”E‘HE .“:?FEF“.$ 3533.!" 2-2.? LEE, SiithLEHi‘; {22323 THE RETENTEGR 05‘ AN 23645333313332?" ‘3. ZESPOEQSE 351122233 3392‘ 3332-3 332225225222 of— Pr: iii m‘iia 6.1523333327333322”. 333‘3: VEMSJT‘.‘ 3‘» a 1 MI 3‘ " “n 3' n 3222212222; 23222222,! 33332222922 WM; This is to certify that the thesis entitled THE EFFECTS OF STIT-‘IULUS SAi-IPLING ON THE RQTENTIOI‘I OF AN AVOIDAHCE RESPONSE presented by James Henry Reynierse has been accepted towards fulfillment of the requirements for Ph.D. degree in Psvcholor‘y . ‘ fl“ 2” ”322% LI BR A R Y Michigan State University (J m...- This is to certify that the thesis entitled w“ THE EFFECTS OF STIP’ZULUS SAMPLING ON THE RETENTION OF AN AVOIDM‘JCE RESPOI‘ISE presented by James Henry Reynierse has been accepted towards fulfillment of the requirements for 43.1411; degree in My .&, fl“ 2” Major gems” Michigan State University ABSTRACT THE EFFECTS OF STIMULUS SAMPLING ON THE RETENTION OF AN AVOIDANCE RESPONSE by James Henry Reynierse In the present series of experiments, rats were trained to avoid shock in a one-way shuttlebox to a criterion of two successive avoidances. Subsequent to the acquisition criterion, gs received either 1, 5, or 20 additional avoidance trials (called sampling trials). In Experiment I, §s were given sampling trials following an appropriate time-out period. The §s that received a sampling trial immediately after reaching criterion extinguished rapidly 24 hours later, that is, when the internal stimuli were associated with a relaxed state rather than an emotional state. The other gs (time-delay groups) received sampling trials approximately 40 min. after reaching criterion when the shock-associated stimuli had dissipated and when the relaxation-associated stimuli that would prevail 24 hours later could be sampled. In contrast to §s that received the sampling trial when the shock-associated stimuli pre- vailed, these time-delay §s were highly resistant to extinction. Furthermore, a single sampling of relaxation-associated stimuli was as effective as 20 sampling trials given over the same time period. This was interpreted as support for a non-incremental learning posi- tion. James Henry Reynierse The retention of an avoidance response was simultaneously investigated for three delay intervals (0 min., 40 min., and 24 hours). A typical retention curve was obtained which was approxi- mately log linear. In Experiment 11, shock-associated stimuli were reinstated after the time-out period by giving‘g an additional shock in the shock compartment. These §s were highly resistant to extinction after a delay of 24 hours even though the sampling trial followed the shock within 120 sec. Experiments III and IV were designed to clarify this finding. In Experiment III, § received an additional shock immediately upon reaching the acquisition criterion. A sampling trial was ad- ministered 120 sec. after this shock when the prevailing internal stimuli were associated with shock. These §s extinguished rapidly after a delay of 24 hours. Experiment III demonstrated that the high resistance to extinction found in Experiment 11 required the presence of the time-out period and was not due to the additional shock pgg_§g. In Experiment IV, shock associated stimuli were reinstated after the time-out period by giving § an additional shock in the shock compartment, as in Experiment II. The sampling trial, however, was administered after only 20 sec. had elapsed, and extinction oc- curred after a delay of 24 hours. An intermediate level of resis- tance to extinction was found for these §s. The results of Experi- ment IV were consistent with the hypothesis that §_may relax within TH ‘wk-i James Henry Reynierse 120 see. when the additional shock is preceded by a long time-out period. Thus the results of Experiment II were interpreted by positing that the avoidance response on the sampling trial was actually associated with relaxational stimuli and could thus maintain an avoidance response 24 hours later. The data from Experiments III and IV were most readily interpreted as indi- cating that a single shock was insufficient to reinstate fully the emotional stimuli of original learning. In general, the results supported a non-incremental learning position and emphasized the importance of stimulus sampling in z} , Approved / “fl? gm ttee Chairman Date /v‘l éiSI / avoidance learning. THE EFFECTS OF STIMULUS SAMPLING ON THE RETENTION OF AN AVOIDANCE RESPONSE By James Henry Reynierse A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1964 ' \ TO JAN AND DANNY ii ACKNOWLEDGEMENTS The author gratefully acknowledges the assistance given him by his major professor, M. Ray Denny. Without his willing assis- tance at every stage of development, this study would not have been possible. I am also indebted to the other members of my guidance committee for their help and encouragement: Doctors Charles Hanley and Stanley Ratner of the Department of Psychology and Doctor John King of the Department of Zoology. in TABLE OF CONTENTS Page ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . iii LIST OF “BLES O O O O O O O O O O O O O 0 O O O O O 0 O 0 v LIST or FIGURES . . . . . . . . . . . . . . . . . . . . . . v1 Chapter I 0 INTRODUCTION 0 O O O O O O O O O O O O O O O O O O 1 II. EXPERMNT I 0 O O O O O O O O O O O O O 0 O O O O 6 Method Results Discussion 111’ EXPERIMENT II 0 o e e e e e o e e e e e e e e o e 21 Method Results Discussion Iv. EXPERMNT III 0 0 O O O O O O I O O O O O O O O 0 25 Method Results Discussion V. EXPERIMENT IV 0 O O O O O O O O O O O O O O O O O 27 Method Results Discussion VI. GENERAL DISCUSSION . . . . . . . . . . . . . . . . 30 VII. SWRY O O O 0 O O O O O O O O O O O O O O O O O 33 REFERENCES 0 O O Q 0 O O O O O O O O O O O O O 0 O O O O O 35 APPENDICES O 0 O O O O O O O O O 0 O O O O O O O O O O O O 37 iv Table II. III. Iv. VI. VII. VIII. IX. X. LIST OF TABLES Differentiation of Groups as to Locus of Sampling Trial, Shock-cue, and Extinction . . . . . . . Mean Trials to Extinction and Standard Deviations for all Groups . . . . . . . . . . . . . . . . . Summary of Statistical Comparisons for Trials to Extinction in Experiment I . . . . . . . . . . . Summary of Analysis of Variance for Number of Trials to Extinction for the Retention Group . . . . . . Summary of Statistical Comparisons for Trials to Extinction in Experiment 11 . . . . . . . . . . Summary of Statistical Comparisons for Trials to Extinction in Experiement III . . . . . . . . . Summary of Statistical Comparisons for Trials to Extinction in Experiment IV . . . . . . . . . . . The Number of Trials to the Acquisition Criterion for all‘gs for all Groups (the Two Criterion Responses are Included) . . . . . . . . . . . . The Number of Trials to the Extinction Criterion for all §s for all Groups (the Two Extinction Criterion Responses are Included but the Sampling Trialisnot).................. Summary of Analysis of variance for Number of Trials to Acquisition Criterion for all Groups. . . . . . Pa ge 13 14 15 22 26 28 38 39 LIST OF FIGURES Figure Page 1. Retention of an Avoidance Response After 0 Min., 40 Min. and 24 Hours . . . . . . . . . 17 vi CHAPTER I INTRODUCTION Theoretical Considerations One trial learning theorists such as Guthrie (1952) and Estes (1959) propose that the total stimulus situation to which an organism responds consists of a set of stimulus elements. On successive trials different elements are sampled and associated in full with the responses that immediately follow them. In simple learning situations the external stimuli to which §_is responding remain relatively stable over time. Because of this relative constancy, the situational cues can be sampled by the organism in a brief period of time. But, the total stimulus complex which is effective in eliciting a response does not con- tain external stimulus elements alone. Internal stimuli are also a part of the total stimulus complex. Typically, these internal cues are less stable than the external stimuli and change with the passage of time. Without doubt, the internal stimuli at the beginning of an acquisition session differ considerably from the internal cues which are present after varying degrees of practice. With many trials most stimulus elements resulting from changes in the organism's internal state become associated with the response: therefore, there is little generalization decrement with the pas- sage of time. In an escape-avoidance learning situation, many of the internal stimuli are response produced stimuli which are related to the response to shock (emotional responses) rather than to the response being learned. Thus, as training progresses, the inter- nal state of the organism and the internal stimuli to which S is responding should change over time. According to elicitation theory (Denny and Adelman, 1955) § begins to relax after shock termination or after removal of the cues associated with shock. Thus, stimuli associated with relaxation are present on later trials. Indirect evidence for the occurrence of relaxation in avoidance learning is present in studies by Knapp (in press), Reynierse, Weisman and Denny (1963) and Denny and Weisman (in press). At the beginning of acquisition, stimuli associated with emotional responses constitute a sizable proportion of the stimu- li for eliciting the learned avoidance response. With the pas- sage of time, however, these stimuli presumably disappear as‘g relaxes. During avoidance learning, unless §_samples the stimuli associated with relaxation,‘§ will not learn to respond when these stimuli prevail. Presumably sampling of the major portion of these relaxation produced stimuli can occur either with a series of suc- cessful avoidance trials frequently presented or with a single trial that follows a non-shock period of comparable length. A pilot study1 showed that a series of successful avoidance trials administered after reaching the acquisition criterion 1 Independent pilot work by Robert K. Knapp yielded the same finding. (typically 3-8 trials) markedly increased resistance to extinction 24 hours later. This-contrasted with low resistance to extinction when no further trials were given after the criterion was reached. But the pilot work did not separate the effects of stimulus sampling from those of total trials. The implication from an incremental learning position is that the additional trials strengthen the habit. The implication from a non-incremental position is that the relax- ation-associated stimuli sampled during the extra trials (early ex- tinction) become associated with the avoidance response and thereby elicit and maintain the response on subsequent occasions. General Design The present study tests an incremental versus a non-incremental interpretation of an avoidance learning situation. Specifically, the question is whether a single sampling of the new internal stimuli strengthens an avoidance habit as well as many additional avoidance trials. According to the non-incremental position, there is an all- or-none association between the response and the new stimulus complex; thus, a single sampling of the new stimuli should maintain the avoidance response. According to the incremental position, the associative strength between a stimulus and a response increases gradually (Underwood and Keppel, 1962). Additional trials should therefore maintain the avoidance response better than a single trial. Should §fs internal state change without administering a series of successful avoidance trials, then a test of the one-trial position is possible. The test depends upon giving §_the opportunity to make the avoidance response after the internal stimuli associated with shock have dissipated, so that new, internal stimuli which occur with relaxation can become associated with the avoidance response. If gs under these conditions do not differ significantly in resistance to extinction from.§s given a series of successive avoidance trials, then a non—incremental position is supported. 0n the other hand, if §s given additional trials are more resistant to extinction, then an incremental position is supported. In addition, a non-incremental position requires that §s responding once in a I relaxed state be more resistant to extinction than gs which have not had the opportunity to do so. ’ Another approach to the incremental versus non-incremental question involves reinstating shock-cues after relaxation takes place, then giving §_an additional trial (sampling trial), and then extinguishing the avoidance response after 24 hours. The non- incremental position would predict that extinction should be rapid after 24 hours since the additional shock should eliminate or decisively attenuate, the relaxation-cue on the subsequent sampling trial. The present study also deals with the retention of an avoidance response as a function of delay interval. Moyer (1958) found essen- tially no differences in resistance to extinction between groups which had variable amounts of delay between acquisition and extinc- tion. He used a lengthy acquisition session (30 trials), however. Thus, relaxational stimuli may have become the cues for eliciting and maintaining avoidance responding even after long retention intervals. The present study provides information about reten- tion over a limited range of delay periods, introduced after the criterion had been attained. Experiments I and II were run simultaneously. Experiments III and IV were begun after the trends from Experiment II were established and were designed to identify the processes under- lying the effects found in Experiment II. The running of the §s overlapped in time, in all experiments. The present series of experiments could have been combined into one experiment containing eleven experimental groups. For expository purposes, however, the present format was considered preferable. Therefore, cross-experimental comparisons were con- sidered to be legitimate whenever such comparisons were necessary. CHAPTER II EXPERIMENT I Method Subjects.--The gs were 70 experimentally naive, male Sprague- Dawley albino rats from the colony maintained by the Psychology Department at Michigan State University. All were between 90 and 130 days old at the beginning of training. The‘gs were maintained in social cages with food and water always available. s; were assigned at random to seven experimental groups of 10 §s each. Procedure.--The apparatus consisted of a one-way shuttle- box having two discriminable compartments separated by a manually operated guillotine door. Each compartment was 18 in. long, 4 in. wide, and 14 in. high. The shock compartment was painted flat black and had a grid floor consisting of l/8 in. stainless steel grids spaced 5/8 in. apart, center to center. The grids were charged independently through a grid scrambler with a current of 1.1 ma. supplied by a C. J. Applegate stimulator, Model 228. The non-shock compartment was painted white and had a wooden floor. A 50 db transistorized buzzer (Malia and Curran, 1960) and the raising of the guillotine door served as the CS. A speaker mounted on the plexiglass top of the shock compartment delivered the auditory CS directly into the shock compartment. 6 For all §s in all groups the intertrial interval was fixed at 120 sec. with §_remaining for 100 sec. in the non-shock com- partment prior to being placed in the shock compartment for 20 sec. The CS-US interval was 5 sec., both CS and US being response terminated when S crossed to the non-shock compartment. All §s ran initially to a criterion of two successive avoidances. After § reached criterion, the shock stimulator was disconnected, so that on subsequent trials all responses became, in effect, avoid- ance responses. The seven groups used in the experiment are described at length below. Summary informatiOn describing these groups appear in Table I. 20-trialfg50up.--After reaching criterion, § received 20 additional acquisition trials. In order to prevent additional escape trials from strengthening the habit after criterion was reached and to insure further the development of a relaxed state, the shock stimulator was disconnected upon reaching criterion. Thus, all responses, whether they occurred before or after the CS-US interval used in training, were avoidances. Such a pro- cedure permitted the shock-associated stimuli present during ac— quisition to dissipate and be replaced by relaxation-associated stimuli. Extinction began after a delay of 24 hours. General Time:out Conditions The purpose of the various timefout conditions was to per- mit the shockeassociated stimuli present during acquisition to TABLE 1 DIFFERENTIATION OF GROUPS AS TO LOCUS OF SAMPLING TRIAL, SHOCKPCUE, AND EXTINCTION Locus of Sampling trial Reinstated After Criterion Group Shock-Cue Or Shock-Cue Extinction Experiment I 20-trials none trials 1-20 24 hour delay Time-out-in-home-cage none 40 min. 24 hour delay Time-out-in-non-shock- compartment none 40 min. 24 hour delay Time-out-five-swmpling- trials none 34-44 min. 24 hour delay 24-hour-retention none immediate 24 hour delay 40-minute-retention none none 40 min. delay 0-minute-retention none none immediate I Experiment IL? 7 Shock-cue-immediate- after 40 extinction minutes immediate immediate Shock-cue-delayed after 40 immediate with extinction minutes 120 sec. ITI 24 hour delay Experiment 131:; Immediate-shock-cue immediate immediate 24 hour delay Shock-cue-ZO sec. ITI (delayed extinction) Experiment IV, after 40 minutes immediate with 20 sec. ITI 24 hour delay ‘_f dissipate, independent of total number of avoidance trials. The time-out period for the various time-out conditions typi- cally included the following sequence. After criterion was reached, § remained for 100 sec. in the non-shock compartment, 36 min. in the home cage, 100 sec. in the non-shock compart- ment, and 20 sec. in the shock compartment. This sequence was immediately followed by a sampling trial that permitted §,to respond to the presumptive relaxational stimuli (the shock stimu- lator was disconnected). A sampling trial was an avoidance response under the relaxed state. FollowingIS's response, S remained for 100 sec. in the non-shock compartment. §_was then returned to the home cage for 24 hours. This procedure made the ZO-trial group and the time-out groups comparable with respect to the time when the twentieth trial or the single additional trial occurred. Deviations from the general time-out procedure, for specific time-out conditions, described that condition and differentiated it from all others. The three time-out groups are described below. Time-out-in»home-cage group.--After reaching criterion and remaining for 100 sec. in the non-shock compartment, §_was re- turned to its home cage for 36 min. Then §_was returned to the non-shock box for 100 sec. and received one sampling trial under the new internal stimulus conditions. Extinction began after a delay of 24 hours. 10 Time-out-in-non-shock-compartment ggoup.--After reaching criterion, § received 19 simulated trials (38 min.). In each simulated trial § spent 100 sec. in the non-shock compartment and 20 sec. in a neutral cage (an individual cage distinct from the social home cages) before being again placed in the non- shock compartment. This group differed from the time-out-in- home-cage group in terms of the place where the change in in- ternal state occurred. Again, one sampling trial was given under the new internal stimulus conditions. Extinction began after a delay of 24 hours. Time-out-five-sampligg;trials group.--After reaching criterion, g was returned to its home cage for 32 min. The ‘S was then returned to the non-shock compartment for 100 sec. and received five successive sampling trials with the shock stimulator disconnected under the new internal stimulus con- ditions. Extinction began after a delay of 24 hours. This group was included as a special control for the potency of a single sampling trial during the relaxed state. Two §s in the 20-trial group reached the extinction cri- terion during the 20 sampling trials. When this occurred training was discontinued. But, after a delay of 24 hours, these .§3 were given further extinction trials and were included in the analysis. In the time-out groups, two gs failed to respond during the sampling trial. When this occurred S was extinguished after a delay of 24 hours and was also included in the analysis. 11 The inclusion of these §s was necessary since excluding them introduces an element of selectivity not present in other groups. 24-hour-retention group.--After reaching criterion, §_was immediately given one sampling trial and was returned to its home cage. Extinction began after a delay of 24 hours. 40-minute-retentionigroup.--After reaching criterion, S was returned to its home cage for 36 min. as in the time-out groups, prior to being placed in the non-shock compartment for 100 sec. Extinction began 40 min. after reaching criterion. 0:9inute-retentionfgrgup.--After § reached criterion, ex- tinction began immediately. For all §s that received delay periods prior to extinction, §_spent the entire time in its home cage. Extinction was con- sidered to be complete when §_failed to respond to the CS for a 60 sec. period on two successive trials. When § did not respond to the 2S, §_remained in the shock compartment for 120 sec. until the next trial was administered. Results All §s were considered to have received one or more sampling trials after the acquisition criterion was reached and before ex- tinction was begun. In analyzing resistance to extinction, both the acquisition criterion trials and the sampling trials were ex- cluded. Only extinction trials following the appropriate delay periods were considered in the analysis of results. For example, 12 the 20 additional avoidance trials in the 20-tria1 group were not included in the analysis. To maximize comparability between groups, the first trial after criterion in the 0-minute and 40-minute reten- tion groups was arbitrarily considered to be a sampling trial and was excluded from the analysis. The results are analyzed exclusively in terms of mean number of trials to extinction. The mean number and standard deviations for all groups are presented in Table 11. Statistical comparisons of import are summarized in Table III and the raw data for all groups are presented in Table VIII and Table IX (Appendix). Through- out, all statistical tests are two-tailed. A single classification analysis of variance (Table X in Appendix) was performed on the number of trials necessary to reach the learning criterion for all eleven groups in the four experiments. The overall test was not significant (F = .665) indicating that dif- ferences in learning could not reasonably account for the obtained effects during extinction. The ZO-trial group took significantly more trials to extin- guish than the 24-hour-retention group (t== 2.83, df== 9,2 p4: .02). Without this finding the remainder of the study would have little meaning. The time-out-in-home-cage group and time-out-in-non-shock- compartment group were treated essentially the same in terms of 2 Where F-tests indicated heterogeneity of variance, Welch's formula (Winer, 1962) was used to obtain the appropriate degrees of freedom. 13 TABLE II MEAN TRIALS TO EXTINCTION AND STANDARD DEVIATIONS FOR.ALL GROUPS v f Group Mean S.D. VF? Experiment I 20-trials 24.7 23.3 Time-out-in-home-cage 20.7 29.409 Time-out—in-non-shock-compartment 25.6 29.239 Cmmbined-time-out 23.15 28.652 Time-out-five-sampling-trials 9.1 7.219 24-hour-retention 3.7 2.495 40-minute-retention 17.7 15.319 0-minute-retention 36.1 32.518 V we. 3 Experiment Iljfi ‘w Shock-cue-immediate-extinction 39.1 14.043 Shock-cue-delayed-extinction 31.7 23.238 Experiment4IIlfi Immediate-shock-cue 11.2 13.782 r“ Experiment IV Shock-cue-ZO sec. ITI 18.5 17.619 14 TABLE III SUMMARY OF STATISTICAL COMRARISONS FOR TRIALS TO EXTINCTION IN EXPERIMENT I Comparisons Mean t df P 20-trials group 24.7 with 2.83 9* 4 .02 24-hour-retention group 3.7 Time-out-in-home-cage group 20.7 with .748 18 N.S. Time-out-in-non-shock- compartment group 25.6 Combined-time-out group 23.15 with 3.02 19* 4f .01 24-hour-retention group 3.7 Combined-time-out group 23.15 with 2.07 23* < .05 Time-out-five-sampling- trials group 9.1 20-trials group 24.7 with .313 28 N.S. Combined-time-out group 23.15 20-trials group with Time- 24.7 out-five-sampling-trials 2.02 11* <:.10 group 9.1 Time-out-five-sampling- trials group 9.1 with 2.24 11* (.05 24-hour-retention group 3.7 * Welch's formula was used to obtain the appropriate degrees of freedom. 15 the locus of the sampling trial and since they did not differ sig- nificantly (t a .748) they were combined for further analyses. This combined-time-out group also took significantly longer to extinguish than the 24-hour-retention group (t = 3.02, df = 19, p.¢L.Ol) but did not differ from.the 20-tria1 group (t.= .313). Thus the non-incremental position is supported. The combined- time-out group extinguished significantly slower than the time- out-five-sampling-trials group (t a 2.07, df== 23, p44 .05), indirectly adding further support to the non-incremental position. The trend becomes somewhat obscured, however, as borderline significance was obtained when the 20-trial group and five-sampling- trials group were compared. That the 20-trials group extinquiahed more slowly than the five-sampling-trials group (ts: 2.02, df = 11, p2£L.10) appears to support an incremental position. But, the time-out-five-sampling-trials group took significantly longer to extinguish than the 24-hour-retention group (t = 2.24, df== 11, 1:41.05) indicating that the effects of sampling are still present. TABLE IV SUMMARY OF ANALYSIS OF VARIANCE FOR NUMBER OF TRIALS TO EXTINCTION FOR THE RETENTION GROUPS Source of variation d.f. Mean Square F Between groups 2 2640.533 6.101* Within groups 27 432.781 Total 29 * pu41.01 16 The results for the 0-minute, 40-minute, and 24-hour-retention groups are presented in Fig. l. A single classification analysis of variance (Table IV) was significant (F = 6.101, df = 2/27, p41,.01). Further comparisons indicated that the O-minute-retention group took longer to extinguish than the 40-minute-retention group (p.4L.01) and the 40-minute-retention group took longer to extinguish than the 24-hour-retention group (p.41.05). The obtained relationship appeared to be 103 linear. Discussion Pilot work previously indicated that a series of successful avoidance trials after criterion increased resistance to extinction 24 hours later as compared with a procedure where there were no further trials after criterion. That the 20-tria1 group was sig- nificantly more resistant to extinction than the 24-hour-retention group confirms this finding. The implication of this effect from an incremental learning position is that the additional trials strengthened the habit. The implication from a non-incremental position is that the relaxation-associated stimuli that were sam- pled during the 20 additional trials (early extinction) became associated with the avoidance response and that one sampling trial, ‘ appropriately placed, should be as good as 20 trials. The §s in the combined-time-out group received one sampling trial after 40 min. Presumably a large portion of the internal stimuli associated with shock have dissipated by this time and have been replaced by relaxation-associated stimuli. On the other hand, .p ‘6’ o MEAN TRIALS TO EXTINCTION . N . O 5 17 6 .6'7 2'4 RETENTION INTERVAL IN HOURS (LOG SCALE) Figure 1. Retention of an avoidance response after 0 min., 40 min. and 24 hours. 18 §s in the 24-hour-retention group received one sampling trial immediately after reaching criterion when the shock-associated stimuli still prevailed. In the 24-hour-retention group the relaxation-associated stimuli could not be sampled and could not be associated with the response. That the combined-time-out group took significantly longer to extinguish than the 24-hour- retention group clearly supports a non-incremental stimulus sampling position. That is, the opportunity to sample relaxation- associated stimuli in the avoidance situation emerges as a criti- cal variable for continued avoidance responding after 24 hours. Furthermore, since the 20-trial group and.the combined-time-out group do not differ significantly in resistance to extinction, the effect appears to be independent of total avoidance trials. According to the incremental position, five sampling trials should result in greater resistance to extinction than a single sampling trial. This was not the case; thus, the incremental position was not supported. The problem with the time-out-five- sampling-trials group is that they did so poorly, extinguishing faster than the groups that received a single sampling trial. At least part of this discrepancy can be explained by assuming that these five sampling trials were highly effective extinction trials occurring when §| was in a relaxed state. According to elicitation theory relaxation is the competing response that is responsible for extinction in avoidance learning. Thus, the gradual development of relaxation, as response, constitutes the competing responses that are responsible for eventual extinction 19 in all groups. Considerable relaxation, chained in over trials, could have been present in the time-out-five-sampling-trials group causing them to extinguish rapidly after 24 hours.3 An apparent inconsistency is present since strengthening is posited for the single sampling-trial groups. In the elici- tation framework, however, relaxation mediates both acquisition of avoidance responses and the acquisition of competing responses during extinction. Presumably one sampling trial would not per- mit sufficient relaxation for extinction effects. It must be admitted, however, that the results from.the time-out-five- sampling—trials group are rather puzzling and requires further research. In this connection, the 20 additional avoidance trials in the 20-trial group could be called extinction trials. But, it is preferable to conceive of most of them at least as sampling trials. If they were extinction trials the 20-tria1 group should extinguish faster than groups that did not receive such trials. This is clearly not the case as the 20-tria1 group was more resise tant to extinction than the 24-hour-retention group. Indeed, what may be operationally considered to be an extinction trial may deviate considerably from "the process of extinction". What we typically call extinction probably contains both stimulus sampling and extinction functions. Pilot work supports this contention since gs that received 3-5 widely spaced sampling trials (10 min. intervals) were as resis- tant to extinction as §s which received only one sampling trial after the time-out period. 20 The results for the retention of an avoidance response are also consistent with the present analysis of stimulus sampling effects. Retention is best immediately after § reaches criterion, that is, when‘g successively samples small changes in its internal state. With a brief delay between criterion and extinction, re- tention is intermediate. With a delay of 24 hours, retention is poorest. Although the relationship appears to be log linear, it is premature to assume log linearity since only three points are represented on the curve. CHAPTER III EXPERIMENT II Experiment 11 was designed to investigate the effects of sampling for internal stimuli when the shock cues are reinstated immediately following a time-out period. seems Subjects.--The‘§s were 20 male albino rats, 90 to 110 days old and assigned at random to two groups of 10 Se each. Procedure.--The apparatus and general procedure were the same as in Experiment 1. After reaching criterion, §s were returned to their home cage for 36 min. as in the time-out con- ditions in Experiment I. § then spent 100 sec. in the non- shock compartment prior to being placed in the shock compart- ment for 20 sec. With the termination of this 20‘sec. period, § received a shock (unpaired with the CS) until §_escaped shock (shock-cue). After remaining in the non-shock compartment for 100 sec., g was again placed in the shock compartment before receiving a sampling trial with the CS alone. Half of the‘gs were extinguished immediately and half after a 24 hour delay. Results The mean trials to extinction and standard deviations are presented in Table 11. Statistical comparisons involving groups 21 22 in both Experiments I and II are summarized in Table V. The shock-cue-immediate-extinction group did not differ significantly from the O-minute-retention group (t = .267) indicating that the effect of the additional shock trial was negligible. The shock-cue-delayed-extinction group did not differ from the shock-cue-immediate-extinction group (t = 1.72). Also, the shock-cue-delayed-extinction group was significantly more resis- tant to extinction than the 24-hour retention group (t 8 3.79, df- 9, p4 .01) but did not differ significantly from the com- bined-time-out group (t = 1.73, df= 28, p4.10). The convincing resistance to extinction in the shock-cue-delayed-extinction group was not predicted and requires further analysis. TABLE V SUMMARY or STATISTICAL COMPARISONS FOR TRIALS To EXTINCTION 1N EXPERIMENT II Comparisons Mean _2 d.f. fifi p Shock-cue-immediate- extinction group 39.1 V with .267 18 N.S. O-minute-retention-group 36.1 Shock-cue-immediate- extinction group 39.1 with 1.72 18 N.S. Shock-cue-delayed- extinction group 31.7 Shock-cue-delayed- extinction group 31.7 with 3.79 9 A1.01 24-hour-retention group 3.7 Shock-cue-delayed- extinction group 31.7 with 1.73 28 ,AL.10 Combined-time-out group 23.15 23 Discussion Following a time-out period and a shock trial, it is apparent that the §s are very resistant to extinction, independent of the delay between the shock and the beginning of extinction. But, it is unlikely that the shock pg£_gg contributed to this resistance to extinction since the additional shock yielded only slight and non-significant superiority over groups that did not receive an additional shock after reaching criterion. These results compare well with the finding that additional errors (shock) after the ini- tial success (avoidance) do not strengthen the avoidance habit (Theios, 1963). The resistance to extinction in the shock-cue-immediate- extinction group is of the same order as the resistance to extinc- tion in the O-minute-retention group. In both groups, extinction is occurring under stimulus conditions that are quite similar to those that prevail during acquisition. Any differences between acquisition and extinction accrue gradually, and generalization decrement is kept at a minimum. The high degree of resistance to extinction in the shock-cue- delayed-extinction group was not predicted but this finding is not necessarily contradictory to a stimulus sampling hypothesis. There are several possible explanations. One explanation is that the additional shock itself may strengthen the habit although the unlike- lihood of this contingency has already been discussed. Another explanation is based on the notion that relaxation pre- sumably occurs faster and faster with successive opportunities to 24 relax. Furthermore, a single shock following a long period of relaxation should result in a faster onset of relaxation, than for a series of repeated shock trials. After a 40 min. period, a single intertrial interval of 120 sec. after shock may result in enough relaxation so that § can sample relaxation-associated stimuli on the subsequent avoidance trial. Experiments III and IV were designed to investigate these alternatives. CHAPTER IV EXPERIMENT III Experiment III was designed to investigate the effects of a shock-cue when it was not preceded by a time-out period. In other words, does an additional shock per se significantly in- crease resistance to extinction. Method Sub]ects.--The §s were 10 male albino rats, 90 to 110 days old. Procedure.-~The apparatus and general procedure were the same as in Experiment 1. After reaching criterion, § remained in the non-shock compartment for 100 sec. prior to being placed in the shock compartment for 20 sec. Then § immediately received a shock-cue (as in Experiment II) and a sampling trial after 120 sec. Extinction began after a delay of 24 hours. Results The mean trials to extinction and the standard deviation for the immediate-shock-cue group are presented in Table 11. Statistical comparisons involving groups from Experiments I, II, and III are summarized in Table VI. The immediate-shock-cue group extinguished significantly faster than the shock-cue-delayed- 25 26 extinction group (t- 4.80, df = 18, p4.001) but did not differ significantly from the 24-hour-retention group (t a 1.69) indi- cating that the additional shock had negligible effects. TABLE VI SUMMARY OF STATISTICAL COMPARISONS FOR TRIALS TO EXTINCTION IN EXPERIMENT III Comparisons Mean t d.f. p Shock-cue-delayed- extinction group 31.7 with 4.80 18 41.001 Immediate-shock-cue group 11.2 Immediate-shock-cue group 11.2 with 1.69 18 N.S. 24-hour-retention group 3.7 Discussion It is clear that the time-out condition and not Shock per se is critical for the high level of resistance to extinction after 24 hours. Again, the effect of the additional shock yields slight and non-significant superiority over comparable groups. CHAPTER V EXPERIMENT IV Experiment IV was designed to examine the effect of an ad- ditional shock following a time-out period under conditions in which a short time period intervened between the shock-cue and the subsequent sampling trial. This is a test of the hypothesis that relaxation can take place within 120 sec. when this interval is preceded byga long time-out period.. Method Sub]ects.--the §s were 10 male albino rats, 90 to 110 days old. Procedure.--The apparatus and general procedure were the same as in Experiment 1. After reaching criterion, §_was returned to its home cage for 36 min. The §_was then placed in the non- shock compartment for 100 sec. prior to being placed in the shock compartment for 20 sec. The §_then received an additional shock trial as in Experiment 11. After 20 sec. (10 sec. in the non- shock compartment and 10 sec. in the shock compartment) §_received a sampling trial. Extinction began after a delay of 24 hours. Results The mean trials to extinction and the standard deviation for the shock-cue-ZO sec. ITI group are presented in Table 11. Statistical 27 28 comparisons involving groups from.Experiments I, II, III, and IV are summarized in Table VII. The shock-cue-ZO sec. ITI group ex- tinguished significantly faster than the shock-cue-delayed-extinc- tion group (t = 2.86, df a 18, p4 .02) and significantly slower than the 24-hour-retention group (t a 2.63, df = 9, pz. .05) pro- viding sufficient support for theflhypothesis that relaxation can take place within 120 sec. when this interval is preceded bya. prior time-out period. The shock-cue-ZO sec. ITI group and the immediate-shock-cue group did not differ significantly (t=:2.07, df = 18, p4..10). TABLE VII SUMMARY OF STATISTICAL COMRARISONS FOR TRIALS TO EXTINCTION IN EXPERIMENT IV hf.— Comparisons Mean t d.f. p ,— Shock-cue-delayed-extinction group 31.7 with 2.86 18 ¢£.02 Shock-cue-Zo-sec. ITI group 18.5 Shock-cue-ZO-sec. ITI group 18.5 with 2.63 9 21.05 24-hour-retention group 3.7 Shock-cue-20-sec. ITI group 18.5 with 2.07 18 41.10 Immediate-shock-cue group 11.2 Discussion The finding that the shock-cue-ZO sec. ITI group extinguished significantly faster than the shock-cue-delayed-extinction group is 29 consistent with the hypothesis that § relaxes within a 120 sec. interval when the additional shock is preceded by a long time-out period. This in turn means that the avoidance response on the sampling trial was associated with relaxational stimuli. With the 20 sec. ITI this was not the case, and §_extinguished faster than when the intertrial interval was 120 sec. The fact that relaxation may occur during a 120 sec. inter- trial interval cannot account for all of the superiority of the shock-cue~delayed~extinction group. This is evident since the shock-cue-ZO sec. ITI group represents an intermediate condition that is significantly more resistant to extinction than the 24- hour-retention group. It is still possible that the shock itself is critical, and this is partially supported by the non-signifi- cant differences between the shock-cue-ZO sec. ITI group and the immediate-shock-cue group. But, the majority of evidence indicates that the effects of the shock pg£_gg_are negligible. Some additional factor then must account for the fact that resistance to extinction in the shock-cue-ZO sec. ITI group is significantly greater than that present in the 24-hour-retention group. An obvious interpretation, from the data and from the observed behavior of §s, is that a single shock did not reinstate the emotional responses that were associated with shock and which were present during acquisition. Such an interpretation is reasonable since 75% of the §s given an additional shock only received a brief shock (one see. or less), In other words, the majority of‘gs received a sampling trial while they were partially relaxed and this presumably increased resis- tance to extinction 24 hours later. CHAPTER VI GENERAL DISCUSSION Avoidance learning research has previously yielded results that can be interpreted as support for a non-incremental learning position. Madsen and McGaugh (1961) used a passive avoidance situation and Maetsch (1959) used an active avoidance situation. They demonstrated one~trial learning under optimal acquisition conditions. Theios (1963) has mathematically described an avoid- ance task with mathematical models which assume all-or-none proper- ties. The present study represents an approach which emphasizes the post-acquisition conditions under which relaxational stimuli are sampled and associated with the avoidance response. The time when §,is permitted to sample relaxational stimuli is critical. This is evident as §s extinguish rapidly after 24 hours if previously they did not receive a sampling trial while in a relaxed condition. It is true that the first extinction trial coming 24 hours later, permits §_to sample relaxational stimuli in the avoidance situation. But, by this time, generalization decrement is maximal, and the avoidance response is so weak, for example, that half the §s in this group never even responded prior to reaching the extinction criterion (see Table IX in Appendix). Thus there is a limited time period during which the effect can occur. The limiting conditions depend upon two related stimulus- 30 31 change gradients. Soon after the last shock is received, § begins to relax. Progressively, over time, relaxation-associated stimuli begin to accumulate. The gradual accumulation of these relaxational stimuli represents one changing stimulus gradient. Paralleling this dimension is another gradient related to shock-associated stimuli. Soon after the last shock is received, shock-associated stimuli begin to dissipate. In other words, as S relaxes, relaxational stimuli replace shock-associated stimuli. The gradually changing stimulus conditions represent gradually increasing generalization decrement since intially only the shock-associated stimuli were conditioned to the avoidance response. At some point in time, the gradient for shock-associated stimuli and the gradient for relaxation-associated stimuli must intersect. Where this point is located is undoubtedly a function of individual differences in §s' capacity to relax as well as other psychological variables which affect the rate at which relaxational stimuli replace shock-associated stimuli. These individual differences may well account for some of the variance that was found. The point at which §.samples the relaxational stimu- li in the avoidance situation will determine the effectiveness of that sampling trial. If the sampling trial occurs where the gradients intersect, then its effectiveness is maximized. If it occurs before the gradients intersect, that is,when the shock-associated stimuli have dissipated somewhat but still predominate, then the greater portion of stimuli sampled will be shock-associated stimuli. After 24 hours when relaxation-associated stimuli prevail there will be 32 considerable generalization decrement. Extinction will be fairly rapid since more shock-associated stimuli than relaxation-associated stimuli have been conditioned to the avoidance response. On the other hand, if the sampling trial occurs after the gradients inter- sect, then the sampling trial can also be ineffective. This is the case when the generalization decrement between acquisition and the sampling trial is considerable. With relaxation-associated stimuli predominating there may be insufficient shock-associated stimuli to mediate an initial avoidance response; and therefore, the relaxation- al stimuli cannot be conditioned to avoidance. Kamin, Brimer, and Black (1963) have shown that there is a lack of parallelism between fear and instrumental avoidance learning. Their study revealed that fear cannot maintain an avoidance response and that some other factor must be responsible for maintained avoid- ance responding. In this connection, the present study shows that relaxational stimuli, when sampled and associated with the avoidance response, can elicit and maintain avoidance responding. CHAPTER VII SUMMARY In the present series of experiments, rats were trained to avoid shock in a one-way shuttlebox to a criterion of two succes- sive avoidances. Subsequent to the acquisition criterion, §s re- ceived either 1, 5, or 20 additional avoidance trials (called sampling trials). In Experiment I,‘§s were given sampling trials following an appropriate time-out period. The §s that received a sampling trial immediately after reaching criterion extinguished rapidly 24 hours later, that is, when the internal stimuli were associated with a relaxed state rather than an emotional state. The other‘gs (time-delay groups) received sampling trials approxi- mately 40 min. after reaching criterion when the shock-associated stimuli had dissipated and when the relaxation-associated stimuli that would prevail 24 hours later could be sampled. In contrast to S; that received the sampling trial when the shock-associated stimuli prevailed, these time-delay g; were highly resistant to extinction. Furthermore, a single sampling of relaxation-associ- ated stimuli was as effective as 20 sampling trials given over the same time period. This was interpreted as support for a non- incremental learning position. The retention of an avoidance response was simultaneously investigated for three delay intervals (0 min., 40 min., and 24 33 34 hours). A typical retention curve was obtained which was approxi- mately log linear. In Experiment 11, shock-associated stimuli were reinstated after the time-out period by giving‘g an additional shock in the shock compartment. These‘gs were highly resistant to extinction after a delay of 24 hours even though the sampling trial followed the shock within 120 sec. Experiments III and IV were designed to clarify this finding. Experiment III demonstrated that the effect required the presence of a time-out period and not the additional shock pgg_gg, Experiment IV was consistent with the hypothesis that §,may relax within 120 sec. when the additional shock is preceded by a long time-out period. Thus the results of Experiment 11 were interpreted by positing that the avoidance re- sponse on the sampling trial was actually associated with relaxa- tional stimuli and could thus maintain an avoidance response 24 hours later. The data from.Experiments III and IV were most readily interpreted as indicating that a single shock was insuf- ficient to reinstate fully the emotional stimuli of original learning. In general, the results supported a non-incremental learning position and emphasized the importance of stimulus sampling in avoidance learning. REFERENCES Denny, M. R. and Adelman, H. M. Elicitation theory: I. Analysis of two typical learning situations. Psychol. Rev., 1955, §_2_, 290-296. Denny, M. R. and Weisman, R. G. Avoidance behavior as a function of length of nonshock confinement. J. comp. physiol. Psychol. (in press). Estes, W. K. The statistical approach to learning theory. Psycholo- gy; A Study of a Science. Vol. 2, McGraw-Hill: New York, 1959. Guthrie, E. R. The s chology of learning. Harper and Brothers: New York, 1952. Kamin, L. J., Brimer, C. J., and Black, A. H. Conditioned suppression as a monitor of fear of the CS in the course of avoidance training. J. comp.4physiol. Psychol., 1963, 22, 497-501. Knapp, R. K. The acquisition and extinction of avoidance with similar and different shock and escape situations. J. comp. physio22Psychol. (in press). Mastsch, J. L. Learning and fixation after a single shock trial. J. comp. physiol. Psychol., 1959, 22, 408-410. Madsen, M. C. and McGaugh, J. L. The effects of ECS on one-trial avoidance learning. J. comp. physigl. Psychol., 1961, 24, 522-523. Malia, J. L. and Curran, C. S. A reliable, low-cost generator for audio stimuli. J. exp. anal. Behav., 1960, 2, 200. Moyer, K. E. Effect of delay between training and extinction on the extinction of an avoidance response. J. comp. physiol. Psychol., 1958, 22, 116-118. Reynierse, J. H., Weisman, R. G., and Denny, M. R. Shock compart- ment confinement during the intertrial interval in avoidance learning. Psychol. Rec., 1963, 22, 403-406. Theios, J. Simple conditioning as two-stage all-or-none learning. Psychol. Rev., 1963, 12, 403-417. 35 36 Underwood, B. J. and Koppel, G. One-trial learning? J. verb. learn. verb. Behav., 1962, 2, 1-13. Winer, B. J. Statisticalgprinciples in experimental desigg. McGraw-Hill: New York, 1962. APPENDICES 37 38 TABLE VIII THE NUMBER OF TRIALS TO THE ACQUISITION CRITERION FOR ALL SS FOR.ALL GROUPS (THE TWO CRITERION RESPONSES ARE INCLUDED) Group Subjects Mean 1, .2 A1 .2 3 :4. 5 6. 7 .8—e9: 10 V - Experiment I ZOvtrials 6 4 4 4 7 6 ll 7 5 5 5.9 Time-out-in-home-cage 6 5 6 7 3 5 5 5 4 5 5.1 Timevout-in-non-shock- compartment 4 -5 6 10 8 5 4 3 9 10 6 . 4 Time-out-five-sampling- , trials 6 7 4 5 .4 4 9 9 3 7 5.8 24-hour-retention 4 8 8 5 6 6 4 3 6 7 5.7 40-minute-retention 3 9 3 5 4 12 6 7 8 5 6.2 O-minute-retention 8 9 6 10 8 5 6 4 5 5 6.6 I Experiment ll. fi~fi Shock-cue-immediate- extinction 12 9 5 8 5 5 5 5 5 7 6.6 Shock—cue-delayed- extinction 10 6 8 6 3 4 6 6 6 5 6.0 2*_ (Experiment III Immediate-shock-cue 6 8 7 5 5 3 4 6 4 7 5.5 " V Experiment 21 Shock-cue-ZO sec. ITI 6 5 6 6 4 4 3 8 5 4 5.1 39 TABLE IX THE NUMBER OF TRIALS TO THE EXTINCTION CRITERION FOR.ALL SS FOR ALL GROUPS (THE TWO EXTINCTION CRITERION RESPONSES ARE INCLUDED BUT THE SAMPLING TRIAL IS NOT) m Group Subjects Mean 12345678910 Expeiyeentgl 20-trials ’9 54 32 15 2* 2* 37 15 71 10 24.7 Time-out-in-home-cage 27 15 3 99 2* 3* 5 27 23 3 20.7 Time-out-in-non-shock compartment 14 7 2 39 49 3 88 2 48 4 25.6 Time-out-five-sampling- trials 23 15 17 2 8 3 ll 6 4 2 9.1 24-hour-retention 2 2 3 9 2 2 7 5 2 3 3.7 40-minute-retention 9 44 2 2 17 15 16 29 3 40 17.7 O-minute-retention 29 119 24 12 62 13 30 20 35 17 36.1 . Experiment I; v f a Shock-cue-immediate-. extinction 56 20 44 21 27 38 29 54 48 54 39.1 Shock-cue-delayed- extinction l6 5 7 54 59 58 17 61 22 18 31.7 Experiment‘III Immediate-shock-cue 42 2 2 24 4 3 2 7 23 3 11.2 W Shock-cue-ZO sec. ITI 44 2 3 7 2 22 12 34 ll 48 18.5 * 2s that either extinguished during the 20 additional trials (20- trial group) or failed to respond.on the sampling trial (time-out-in- home- cage group) . 40 TABLE X SUMMARY OF ANALYSIS OF VARIANCE FOR NUMBER OF TRIALS TO ACQUISITION CRITERION FOR.ALL GROUPS Source of variation d.f. Mean Square F Between groups 10 2.82 . 665 Within groups 99 4. 24 Total 109