:RELAXATLONMEDIATEE APPROACH AS A NECESSARY COMPONENT 8N SIMPLE AVOiDANCE LEARNING Thesis for the Dayna of Ph. D. M3CHEGAN STATE UNWERSWY Dominic Joseph Zerbofio, Jr. 3965- masrs LIBRARY Michigan State University This is to certify that the thesis entitled REIAXATION-MEDIATED APPROACH AS A NECESSARY COMPONENT IN SIMPLE AVOIDANCE LEARNING presented by Dominic Joseph Zerbolio, Jr. has been accepted towards fulfillment j of the requirements for 43141;... degree in My Major prgfiés/sor Dam May 19, 1965 0-169 ...._... ._’..p 4 . Cl 85 ti e1 81' dj di 1e b1 to Co th ABSTRACT RELAXATION-MEDIATED APPROACH AS A NECESSARY COMPONENT IN SIMPLE AVOIDANCE LEARNING by Dominic Joseph Zerbolio, Jr. In the simple avoidance learning situation, Elicitation Theory posits that gs confined long enough to relax in the presence of the non-shock cues of an avoidance apparatus learn both to approach those cues and to escape from the cues that are associated with shock. Theoretically, relaxation occurring in the non-shock area after escaping shock or shock-associated cues mediates the development of the approach component. Three experiments were designed to test the effect of the posited approach component in a simple avoidance learning situation. In avoidance learning, all gs were trained to run in one direction in a "shuttle" box situation. The shuttle-box had distinctively colored chambers, one black and one white, color serving as differential cue complexes. Experiment I employed a reversal learning design. In original learning (0L),'§s were trained to run in one direction (e.g., from black to white) and were then confined for 30 sec. (not long enough to relax) or 150 sec. (long enough to relax) in the safe (non-shock) compartment on each trial. A third group was confined for 30 sec. in the safe chamber and then spent 120 sec. on a neutral open platform (controlli; confinenen In It (e.g., fro: ITI being I successive In g learned fa However, t 0f signif: masked in. Exp omitted. Significa sec, and Direct 01 IEarning CUES (Eh shocking t0 one 0 sec. Con learned aVOldanC training from the Dominic Joseph Zerbolio, Jr. (controlling for the III difference between the 30 and 150 sec. confinement groups). In reversal learning (RL), §s were run in the reverse direction (e.g., from white to black) but were not differentially confined, the III being 30 sec. for all gs. All §s were run to a criterion of three successive avoidances in both 0L and RL. In general, §s confined in the safe region long enough to relax learned faster in 0L, but reversed slower than Es not allowed to relax. However, these differences were not statistically significant. The lack of significance was attributed to the presence of a buzzer-CS which masked the effects of the color cues of the chambers. Experiment II was a replication of Experiment I with the buzzer omitted. In Experiment II, the 150 sec. group learned the OL CAR significantly faster, but reversed significantly slower than the 30 sec. and 30-120 sec. groups, which were not statistically different. Direct observations of §s' behavior indicated that part of the reversal learning effect was due to fear responses associated with the RL safe cues (the OL shock cues) in OL. Experiment III precluded fear association to shuttle-box cues by shocking §s outside the shuttle-box and then transporting them by hand to one of the distinctive chambers for 150 sec., 30 sec., or 30-120 sec. confinement periods. In the above are-training phase, §s neither learned a CAR nor fear responses to the shuttle-box cues. In the avoidance learning phase, §s were either run toward the cues of pre- training confinement (e.g., confined black; run toward black) or away from the cues of pre-training confinement (confined black; run toward white). in pre-tr groups le cantly fa cantly 51 showed no All of respon situation tation ef the escap impeding When both In mediated Elicitati Dominic Joseph Zerbolio, Jr. white). The results indicated that only groups confined for 150 sec. in pre-training showed differential effects. The 150 sec. confinement groups learned a CAR toward the pre-training confinement cues signifi- cantly faster, but away from the pre-training confinement cues signifi- cantly slower than the other groups. The other confinement groups showed no such effect. All of the results described above can be interpreted in terms of responses conditioned to the specific cue complexes in the avoidance situation which either facilitate or impede learning a CAR. The facili- tation effect of relaxation-approach responses occurs when it sums with the escape conditioned responses to move g in a single direction. The impeding effect from relaxation-approach and escape responses occurs when both are elicited by the same cue complex. In sum, these results are interpreted as confirming a relaxation- mediated approach component in avoidance learning as postulated by Maze. as, Committegyfihairman Date @271 /j [746-— Elicitation Theory. L/ RELAXATION-MEDIATED APPROACH AS A NECESSARY COMPONENT IN SIMPLE AVOIDANCE LEARNING BY Dominic Joseph Zerbolio, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1965 ACKNOWLEDGEMENTS The author gratefully acknowledges the assistance given him by his major professor, M. Ray Denny. Without his assistance in the development of this thesis or without his theoretical position, this study would not have been possible. The author is also indebted to the other members of his guidance committee for their help and encouragement: Doctors Stanley Ratner and Paul Baken of the Department of Psychology, and Doctor John King of the Department of Zoology. In addition, the author would also like to acknowledge the cooperation of Rattus Albinicus, which was of fundamental importance in the preparation of this thesis. ii TABLE OF CONTENTS ACKNOWLEDGMENTS LIST OF TABLES LIST OF FIGURES LIST OF APPENDICES INTRODUCTION EXPERIMENT I Method Results Discussion EXPERIMENT II Method Results Discussion EXPERIMENT III Method Results Discussion GENERAL DISCUSSION AND SUMMARY REFERENCES . . . . . . . . . . . . . . . APPENDICES iii Page ii iv vi . viii 17 27 40 43 45 Table 10. ll. 12. 13. LIST OF TABLES Analysis of variance for the in Experiment I . Analysis of variance for the in Experiment I . Analysis of variance for the Experiment I Analysis of variance for the on the first trial of 0L and Analysis of variance for the on the first trial of 0L and Analysis of variance for the in Experiment II Analysis of variance for the in Experiment II Analysis of variance for the Experiment II . Analysis of variance for the on the first trial of 0L and Analysis of variance for the on the first trial of 0L and Analysis of variance for the for the confinement control groups in Experiment III. Analysis of variance for the the experimental confinement Experiment III Analysis of variance for the OL trials to criterion RL trials to criterion savings scores in latency of response RL in Experiment I . regressions occurring RL in Experiment I . 0L trials to criterion RL trials to criterion savings scores in latency of response RL in Experiment II regressions occurring RL in Experiment II trials to criterion trials to criterion for groups in trials to criterion to like vs. different colors for all groups in Experiment III iv Page 47 47 48 48 49 49 50 . 51 51 52 52 53 Table 10. 11. 12. 13. LIST OF TABLES Analysis of variance for the in Experiment I Analysis of variance for the in Experiment I . Analysis of variance for the Experiment I Analysis of variance for the on the first trial of 0L and Analysis of variance for the on the first trial of 0L and Analysis of variance for the in Experiment II Analysis of variance for the in Experiment II Analysis of variance for the Experiment II Analysis of variance for the on the first trial of 0L and Analysis of variance for the on the first trial of 0L and Analysis of variance for the for the confinement control groups in Experiment III. Analysis of variance for the the experimental confinement Experiment III Analysis of variance for the OL trials to criterion RL trials to criterion savings scores in latency of response RL in Experiment I . regressions occurring RL in EXperiment I . 0L trials to criterion RL trials to criterion savings scores in latency of response RL in Experiment II regressions occurring RL in Experiment II trials to criterion trials to criterion for groups in trials to criterion to like vs. different colors for all groups in Experiment III iv Page 47 47 48 48 49 49 50 . 51 51 52 52 Table Page 14. Analysis of variance for the first trial response latency for all groups in Experiment III . . . . . . . . 53 15. Analysis of variance of the regressions occurring in avoidance learning for all groups in Experiment III . . . 54 16. The number of trials to criterion, number of shocks, and the latency of response on the first trial for each S in both the 0L and RL phases of Experiment I . . . 56 17. The number of trials to criterion, number of shocks, and the latency of response on the first trial for each S in both the 0L and RL phases of Experiment II. . . 57 18. The number of trials to criterion, number of shocks, and the latency of response on the first trial for each experimental g in the avoidance learning phase of Experiment III . . . . . . . . . . . . . . . . . . . . 58 19. The number of trials to criterion, number of shocks, and the latency of response on the first trial for each control S in the avoidance learning phase of Experiment III . . . . . . . . . . . . . . . . . . . . . S9 Figure 1. LIST OF FIGURES The mean number of trials to criterion in original learning, reversal learning, and the number of trials saved in reversal learning for the 30 sec., 30-120 sec., and 150 sec. confinement groups in Experiment I . The mean latency in seconds to respond on the first trial of original learning and reversal learning for the 30 sec., 30-120 sec., and 150 sec. groups in Experiment I . . . . . . . . . . . . . . . The mean number of regressions in reaching the 0L and RL criterions for the 30 sec., 30-120 sec., and 150 sec. confinement groups in Experiment I . . . . . . The mean number of trials to criterion in original learning and reversal learning for the 30 sec., 30- 120 sec., and 150 sec. confinement groups in Experiment II The mean latency in seconds to respond on the first trial of original learning and reversal learning for the 30 sec., 30-120 sec., and 150 sec. confinement groups in Experiment II The mean number of regressions in reaching the 0L and RL criterions for the 30 sec., 30-120 sec., and 150 sec. confinements in Experiment II . The mean number of trials to criterion for the experimental and confinement control groups in Experiment III for groups run to the color of pre- training confinement (LC) and groups run away from the color of pre-training confinement (DC) . The latency for the first trial of avoidance training for the experimental and combined control groups (CC) in Experiment III. The figure shows the means for the experimental confinement groups run to the color of pre-training confinement (LC) and away from the color of pre-training confinement (DC) . vi Page 12 l4 19 21 23 33 35 Figure Page 9. The mean number of regressions for the experimental and combined control groups (CO) in Experiment III. The figure shows the means for the experimental confinement groups run to the color of pre-training confinement (LC) and run away from the color of pre-training confinement (DC) . . . . . . . . . . . . . 37 vii Appendix A. B. LIST OF APPENDICES Summary Tables for the Analyses of Variance . Tabled Raw Data . . . . . . viii Page 46 55 RELAXATION-MEDIATED APPROACH AS A NECESSARY COMPONENT IN SIMPLE AVOIDANCE LEARNING INTRODUCTION In simple avoidance learning situations, traditional learning theorists stress the importance ofug learning to fear the stimulus complex associated with shock (the CS). After fear to the CS is con- ditioned, S may anticipate the US by making the appropriate response (avoidance) prior to its onset. The appropriate response is immediately followed by the removal of the CS and a reduction of the CS-associated fear state. It is the reduction of conditioned fear which is tradi- tionally considered to mediate the acquisition of a Conditioned Avoidance Response (CAR) (Hull, 1943; Miller, 1958; Mowrer, 1960). In addition, the maintenance of the fear response to the CS complex is considered responsible for the maintenance of the CAR over time (or trials) (Solomon and Wynne, 1954). Elicitation Theory, another way of conceptualizing avoidance learning, proposes that S learns not only to escape from shock and shock-associated cues, but also learns to approach cues associated with non-shock confinement or "safe" region (Denny and Adelman, 1955 and 1956). The first, or escape-fear component of the CAR, is fundamentally the same as the traditional View. The second component, learning to approach cues associated with non-shock confinement bears explanation. When placed in an area where shock occurs, the animal eventually escapes from shock and the shock-associated cues. Sometime after escaping, the animal begins to relax and to approach and/or investi- gate the cues surrounding him. Relaxation, in the Elicitation Theory framework, is considered to be a fairly long-latency response with reference to the change in the shock-produced emotional state of the animal. It is posited that the stimulus aspects of the safe region become conditioned to relaxation-approach responses by a contiguity principle and thereby elicit relaxation-approach behaviors on sub- sequent trials. Consequently, according to the elicitation point-of- view, the animal learns to approach those cues that are associated with non-shock confinement as well as to escape from shock and shock-associated cues. Several studies have been done which provide support for the interpretation of avoidance learning in terms of Elicitation Theory. Denny, Koons, and Mason (l959) found that extinction of a CAR was much more rapid when shock and safe confinement areas were similar than when they were dissimilar. The interpretation of these data rests on an analysis of the behaviors elicited by the respective cues in the safe and shock areas. When the cue complexes of the safe and shock areas are similar, relaxational responses which have been conditioned to the cues of the safe area are also elicited by the similar cues in the shock area. This generalized relaxation competes with the CAR, which thereby facilitates extinction. When the safe and shock area cues are dissimilar, conditioned relaxation is not elicited in the shock area and the CAR continues unimpeded for many more trials. [‘lI'lllllll-ll‘ll Knapp (1960) not only found that similar shock and safe regions facilitate extinction, but that dissimilar areas facilitate acquisition. This presumably occurs because no conditioned relaxation is elicited by the dissimilar shock cues of the shock area, consequently, the acquisition of the CAR is unimpeded. Knapp's data also suggested that length of non-shock confinement was an important variable. Denny and Weisman (1964) varied the proportion of a constant 230 sec. intertrial interval (III) that SS spend confined in a safe area. In acquisition, with shock and safe areas dissimilar, §s confined for 150 sec. or more required significantly fewer trials to reach an acqui- sition criterion than SS confined for lesser proportions. In extinction, with safe and shock boxes similar, both the least length (10 sec.) and the longest length (225 sec.) of confinement produced significantly fewer trials to an extinction criterion than for the other confinement groups. A post-hoc analysis of these data suggested that relaxation occurring in the neutral zone where §s were placed during the non- confined portion of the III may generalize back to the shock area and facilitate extinction. Theoretically, if relaxation occurring in the neutral zone was prevented, then the 10 sec. group should require the most number of trials to extinction. A second 10 sec. confinement group, where the neutral zone was changed every 40 sec. to prevent relaxation did require more trials to extinguish than any other group. The studies cited above support the existence of an approach component in avoidance learning, but they all have the following characteristic: All of them have used a jump-out apparatus in which § could not see the safe area cues from the shock area. Under these conditions, it is possible to interpret poorer learning with similar regions as due to a competition between fear and relaxation in the shock region,which impedes acquisition, rather than in terms of the facilitating effect of a distinctive place to approach. The same sort of competition presumably occurs during extinction, and this is why similar boxes yield much faster extinction than dissimilar boxes. In addition, both of these results depend on conditioned relaxation being elicited by similar cues, but 325 the identical cues, to which it was originally associated. Due to generalization decrement, one would certainly expect that conditioned relaxation elicited by similar cues to be weaker than conditioned relaxation elicited by the identical cues to which it was originally associated. The purpose of the following research is to carry out a direct investigation of the escape and approach components of avoidance learning as posited by Elicitation Theory. The first two experiments are designed to show the combined effects of learning to escape shock- associated cues and learning to approach cues associated with non~ shock confinement. The third experiment deals directly with the approach component. All experiments utilize a "shuttle" box situation where S can see the non-shock or safe cues from the shock area. Where conditioned relaxation is pitted against a CAR, the cue complex which elicits conditioned-relaxation is the identical complex to which relaxation was originally conditioned. Thus, the effects of relaxation are not weakened by generalization decrement. EXPERIMENT I Experiment I has two purposes. The first is to demonstrate the combined effect of the escape and approach components of avoidance learning as posited by Elicitation Theory. The second aim is to show that conditioned relaxation, elicited by a specific cue complex as a result of prior conditioning, can effectively compete with the acqui- sition of a CAR to leave (escape from) that cue complex. In other words, once an animal has learned to relax in the presence of certain cues, it will be reluctant to leave those cues even when shocked there. To test both of these hypotheses, a one-way shuttle-box with distinctive chambers and a reversal-learning design is used. The experiment is run in two phases. The Original Learning (0L) phase consists of running §s from one chamber to the other (e.g., from black to white). In the Reversal Learning (RL) phase, gs are run in the reverse direction; i.e., §S run from black to white in original learning are reversed and run from white to black. Under reversal conditions, the original safe cues become the reversal shock cues and the original shock cues become the reversal safe cues. Under these conditions, the specific predictions according to Elicitation Theory are: (A) In original learning, §s confined in the safe area long enough to relax require the fewer trials to reach the avoidance learning criterion than §s confined for shorter periods; (B) In reversal learning, §s confined long enough to relax in OL require more trials to reach the RL avoidance criterion than §S confined for shorter periods in OL. Method Subjects.--The §S were 30 experimentally naive male Sprague-Dawley albino rats obtained from Spartan Research Animals, Inc., in Haslett, Michigan. All Sp were between 85 and 100 days old at the beginning of training. The Es were maintained in pairs in 11 in. long, 8 in. wide, and 9 in. high wire mesh cages with food and water always available. The §S were assigned randomly to six experimental groups of five §s each. Apparatus.--A shuttle-box with two discriminable chambers, one white and one flat black, separated by a manually operated guillotine door was used. The closed door was the same color as the chamber it faced. Each compartment was 18 in. long, 4 in. wide, and 14 in. high. The floors of both compartments consisted of 1/8 in. stainless steel grids spaced 5/8 in. apart, center to center. Each grid could be charged inde- pendently through a Grayson—Stadler grid scrambler (Model E1064GST) with current supplied at 2.0 ma. supplied by a C. J. Applegate stimulator (Model 250). A transistorized buzzer (Malis and Curran, 1960) and the raising of the guillotine door served as the CS. A speaker mounted on the plexiglass top of the black chamber delivered an auditory CS at approximately 60 db directly into the black chamber. The CS-US interval, the ITIs, and the latency of response were measured by Standard Electric timers, Hunter timers, and appropriate relay circuitry. Procedure in Original Learning (OL).--The §s were divided into three groups on the basis of length of confinement after making the response; and each group was counterbalanced for the color serving as the safe compartment, making a total of six subgroups. The relaxation group (confined long enough to relax: Denny and Weisman, 1964; Platt, 1964) was confined 150 sec. on the safe side after the shuttle response. One-half of the group was run toward the black compartment (OLTB 150), i.e., the black compartment was the safe side. The other half was run toward the white compartment (OLTW 150). A second group was confined for 30 sec. on the safe side (not long enough to relax), after responding (OLTB 30 and OLTW 30). A third group spent 30 sec. confined in the safe area and 120 sec. on a neutral Open 12 in. square platform (OLTB 30-120 and OLTW 30-120). The 30-120 groups were a control for the difference in ITI between the 30 and 150 sec. confine- ment groups. At the end of the confinement period, S was picked up and placed back in the CS side and the next trial begun. The CS-US interval was 5 sec., both CS and US being response terminated when S crossed to the safe compartment. All §S were run to a criterion of three successive avoidances. Immediately prior to OL, all §s were given an habituation period of one minute in the shuttle- box with the guillotine door raised. Procedure in Reversal Learning (RL).-~After §_reached the acqui- sition criterion and had been confined according to the conditions of its group, it was picked up from the confinement area and placed on a 7 in. by 5 in. open platform for 30 sec. At the end of this period, S was placed in the chamber that had served as the safe chamber in 0L; and the CS period was initiated. In other words, if S was run from black to white in original learning, it was run from white to black, the reverse direction, in reversal learning. The S3 were not differentially confined in reversal: All §s were confined for 30 sec. in the reversal "safe" chamber on each trial regardless of the OL conditions. All other conditions were identical to CL. Results Five different analyses of the data were performed: (1) the number of trials to criterion in 0L, (2) the number of trials to criterion in RL, (3) the savings between 0L and RL in trials to criterion, (4) an analysis of the latency of response on the first trial of 0L and first trial of RL, and (5) the differences in regressions1 between 0L and RL. Homogeneity of variance was present in all statistical analyses. All differences at or less than the .05 level are considered significant. (1) The number of trials to the OL criterion. The mean number of trials to criterion for the different con- finement groups appear in Figure 1. Since no color differences appeared at any point in the statistical analysis, groups were summed across colors in all graphical presentations, and individual comparisons dealt only with the three main confinement groups. A 2 x 3 analysis of variance (Winer, 1962) revealed no major effects for color of confinement (F=l.795, df=l/24), length of confinement (F=l.l37, df=2/24) or their interaction (F.05) but the trials to criterion in RL were significantly more variable in Experiment I than II (F=5.681, df-24/12, p<:.001). 17 (1) (2) 18 The number of trials to the OL criterion. A 2 x 3 analysis of variance revealed no differences due to color of confinement (F< 1) but showed a significant effect for the length of confinement (F=7.70, df=2/12, p<(.01)(see Table 6, Appendix A). The color by length of confinement interaction was not significant (F.05)(see Table 9. Appendix A). Individual comparisons showed that the 150 group required significantly longer to respond on the first trial of reversal than on the first trial of original learning (t=3.461, df=30, p<:.01). No significant differences were found in first 1“ ‘1 l l I 1!. '- ‘l‘ l' l 1 II {filial Seconds 14 13 12 11 10 21 ORIGINAL REVERSAL LEARNING LEARNING D 12.83 ifiv h 8.58 p b 6.25 h -6_'1.7..-—r_.fl3_ I D 30 30- 150 30 30- 150 120 120 0L Confinement Conditions Figure 5. The mean latency in seconds to respond on the first trial of original learning and reversal learning for the 30 sec., 30-120 sec., and 150 sec. confinement groups in Experiment II. t‘lrl (5) 22 trial latencies for 0L and RL for the 30 and 30-120 groups (t=l.067 and t=1.143 respectively). No significant differences between confinement groups were found in OL. In reversal, the 30 and 30-120 groups did not differ from each other in latency on the first trial (t=.l69) but both responded in significantly less time than the 150 confinement group (t=2 348, df=30, p<:.05 and t=2.179, df=30, p<.05 respectively). The differences in regressions between 0L and RL. A 2 x 3 analysis of variance with repeated measures on one dimension revealed no major effect for length of confinement (F<:l), no differences between regressions in 0L and RL (F=l.276, df=l/15, pt; 05), but a significant confinement by 0L vs. RL interaction (F=10.7l9, df=2/15, p«(.01)(see Table 10, Appendix A). Means for the number of regressions for 0L and RL according to confinement conditions appear in Figure 6. Individual comparisons found that the 30-120 sec. confinement group made significantly more regressions in OL than the 150 group (t=2.278, df=30, p‘<.05) but did not differ from the 30 sec. group (t=1.139). The 30 and 30-120 sec. confinement groups did not differ in number of regressions between 0L or RL (t=.752 and t=1.139 respectively), but the 150 group made significantly more regressions in RL than they did in 0L (t=3.4l8, df=30, p<.01). In reversal, the 30 and 30-120 sec. confinement groups did not differ in number of regressions (t=.752) but both made significantly fewer regressions than the 150 sec. group (t=3.030, df=30, p‘<.01 and t=2.278, df=30, p< .05 respectively). Number of Regressions ORIGINAL REVERSAL LEARNING LEARNING 2.5 2,0» 1-5 I nlfifigL. 1.0L M .50 .5 ' % fi .17 30 30- 150 30 30- 150 120 120 23 0L Confinement Conditions Figure 6. The mean number of regressions in reaching the 0L and RL criterions for the 30 sec., 30-120 sec., and 150 sec. confinements in Experiment II. 24 Discussion The effect of removing the buzzer component from the CS cue complex produced exactly the results predicted. In general, the effect of long confinement is to enhance learning. This is evident from the finding that the 150 sec. confinement group reached the OL criterion significantly faster than the 30 and 30-120 sec. confinement groups which did not differ in their learning rates. One discrepant finding is that the 30~120 sec. confinement group had significantly more regressions in OL than the 150 sec. group, although the 30-120 group did not differ from the 30 sec. group. There does not seem to be a good theoretical basis for this finding, and in this analysis it is attributed to chance- But the effect of confinement on reversal learning is very clear. Even though the reversal learning experimental conditions, including confinement, were identical for all groups, the group confined for 150 sec, in OL performed significantly poorer on every RL measure taken. Specifically, §S in the 150 confinement group: (1) required significantly more trials to reach the RL criterion than §S in the shorter confinement conditions, (2) showed signii cantly less savings in acquiring the RL criterion than £8 in the shorter confinement conditions, (3) required Significantly more time to leave the RL Shock area on the first trial of reversal (the OL relaxation cues) than §S in the shorter confinement conditions, and (4) regressed in RL signifi- cantly more often than §S in the shorter confinement conditions. The SS in the 30 and 30-120 conditions did not differ on any of these measures. These results are especially convincing because the 150 sec. 25 confinement group learned significantly faster in 0L and, therefore, had significantly fewer trials to acquire a fear reSponse to the RL safe area (the OL shock cues) than the §s of the other two confinement conditions. That an initial fear response to the RL safe cues (the OL shock cues) contributed to the effects found in reversal learning is clearly indicated by the latency data where all groups took longer to leave the shock area in RL than in OL. Observation of gs' behavior support this contention. In reversal, after shock onset, §_would typically start to leave the area. Upon approaching midline, and the OL shock cues, § would turn around and run back into shock. The effect was more pronounced in the 150 confinement gs, but it was evident in all groups. It looked as if, when the g saw the OL shock cues, "he thought he was going the wrong way”. Of course, it is more appropriate to say that the OL shock cues elicited escape/avoidance responses from § even though those responses turned him back into the RL shock area. This is exactly what one would expect: Elicitation Theory posits that §s learn to escape from specific cues as well as approach others. A generalization decrement analysis of these data would suggest that the least change from OL to RL occurs for the 30 sec. groups. Thus, in terms of responses associated with specific cues, this group would be expected to show the most difficulty in reversing the responses associated with the OL shock cues since they have the least total Change. In this analysis, the 150 sec. and 30-l20 sec. groups, both which have more total change from OL to RL, would be expected to reverse faster. The fact that the 150 sec. group reverses slowest and that the 30 sec. and 30-120 sec. groups are fundamentally equivalent argues against the generalization decrement interpretation. 26 Since the 150 sec. confinement groups required significantly fgwgr trials to reach an 0L avoidance criterion than the short con- finement groups, it seems reasonable that conditioned fear and escape tendencies to the OL shock cues in the short confinement groups were stronger than in the 150 sec. group. It is impossible to separate the effects of conditioned fear-escape and conditioned relaxation within the confines of the reversal learning design of this experiment. However, if one could condition relaxational responses without con- ditioning fear and escape responses, then one could measure the effects of the relaxation-approach component alone. Experiment III attempts to achieve this end. EXPERIMENT III According to Elicitation Theory, the contiguous pairing of shock termination and a specific cue complex is sufficient to con- dition relaxational and approach responses to that complex, providing the animal is confined in the presence of the cues long enough for relaxation to occur. Since the latency of the relaxational response is relatively long (Denny and Weisman, 1964; Platt, 1964), responses which occur early in the confinement period should have relatively little effect on the conditioning of relaxation. These early reSponses can be thought to include the response by which g escapes/avoids from the shock area into the safe area. Thus, according to this analysis, the escape or avoidance response serves only to transport g from the shock area into the safe area. Theoretically then it should matter very little how § is transported from the shock to the safe region with respect to the development of the relaxation-approach component. Therefore, if gs are shocked in an area outside a shuttle-box and immediately transported by hand into one of the distinctive chambers of the shuttle-box, the conditioning of relaxational responses to the cues of that chamber should occur in much the same manner as if the animal had transported itself. This technique precludes the association of escape or fear responses to the distinctive cues of the other chamber and also pre- cludes the acquisition of a CAR during the time when the relaxational 27 28 responses are being conditioned. But the rate of acquisition of a subsequent CAR can be used as a measure of the effect of the relaxation which was conditioned by such pre-conditioning. Experiment III will be run in two phases, a pre-training phase, and an avoidance learning phase. The pre-training phase will involve shocking §s outside the shuttle-box in a distinctive shock compartment and then transporting them, by hand, to one of the distinctive compart» ments of the shuttle-box. The §s will then be confined for different lengths of time in that compartment to condition differential amounts of relaxation. In the avoidance learning phase, §s will be run either toward the relaxation-conditioned cues (to the compartment of pre- training confinement) or away from the relaxation-conditioned cues (away from the compartment of pre-training confinement). Hypotheses: For the groups that are not confined long enough in preotraining to relax, there should be no difference in trials to criterion toward or away from the pre-training confinement compartment. For the groups confined long enough to relax in pre-training, acquisition of a CAR should be facilitated when they are run toward the pre-training confinement compartment, and impeded when they are run away from it. The theoretical bases for these predictions are given below. In the long confinement groups run toward the confinement compartment, §.is required to approach cues which, by previous conditioning, already elicit relaxational-approach responses. This should facilitate the acquisition of the CAR. When g is trained to go to the other compartment, § is required to leave the cues conditioned to relaxation-approach which should impede the acquisition of the CAR through reponse-competition. In addition, with the design proposed, the number of pairings o: shock and periods of confinement can be rigidly controlled for all groups. MCEhOd §2§lg££§.--Ihe is were 52 male albino rats of the same age, strain, stock, and source as those used in the previous experiments, and maintained under the same conditions. The is were randomly assigned to 12 experimental groups of three each, and eight control groups of two each. Apparatus.--The shuttle‘box apparatus was identical to the one used in the previous experiments. In addition, a chamber, measuring 11 in. long, 9 in. wide, and 10 in. high was used to deliver shock to gs outside the shuttle-box. The external chamber had three walls painted with diagonal l in. wide black and white stripes and a forth wall of unpainted brown masonite. The floor consisted of 1/8 in. brass rods spaced 1/2 in. center to center. The top was a clear plexiglass lid hinged on the 9 in. side\ Shock was delivered to the grid of tme external shock chamber by the same system used to charge the grid 03 the shuttle-box. The duration of shock delivered to the external shock chamber was controlled by a hunter timer. Procedure.-~Twelve experimental groups were formed on the basis of conditions of confinement (3), color of confinement cues (2;, and avoidance conditioning to or away from the confinement cues (2. lhe three confinement conditions were 150 sec. of confinement in the shuttle: box on each pre-training trial, 30 sec of confinement in the shuttlevhox 30 on each pre-training trial, or 30 sec. in the shuttle-box and 120 sec. on an Open platform on each pre-training trial. These confinement conditions were comparable to those used in the two previous studies. In pre-training, §$ were confined either in the black (CB) or the white (CW) chamber of the shuttle-box. In avoidance learning, gs were either run to the black chamber (TB) or to the white chamber (TW). The 150 CBTW group was confined for 150 sec. in the black chamber in pre-training and run in avoidance learning to the white side, or away from the chamber where gs were confined in pre-training. This treatment permitted complete counterbalancing for cue colors. All gs were given one minute of habituation in the shuttle-box prior to the start of experimental manipulations. Procedure in Pre-Training.--At the end of the habituation period, g was lifted by hand and placed in the external shock chamber. g was then given one sec. of unescapable shock. At the end of shock, g was immediately picked up and transported to one of the chambers of the shuttle-box. During confinement in the shuttle-box, the guillotine door was closed so that g only had access to the chamber of confinement. Each g was given eight pre-training trials. Shock was delivered on the first five pre-training trials (the approximate mean number of shock trials taken by all §s in the OL phases of the two previous experiments), and the last three trials were performed without shock. This is com- parable to five escape trials and three avoidance trials in terms of shock delivery. The last three trials without shock is derived from the avoidance criterion used in the previous experiments. 31 Procedure in Avoidance Learning.--At the end of the confinement period on the eighth pre-training trial, g was placed on an open 7 in. long and 5 in. wide platform for 30 sec. At the end of this time, g was picked up and placed in the shuttle-box chamber appropriate to his avoidance learning group, the door raised, and the CS-US period started. After making an escape/avoidance response, each g was confined in the safe region of the shuttle-box for 30 sec. on each trial. Each § was run to a criterion of three successive avoidances. This avoidance situation was identical to the reversal-learning conditions of the two previous studies. Confinement control groups of 150 sec, and 30 sec. confinement were also run. Each control group was equally divided into four groups of two gs each on the basis of color of confinement and direction run in avoidance learning. The control groups were treated exactly as the experimental groups except that shock was omitted on all eight pre- training trials. Results As before, no significant heterogeneity of variance was found. The effects of confinement alone had no effect on gs' avoidance learning behavior. A 2 x 2 analysis of variance found no effects on number of trials to the avoidance criterion for direction of running (F«.05)(see Table 11, Appendix A). A 3 x 2 x 2 analysis of variance was performed on the experimental data on number of trials to the avoidance criterion. No simple effects for the color confined in 32 (rate. df=l/24, p>.05), length of confinement (F.05) were found. But the color confined in by color run to interaction (F=l7.586, df=l/24, p«(.Ol) and the color confined in by length of confinement by color run to interaction (F=7.330, df=2/24, p«<.Ol) were highly significant. None of the other two-way interactions were significant (see Table 12, Appendix A). The means for trials to criterion for both the control conditions and experimental conditions are presented in Figure 7. Since no color differences were found in the above analysrs, groups were Summed across like colors and different color groups. Groups run to like Colors in avoidance were run toward the color they had been confined in in pre-training (the C818 and CWTW groups . Groups run to Different Colors in avoidance were run toward a dlfltrent color in avoidance training than the color of pre-training confinement, or away from the color of pre-training confinement (the Chin and CWTB groups). A 2 x A analysis of variance including the three exyerimental groups and the combined control groups was then performed on the number of trials to criterion data (Lindquist, 1953). A main effect for Like vs. Different colors was found (F=ll 209, df=l/44, p«<.01) but length of confinement alone was not significant (F<(l). There was a significant direction of running (Like vs. Different colors) by length of confinement interaction (F=8.914, df=3/44,‘p<:.01)(see Table 13, Appendix A). Individual comparisons found that for Like vs. Different colors, the control of groups, the 30 sec. confinement groups, and the 30~120 sec. confinement groups did not differ significantly (t=.739, t= 853, and t=.996 respectively). 33 .Aoov ucmEmcchoo meanwmno umpa wo uoHoo osu Eouw xmBm can mQDOpw pom moqv osoEoCfiwcoo moficwmnuaowa wo uoHoo ago On one mascuw no“ HHH acmeummxm cfi mosoow Houuooo uooEmcchoo pom Hmucmeumaxm ecu How cowumuwho Ou mfimwuu wo Hogans some mLH .m mpowwm .moowuwvcou wcflcMmuH mucmcwo>< vow ucmEmoncou wCHCflmuHumHm oma OmH om on 8g III—.IL cog . omg mm.m mm.m mmzomo QOMHZOU HZMZMZHmZOU ONH oNH om Omfi omH no uom . om om i a Om.q ,l_ r. M®.N J NH.w . , m so on on x— on , oo oo o4 \xx x A\ \ xix \ X \W FL Cm.HH mmbomo A.05), or their interaction (17(1) (see Table 14, Appendix A). Individual comparisons showed that all experimental groups required slightly more time to leave the avoidance training shock chamber when it was the same as the pre-training confinement chamber, but none of these differences reached statistical significance. Comparisons between the latencies of the control conditions against the experimental confinement conditions were also made, The control conditions did not Seconds 35 8.67 8.08 7.33 6 921—1 LC DC. LC DC LC DC 30 3O 30- 30- 150 150 120 120 Pre~Training Confinement and Avoidance Training Conditions Figure 8. The latency for the first trial of avoidance training for the experimental and combined control groups (CC) in Experiment III. The figure shows the means for the experimental confinement groups run to the color of pre-training confinement (LL) and away from the color of pre-training conlinement (Dtk. 36 uiiiet significantly from any of the experimental conditions regardless of direction of running. An analysis for the number of regressions was also performed. The mean number of regressions in avoidance training for the combined confinement control and experimental conditions appear in Figure 9. A 2 x 4 analysis of variance including the control conditions found that neither length of confinement (F=i 026_ df=3/AA, p)? 05) nor color run to (F<:l) were significant however} the length of con- finement by color run to interaction was significant (F=2 926. df=3/44. p<:.05)(see lable 15, Appendix A). Individual comparisons for con~ finement groups run to the Like Colors or to Different Colors showed that the confinement control. the 30 sec confinement, and the 30- 120 sec. confinement conditions did not differ significantly (t= 879. t=l.685, t=l,015 respectively) However the 150 sec. confinement groups run to Like Colors made significantly fewer regressions than the 150 sec. confinement groups run to the Different Colors (t=2.0Dl. df=44, p‘