in Fish Until quite recently, little systematic experi- Inentation had been done on habituation in fish (Thorpe, 21963). Within the last few years, however, many workers Ilave demonstrated habituation involving various stimulus- Jresponse systems in fish. Rodgers §t_2t, (1963) found that the "tail-flip" response in goldfish, Carassiusauratus, habituated to Eiressure waves induced by repeated tapping on the side of 16 the aquarium. A decrement in response also occurred with the repeated passing of a shadow over the top of the test aquarium. Both stimuli presented together produced a higher rate of response and slower habituation than either one alone. Habituation of fear responses (freezing; "jerk response") in guppies, Lebistes reticulatus, has been shown also utilizing an overhead moving shadow as the eliciting stimulus (Russell, 1967). Fear responses de- creased in number and intensity during a series of 40 stimulus presentations. Forty-eight hours after testing, however, recovery was practically complete for both de- pendent variables. In an investigation of the habituation of aggres- sive behavior in male three-spined sticklebacks Gasterosteus aculeatus 9,, Van Den Assem and Van Der Molen (1969) used a novel method for measuring the effects of habituation. Stickleback males which were visually isolated gave more (or less constant aggressive responses toward a test stimu- llus (a live male stickleback in a plastic tube) when it vvas presented once or twice daily during a four day period. UPhe response to the test stimulus waned for a group of Erubjects that had a continuous View of a male in a neigh- l3<>zring territory behind a glass partition. Waning also <>e noted, however, that the absolute amount of recovery of Jresponsiveness between stimulus presentations was greater for the live male group (the stronger elicitor) . In another investigation of the same species, Peeke SEE_§$, (1969) focused their attention on the differential IIIIIIIIIIIl--::________ , r (V \leF ‘ 18 eliciting value of five moving wooden models of a male stickleback. The models were similar to each other except for different amounts of red color on the ventral anterior surface. Red in that anatomical locus is an important component of the eliciting stimulus for aggressive behavior in the stickleback. Each of the models was presented to each S for two minutes each day for 12 successive days. Using the frequency of biting as the measure of aggressive behavior, they found that the initial level of response did not differ for the different models. Pooling the data across models, revealed clear evidence for habituation over the 12 day period. With respect to work done specifically on habitua- tion in Betta splendens, only a few studies have appeared in available literature. Baenninger (1966) reported a waning tendency to approach a mirror or another live male EEEEE.WhiCh had habituated to its own mirror image. Each §_spent 32 consecutive hours in the test tank in two con- ciitions. In the control condition, neither the mirror nor 'the stimulus fish was present at the ends of the aquarium. (Over the 32 hour session, the fish in the experimental <:ondition showed a definite trend toward spending more time :in.the middle of the tank, out of sight of both stimuli. TPhe changing preference could not be considered avoidance (bf the ends of the apparatus, since time in the center of 1:he tank in the experimental condition was never significantly 19 greater than in the control condition. Although the two stimuli, at opposite ends of the test tank, elicited threat displays, all subjects spent more time viewing their own reflections than viewing the live, relatively unresponsive stimulus fish. The dependent variable in this study, how- ever, was the time spent firparticular portions of the tank, and not changes in the various components of the display itself. Clayton and Hinde (1968) exposed male EEEEES to a mirror for ten days. The mirror was then removed for re- covery periods of 15 minutes, 6 hours, 24 hours, 48 hours, and 4 days. Each §_was given all recovery period lengths, which were counterbalanced across subjects. At the end of each recovery period, the mirror was replaced for 48 hours and the display observed for the first 26 minutes. Various components of the display habituated over the course of time, although the process was an extremely slow one. Gill cover erection frequency (GCEF) had not disappeared com- pletely, even after the first 10 days with the mirror. Re- covery of this particular response after original waning ‘was slow, with some long term effects being evident. There ‘was a positive relation between the length of the recovery period and the amount of recovery for the response (GCEF) . 1E major difficulty with the study was that the data for three of the eight gs were thrown out and the data that ‘were presented were usually for one § used as "representa- tive" of the five other gs. 20 Both of these habituation studies had methodologi- cal problems. Baenninger used a very small sample size (§.= S), but more serious is the fact that water temperature was kept at 22.2°C. Hess (1952) and Lissmann (1932) have found that aggressive behavior in the EEEEE is positively related to water temperature, 22.2°C. being a low tempera- ture for this species. Baenninger's results may have been confounded by the fact that the waning of the approach was partially due to the low temperature. Clayton and Hinde also had problems with sample size as discussed above, and also kept water temperature quite low, with a rather wide range (24-27°C.) of variation. Statement of the Problem It seems apparent that qualitatively different stimuli (e.g. a live male stimulus, mirror-image, model, or two dimensional cut-out) elicit the complete threat display in male Egttg splendens, but to varying degrees. No previous investigation has looked at the comparative (eliciting values of a number of different stimuli in terms (bf the intensity of various components of the threat dis- };lay. Also, it is evident that the display itself is sub— Ziect to habituation, although a zero level is rarely Ineached. What interests the present investigator is trying tn: ascertain whether a relation exists between the strength CNE different eliciting stimuli and the intensity of various 21 components of the threat display and their consequent amount and proportion of habituation. Also of interest is whether the retention and recovery of habituation are affected by eliciting stimulus strength. Since there is no satisfactory way of ranking qualitatively different stimuli as to their "adequacy", other than looking at the initial response level, the ini— tial response level is used as the measure of stimulus adequacy in the present study. However, there is strong evidence for expecting live stimuli to elicit higher ini- tial response levels than models (Tinbergen, 1951; Martin and Melvin, 1964; Peeke, 1969). This study also seems to provide an opportunity to evaluate portions of the elicitation and interference theory of habituation presented by Ratner (1970). Accord- ing to this position, the degree of habituation and degree of retention of habituation is inversely related to the strength of the association between the original eliciting stimulus and its response (e.g. eliciting value of the stimulus). A direct extension of this theory is that the degree of recovery of habituation after a rest period is directly related to the eliciting value of the stimulus. In other words, stronger elicitors are assumed to be asso- ciated with a lesser degree of habituation and retention Of habituation, and therefore a greater degree of recovery of habituation. 22 A final purpose of the present study is to investi— gate the relations between the various measures of threat behavior (e.g. medial fin erection, gill cover erection) in male Betta splendens. METHOD Subjects Fifty adult male Egttg splendens were purchased from a commercial breeder and were brought to the laboratory in separate unmarked, sealed containers. The fish were un- packed in random order and placed into individual jars which were numbered from one to fifty. The first forty fish served as experimental gs, while the remaining ten fish served as extra subjects in case of deaths of any of the forty sub- jects. Since all fish were unpacked blindly, no bias was present in choosing the first forty subjects. There was some variability among the gs with regard to color, but this is common even within the same brood. The sizes of the animals used in the experment were very similar, ranging between 3.5 and 4.0 cm. from the tip of the snout to the base of the caudal fin. All fish were maintained in the laboratory for one month prior to testing. The fish were maintained and tested in a special research laboratory located on the third floor of the Natural Science building at Michigan State University. This room contained steam heating apparatus plus an auto- Inatic air conditioner which kept the room temperature rela— tively uniform (27-29°C). A fan kept the air in continuous 23 24 circulation. The three windows, facing north, were covered with venetian blinds in order to reduce the amount of natural light entering the laboratory which was lighted with sixteen 40-watt fluorescent bulbs. During the period of the present experiment an automatically controlled timer provided 12 hours of light (8:00 a.m.-8:00 p.m.), and 12 hours of darkness. The continuous operation of the filter- ing equipment in the laboratory provided a natural "white noise" effect which tended to mask most extraneous auditory stimuli. The experimental subjects and stimulus fish were housed on the top shelf along the south wall of the room. The shelf, itself, measured 302.26 cm. (length) by 81.28 cm. (depth). The fish were individually maintained in 4.23 liter grayish plastic salad dressing jars which were pre- sterilized with potassium permanganate crystals. Each transparent jar was sanded on the outside with a fine sand paper, making it translucent. The jars were placed about 2.54 cm. apart in rows of four jars per row. The subjects were in complete visual isolation, since one could neither see out of his home jar nor view his own reflections. Each jar was filled to about two-thirds capacity with conditioned tap water, having been passed through a commercial charcoal filtering apparatus. Upon arrival in the laboratory, each fish was placed into one of these jars which was treated with an auromycin solution to prevent bacterial invasion. 25 A 12.5 cm. square piece of heavy gauge galvanized steel was placed on the top of each jar to retard evaporation and protect the environment from dust. All waste products and uneaten food were siphoned out weekly, and the jars filled to two-thirds capacity again. The fish were fed daily with fresh brine shrimp which were cultured in an adjacent room in the laboratory. The brine shrimp were fed to the fish via a medicine dropper, each getting about 0.5 c.c. of highly concentrated food. Apparatus The study was conducted in a 106.68 cm. long by 76.2 cm. deep cubicle located on the right side of the bot- tom shelf immediately below the shelf upon which the jars were placed. The height of the experimental cubicle was 54.61 cm. A 1.27 cm. thick sheet of flat gray enameled plywood provided the base of the cubicle. The fish were tested in a plexiglas aquarium sur- rounded by a rectangular wooden amphitheatre which was also painted flat gray. The test aquarium had outside measurements of 24.13 cm. (length) by 12.7 cm. (width) by 20.96 cm. (height) (See Figure 1). This clear .64 cm. thick plexiglas test aquarium was partitioned off into two compartments whose inside measurements were 19.69 cm. long and 2.54 cm. long. The larger area served as the subject compartment while the smaller area enclosed the various 20. 96 cm. Figure 1. Test aquarium. stimuli used in the experiment. The partition also was constructed of .64 cm. thick, clear plexiglas. All seams were first joined together with a plastic cement and then made water tight with Dow "silastic" cement. In other words, only visual cues were passed between the two com- partments. A vertical notch .32 cm. wide ran down each side of the stimulus compartment. These notches were machined .32 cm. behind the plexiglas partition and accom- modated two of the stimulus conditions in the experiment. A 17.78 cm. long by 12.07 cm. wide by .32 cm. thick one way mirror could be slipped into the notches thereby making the far wall into a reflecting mirror. The other stimulus condition which utilized these notches was a piece of translucent plexiglas which had a cut-out of a male EEEEE secured to it. The dimensions of this plexiglas sheet were identical to that of the mirror. The bottom and three sides of the aquarium were sanded on the inside to reduce reflections and ambient stimuli. The test aquarium was placed within the four sided viewing amphitheatre, with the long transparent side of the aquarium facing the experimenter. The outside length of the enclosure was 36.83 cm., the width was 30.48 cm., and the height was 23.5 cm. There was no top or bottom and the entire enclosure was painted in a flat gray enamel to minimize reflections. The test aquarium was viewed through a 7.62 cm. long by 5.08 cm. high slot cut into the long 28 side of the enclosure nearest the experimenter. A piece of blue transparent plastic sheeting was secured over the viewing slot, thereby lessening the opportunity of the fish seeing the experimenter. Illumination was provided by a gooseneck lamp with a white 25-watt soft glow bulb. During testing, the illumination at the top of the test aquarium (within the viewing amphitheatre) was approximately 160 ftc. Outside the amphitheatre at the viewing slot, the illumina- tion was only about 6 ftc. The combination of the lighting differential between the inside and outside of the amphi- theatre, and the blue plastic over the viewing slot pre- cluded the chance of a subject seeing the experimenter, while the subject could still be easily observed. Water for the test aquarium was changed for each test session. It was supplied from a tank immediately to the left of the experimental cubicle. This supply source had a capacity of 42.3 liters and was filled with condi- tioned tap water and was equipped with a thermometer. Throughout the entire experiment, water temperature was maintained at 25.6-27.2°C. The recording apparatus consisted of a model 92 Rustrak four channel event recorder which pulled 300 inches of recorder paper across the pens every hour. Each pen was exclusively activated by an individual standard door- bell button embedded in a wooden holder. Each of the but- tons was also wired to the 24 VDC power source which 29 activated the recording pens. Both the event recorder and the power source were connected to a model 171 Universal timer which automatically controlled the duration and termination of the experimental sessions. Procedure The forty experimental fish were randomly assigned to one of the four independent stimulus groups. There were ten subjects assigned to each of the following groups: A. Each g in group A was exposed to a standard reflecting one way mirror which was described in the pre- vious section. B. Each g in group B was exposed to a live, 3.7 cm. long red male EEEEE which was in visual isolation for one month prior to testing. C. Each g in group C was exposed to a live male E2222 which was as similar as possible to stimulus B ex- cept that it was continuously exposed to other live males for approximately six weeks prior to testing. In all other visible aspects, it was almost identical to stimulus B. D. For each S in group D the elicitor was a two dimensional cut-out of a 3.7 cm. red male Egttg which was secured to a piece of gray translucent plexiglas having the exact dimensions as the mirror (stimulus A). The cut- out was about the same size as the other stimuli and subjects. The figure, itself, was a lateral view of a male in full display (medial fins and gill covers erected).‘ The entire stimulus configuration slid into the notches in the test aquarium in exactly the same way as did the mirror. The unhabituated stimulus males (three different, ones were used) were kept in translucent jars as were the subjects. Whenever this stimulus group was used, one of the three group B stimulus males was randomly selected. When the next session of this stimulus condition arose, the stimulus fish for it was randomly selected from the is} two unused stimulus fish. For succeeding sessions the stimulus male which had not been used for the longest period of time was chosen. For a particular subject, the same stimulus fish was utilized for two 40 minute sessions, spaced 24 hours apart. Immediately upon the end of a ses- sion, the stimulus fish was put back into isolation. Group C habituated stimulus males (three different ones were used) were housed separately in transparent battery jars that measured 13.7 cm. (length) by 7.1 cm. (width) by 14.5 cm. (heighth). Other than the 15 minutes it took to transport them to the laboratory, they had been maintained for over a month within full view of at least one other live male EEEEE° These stimulus males were chosen for a particular session in exactly the same manner as the unhabituated (Group B) stimulus males, and were as closely matched as possible to each other and the unhabituated 31 stimulus males in size, length, and color. It should be noted that the habituated stimulus males were always in full view of one another, since the battery jars were adjacent to one another with the long sides touching. No control group utilizing an empty stimulus com- Ipartment was used, since it XEEEE threat displays were non-existent in pilot work and in available literature. Other than the choice of the eliciting stimulus, the ex— perimental procedure was identical for every subject in the experiment. Each subject was placed in the test aquarium which was filled to 15.24 cm. in depth and covered by a .64 cm. translucent plexiglas sheet. The g was allowed to habituate to the surroundings for 10 minutes while an opaque sheet of gray plastic separated him from the stimulus which also had been placed in the other compartment. Observations were made during this period to detect any tpryggpg dis- playing. The gooseneck lamp remained on throughout this acclimation period and the duration of the session. At the end of the acclimation period and when S was facing the partition, the experimenter lifted the opaque screen, re- vealing the stimulus compartment to the fish. At the same time, the timer was activated which automatically started the event recorder. Each session lasted 40 minutes with observations recorded for ten minute periods during the first, middle, and last ten minutes of the sessions. At 32 the end of the session, the recording equipment was auto- matically turned off by the timing circuit and the subject was netted and put back into his home jar. Netting the fish was always accomplished in less than two seconds. Then the entire aquarium was drained and flushed out with conditioned water. Water from the adjacent holding tank was then siphoned into the stimulus tank to a depth of 15.24 cm. For conditions involving live stimuli, the stimulus compartment was filled to an identical depth. No water was present in the stimulus compartment for the mir- ror or the cut-out conditions. Water temperature was re- corded before and after each session. Each fish was given two 40 minute sessions spaced by 24 hours. Testing took place from October 6 to October 30, 1969. Sessions were conducted Monday through Saturday at 8:30-9:20 a.m., 9:50- 10:30 a.m., 11:00-11:40 a.m., and 12:10-12:50 p.m. The order of subject testing was randomized across groups prior to the experiment. As described above, 3 ten minute recordings, separated by 5 minutes were taken during each of the two sessions. Each g therefore was observed for 6 ten minute periods. During these observation periods, five components of the subject's threat display were simultaneously recorded: 1. Total gill cover erection duration (GD): total amount of time that the gill covers were erected during the 10 minute observation period. 33 2. Gill cover erection frequency (GF): number of times the gill covers were erected during the 10 minute period. 3. Total medial fin erection duration (FD): total amount of time that the medial fins were erected during the 10 minute period. 4. Medial fin erection frequency (FF): number of times the medial fins were erected during the 10 minute period. 5. Air gulping frequency (AG): number of times the fish came to surface to gulp air during the ten minute period. AG was chosen as a measure because it is a purported activity level indicator (Simpson, 1968; Clayton and Hinde, 1968). This response was easily discernible and occurred in an all—or-none manner. The frequency and duration of a particular response were both recorded on one channel of the recorder by holding the push button down for as long as the response occurred. The operational definition of gill cover erection was taken from the extensive work of Simpson (1968): Gill cover erection is a sudden increase in the dis- tance between the distal edge of the operculum and the body, regardless of how far erect the operculum was to start with. The end of a gill raising bout is marked by a sudden decrease in this distance, even if the opercula are not completely closed thereby. In practice, such changes are clear—cut, although differ- ent individuals have different degrees of opening for 34 what are here recognized as "open" and "closed" posi- tions. When the gill covers are erected the brachio- stegal membranes are usually protruded. Different degrees of protrusion were not distinguished (p. 16). This response has been mentioned in nearly all descriptions of the Betta's threat and courtship displays. Some investi- gators use the frequency (GF) measure (Clayton and Hinde, 1968; Laudien, 1965), while others use some form of the duration (GD) measure (Adler and Hogan, 1963; Simpson, 1968). For the purposes of this study both frequency and duration measures were utilized in order to be able to evaluate the relationship between them. The operational definition of medial fin erection was also taken from the work of Simpson (1968): As a fish turns toward a display—eliciting stimulus for the first time, the vertical fins are spread . . . The dorsal is the last fin to be erected fully, prob— ably because that is the fin that has to be lifted furthest against gravity. The criterion for fins spread is thus an erect dorsal. However, when a dis— playing fish approaches another one, the first few dorsal rays may be slightly lowered, but the remaining ones remain rigidly erect (p. 14). Although this response has not been looked at as closely in the literature as gill cover erection, it is an integral part of the display and easily recorded. Previous investi— gators have broken this response into finer components (e.g. tail flashing, carouseling) but for the purposes of the present study, the above measurement was taken (Simpson, 1968). 35 Although there are other finer—grained components to the threat display, the above ones were chosen for their discreteness, wide usage in previous studies, reli- ability of elicitation, and accuracy with which they could be recorded. RESULTS The data analysis consisted of scoring each ten- minute observation period for each g in terms of each of the five dependent variables. Therefore, each g had six scores for each dependent variable (i.e. a total of 30 scores for each S). Then the data were analyzed using a three—factor multivariate analysis of variance and a consequent three- factor repeated measures (Case I) univariate analysis of variance (Winer, 1962) for each dependent variable, sepa- rately. The data analyses could have been undertaken using only the five separate repeated measures three-factor analyses of variance, but the true overall alpha level would have been higher than that specified for each test. When using univariate tests, a single probability statement ap- plicable to all variables jointly cannot in general be ob- tained from separate Es (Bock and Haggard, 1968): Because all . . . variables have been obtained from the same subjects, they are correlated in some arbi- trary and unknown manner, and the separate F-tests are not statistically independent. No exact probability that at least one of them will exceed some critical level on the null hypothesis can be calculated. The multivariate tests, on the other hand, are based on sample statistics which take into account the correla— tions between variables and have known exact sampling distributions from which the required probabilities can be obtained (p. 102). 36 37 Preliminary inspection showed that the assumptions of homogeneity of dispersion and normality were not ap- preciably violated. Also, since an N of 10 is of moderate size and since all groups in the study had equal Ns, mild violations of the above assumptions would not affect the true probability of making a type I error (Boneau, 1960; Hays, 1963). A level of significance of .05 was used throughout, with respect to both multivariate and univari- ate effects. If, in any case, the multivariate overall, with respect to a particular effect does not reach signifi— cance, the usual procedure is to dispense with the analysis of each dependent variable, separately. This procedure was followed in the present study. The general organization of the results section will involve a brief discussion of the multivariate re— sults and then a more complete discussion of each depend- ent variable, separately. Then the relation among the various dependent variables will be discussed. Finally, a further discussion of some selected dependent measures will be undertaken. Since the analyses become quite in- volved, a general summary will follow each section. To begin with, it is quite clear (see Table 1) that the multivariate F for each and every effect reached significance. Therefore, a true overall alpha level of at least .05 was attained for all main effects and the various interactions. Since the prime purpose for using coammom mo. .m.z so. .m.z .m.z Ho. mm x ma.~ m~.H mm.m so.a ao.~ mm.H In uoanwm mafia os\m osxm chxm osxm oexm mm.ms\om an x masseHum mo. .m.z Ho. mo. .m.z Ho. vm sonmwm mv.e am. m~.m mm.¢ ms. Hm.m um x mmxm mmxm mmxm mmkm mm\~ h~\oa mm\ woflumm mafia Owoz Imoz Omoz mo. Omoz Ho. vm gounmwwm mm.~ mm. am.s m~.m mm.~ sm.~ um x wmxm mm\m mm\m mmxm mm\m vh.mm\ma we masseHum Ho. Ho. Hooo. .m.z .m.z Hoe. vm mm.sa mm.m Nv.as mm.m mm.m Hm.m 1m wmxH mmxs mm\H mm\s mm\H mmxm mm aoammmm Ho. Hoe. Ho. Ho. .m.z mo. vm pofluwm mafia mv.m H¢.s sm.m os.¢ mH.H Hm.H It x otxo oa\m os\m oa\m oaxm mm.ms\om we maHaeHum w Hooo. .m.z Hooo. .m.z .m.z Hoe. vm am.vH a~.~ as.ma mm.~ Nw.~ mm.s It mmxm mmxm mm\~ mm\~ mm\~ Faxes we wosuwm mass Hoe. Hooo. Hoe. Hooo. mo. so. vm oo.a mm.os mm.s mm.oa mH.m m~.~ um mmxm Gm\m om\m om\m mmxm sa.mm\ma «a masseHum HHMHO>O as am no mo om wumaum>auass powwow momhamc< mumflHm>HCD .Aomo aoflpmuao coauomum cum mam .Ammv mocwswmum coauomum cam .Aowv soflumusa sofluomum Hm>ou HHflw .Amwv soap locum Hm>oo HHHU .Awdv aoswcwmum msflmasw Hfid "wHQMHHm> ucmwcwmwd nomm How mocmfium> mo momMHmcd sunfium>flsb can mflmhamsd museum>fluasz mo 0 . m Home mumsssmuu.a «Home 39 the multivariate analysis was to ensure the accuracy of the overall alpha levels, the presentation of results will move immediately into the treatment of each dependent variable, separately. Air Gulping Frequency (AG) From an inspection of the AG variable in Table l and Figure 2 which show the mean frequency per ten-minute observation for each stimulus group, it can be seen that only the Stimulus effect reached significance (F = 3.15, dt = 3/36, p < .05). Across both sessions, the different threat stimuli elicited differential air gulping frequencies. Figure 2 shows that the unhabituated male stimulus (Unhab. male) was the strongest elicitor, followed by the mirror stimulus (Mirror), habituated male stimulus (Hab. male) and cut-out stimulus (Cut-out), respectively. A Newman— Keuls paired comparisons test uncovered no .05 differences between pairs of treatment means, although the Unhab. male- Cut-out and Unhab. male-Hab. male pairs approached significance. Earlier in the data analysis, a multivariate analy— Sis for the vector of all dependent variables at T1 (first ten-minute observation) was undertaken. It proved to be highly significant (3 = 2.33, g = 15/88.74, P. «.01). A One-way analysis of variance was then performed on the AG data at T1. Again the effect was significant (F — 4.76, MEAN AG 40 Session I Session 2 30- 25L- 20L .Unhab. male OMirror AHab.male '5_ ACut—out NH“ I é: 5,h- 0;. . . . . L TI 72 134714 T5 T6 OBSERVATIONS Figure 2. Mean air gulping frequency per ten minute observation period. 41 Q: = 3/36, p < .01). Using the Newman-Keuls test, it was found that the Unhab. male group had a significantly higher frequency of air gulping during the first observation (T1) than did the Mirror, Hab. male or Cut—out groups. However, the other stimulus groups did not differ from one another (all Newman-Keuls reported at the .05 level). No Time Period effect (habituation) nor Session effect was evident with respect to this particular dependent variable. Also, none of the interactions attained significance. Summary of AG data analysis In general, air gulping frequency was strongly in- fluenced by the differential eliciting stimuli. At both Tl (initial observation) and generally throughout the dura- tion of the experiment, the Unhab. male group was the strongest elicitor, followed by the Mirror, Hab. male and Cut-out groups, in that order. No systematic habituation of the air-gulping response was evident during the present (experiment. Also, no significant frequency differences ivere present between the two 40 minute sessions. Gill Cover Erection Frequency (GF) Table l and Figure 3 provide an overall picture of thka GF data. There was a highly significant Stimulus ef- fect (g = 10.95, g = 3/36, E < .0001). The Newman-Keuls 42 6 Session I Session 2 5 .U.nhab male OMirror . __ AHab. male ACut— out TI T2 T3 T4 T5 T6 OBSERVATIONS A o I MEAN GF 00 o h) C) (éfg/SD F? ) Figure 3. Mean gill cover erection frequency per ten minute observation period. 43 test showed that the Unhab. male group had a significantly higher GF during the experiment than the other three groups (which did not differ from one another). Notice from Figure 3 that the order of strength of eliciting stimuli is again Unhab. male, Mirror, Hab. male and Cut-out groups, in that order. A simple analysis of variance of the GF data at T1 was highly significant (F = 5.81, dt = 3/36, p < .01). The Unhab. male group had a significantly higher GF than the Cut-out group. The Mirror group was also significantly greater in response than the Cut-out group, as shown by the Newman-Keuls test. The other paired comparisons were not significant. The Time Period effect was not significant for the GF data; therefore, no systematic decrement occurred when stimulus groups were pooled. However, the Stimulus x Time Period interaction was significant (F = 4.10, dt = 6/70, p'< .01), reflecting an increase in GF across time for the Unhab. male group and a decrement in response for the other three stimulus groups. Although there was no difference laetween sessions in GF level when treatments were pooled, a. Stimulus x Session interaction (3 = 3.26, dt = 3/36, E: <1 .01) reflected an increase in session 2 responding for tflle Unhab. male group while the other three groups showed a unopcomod How mu GOmHmwmil.N canoe 56 were no GP or GD responses for the Cut-out group during session 2. Reliability Product-moment correlations comparing the same de- pendent variable across sessions showed a high level of reliability for every measure. Although the correlations reported are derived from averaging across the stimulus groups, the reliability levels almost uniformly hold for each stimulus group separately (see Appendix). For those correlations based upon the combined groups, an 5 equal to or greater than .312 is significant at the .05 level for the first correlation used. An 5 equal to or greater than .404 is significant at the .01 level. Those measures (AG, FF, FD) based upon all four groups will be discussed first. ‘The session l-session 2 correlation for the AG variable was .718, reflecting a high degree of consistency of re- sponse across the two sessions. The E for the FF variable was .888. The FD measure was also extremely reliable across the two sessions (E = .932). These three dependent vari- ables attained ts far beyond the .01 level of significance. Since there were no GF or GD responses for the Cut— out group, the following ts are based upon the three re- maining groups (E = 30). With this sample size, ts equal to or greater than .361 are significant at the .05 level for the first correlation used, while ts equal to or greater 57 than .463 are significant at the .01 level. The E for the GF data was .890, while the GD measure had a session 1— session 2 E of .926. Both of these measures are signifi- cant far beyond the .01 level. In summary, every dependent variable proved to be highly reliable, reflecting both the ease and accuracy of recording and the relative constancy of the experimental situation from session to session. Interrelationshi s amon dependent variaEIes The following discussion is again based upon the data averaged across stimulus groups. However, as before, these general relationships hold up well when treating each stimulus group independently. First of all, it may be seen from Table 2 that there are significant positive intercor- relations among all of the following four dependent vari- ables: GF, GD, FF and FD. For example, the correlation between the FFl measure and GFl, GF2, GDl, GD2, FDl and FD2 ranged from ts of .486 to .910. The GDl measure had a high degree of relationship with those variables mentioned above, as witnessed by correlations from Es = .866 to .954. The point to be made here is that the four threat components used in the present study have high positive interrelation- ships with one another. Looking again at Table 2, it is apparent that the IXG variable is not significantly correlated with any of the 58 other four dependent variables in either session. The in- tercorrelations ranged from ts of -.014 to .288, all being non-significant values. Very few of the values even reached an r of .20. Therefore, it was shown in a quantitative fashion that air gulping is not an integral part of the threat display, itself. This finding supports the conten- tions in previous sections. Further discussion will be deferred until the next section. Further Analyses of GD and FD Data Only two of the dependent variables, GD (Figure 4) and FD (Figure 6) showed systematic habituation for every stimulus condition. The other three variables reflected either no decrement at all (AG) or increments for certain treatment groups and decrements or no time changes at all for other groups (FF and GF). The probable reasons for these differences will be discussed later. The following discussion will focus solely on the GD and FD variables, for the reasons listed above. In order to investigate further the relationship between initial eliciting stimulus strength and the degree of habituation,retention,and recovery of habituation the following analyses were performed for the GD and FD data. The procedure was identical for each measure. Proportions Tvere used in this phase of the analysis in order to equate 'the initial response levels for the different stimulus 59 groups in some manner. It was already shown that the Unhab. male group had the highest initial level (T1) of response followed by the Mirror, Hab. male and Cut-out groups, re- spectively. It was also shown that there was a positive relationship between the initial level of response (elicit- ing stimulus strength) and the absolute amount of decrement. The above relationships held for both the GD and FD data. An attempt was made, via proportions, to see if initial eliciting stimulus strength was related to the proportion of decrement, and retention and recovery of this decrement. Only §s which responded during the appropriate sessions were used in the following analyses. There were four fish in the Cut-out group, nine in the Hab. male group and ten in both the Mirror and Unhab. male group which responded during session 1. A chi—square analysis of this raw frequency data was highly significant (g? = 17.10, gt = 3, p < .01). During session 2, the number of Es which responded in each group were as follows: Unhab. male, nine; Mirror, six; Hab. male, five; and Cut-out, one. A chi-square analysis of this raw frequency data was also highly significant (£2 = 13.10, d_f = 3, 2 < .01). Notice that there is a direct relationship between elicit- ing stimulus strength and the number of gs which showed threat response: the Unhab. male group had the largest number responding in session 2, followed by the Mirror, Hab. male and Cut-out groups, respectively. A chi-square 60 analysis with the data pooled across sessions was not attempted, since there was no E which responded in session 2 that did not respond in session 1. In other words, the data for the pooled sessions was identical to that of ses- sion 1. Proportional analyses of GD data A session 1 decrement proportion was calculated for each S in each group using the formula -lfir——;1 (first ob- _ l servation minus the third observation divided by the first observation). The data was analyzed with a Kruskal-Wallis analysis of variance by ranks, because of unequal cell sizes, small E, and violations of the assumptions for the parametric analysis of variance. The analysis did not reach the .05 level of significance: the four stimulus groups did not differ in the proportion of decrement during session 1 (see Table 3). A three-group analysis, without the Cut—out group also failed to reach significance. Since the effect of initial eliciting stimulus strength on the degree of habituation might be made appar- ent with longer stimulus exposure, a proportion of across session decrement measure was calculated for each E using the formula Ei¥:;3§ . This time the Kruskal-Wallis test was significant (E = 10.44, gt = 3, p < .05). Notice in Table 3 that the mean group rank is the largest (greatest decrement) for the Cut-out group, followed by the Hab. male, Iflvlo'iirn- I-lrl Iliii . I . I II . II II lii- TTIII i n: (Fr—flu... IW6+HM>IHMCOH0M0QOHA DU L:C.u IIH .. mud .mHthmsm one Eouw pwuoHop mmz msoum psolusoa *3; SJ mo.v .m.z m om.NH V cm nm e mOH om.vN v mm mm.>m v om.mOH usoluso O O O I mHmE mm.MH m mNH HH om m HmH mm Hm m om mmH mm SH m mmH .Qmm OH.NH 0H HNH 0H.wH 0H HmH on.mH 0H an mm.mH OH om.mmH HOHHHZ a . o . o i . . . a 2...... om mm 0H mmm H m H m mm OH OH om moH or NH OH SN .omsso xcmm I mxsmm xcwm I mxsmm scam I mxcmm Room I mxswm msonw msouw 2 mo mdouw 2 mo QDOHD 2 mo mdouo 2 mo msHsEaDm coo: Edm and: Sam cmoz 85m coo: Sam . >Hw>ooom coducopom mconwom mmouos H conmom seauuomoum coawuomoum "usofiwnooo "ucofiouooo . . coHDHomoum coHuuomoum moHanum> HmcoHpnomoum Q0 .msouw desEHum comm now mxcmm com: com wxcmm mo Sam HA>Mm>ooom mo coHuHomoum pom .QOHucouwm mo COHuuomoum .ucoE Imeomo mGOHmmmm mmouod mo GOHuMomonm .H sonmom cH ucofiouoob mo coHuuomoumv moHQoHHm> HmcoHDuomoum Q0 boom How wHome mumEEdm umme m mHHHmzimemsuM||.m oHome 62 Mirror and Unhab. male groups, respectively. Post-hoc paired comparisons using the Mann-Whitney E test follow- ing the method of Lewis and Cotton (1958), and Peeke (l969),showed that the Unhab. male group had a significantly smaller proportion of across session (T1 — T6) decrement when compared to either the Hab. male or Cut-out groups. The other pairs were not significant at the .05 level, al- though the general trends show that the stronger the eliciting stimulus strength, the smaller the proportion of session 1 and across session decrement. Next, a retention score was calculated for each E using the formula TlT- T4. A Kruskal-Wallis test was highly significant (E = 12.85, gt = 3, p < .01). Post-hoc Mann-Whitney E tests showed that the Unhab. male group had a significantly smaller proportion of retention than either the Mirror, Hab. male or Cut—out groups. Table 3 shows that the mean group rank (proportion of retention) was smallest for the Unhab. male group, followed by the Mirror, Hab. male and Cut-out groups, respectively. The relation between the initial eliciting strength of the stimuli and the proportion of retention of habituation after a 24 hour rest period is thus a monotone decreasing one. Finally, to investigate the proportion of recovery from the last observation of session 1 (T3) to the first observation of session 2 (T4), the following proportion T _ 4____ T1'T3 was calculated for each E: This proportion gives I!“ III IIIII , 63 a rough ratio of the amount of recovery (T4 - T3) to the amount of decrement during session 1 (T1 — T3). Due to the small, unrepresentative sample for the Cut-out group (E = 4), this group was deleted from this part of the analysis. The three-group Kruskal-Wallis test reached significance (E = 6.33, gt = 2, p < .05). The mean group rank (proportion of recovery) was the largest for the Unhab. male group followed by the Hab. male, Cut-out and Mirror groups, respectively. However, the post-hoc Mann- Whitney E tests showed that the only significantly differ- ent pairs were in the expected direction: the Unhab. male group had a significantly higher proportion of recovery than either the Mirror or Hab. male groups. Summary of proportional analyses of GD data Although conclusive data are not available due to the sample size of the Cut-out group (E = 4) in these analyses, there were some general relationships that were uncovered. The proportion of across session decrement was related to initial eliciting stimulus strength in a monotone- decreasing manner: the stronger the elicitor, the smaller the proportion of decrement. The proportion of decrement data for session 1 did not reach significance, although the trend was in the same direction. The proportion of reten- tion of habituation after a 24 hour rest period was also strongly related to initial eliciting stimulus strength in 64 a monotone-decreasing manner: the stronger the elicitor, the smaller the proportion of retention. Proportion of recovery of habituation after a 24 hour rest period was related to initial eliciting stimulus strength in a monotone-increasing manner. Proportional analyses of FD data The analyses of the FD measures were dealt with in exactly the same way as the GD analyses above. The same calculations, alpha levels, and Mann-Whitney post-hoc com- parison procedures were used throughout. The Kruskal- Wallis summary data is presented in Table 4. A proportion of session 1 decrement measure was _ii Tl ' The Kruskal-Wallis test did not reach significance for calculated for each E, again using the formula either the four-group analysis or when the Cut-out group (E = 2) was deleted (three-group analysis). An across session decrement proportion was then calculated for each E using the formula E$T:—E§ . The Kruskal-Wallis test was significant (E = 9.42, gt = 3, p < .05). Since there were only two Es in the Cut-out group which could be used in this analysis, little conclu- sive data can be obtained for this particular stimulus group. In general, it can be seen that the Unhab. male group had the smallest mean rank (proportion of across session decrement), followed by the Mirror group. Although .mHthmso on» EOHM oouoHoU mos mooum uselpsos :mo.v mo.v mo.v .m.z m «H N ON ON N ow OH N mm mN.ON N om.O¢ uncluso E o: m om mN.oN m was om.Hm m was m~.sa m mm: wmmm O¢.NH OH vNH O0.0H OH OOH OH OH OOH mH.hH OH om.HhH HouuH: 5 6 O O C . ”HME om ON OH mON ov O OH em on O OH hm om HH OH mHH .omoco scam I meow: scam I mxcmm xsom I mxsmm scam I mxcmm msouw macho 2 mo moouo Z we macaw 2 mo muonw 2 m0 m5 sad coo: sow coo: Sow coo: Edm coo: Edm H .um huo>ooom coaucouom mconmom wooed: H :OHmmom . "usosouooo uucofiouooo GOHuHomoum coHuHomoum COHuHomoum QOHuHomoum moHooHHm> HoquuHomoum om .msouw manEHum comm How oxen: coo: can mxsmm «o How "Ahuo>ooom mo soHuuomoum can .soHucouom mo cOHuHomoum .ucofi Iouooo chHmmom mmouod mo coHuHomoum .H sonmom cH ucofiouooo mo coHuHOQoumO moHQMHHM> HocoHuHomoum am noon Mom oHooB MHMEESm umoa m mHHngleu—msuVH .v mHnm " B 66 the Hab. male group had a larger mean group rank than the Cut-out group, the difference is negligible and the Cut- out data was based on an E of 2. The Mann-Whitney E tests showed that only one pair reached significance. The Unhab. male group had a significantly smaller proportion of across session decrement as compared to the Hab. male group. Then a proportion of retention was calculated for T -T l 4 T O Wallis test was significant (E = 8.57, gt = 3, p < .05). each E, again using the formula The Kruskal— The Mann—Whitney E tests showed that the Unhab. male group retained significantly less than either the Mirror or Hab. male group. The other pairs did not reach significance. Any comparison with the Cut-out group was almost impossible to make due to the small sample size. Finally, the proportion of recovery measure was -T 3 T - T ' 1 3 A three-group Kruskal-Wallis test was performed. The Cut- calculated for each E, again using the formula out group data was discarded because of the small, unrepre- sentative sample (E = 2). The three-group analysis reached significance (E = 6.17, gt = 2, p < .05). The Mann-Whitney E procedure on the remaining three groups showed that the Unhab. male group had a significantly higher proportion of recovery than either the Mirror or Hab. male groups. This trend is clearly shown in Table 4. B“ ._ 1. . 67 Summary of proportional analyses of FD data Again this part of the analysis suffered from the small sample size of the Cut-out group (E = 2). This particular stimulus was an overall poor elicitor. However, from the three remaining groups it is quite clear that eliciting stimulus strength strongly affects the degree of habituation, retention and recovery of habituation. There was a monotone-decreasing relation between initial elicit- ing stimulus strength and the proportion of across session decrement and the proportion of retention of habituation. The stronger the elicitor the less the decrement, and the poorer the retention of the decrement. There was a general monotone-increasing relation between initial eliciting stimulus strength and the proportion of recovery of habit- uation after a 24 hour rest period. The stronger the elicitor, the larger the proportion of recovery. Of course, the retention and recovery measures are strongly related in a negative fashion. It is interesting to note that the above relations held up quite well for both the GD and FD analyses. The Unhab. male stimulus elicited the greatest initial level of response (eliciting stimulus strength) followed by the Mirror, Hab. male and Cut-out groups, in that order. This order of the strength of the elicitors was consistent for all investigated dependent variables. Finally, initial 68 eliciting stimulus strength was found to be strongly re- lated to the absolute amount of habituation, the proportion of across session habituation, the proportion of retention of habituation and the proportion of recovery of habituation. DISCUSSION The present study provided strong evidence that components of the threat display of male Egttg splendens are subject to habituation and that the degree of habitua- tion and retention of habituation are strongly influenced by the initial strength of the eliciting stimulus. Eliciting Stimulus Strength For all five dependent variables used in the study, the ordinal strengths of the four elicitors were the same. The group exposed to the Unhab. male stimulus responded to the greatest degree, followed by the Mirror, Hab. male and Cut-out groups, respectively. Why this particular ordinal relationship occurred may be explained by means of past literature and some speculation. That a live stimulus is a better elicitor than a stationary one (Cut-out) has been quite firmly established (Tinbergen, 1951; Martin and Mel- vin, 1964; Peeke, 1969). With respect to the Unhab. male, Mirror and Hab. male stimuli, the following argument is pr0posed. The Unhab. male stimulus provided movement and relatively prolonged, positive feedback to E, since the responding of both stimulus and subject in this particular group was quite consistent throughout the experimental 69 70 sessions. Although the Hab. male stimulus typically dis— played and vigorously interacted with the subject for the first few minutes, the threat display and attention directed toward the subject waned extremely rapidly in this particu- lar stimulus. Even though no quantitative measurements were taken, the differential responding of the Unhab. male F“ and Hab. male stimuli was extremely evident beyond any L question. Most Unhab. male stimuli were still vigorously responding to the subjects even at the end of the second session. The differences between the two live male stimuli (Unhab.-Hab.) therefore vary along a feedback dimension, including the intensity of the threat display, changes in body color, movement, etc. The Hab. male stimulus was, however, mobile, three dimensional and somewhat variable in its movements which made it a better elicitor than the two dimensional Cut-out stimulus. The most difficult difference to explain is that between the Unhab. male stimulus and the Mirror stimulus. A Mirror gives invariable feedback of a subject's actions. That is, whatever the subject does is immediately reflected in the mirror. If the fish raises his gills at his "adver- sary," the "adversary" erects his gills at the same moment and in exactly the same way. As Peeke and Peeke (1970) state with respect to this simultaneous feedback situation: While this might happen in a real encounter, it would certainly not be a frequent occurrence. If the behav- ior elicited in the animal is facilitated by having 71 his adversary perform the same act (e.g., if the prob- ability of biting is increased by having just been bitten at) then the mirror image situation is one of perfect positive feedback which will increase the rate of the elicited response at a steeply accelerated rate (p. 233). Another problem with the Mirror stimulus, especially with respect to the display of Betta splendens concerns the relative spatial orientation of two displaying fish. While the subject is broadside to the "adversary" in the mirror, the "adversary" is also broadside to the subject. Any changes on the part of the subject will bring about the exact same changes in his "adversary." This situation is extremely rare as Simpson (1968) demonstrated in his exten- sive research on the display of the Egttg. He found that when two fish are displaying at each other, a frontal posi- tion in one fish usually elicits a broadside orientation in the other, and vice-versa. It is an extremely rare occur- rence that both fish remain in a face-to-face position, or a broadside—to-broadside position. In general, as Gallup (1968) has stated, mirror image stimulation provides for many situations hardly ever encountered in the natural habitat. The Unhab. male stimulus simulates most closely the type of adversary which would be encountered in a natural situation. It vigorously responds to a subject "adversary," but not in a perfectly invariant way as does the perfectly fed-back Mirror stimulus. The Unhab. male provides a source 72 of variable feedback in that it displays and interacts with the subject, but does not duplicate its every move. With regard to stimulus variability, Lovibond (1969) found that the degree of habituation of the orienting response (OR) to multiple stimulus sequences is a negative function of the uncertainty in the stimulus series. The relation between uncertainty of stimulus and the degree of habitua- tion seems to be a curvilinear one. According to Sokolov's theory of habituation, a "neuronal model" of a stimulus which is variable in its attributes, is difficult to form as compared to some invariant stimulus (Lynn, 1966). Lorenz (1965) has also spoken on the problem, stating that the ease with which habituation of a response is attained in the laboratory as compared with a field situation is due to the relative constancy of the experimental situa- tion, itself. The Mirror stimulus is a better elicitor than the Hab. male stimulus because the Mirror "adversary“ responds back as long as the subject responds, providing positive feedback to the subject. The Hab. male stimulus responded only briefly, and usually appeared as rather innocuous (fins folded, pale color, and ignoring direct contact with the subject). Apparently the constancy of feedback inherent in the Mirror stimulus has some eliciting qualities, itself. 73 Habituation Retention, and Recovery of HaEituation Perhaps the most important finding in the present study was that the degree of habituation and retention of habituation is related to the initial eliciting strength of the stimulus. It should be mentioned here that for both of the threat components (GD and FD) that exhibited systema- tic habituation, the greatest absolute amount of habituation was associated with the Unhab. male group followed by the Mirror, Hab. male and Cut—out groups, respectively. That is, the greater the initial eliciting strength of a stimu- lus (as measured by the first 10 minute observation) the greater the absolute amount of habituation. The absolute amount of habituation is defined as the initial response level minus the terminal response level. This way of measuring habituation obviously does not take into con- sideration the different initial response levels to dif- ferent elicitors. In other words, the response to a strong elicitor (measured by initial response level) has a greater potential absolute decrement than the responses elicited by stimuli of lesser initial strength (See Figure 6 for an example). A proportional measure of decrement, using the initial response level as the anchor point, corrects for the different initial eliciting stimulus strengths. When these proportional measures of habituation were used in .1} .III I II. 74 the present study for GD and FD data, the relation to eliciting stimulus strength was different from that when the absolute amount of habituation was used as the measure of habituation. It was found that initial eliciting stimu- lus strength was inversely related to the proportion of across session decrement. That is, the stronger the ini- tial response level, the smaller the across session decre- ment. The proportion of retention of habituation after a 24 hour rest interval was also related to initial eliciting stimulus strength: the stronger the initial level of re- sponse, the smaller the proportion of retention after 24 hours. The prOportion of recovery of habituation after 24 hour rest was, therefore, positively related to initial stimulus strength. A Theoretical Framework The results of this study using proportional data, nicely fit an elicitation and competing response theory of habituation like that of Ratner (1970). According to this theoretical framework, the elicitation of a response is followed by a refractory period during which response strength diminishes. This refractory period is a conse- quence of temporary changes in sensory, transmission, and effector processes. Also affecting the originally elicited response are stimulus conditions that Operate concurrently ‘with the original eliciting stimulus. This concurrent 75 stimulation may facilitate or interfere with the originally elicited response, depending on the stimulus and the topo- graphy of the responses elicited by concurrent stimuli. Habituation involves the intrusion of competing re— sponses that arise from the action of stronger (more dominant) eliciting stimuli during the refractory periods. The dominance of the competing response may be temporary but if a number of such responses are elicited they may combine to equal the strength of the original response and thus cause habituation (Ratner, 1970, ch. 3). Therefore, the greater the strength of the elicit— ing stimulus, the less the degree of habituation. This supposition follows from the argument that the greater rela- tive strength of the elicitor will be dominant over other concurrent stimuli eliciting competing responses. This portion of the theory seems supported by the results of the present experiment: the greater the initial eliciting stimulus strength the smaller the proportion of habituation. As previously mentioned, however, increasing initial elicit- ing stimulus strength was associated with an increasing absolute amount of habituation. This finding concurs with the findings of much previous research involving many dif- ferent organisms and S-R systems (Martin and Melvin, 1964; Miller and Murray, 1966; Askew, 1969; Melvin and Cloar, 1969; Peeke, 1969). Although Ratner's (1970) theory con- cerns itself with relative amounts of decrement, it might be restated as follows: the greater the initial eliciting stimulus strength, the greater the absolute amount of 76 decrement and the smaller the relative amount of habitua- tion. These contentions obviously require further valida- tion through research using different organisms and S-R systems. The proposition that greater eliciting stimulus strength is associated with a relatively smaller amount of retention of habituation also was supported by the data of the present study. Non—Habituating Indices of the Display A few comments are necessary, at this point, in order to explain the lack of systematic habituation for all stimulus groups when using GF or FF as the indices of decre- ment. Apparently the frequency measures operate along a different time dimension when compared to the quickly habituating duration measures (GD and FD). That is, if the experimental sessions were of longer duration, the frequency measures might have shown systematic habituation. Evidence for this supposition is available from Clayton and Hinde (1968). Using male Bettas, they found that gill cover erection frequency showed little systematic change during the first day (26 minute observation) of exposure to a mirror image, but showed a significant decrement as early as the second day of their ten day exposure period. What is of importance in the present experiment is the fact that different threat components were differentially reflective (of habituation when the same test period length was used. 77 Therefore, whether or not habituation of a component occurs is related to both the duration of stimulus exposure and the particular way of measuring the component. It was also noticed that there was an increment in both within and across session responding for the GF and FF data for the group exposed to the unhabituated male stimulus (Unhab. male). Since this stimulus was the strongest elicitor, the responses were much more resistant to habituation. Hinde (1954) has shown that habituation involves both incremental and decremental processes, with incremental processes predominating during the early phases of stimulus exposure. Apparently the Unhab. male group exhibited most conspicuously the incremental com- ponents of the habituation processes during the time span of sessions used in the present study. Relation Among the Indices of t e Disp ay There were large individual differences in respond- ing within each stimulus group and for all measures used in the present study. These highly reliable individual dif- ferences are a general characteristic of the habituation process (Ratner, 1970). However, these individual differ- ences were overshadowed by the group differences in respond- ing associated with different stimulus conditions. For instance, air gulping frequency (AG) was strongly affected 78 by the different elicitors: the Unhab. male stimulus elicited the highest level of AG responding, followed by the Mirror, Hab. male and Cut-out stimuli, respectively. Perhaps of greater significance was the fact that no sig- nificant changes in the level of responding were noticed for this particular measure. This response is a purported measure of activity level (Simpson, 1968) and thus was not expected to show a decrement during the present experiment. The data confirmed this expectation. That AG is not an in— tegral component of the threat display was shown by the uniformly negligible correlations with the other threat measures. However, the four other measures (GF, GD, FF, FD) were highly correlated in a positive direction, re- flecting the integral relation among these components of the Betta's threat display. Controls for Adaptation and Fatigue The consistency of the AG data also provided evi- dence that the decrement in GD and FD responding was not due to the effects of general fatigue, since this component remained fairly constant throughout the duration of the experiment. Other factors support the interpretation of the response decrement (GD, FD) in the present study as being due to habituation rather than sensory adaptation and/or general fatigue. First of all, significant retention of 79 habituation after a 24 hour rest period rules out these effects. Also, immediately upon the termination of session 2, any E which either had not displayed to the appropriate elicitor during the experimental sessions or had approached a near-zero level of responding, was exposed to an unhabit— uated male stimulus (Unhab. male) placed for 30 seconds in the same compartment as the E. Every E which fitted into the categories described above, responded immediately, completely and almost constantly during the 30 second post- experimental session. Therefore, as far as the Es which did not respond at all during the experiment are concerned (almost all were in the Cut-out group) the lack of respond- ing was not due to an inability to perform the complete threat display. The Threat Display as a Consummatory Response Of particular interest is interpreting the results of the present experiment when the fighting sequence of the male Egttg is placed in the appetitive, consummatory, and post-consummatory framework of Denny and Ratner (1970). The threat display probably fits somewhere in the middle of the consummatory sequence, culminating in direct attack (biting, tail-slashing, etc.). Although quantitative measurements were not taken, it is interesting to note that the only instances of biting at the partition between the 80 stimulus and E were noticed in Es in the Unhab. male group and the Mirror group (the two strongest elicitors). Biting by the Unhab. male group was especially apparent. Stronger elicitors may elicit the complete consummatory sequence in less time than weaker stimuli. Perhaps weaker stimuli do not even elicit the late components of the consummatory sequence at all. These speculations are particularly amenable to experimental validation. Conclusions The threat display of male Egttg splendens is an easily measured, highly reliable phenomenon, consisting of many different S-R components. It was found that certain components of the display are subject to habituation and that the decrement is retained over a 24 hour period. For those threat components that exhibited systematic habitua- tion (GD, FD), there was a strong relation between the initial eliciting strength of the various threat elicitors and the absolute and relative amounts of habituation, and retention of habituation. Portions of Ratner's (1970) elicitation and competing response theory of habituation were supported. REFERENCES Adler, N. and Hogan, J. A. Classical conditioning and punishment of an instinctive response in Betta splendens. Anim. Behav., 1963, it, 351-354. .Askew, H. R. Effects of intertrial interval and stimulus intensity on habituation of the head-shake response in the rat. Unpublished doctoral dissertation, Michigan State University, 1969. Baenninger, R. Waning of aggressive motivation in Betta splendens. Psychon. Sci., 1966, 2, 241—242. Beanninger, R. Catechol amines and social relations in Siamese fighting fish. Anim. Behav., 1968, 16, 442-447. ‘— Bock, R. D. and Haggard, E. A. The use of multivariate analysis of variance in behavioral research. In D. K. Whitla (Ed.), Handbook of measurement and assessment in behavioral sciences. Reading, Mass.: Addison Wesley, 1968. Boneau, C. A. The effects of violations of assumptions underlying the t test. Psychol. Bull., 1960, El, 49—64. Braddock, J. C. and Braddock, Z. I. Aggressive behavior among females of the Siamese fighting fish. Physiol. Zool., 1955, EEJ 152—172. Clayton, F. L. and Hinde, R. A. The habituation and re- covery of aggressive display in Betta splendens. Behaviour, 1968, EE, 97-105. Denny, M. R. and Ratner, S. C. Comparative Psychology, (Rev. Ed.). Homewood, IIIinoiE: Dorsey Press, 1970. Dunlop, C. W., Webster, W. R., and Rodger, R. S. Amplitude changes of evoked potentials in the auditory system of unanesthetized cats during acoustic habituation. J. Aud. Res., 1966, E, 47-66. 81 82 Evans, L. T., Abramson, H. A., and Fremont-Smith, N. Lyser- gic acid diethylamide (LSD-25): XXVI. Effect on social order of the fighting fish, Betta splendens. J. Psychol., 1958, 12, 263-273. Evans, L. T., Geronimus, L. H., Kornetsky, C., and Abram- son, H. A. Effect of ergot drugs on Betta splendens. Science, 1956, 123, 26. Forselius, S. Studies of anabantid fishes: I, II, and III. Zool. Bidragp Fran. Uppsala, 1957, EE, 93-597. Gallup, G. G. Mirror image stimulation. Psychol. Bull., Goldstein, S. R. Mirror image as a reinforcer in Siamese fighting fish: a repetition with additional con- trols. Psychon. Sci., 1967, 1, 331-332. Gordon, M. and Axelrod, H. Siamese fighting fish. Jersey City, New Jersey: T.F.H. Publications, 1968. Harris, J. D. Habituatory response decrement in the intact organism. Psychol. Bull., 1943, 3E, 385—423. Hess, E. H. Temperature as a regulator of the attack- response of Betta splendens. Z. Tierpsychol., 1952, E, 379-382. Hinde, R. A. Factors governing the changes in strength of a partially inborn response, as shown by the mobbing behavior of the chaffinch (Fringilla coelebs). II. The waning of the response. Proc. Roy. SOC. B., 1954, 142, 331-359. Hogan, J. A. Fighting and reinforcement in the Siamese fighting fish (Betta splendens). J. Comp. Physiol., Psychol., 1967, E1, 356-359. Kuenzer, P. P. Verhaltensphysiologische Untersuchungen fiber das Zucken des regerwurms. Z. Tierpsychol., 1958, EE, 31-49. JLaudien, H. Untersuchungen fiber das Kampverhalten der Mann- chen von Betta s lendens Regan (Anabantidae, Pisces). Z. Wiss. Zool., I965, I72, 134-178. Iiewis, D. J. and Cotton, J. W. Partial reinforcement and non-response acquisition. J. Comp, Physiol. Psychol., 1958, El, 251-257. I I .II I . IIIIJ 83 Lissmann, H. W. Die umwelt des kampfisches (Betta splendens Regan) Z. Vergl. Physiol., 1932, EE, 653II1. Lorenz, K. ‘Evolution.and Modification of Behavior. Chi- cago: Uan. of Chicago Press, 1965. Lovibond, S. H. Habituation of the orienting response to multiple stimulus sequences. Psychophysiology, Lynn, R. Attention, arousgl, and the orientation reaction. Oxford: Pergamon Press, 1966. Marrone, R. L. Effects of food deprivation on thefighting response of Betta splendens. Psychol. Rep., 1965, 17, 632. Marrone, R. L., Pray, S. L. and Bridges, C. C. Norepine- phrine elicitation of aggressive display responses in.Betta splendens. Psychon. Sci., 1966, E, 207- 208. ' Martin, R. C. and Melvin, K. B. Fear responses of bobwhite quail (Colinus Virginianus) to a model and a live red-tailed Hawk (Buteo jamaicensis). Psychol.. Forsch., 1964, El, 323-336. McDonald, A. L. and Heimstra, N. W. Modification of ag- gressive behavior of green sunfish with D-lysergic diethylamide. J. Psychol., 1964, E1, l9-23.- Melvin, K. B. and Anson, J. E. Facilitative effects of punishment on aggressive behavior in the Siamese fighting fish. Psychon. Sci., 1969, EE, 89-90. Melvin, K. B. and Cloar, F. T. Habituation of responses of quail (Colinus Virginianus) to a hawk (Buteo swainsoni): Measurement through an 'innate sup- pressionTtechnique. Anim. Behav., 1969, El, 468-473. Ddiller, J. D. and Murray, F. S. Guinea pig's immobility response to sound: Threshold and habituation. J. Comp. Physiol. Psychol., 1966, El, 227-233. IPeeke, H. V. S. Habituation of conspecific aggression in the three-spined stickleback (Gasterosteus aculea- tus E;)° Behaviour, 1969, EE, I37-156. 84 Peeke, H. V. S. and Peeke, S. C. Habituation of conspecific aggressive responses in the Siamese fighting fish (Betta splendens). Behaviour, 1970, EE, 232-245. Peeke, H. V. S., Wyers, E. J., and Herz, M. J. Waning of the aggressive response to male models in the three-spined stickleback (Gasterosteus aculeath £1" Anim. Behav., 1969, ll, 224-228. Ratner, S. C. Habituation: Research and theory. In J. Reynierse (Ed.), Current issues in animglilearning. Lincoln, Nebraska: University of Nebraska Press, 1970. Ch. 3. Rodgers, W. L., Melzack, R., and Segal, J. R. "Tail flip response" in goldfish. J. Comp. Physiol. Psychol., Russell, E. M. Changes in the behavior of Lebistes reticu- latus upon repeated shadow stimulus. Anim. BeEav., I E, 574-585. Siegel, S. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill, 1956. Simpson, M. J. A. The display of the Siamese fighting fish Betta s lendens. Animal Behaviour Monographs, I968, t, wEole issue. Smith, H. M. The fighting fish of Siam. Am. Mus. Nat. Hist., 1937, E2, 265-271. Thompson, R. F. and Spencer, W. A. Habituation: A model phenomenon for the study of neuronal substrates of Thompson, T. I. Visual reinforcement in Siamese fighting fish. Science, 1962, 141, 55-57. Thompson, T. I. Visual reinforcement in fighting cocks. J. Exp. Anal. Behav., 1964, 1, 45—49. thompson, T. I. Operant and classically conditioned ag- gressive behavior in Siamese fighting fish. Am. Zool., 1966, E, 629-641. Irhompson, T. I. and Sturm, T. Classical conditioning of aggressive display in Siamese fighting fish. E; Exp) Anal. Behav., 1965a, E, 397-403. 85 Thompson, T. I. and Sturm, T. Visual-reinforcer color, and operant behavior in Siamese fighting fish. J. Exp. Anal. Behav., 1965b, E, 341-344. Thorpe, W. H. Learning and instinct in animals. London: Methuen, 2nd ed., I963. Tinbergen, N. The study of instinct. Oxford: Clarendon Press, 195I. Ulrich, R. Pain as a cause of aggression. Amer. Zool., 1966, E, 643-662. Uno, T. and Grings, W. Autonomic components of orienting behavior. Psychophysiology, 1965, t, 311—321. Van Den Assem, J. and Van Der Molen, J. N. Waning of the aggressive response in the three-spined stickle— back upon constant exposure to a conspecific. I, A preliminary analysis of the phenomenon. Be— haviour, 1969, EEJ 286—324. ——— Winer, B. J. Statistical principles in experimental design. New York: McGraw-HilI, 1962. S. v N... mo. v m. Now. vmm. ONw. mam. ONO. wmw. wen. mmo.l ONO. Nam «a a « vmh. ONO. mam. com. New. emm. mmo. ONH. Ham r r «a «a a vum. mom. mmn. vom. mom. vmo. who. Nhh a a as as Nmm. ome. web. «mm. OHH. NMH. Hhh * as vmm. vmh. owm. mmH.I OOH.1 Now «a a r % mun. th. owo.l mHo.| Haw «r r Nmm. mwo. owo. wa as OMH ovN Hhu mmm. N0¢ or Hw< Nah Ham Nmm Hhh NDU How th Hmo NU< de . 2: n his macho msHssHum 3...: ooumflHomEO “N sonmom poo H conmom 9." moHnoHHo> unopsomoo Mom mu GOmHmom .a was HO. v m: mo. v mar mwm. vow. NNm. ewm. ONM. wmm. va. wow. mmm. Nam re re www. vHv. ONO. OOO. OMH. mmm. mmm. Nmm. Ham r * mmh. mvv. Nam. www. wa. OHv. mmv. Nhh * r r ONv. Ohm. wa. vwm. Nvm. VHV. Hmh « r r th. mHm. web. vHH.I OOH. Now as re a m me. ONO. mHO. ONN. HQO as as Ohm. VNO.I me. th «a VHO.I mmN HmU mmw N0< rs H04 Nah Ham Nmm Hmh N00 HQU th HMO Nwfi HUN .AOH u mv QDOHO msHDEHWm HOHHHE 1N conmom can H conmom :H moHannm> usoocomoo How my GOmHmmm . m 031...: .l‘llil l I‘l". 8. v a... mo. v m. owm. wmm. 0mm. cam. vvm. mom. mmh. OMH. «OH.I Nah «a he «a «r as as omm. vem. Nom. mHm. whm. ONO. omO. omN.I Hum as «a as re as {a wow. Nam. wwm. wwmd Nah. OMH. OON.I Nhh as «a as as Nmm. 0mm. mew. ONO. NHH. mNm.I Hhm « re r as mmm. vmm. VNm. omH. mmH.I New «r as «r M mvm. mvm. me. OmH.| How re as New. NOH. mmH.I wa as mmH. NmH.I Hmw cmm. Nwd Had Nam Ham th Hhh Now Haw Nho Hmw Nwd HUN .uAOH 2O moonw msHsEHum on: ploumnfiHnmm .N conmom can H GOHmmom cH moHQMHHm> ucoocomon How mu acmummmll. U OHQME Ho. v.ms« mo. v.me .omcommou ozo mmm. mmm. mmm. vmm. omm. ovv. mmO.I Nah «« as «a m or m or mmm. mma. m mmm. M ”NO. OHO. O¢O.I Ham mmm. o ”mm. m ”mm. mow. OOO.I Nah o mmm. o ”mm. va. meo.l Hhm M M M M .0 N00 M o mmm. mmm. OOO.I Haw o o o wa «hm. mmO.I Hmo was. «we HON Nam Ham th Hhh NQU HDU NhU th NON HUN . 8H u .5 moons dessHuM usoiuso 1N sonmom can H conmom cH moHAMHHo> ucopsomoo Mom on acmumomli To oHooB