ESCAPE VARLQBLES AR’D fiVGmANCE CQNDH‘IGNING: E‘Wé} @5133 CL GN E3EGCESSES THESES FOR THE DEGREE OF M. A. MICHIGAN STATE UNIVERSITY CURTIS ALAN BAGNE ‘3 LIBRARY "“ 1 Michigan State ' 1..., University :3 1H ES‘S ABSTRACT ESCAPE VARIABLES AND AVOIDANCE CONDITIONING: TWO EXTINCTION PROCESSES by Curtis A. Bagne The role of an escape contingency on behavior under aversive control is not clear. A number of studies suggest that aversive events which are response terminated are less aversive than those from.which escape is prevented. Experiment 1 demonstrated that hooded rats trained to avoid shock in a one-way box under conditions of no escape showed the same resistance to extinction as §s allowed to escape during training. These results are not readily interpretable because of a possible effect of method of transferring §s from the shock area to the safe area fol- lowing the aversive US. In Experiment 2 several treatments were interpolated between the acquisition and extinction phase of avoidance learning. The treatment variables yielded results that can be summarized as follows: 1. Interpolated escapable shock increases resistance to extinction more than the same amount of in- escapable shock; 2. Method of transferring §s from the shock area to the safe area following inescapable shock has Curtis A. Bagne an effect that could account for the negative results of ExPeriment 1; 3. In general, resistance to extinction is greater when conditions in the shock and safe areas are similar during interpolated shock than when they are different; 4. In general, resistance to extinction is increased when line of sight from shock to safe areas is blocked. An analysis of response latencies for trials during extinction suggests that all §s can be classified into one of two extinction process groups. Some §s consistently respond rapidly until the extinction criterion is reached and extinguish during the first few trials of an extinction session. These §s are identified as belonging to the freeze group. The extinction response latencies of §s iden- tified as belonging to the relax group respond less rapidly and with greater variability as extinction progresses. These §s are likely to extinguish at any time during an extinction session. Thus, the assumption that number of trials to extinc- tion of an avoidance response constitutes a unitary measure of the effectiveness of certain treatments on the acquisi- tion and extinction on an avoidance response is untenable. In the experimental analysis of avoidance learning it is important to remember that treatment variables may have effects that are process specific. ESCAPE VARIABLES AND AVOIDANCE CONDITIONING: TWO EXTINCTION PROCESSES By . wfi Curtis A. Bagne A THESIS Submitted to Michigan State university in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1967 DEDICATION TO FARIDEH ii ACKNOWLEDGMENTS The writer wishes to express special thanks to Dr. M. Ray Denny, chairman of his committee, for his assistance throughout the preparation of this report. Also he wishes to convey thanks to Drs. R. Levine and M. E. Rilling for their helpful criticism and advice. iii ACKNOWLEDGMENTS . 0 LIST OF TABLES . . LIST OF FIGURES . . INTRODUCTION . . . EXPERIMENT 1 . Subjects 0 0 Apparatus . . Procedure . . Restllts Q ¢ . Discussion . EXPERIMENT 2 . o . Introduction Subjects . . Apparatus . Procedure . ROSUltS o 0 Discussion 0... SUMMARY . o . . . . BIBLIOGRAPHY . . . TABLE OF CONTENTS 0 0.... 00.... O .00.. 0.0... O .0... 0.0... Page OOOOOOOOOOCC 111 0000000060.. v ............ vi 0 O O O O O O O D O I O O O O O D O O O O O O C O O O I O O O O O O O O O D O O D O O O O O 0 O‘ U 0' O O D O O D O 9 I O O O O O O o I t D O O o bob-I OONNNG O\ mbwww b.) l—‘ O O O O O O O O O O I D o O O 0 C I Q 0 O O Q 0 C O O O O O O O O O I O O O O O O C O O O O I O O O C O O O O O C O 0 O o o o o o o o o o a 0 b O I o a o o o o 0 O o O 41 I O O O I O O Q 0 O O G 43 iv Table 10 11 12 LIST OF TABLES Measures of Acquisition - All Groups . . . ANOVA: Number of Shocks During Acquisition ANOVA: Total Shock Received During ACQUiBition 6 o o o o o o o o o o o a . ANOVA: Total Interpolated Shock . . . . . Trials to Extinction - Individuals by Group a o o o o o o o o a o n o o o o . Sums of Ranks by Group . . . . . . . . . . Individual Comparisons . . . . o . . . . . Trials to Extinction - Dichotomized by Extinction Process Group . . . . . . . . Acquisition Comparisons - Freeze and Relax Groups 0 o a o b o o a o o o o o o o a o ANOVA: Trials to Extinction - Relax Group Only a a o o o o o o o o o o t o o o o o Individual Comparisons - Relax Group Only ANOVA: Blind and Conditions in Safe - Relax Group only 6 o o o o O o o a o o o Page 10 10 11 11 12 12 13 23 25 27 28 30 Figure 2a 2b LIST OF FIGURES Page Mean Response Latencies for §s in Group ED Plotted by Blocks . . . . . . . . . 16 Mean Response Latencies for Last Block of Extinction Trials . . . . . . . . . . 18 Normalized and Standardized Distributions after Dichotomization at the 5 sec. Mark O I O O O O O O O O O O O O O O O 1 8 Average Extinction Response Latency Patterns for Relax and Freeze §s Plotted by Treatment Group . . . . . . . 19-20 Freeze and Relax Extinction Process Groups Contrasted in Terms of (a) Average Mean Response Latencies and (b) Average Response Latency Standard Deviations . . 22 Relax and Freeze Groups Compared Trial by Trial in Terms of (a) Average Avoidance Latency, (b) Average Escape Latency and (c) Percentage Avoiding During Acquisition . . . . . . . . . . . 26 vi INTRODUCTION A number of experiments suggest that escapable aver- sive events are less aversive than those from.which escape is prevented. Leitenberg (1967) found that punishment is less effective in suppressing a platform pressing response when §s are allowed to escape the aversive stimulus. Study- ing rats in a shuttle box, Marx and Hellwig (1964) found that when escape from.shock is prevented both acquisition and extinction of the avoidance response proceeds more slowly. Mowrer and Viek (1948) found that a CS preceding in- escapable shock more effectively suppressed an eating response than a CS preceding escapable shock. This effect was labled "fear of a sense of helplessness." A study by Brimer and Kamin (1963) did not replicate this finding. They found that the CS used in Shuttle box avoidance training suppressed a previously learned bar- pressing response to the same extent whether or not escape was permitted during avoidance training. These aufihors conclude that fear of the CS in avoidance conditioning is independent of the S's instrumental behavior. Studies on the effect of an escape contingency yield apparently conflicting results. Theoretical explanations of the role of the escape contingency are weak. For in- stance, both the Leitenberg (1967) and the Marx-Hellwig (1964) studies suggest that inescapable shock is more motivating. Mowrer and Viek (1948) suggest that a sym- bolic response for leaping may lessen fear of shock when escape is permitted. Methodologically, most studies do not adequately match total duration and variability of the aversive US for SS trained under conditions of escape with §s trained under conditions where escape is prevented. In Experiment 1 rats were used in an attempt to determine if the escape variable affects acquisition and extinction of an avoidance response in a one-way box. Experiment 2 explores the effect of conditions in the safe area following escapable shock when line of sight from the shock area to the safe area is either open or when it is blocked. The effect of the method of transferring §s from shock to safe areas following inescapable shock is also studied. EXPERIMENT 1 Subjects The gs were 10 experimentally naive male hooded rats from.the colony maintained by the Psychology Department of MIChigan State University.. All §s were 160-190 days of age at the beginning of training. Apparatus The basic apparatus was a one-way bOX‘With two cham- bers 18 in. long by 14 in. high by 4 in. wide. The floors were 1/8 in. stainless steel rods 5/8 in. center to center. The shock and safe areas were separated by a barrier 2 1/2 in. high and a manually operated guillotine door that opened a distance of 2 3/4 in. A shock of 1.1 ma. from.sn Applegate Stimulator was delivered through a Grayson-Stadler scrambler. The CS-US interval and shock duration were con- troled by Hunter timers. Stop clocks facilitated the con- trol of ITI's and the recording of response latencies. Procedure The §s were randomly divided into two groups of five each. A CS-US interval of 5 sec. was used during acquisi- tion and an ITI of 30 sec. remained constant throughout the experiment. The CS consisted of handling and Opening the door between shock and safe areas. 4 The SS in the control group were run first and re- ceived 12 regular acquisition trials. At the end of these trials, the shock was turned off and extinction trials were begun without interruption. Extinction trials were run in blocks of 50 trials per day until the extinction criterion of two consecutive 60 sec. latencies was reached. At the end of 60 sec. §s were boosted into the safe area. All latencies were recorded. The §s in the experimental group received inescapable shock if they failed to avoid within the CS-US interval of 5 sec. Control and experimental §s were matched in regard to total shock received and shock variability. Escape was prevented by closing the door between the shock and safe areas at the end of the CS-US interval. Within a second after shock termination, the guillotine door was again opened and SS were manually boosted over the barrier and into the safe area. Results The SS receiving inescapable shock were shocked less frequently than controls although the difference was not significant (t==.805). Matching for total shock was suc- cessful as indicated by a t test (t==.134). The shock standard deviations for the control and experimental groups were 12.96 sec. and 15.67 sec.,respectively. Experimental and control groups were almost identical in terms of the number of trials to extinction. Experimen- tal §s extinguished in an average of 42.2 trials (s==27.2). Control §s extinguished after an average of 38.6 trials (s==27.7). The difference between the groups is not significant (t==.207). Discussion The presence of an escape contingency does not affect acquisition or extinction of an avoidance response for rats in a one-way box under the conditions of this experiment. The absence of an effect would be explained if boosting §s over the barrier between shock and safe areas served the same function as an escape response. This and other variables associated with the role of escape in avoidance conditioning are explored in Experiment 2. EXPERIMENT 2 Introduction Experiment 1 revealed that blocking escape from the US during acquisition of an avoidance response had no ef- fect on either acquisition or extinction of that response. But §s receiving inescapable shock were manually assisted into the safe area shortly after the termination of the US. If this method of transferring §s from shock to safe areas serves the same function as an escape response, the nega- tive results of Experiment 1 would be explained. A dif- ferent method of transfer might reveal the effect of an escape contingency. In Experiment 2 a method was used in which a treat- ment was interpolated between the acquisition and extinc- tion phase of avoidance learning. All experimental §s received either escapable or inescapable shock. Following inescapable shock §s were transferred from the shock region to the safe region by one of two different methods. Follow- ing shock, §s were transferred from.the shock region to the safe region by one of two different methods. Following shock, §s of one group were boosted in the same manner as those in Experiment 1. The SS of the other group were lifted out of the shock area (completely out of the appara- tus) and immediately placed in the safe area. 6 Conditions following escape were also manipulated. Two groups of Se escaped to a familiar safe area and two groups of gs escaped to a different, unfamiliar safe area. One group of §s escaping to a familiar area and one escap- ing to a different area were trained with a blind that blocked the line of sight from shock to safe areas. All experimental groups were compared with a control receiving no interpolated shock. Subjects The §s were 56 experimentally naive male hooded rats from the colony maintained by the Psychology Department of Michigan State University. .All §s were 114-192 days of age at the beginning of training. Apparatus The apparatus consisted of a one-way box.with the same specifications given for Experiment 1. Procedure The §s were divided into seven groups of eight each. Training and testing of all grOups was divided into three phases: acquisition, a treatment interpolated between ac- quisition and extinction, and regular extinction trials. A shock level of 1.1 ma. was used for all groups dur- ing acquisition and interpolated shock. A CS-US interval of 5 sec. was used during acquisition and an ITI of 30 sec. remained constant for all §s throughout acquisition and extinction. The CS consisted of handling and opening of the guillotine door. All §s received twelve acquisition trials after 60 sec. of habituation in the apparatus. The walls of the shock and safe areas were covered with white cardboard throughout acquisition and extinction. Two groups were trained and tested without a blind. The remaining groups were trained and tested with the blind. The blind was a white cardboard attachment to the guillotine door that extended into the safe area in such a way that it blocked the line of sight into the safe area without seri- ously hindering access. Groups were differentiated primarily by treatments interpolated between acquisition and extinction. These treatments were separated from both acquisition and extinc- tion for all §s by 90 sec. of confinement in a holding cage. All experimental groups receive two additional shock trials. The SS of the four groups receiving escapable shock were placed in the shock area, the door opened and shock turned on simultaneously. One of these groups trained with a blind and one group trained without the blind escaped to the familiar safe area which was very similar to the shock area. One group trained with a blind and one without es- caped to an unfamiliar safe area very different from.the shock area. On one of the two interpolated trials in which §s escaped to a different area, the walls and floor of the safe area were covered with black cardboard. Escape after the other interpolated trial was to a black chamber with vertical white stripes and a white cardboard floor. The order in which these conditions were presented was alter- nated for SS. Ten seconds on a stool separated the inter- polated trials. All §s remained in the safe area for 30 sec. after interpolated shock. Two groups of Bs received two inescapable shocks of the same average duration as Be that were permitted to escape. Escape was prevented by not opening the door be- tween the shock and safe areas. For one group, the door was Opened within a second after shock termination and Bs were manually boosted over the barrier into the safe area. For another group, Bs were manually removed from the shock area without opening the door separating the two compart- ments and placed in the safe area. Timing of these proce- dures was made as similar as possible to the other inter- polated shock groups. The Bs in the control group received no additional shock but were handled on the same schedule as Bs receiv- ing shock with placement and confinement in a holding cage rather than in the apparatus. After interpolated treatments all Be were extinguished with shock and safe areas similar. Extinction trials were run in blocks of 50 per day until the criterion of two con- secutive 60 sec. latencies was reached. All response laten- cies were recorded. The major treatment variables are summarized and the groups coded for later identification as follows: Group Interpolated Code Acquisition Treatments Extinction ES No blind Escape to Bimilar No blind ED No blind Bscape to Eifferent No blind BES Blind present Bscape to Bimilar Blind BED Blind present Bscape to Bifferent Blind BIB Blind present inescapable - Boosted Blind BIC Blind present lnescapable - Qarried Blind C Blind present No shock Blind 10 Results Table 1 summarizes the acquisition data for all groups. Included are the average number of shocks received during the 12 regular acquisition trails, the average total shock re- ceived, and the average amount of shock received during the interpolated trials where escape was permitted. Table 1. Measures of Acquisition - All Groups GROUP ES ED BES BED BIB BIC C Number of _ Shocks x== 4.25 6.00 5.63 5.75 5.50 6.50 6.50 s== 1.91 2.45 2.07 2.38 1.19 2.20 1.85 Total Shock _ (sec.) x==13.50 15.84 14.73 16.10 11.80 9.20 16.14 s== 6.12 4.21 6.25 10.27 9.08 4.43 13.70 Interpolat- _ ed Shock x== 2.10 1.88 1.95 1.55 (sec.) s== .46 .42 .58 .25 A one-way analysis of variance was performed on each of these measures of acquisition variables. The results are summarized in Tables 2, 3, and 4. Table 2. ANOVA: Number of Shocks During Acquisition Source SS df MS F Between 28.11 6 4.68 1.12 Within 204.88 49 4.18 Total 232.99 55 11 Table 3. ANOVA: Total Shock Received During Acquisition Source SS df MS F Between 327.53 6 54.59 .78 Within 3426.89 49 69.94 Total 3754.42 55 Table 4. ANOVA: Total Interpolated Shock Source SS df MS F Between 1.29 3 .43 2.22 Within 5.46 28 .19 Total 6.75 31 None of the obtained F ratios are significant at the 5% level. The Pearson product moment correlation between number of shocks received during acquisition and number of trials to extinction for all Be was insignificant (r==-.017, t==-.125). Similarly, the correlation between total shock received during acquisition and the number of trials to extinction for all B3 was also found to be insignificant (r= .069, t = .508). Together these results indicate that group differ- ences in reaching the extinction criterion cannot be reason- ably attributed to differential learning during acquisition. As would be expected, the correlation between number of shocks received and total shock was significant (r==.337, t = 2.943, df = 54, p (.005) although the linear relationship is not very strong as indicated by the coefficient of de- termination (r2 = . 1 l4) . 12 Table 5 lists the number of trials to extinction for all Be by group together with apprOpriate summary statistics. Table 5. Trials to Extinction - Individuals by Group GROUP ES ED BES BED BIB BIC C 51 50 43 102 38 0 3 53 50 51 102 50 0 53 56 51 54 122 52 0 62 129 52 154 127 52 0 68 183 53 338 172 54 16 83 229 67 362 213 55 48 89 339 70 468 215 167 51 120 382 81 480 218 251 125 142 i==177.75 59.25 243.75 158.87 89.87 30.00 77.50 s==130.49 11.83 189.16 51.55 77.06 44.06 42.44 Cochran's test for homogeniety of variance yielded a C= .535 (k= 7, df= 7, p<.01). Some degree of bimodality in the sample distributions was also noted. For these reasons, the extinction data was analyzed by the Kruskal-Wallis one- way analysis of variance by ranks corrected for ties (Siegel, 1956). The effect of treatments was highly signifi- cant (H = 1383.5, df=6, p(< .001). Table 6 presents the sums of ranks for all groups. Table 6. Sums of Ranks by Group GROUP SUM OF RANKS ES 301.5 ED 164.5 BES 303.0 BED 330.0 BIB 195.5 BIC 77.5 C 224.0 13 Individual comparisons were made with the Mann-Whitney U test. All tests are of the two-tailed hypothesis of no difference. The values of U and their associated proba- bilities are given in Table 7. The effect of an escape contingency is clearly evident in these comparisons. The multiple comparison between the two groups receiving inescapable shock (both trained with the blind) and the group trained with a blind and escaping to a similar area is significant (U = 24, p4<.02) with Be es- caping being more resistant to extinction (BIB & BIC vs. BES). Individual comparisons also support this conclusion. The Bs trained with a blind and escaping to either similar or different areas are more resistant than Bs receiving in- escapable shock and carried to safe (BES vs. BIC; BED vs. BIC). This effect is also present when Bs trained without a blind and escaping to either similar or different areas are compared with Be receiving inescapable shock and carried to saftey (ES vs. BIC; ED vs. BIC). The Be trained with a blind and escaping to a different area are more resistant to extinction than Bs receiving inescapable shock and boosted to saftey (BED vs. BIB).° Table 7. Individual Comparisons (Mann-Whitney U Test) ED BES BED BIB BIC C ES U = 11 U = 29 U I 31 U = 16 U = 3 U = 20 p <.028 p < .798 p < .960 p < .104 p I. .002 p 4 .234 ED U = 17 U = 0 U = 30 U = 10 U = 16 P < 0130 P ( .000 P < .878 P < .020 P (- e104 BES U = 28 U = 18 U = 5 U = 20 p< e720 PC5160 P‘ .002 P( .234 BED U = 12 U = 3 U = 6 p < .028 p c .002 p 4 .004 BIB U = 9 U = 24 p < .014 p < .442 BIC U = 10 p < .020 14 The Be boosted into the safe area after inescapable shock are more resistant to extinction than Bs manually re- moved from.the shock area and placed in the safe area (BIB vs. BIC). Boosted Bs do not differ from no shock controls (BIB vs. C). The Bs manually moved after inescapable shock are less resistant to extinction than no shock controls (BIC vs. C). When conditions in the safe area are different and a blind is present, escapable interpolated shock increases the resistance to extinction of Bs compared with no shock controls (BED vs. C). When the blind is not present or when the conditions in the shock and safe areas are simi- lar, escapable interpolated shock has no effect when com- pared with controls (ED vs. C; ES vs. C; BES vs. C). Different conditions in the safe area after escape from interpolated shock decrease resistance to extinction only when the blind is not present (ES vs. ED; BES vs. BED). The blind has no effect when conditions in the shock and safe areas are similar but increases the resistance to extinction when conditions are different (ES vs. BES; ED vs. BED). Behavioral observations suggested the presence of two extinction processes. Some Be appeared fearful at the time the extinction criterion was reached. Others ap- peared relaxed. Freezing evidenced fear; exploration and grooming evidenced relaxation. The Bs of the first kind 15 responded rapidly until they froze during the criterion trials. The Bs that extinguished in this way will be identified as belonging to the "freeze" group. Other Bs showed a pattern of gradually increasing response latencies until criterion was reached. These will be labled as be- longing to the "relax" group. Response latencies during extinction were analyzed to see if the two patterns could be detected more precisely. ' The extinction response latencies of all Be were divided into successive blocks each containing 20% of the extinction trials for that B. The last block was further subdivided into two blocks of 10% each. The criterion trials as well as all individual 60 sec. "latencies" were excluded from this analysis because after 60 sec. in the shock area Bs were manually assisted into the safe area. Also, 5 Bs were excluded from this analysis because they extinguished in less than 5 trials. These 5 B3 froze early in extinction and are included in-the "freeze" group on the basis of behavioral observations alone. The mean and standard deviation of each block of extinction response latencies was calculated. The mean for each block was plotted for each B by group. Two rather distinct patterns of response latencies during ex- tinction emerged. These were most clearly observed in group ED and are presented in Figure 1. v—MOO LnLnLnLn mm Fm Aim 1’ .ocw_ some we use 0:» pm poumo_pc_ mum comwwc_uxe ou m.m_te .mxuo_m >n pouuopa om anotw cm mm LOe mo_ocoumu omcoamom cmez ._ ocam_a V mmmznz xuoum m H O_ ,3 on .UOm Zw- 17 Figure 2(a) is a bar graph of the means of the last block (10%) of the extinction response latencies for all Bs. This graph is sufficiently bimodal to warrant further exploration for evidence of two extinction processes. All Bs were dichotomized into two groups on the basis of the mean latency of the last block of extinction trials. Those which, on the average, responded in less than 5 sec. (the CS-US interval) tended to show a freeze pattern during extinction and were classified as belonging to the freeze group. The Be which averaged more than 5 sec. tended to relax during extinction and were so classified. The distribution of means for the freeze group has a mean equal to 2.53 and standard deviation equal to 1.12. The values of these statistics for the relax group are 14.56 and 5.24, respectively. These distributions are nor- malized, standardized and plotted in pr0per relationship to each other in Fibure 2(b). This dichotomization was applied to Be in each ex- perimental and the control group. The average means for the freeze and relax groups are plotted block by block for all groups in Figure 3. The contrast between freeze and relax extinction process groups for all experimental and the control group support the dichotomization. (One B is not included in this and the next figure. This B's average response latencies for the first, second and third extinction sessions are .87 sec., 10.0 sec., and 2.0 sec., respectively.) .xumz .oem m ecu um comum~_Eouoromo Leuem companmcum_o pe~_numpcmum ucm peN__mELoz ADV m_m_ch co_uoc_uxM mo xoo_m ummu to» mo_ocmumu emcoamom cmez Amy .N mu:m_m A.ummv >ozue L _ o z _ m a E o A.uem my A LN% s. m; N; e. o e: aria..- . o e: N..- m.T {TLN j, ._ .. m a 4 .. _. 3 fl? . i, . _ > ////, _ W. o z 7%» ._ a r/ -0: 3V . N new: «was» Ul’l'" 0W -19... .,_, ,3.) s. S .1 LJ . BES ' BED 1,. - /3 : all A .3 l*56'1,2 ‘3 #5 Average Extinction Response Latency Patterns for Relax and Freeze Be Plotted by Treatment Group. The upper curve always represents relax Be. Number at end of each plot gives the number of Be upon which pattern is based. w l -20.. BIB l6 ' l6 ’5": 'td'dtb wa or! 9:13 ec. I 2 3 1+ 5 —6 J) ‘ Figure 3. Continued 21 Figure 4 contrasts the pattern of response latencies of 17 Be demonstrating the freeze pattern with 33 Bs demon- strating the relax pattern. Performance for the first 20% of the extinction trials is almost identical for both groups. But the differences between the two groups in- crease greatly as extinction progresses. The two groups differ significantly beginning with the second block of trials (t=2.35, p (.05) and the significance of the dif- ferences increases as extinction progresses. The dichoto- mization between groups was based only upon the last block (10%) of the extinction latencies; a significant difference between the two process groups for the last block is guar- anteed. But the significant differences for blocks 2, 3, 4, ‘and 5 give independent support for the dichotomization of extinction processes. A similar didhotomization could be made on the basis of standard deviations. The standard deviations of the last block of extinction trials also yield a bimodal distribution when plotted as the means were plotted in Figure 2(a). The trough between the modes is at approximately s==3. A dichotomization based upon standard deviations would change the classification of 2 Be as compared with a dichotomiza- tion based upon means. These two bases of dichotomization are only moderately independent because of a tendency for positive skewness in the latency distributions. Figure 4(b) parallels Figure 4(a) except that average standard deviations are plotted rather than means. Both figures are based upon the same Bs dichotomized in the same way. Taken together 15 Hfibhfitd<> H 0 >0 (sec.) BLOCK (b) 10 ' A V’ E R A G. 135» 5 (sec.) “5 j r l . 2 3 h 5 6 BLOCK Figure #. Freeze and Relax E :tinction Process Groups Contra.ted in Terms of (a) Average-Lean Response Latencies and (b) Average Response Letenc cy Stcndcnd Deviations. Tee upper curve n so tn fiPPDPS is easei upon 33 relax Bs totaling 561: reSponse le tenc .es. T10 lever curves are-based upon l7 freeze Bs toss ling 929 res pans e latencies. MC'J.‘ 23 these figures reveal two rather distinct patterns of extinc- tion response latencies. Some Bs respond rapidly with little variability until the extinction criterion is reached. The latencies of Be in the relax group increase and become more variable as extinction progresses. The correlation between the standard deviations and the means of all blocks of extinction trials for all Bs is highly significant (r= .71, t= 17.3, p«.001). This evidence for two extinction processes provides the rationale for a re-analysis of the data dichotomized by process groups. Table 8 presents the extinction data for each group of the experiment dichotomized by process. Table 8. Trials to Extinction - Dichotomized by Extinction Process Group GROUP as ED BES BED BIB BIC c 51 5o 43 102 50 o 3 gr°°z° 53 50 51 52 o 53 r°“P 56 51 54 52 o 53 154 54 o 51 R 1 129 52 338 102 38 17 62 6° ax 183 67 362 122 55 48 68 r°“P 229 70 468 127 167 125 83 339 81 480 172 251 89 382 213 120 215 142 218 n:- 5 4 4 7 4 3 6 i==252.40 67.50 412.00 167.00 127.75 63.33 94.00 s==105.93 11.96 72.42 49.83 100.13 55.61 30.90 24 Examination of the extinction data yields further sup- port for the dichotomy between process groups. All 23 Bs belonging to the freeze group extinguished within six trials of the beginning of an extinction session. Only four of the 33 Bs relaxing during extinction met the criterion within six trials of the beginning of an extinction session. Chi- square for this double dichotomy contingency table is highly significant (x2==41.9, p4<.001). The Be that freeze extin- guish early in an extinction session. (Extinction for B that extinguished after 43 trials was interrupted by an electrical power failure during the 418t trial but was con- tinued as normal on the next day.) The B3 in the freeze group extinguish more rapidly than those in the relax group as evidenced by an over all Mann-Whitney U test (U=84, p (.001). The acquisition data for both process groups were searched for differences. Student's t tests were made on eight measures designed to detect differences between the freeze and relax groups during acquisition. They are: a) Number of escape responses (shocks) made during the 12 acquisition trials; b) Total amount of shock received during regular acquisition; c) Average shock per escape; d) Number of escapes in last 6 acquisition trials; e) Trial number of first avoidance; f) Trial number of last escape; g) Number of reversals - a reversal defined as an es- cape after an avoidance or an avoidance after an escape; 25 h) Longest shock. The results of these measures are presented in Table 9. As a final check for differences during acquisition the graphs in Figure 5 are presented for inspection. Groups re- presenting the two extinction processes are compared in terms of the average avoidance latency during acquisition, trial by trial, average escape latency, trial by trial, and percentage avoiding, trial by trial. Table 9. Acquisition Comparisons - Freeze and Relax Groups GROUP Measure Freeze Relax Number of Escapes i= 5.61 i= 5.82 s = 1.95 s = 2.16 t = .455 Total Shock §c= 14.83 §= 13.25 s = 9.32 s = 7.52 t = .701 Average Shock per Escape 5': = 2.77 ii = 2.44 s = 1.55 s = 1.45 t = .255 _ Number of Escapes in Last 52 = 1.17 x = 1.21 Six Acquisition Trials 3 = 1.23 5 s = 1.45 t = .10 Trial Number of First 5': = 3.52 SE = 5.27 Avoidance s = 2.64 s = 2.82 t = 2.359* Trial Number of Last 1": = 7.48 i= 7.06 Escape 3 = 2.57 s = 2.56 t = .652 Number of Reversals 5E= 2.96 5E= 2.33 s = 1.77 s = 1.45 _ t = 1.455 _ Longest Shock x= 8.08 x== 7.03 s = 5.86 s = 5.52 t = .683 *P k .05 ‘_ VI ' r “O s M ”fi3>t‘tjhu3<flb°mh<3> 14 a» 1 2- 3 1+ 5' 6 7 10 1a. 12 TRIAL KUIBER (I) \o . (b) .r oa>woommotr1 N LU (sec.) lOO TRIAL 211122—3233 (C) 00 O QZHUHo is. .1? ex 0 O N O , . , '1 2 ' 3 4 5 6 7 8 9 210 ll 12 TRIAL IYLAER Figure 5., Relax and Freeze Groups Compared Trial by Trial in ' Terms of (a) Average Avoidance Latency, (b) Average Escape Latency and (0) Percentage Avoieing T‘ _- 27 No over-all measure of learning detects a reliable difference between the acquisition behavior of Bs that freeze and Be that relax during extinction. The Bs that freeze make their first avoidance sooner although they do not make more avoidances during acquisition. The B8 in the freeze group also tend to take longer to escape on trials 2 and 3 (t= 1.95, p<.10 and t3 1.69, p<.10 , respec- tively). When only Bs in the relax group are considered, the variances of the experimental and control groups are suf- ficiently homogeneous to justify a parametric analysis (C= .338, p).05). Table 10 presents the results of a one- way analysis of variance for the six experimental groups and the control group. The effect of treatments for the group relaxing during extinction is highly significant (F==14.30, df= 6/26, p<.01). Table 10. ANOVA: Trials to Extinction - Relax Group Only SOURCE SS df MS F Between 386221.90 6 64370.32 14.299* Within 117037.62 26 4501.45 Total 503259.52 32 *F 99(6,26) = 3.59 Individual comparisons were made with the Tukey (a) procedure (Winer, 1962). The differences between the means of all groups and their significance are presented in Table 11. Certain theoretically meaningful multiple com- parisons were also made. 28 Table 11. Individual Comparisons - Relax Group Only Group BIC ED C BIB BED ES BES Mean 63.33 67.50 94.00 127.75 167.00 252.40 412.00 BIC 4.17 30.67 64.42 103.67 189.07** 348.67** ED 26.50 60.25 99.50 184.90** 344.50** C 33.75 73.00 158.40* 318.00** BIB 39.25 124.65 284.25** BED 85.40 245.00** ES 160.00* *P<005 **p( .01 The following conclusions are based upon comparisons of groups of Be that relax during extinction. The multiple comparison between the two groups receiv- ing inescapable shock and the group trained under similar conditions but receiving escapable shock is significant (p<.01) with escapable shock increasing resistance to ex- tinction (BIC & BIB vs. BES). The Be trained with a blind and escaping to a similar area are far more resistant to extinction than an B receiving inescapable shock (BES vs. BIB; BES vs. BIC). This effect is less pronounced when the blind is not present (ES vs. BIB; ES vs. BIC). Method of getting Bs from.the shock area to the safe area after inescapable shock has no effect on resistance to extinction (BIB vs. BIC). _fered from.the control (BIC vs. C; BIB vs. C). Neither of these groups dif- 29 Conditions in the safe area after escape from inter- polated shock have an effect on resistance to extinction. The Be escaping to a safe area similar to the shock area are more resistant to extinction than those escaping to a different safe area (ES & BES vs. ED & BED). Both groups escaping to a similar safe area are more resistant to ex- tinction than a control group that received no additional shock after regular acquisition (ES vs. C; BED vs. C). The Be escaping to a different area without a blind are far less resistant to extinction than those escaping to a similar area with a blind (ED vs. BES). There is a tendency for the presence of the blind to increase resistance to extinction but the over-all effect does not reach significance in this analysis (ES & ED vs. BES & BED). The blind increases resistance to extinction when Bs escape to a similar chamber (ES vs. BES). When they escape to a different chamber, the effect is not significant even though the groups do not overlap and are separated by 21 trials (ED vs. BED). In order to explore the variables of blind and safe area conditions still further, a two-way analysis of vari- ance was performed on these groups alone using the harmonic mean (Winer, 1962). The results of this analysis are pre- sented in Table 12. 30 Table 12. ANOVA: Blind and Conditions in Safe - Relax Group Only Source SS df MS F Conditions in Safe 219466.64 1 219466.64 46.24* Blind 79720.21 1 79720.21 16.79* Conditions X Blind 903.00 1 903.00 .19 Error 75948.20 16 4746.76 Total 376038.05 19 *F 99(1,16)=8.53 The variables of condition in safe area after inter- polated shock and presence or absence of a blind are highly significant. Interaction between the variables is negli- gible. All individual comparisons are significant beyond the .01 level except one comparison (ES vs. BED) which is sig- nificant at the .05 level. These comparisons indicate that conditions in the safe area are a more important variable than the presence of a blind and that the effects of treatments are additive. Discussion The design of this experiment rests upon an assumption; namely, that number of trials to extinction constitutes a unitary measure of the effectiveness of certain treatments on the acquisition and extinction of an avoidance response. Thorough analysis of extinction response latencies suggests that this assumption is unwarranted. 31 A dichotomization of Bs based upon the last block (10%) of extinction response latencies identifies significant dif- ferences in the mean response latencies of blocks 2, 3, 4 and 5. A moderately independent measure, response variabil- ity, yields an almost identical pattern of differences be- tween the dichotomized groups. The Be in the freeze group are almost sure to extinguish during the first few trials of an extinction session. The B8 in the relax group extin- guish any time during an extinction session. The Be that freeze extinguish faster. Thus, there is good evidence for two different extinction processes. These different extinction processes are probably in- fluenced by different variables and by the same variables in different ways. These results indicate that when a pre- diction is being made about the effect of a treatment on the resistance to extinction of an avoidance response it is important to specify the extinction process involved. For example, it would be inappropriate to test an hypothesis about relaxation during extinction with Bs that freeze during extinction. Yet the hypothesis may receive firm support when only Bs from.the relax group are considered. These data suggest that need for isolating and con- troling the variables that determine which extinction process will be operative for a given B_or at least a provision in the design of an experiment to treat the freeze and relax groups separately. 32 At least one B in each experimental and the control group showed the freeze pattern during extinction. This suggests that no specific aspect of the treatment condi- tions can be identified as the determinant of which extinc- tion process will be operative for a given B. No systematic differences marked the pre-experimental history of Be. A Thus, genetic variables are suggested. The suggestion of two extinction processes is not disparate with published findings on genetic differences in emotional reactivity, the effect of reactivity on avoid- ance performance, the facilitating effect of electroconvul- sive shock on avoidance performance, and the Kamin effect. Emotionally reactive and emotionally non-reactive strains of rats have been bred on the basis of defecation scores on a version of Hall's open-field test (Broadhurst, 1960). Bignami (1965) has been successful in selectively breeding rats specifically for high and low rates of avoid- ance conditioning. Owen (1963), using Bs from.the strains developed by Broadhurst found that non—reactives learned avoidance more efficiently as measured by the number of trials needed to extinguish. Joffe (1964) found this effect to continue over a long period of time. Broadhurst and Levine (1963) also tested rats of these two strains in an avoidance situation. The Bs of the reactive strain showed superior conditioning of emotional responses, as measured by fre- quency of defecation, but learned to avoid less efficiently 33 than non-reactives. These researchers suggested that freez- ing to shock and the CS interferes with efficient avoidance responding - especially for reactives. While studying performance decrement at high levels of motivation Kaplan, Kaplan and Walker (1960) observed wide individual differences in the behavior of rats in a T maze with a grid floor. Fixated behavior was associated with high emotionality. Reynierse, Zerbolio and Denny (1964) studied the de- crease in avoidance responding with continued training. Two groups were observed - decrementers and non-decrementers. Decrementers showed an increased tendency to freeze after continued training in a fear arousing situation. Genetic differences in emotionality are well estab- lished. Reactive §s do more poorly in an avoidance learning situation - a decrement frequently associated with freezing or some type of fixated behavior. Electroconvulsive shock has been found to facilitate shuttle box performance (Vanderwolf, 1963). Vanderwolf suggests that a series of convulsions damages the neural system underlying freezing behavior facilitating the ac- quisition of avoidance responding. Evidence supporting the mechanism is cited. Delprato (1966) found that gs receiving 16 electro- convulsive shocks were inferior in inhibiting a previously learned avoidance response. Again, the effect was explained 34 in terms of impairment of a neural mechanism underlying freezing behavior. Cassaday (1966) found that this effect was not caused by the fear arousing properties of ECS. Rats which re- ceived ECS 10 sec. after grid shock learned an avoidance response more efficiently than §s which received only grid shock or only ECS. Out of 23 §s constituting the freeze process group of the experiment here reported, 16 froze at the beginning of the second extinction session 4 approximately 24 hours after training. The timing of the appearance of the freeze reaction needs to be explained. Relearning of a partially learned avoidance response has been found to be a U shaped function of intertraining interval. This phenomenon has been labled the "Kamin effect" and has been reliably observed (Kamin, 1957; Denny, 1958; Denny, 1962). Brush (1964a) studied the joint effects of intertrial and intercession interval upon relearning a partially learned avoidance response with particular emphasis on the first 10 relearning trials. He found maximum.interference for §s trained with a 30 sec. ITI after an intercession in- terval of 24 hours-exactly'the conditions of this experiment. In another series of experiments, Brush (1964b) found that the fear component of original training is the neces- sary and sufficient condition to produce the U shaped func- tion of the Kamin effect. 35 The Kamin effect, the effect of ECS on avoidance per- formance, freezing and emotionality have all invited explan- ations in terms of the autonomic nervous system. Brush (1963) suggests that a parasympathetic over-reaction follows fear conditioning and that this renders a g ill- equiped to relearn the avoidance response when the over- reaction is at a peak. He also cites evidence on the re- lationship of fear conditioning, parasympathetic over-re- action and ulcer formation to support the suggestion. Using injections of adrenaline, a placebo and chloro- promazine to obtain descending levels of sympathetic acti- vation, Singer (1963) measured manifestations of fright in a fear and a non-fear situation. The reliable (r-=.92) measures of fear effectively discriminated the fear from non-fear situations. Singer concluded that amount of emotional behavior is a direct function of the degree of sympathetic activity. Doyle and Yule (1959) studied freezing behavior and grooming activities in relation to emotionality. Freezing was found to be a valid measure of emotionality in the rat but no correlation was found between grooming activities and emotionality. The extensive studies by Gellhorn (1957) on the auto- nomic system.of the cat help connect these diverse strands of evidence into an explanation of the appearance and timing of freezing behavior during the extinction of an avoidance response. Gellhorn observed two types of after-effects following stimulation of the sympathetic nervous system. 36 First, there may be a persistence of sympathetic discharge which reaches a peak after the cessation of stimulation. Second, and most important here, a sudden change from sympathetic to parasympathetic discharge may be observed after sympathetic stimulation. This "parasympathetic after- discharge, referred to as successive autonomic induction, increases with increased effectiveness of the preceding sympathetic stimulation..." (P0 72). Also, Gellhorn found that the "law of reciprocal innervation remains valid in states of a reflexly altered imbalance of the autonomic system" (p. 262) although sympathetic activity may dominate one organ system or set of sturctures while simultaneous parasympathetic activity dominates another. In these studies, stimulation of the sciatic nerve was frequently used to in- duce the parasympathetic reflex. This fact may clarify the relationship between freezing and parasympathetic activity. It thus appears that animals which freeze during ex— tinction are emotionally reactive. These §s apparently made a stronger sympathetic response to training and showed a greater parasympathetic rebound during extinction. Such a rebound became evident 111 the form of freezing behavior at the beginning of an extinction session. These ideas are readily testable. The Kamin effect should be more pronounced for emotionally reactive'gs and should be reduced or eliminated by electroconvulsive shock. 37 Experiment 1 demonstrated that the presence of an es- cape contingency had no effect on the acquisition or ex- tinction of an avoidance response. But the effect of the escape variable is clearly evident in Experiment 2 even before freeze and relax groups are separated. Here, §s permitted to escape interpolated shock are more resistant to extinction. These results appear contradictory, but the experimental situations are also different. In EXperiment 1 escape was consistently prevented during acquisition. In Experiment 2, two inescapable shock trials followed regular acquisition. In both cases, escape was blocked. But only §s in Experiment 2 were blocked from escape after, presumably, learning the escape response. The interpolated shock treatment represents a change in conditions that could facilitate the discrimination of the acquisition and extinction phases of training and thus speed extinction. The §s of the relax group receiving inescapable inter— polated shock do not differ in resistance to extinction when compared with no shock controls. When escape is prevented, additional shock does not strengthen the avoidance response. When §s of the freeze group are included in this comparison additional shock has no effect on boosted §s. But addi- tional shock decreases resistance to extinction of §s carried into the safe area. It is noteworthy that all four §s freezing during the first extinction trial are from this group. It is suggested that additional inescapable shock 38 increases fear without strengthing the avoidance response. The §s boosted to safety after inescapable shock are more resistant to extinction than §s carried to safety. The significance of this difference is due mainly to the four carried-Se that froze during the very first extinction trial. When only relax-§s are considered, this comparison does not yield a significant difference, but the four boosted- §s did require, on the average, twice as many trials to ex- tinguish as the three carried-§s. Thus,.these findings seem to suggest that the failure to find a significant effect from the escape variable in Experiment 1 can, in part at least, be explained in terms of the method of transferring §s to the safe area following inescapable shock. Conditions in the safe area following interpolated shock have an effect on resistance to extinction which is even more evident when only §s that relax during extinction are compared. Escape to an area similar to the shock area and familiar from previous training consistently produces greater resistance to extinction than escape to a different area. When only §s that relax are considered, interpolated shock increases resistance to extinction only when escape is to an area similar to the shock area. Since this effect was demonstrated even when line of sight from shock to safe areas was blocked, it seems necessary to offer an explana- tion in terms of what happens to §s while in the safe area. At first sight, these results appear to be inconsistent with those obtained by Denny, Koons and Mason (1959) and 39 Knapp (1965). Both of these studies report more rapid ex- tinction of an avoidance response when shock and safe areas are similar than when these areas are different. This ef- fect is explained by the authors in terms of elicitation theory (Denny & Adelman, 1955). According to this theory, relaxation provides the main competing response for extinguish- ing avoidance. The results of the studies mentioned above are explained if it is assumed that relaxation, which occurs in the safe area, chains back more rapidly when shock and safe areas are similar than when they are different. Relaxation in the shock area interferes with avoidance responding. This theoretical position receives additional empirical support (Weisman, Denny, Platt & Zerbolio, 1966; Denny & Weisman, 1964). In the present study, extinction was prolonged when shock and safe areas were similar. It thus appears that fear generalizing from the shock area to the safe area interferes with the development of the relaxation needed to extinguish the avoidance response. Comparison of the different experi- mental situations in these studies makes this interpretation reasonable. Conditions in both the Denny, Koons and Mason (1959) study and the Knapp (1965) study favor the develop- ment of moderate fear and considerable relaxation when com- pared with the conditions of the present study. The §s in the present study received more shocks during acquisition and had far less time to relax in the safe area. It is suggested that two processes may be operative in avoidance conditioning when similarity of shock and safe 40 areas is a variable. Relaxation, associated with cues in the safe area, may chain back to speed extinction. And fear, associated with cues in the shock area, may generalize to the safe area and retard extinction. Both processes depend on relaxation even though they have an opposite effect on rate of extinction of an avoidance response. The learning situation determines which process predominates. It could also be suggested that escape to a different area interpolated between acquisition and extinction would facilitate the discrimination of the two phases of learning and speed extinction. The presence of a blind tended to increase resistance to extinction. This effect can be expected if the oppor- tunity to observe distinctive aspects of the safe area facilitates the back-chaining of relaxation. It is also possible that the effect of the blind may be a reflection of a fearful rat's preference for a restricted area. Several §s repeatedly placed their heads under the blind after escape and were difficult to remove from the appara- tus. The §s frequently pressed their heads into a corner of the apparatus especially during the early stages of training and extinction. SUMMARY The role of an escape contingency on behavior under aversive control is not clear. A number of studies suggest that aversive events which are response terminated are less aversive than those from which escape is prevented. Experiment 1 demonstrated that hooded rats trained to avoid shock in a one—way box under conditions of no escape showed the same resistance to extinction as §s allowed to escape during training. These results are not readily interpretable because of a possible effect of method of transferring §s from the shock area to the safe area follow- ing the aversive US. In Experiment 2 several treatments were interpolated between the acquisition and extinction phase of avoidance learning. The treatment variables yielded results that can be summarized as follows: 1. Interpolated escapable shock increases resistance to extinction more than the same amount of ines- capable shock; 2. Method of transferring §s from the shock area to the safe area following inescapable shock has an effect that could account for the negative results of Experiment 1; 41 42 3. In general, resistance to extinction is greater when conditions in the shock and the safe areas are similar during interpolated shock than when they are different; 4. In general, resistance to extinction is increased when line of sight from.shock to safe areas is blocked. An analysis of response latencies for trials during extinction suggests that all §s can be classified into one of two extinction process groups. Some §s consistently respond rapidly until the extinction criterion is reached and extinguish during the first few trials of an extinction session. These §s are identified as belonging to the freeze group. The extinction response latencies of §s identified as belonging to the relax group respond less rapidly and ‘with greater variability as extinction progresses. These .§s are likely to extinguish at any time during an extinc- tion session. Thus, the assumption that number of trials to extinc- tion of an avoidance response constitutes a unitary measure of the effectiveness of certain treatments on the acquisi- tion and extinction of that response is untenable. In the experimental analysis of avoidance learning it is important to remember that treatment variables may have effects that are process specific. BIBLIOGRAPHY Bignami, 0., Selection for high rates and low rates of avoidance conditioning in the rat. Anim. Behav., 1965, 13, 221-227. Brimer, C. J., & Kamin, L. J. Fear of the CS in avoidance training and fear from a sense of helplessness. Canad. g. Psydhol., 1963, 11, 188-193. Broadhurst, P. L. Experiments in psychogenetics: Applica- tion of biometrical genetics to the inheritance of behaviour. In H. J. Eysenck (Ed.), Experimengg in Personality, 22;..13 Psychogenetics and Psychtharma- cology. London: Routledge & Kegan Paul, 1960, 1-102. Broadhurst, P. L., & Levine, S. Behavioral consistency in strains of rats selectively bred for emotional elimin- ation. Brit. g. Psychol., 1963, 24, 121-125. Brush, R. F., Myer, J. S., & Palmer, M. E. Effects of kind of prior training and intersession interval upon subsequent avoidance learning. g. comp. physio . Psychol., 1963, 56, 539-454. Brush, R. F., Myer, J. S., & Palmer, M. E. Joint effects of intertrial and intersession interval upon avoid- ance learning. Psychol. Rep., 1964, 14, 31-37. Brush, R. F. Avoidance learning as a function of time after fear conditioning and unsignalled shock. Psychon. Sci., 1964, 1, 405-406. Casseday, J. H. Retroactive facilitative effects of ECS on learning an avoidance response in a two-choice situation. Psychon. Sci., 1966, 4, 19-20. Delprato, Dennis J. Effects of electroconvulsive shock on inhibition of an active avoidance responsive. Psychon. Sci., 1966, 4, 15-16. Denny, M. R., & Adelman, H. M. Elicitation theory: I. An analysis of two typical learning situations. Psychol. Rev., 1955, 6;, 290-296. Denny, M. R. The "Kamin Effect" in avoidance conditioning. Amer. Ps cholo ist, 1958, 1;, 419. (Abstract) 43 44 Denny, M. R., Koons, P. B., & Mason, J. E. Extinction of avoidance as a function of the escape situation. g. comp. physio . Psychol., 1959, 52, 212-216. Denny, M. R., & Ditchman, R. E. The locus of maximal "Kamin Effect" in rats. g. comp. physiol. Psychol. 1962, §_5_, 1069-1070. ’ Denny, M. R. & Weisman, R. G. Avoidance behavior as a function of length of nonshock confinement. g. comp. physio . Psychol., 1964, 28, 252-257. Doyle, G., & Yule, E. P. Grooming activities and freezing behaviour in relation to emotionality in albino rats. Anim. Behav., 1959’ 1’ 12‘18; Gellhorn, E. Autonomic Imbalance and the Hypothalamus. London: Oxford University Press, 1957. Joffe, J. M. {Avoidance learning and failure to learn in two strains of rats selectively bred for emotionality. Psychon. Sci., 1964, 1, 185-186. Kamin, L. J. The retention of an incompletely learned avoidance response. .J. comp. physiol. Psychol., 1957, 29, 457-460. Kaplan, R., Kaplan, S., & Walker, E. L. Individual dif- ferences in learning as a function of shock level. g. ex . Psychol., 1960, pg, 404-407. Knapp, R. K. Acquisition and extinction of avoidance with similar and different shock and escape situations. :1. comp. 211281.01. PSZCIIOIQ, 1965, 9Q, 272-273. Leitenberg, H. Punishment training with and without an escape contingency. J. exp. Psychol., 1967, 13, 393-399. Marx, M. H., & Hellwig, L. R. Acquisition and extinction of avoidance conditioning without escape responses. .g. comp. physiol. Psychol., 1964, 58, 451-452. Mowrer, O. H., & Viek, P. An experimental analogue of fear from a sense of helplessness. g. abnorm. soc. Psychol., 1948, 5;, 193-200. Owen, S. The effect on avoidance response extinction in rats of CS continuation and emotional constitution. g. genet. Psychol., 1963, 103, 147-151. 45 Reynierse, J. H., Zerbolio, D. J., & Denny, M. R. Avoid- ance decrement: Replication and further analysis. Psychon. Sci., 1964, 1, 401-402. Siegel, S. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill, 1956. Singer, J. E. Sympathetic activation, drugs, and fear. g, comp. physiol. Psychol., 1963, §§, 612-615. Vanderwolf, C. H. Improved shuttle-box performance follow- ing electroconvulsive shock. J. comp. physiol. Psychol., 1963, g9, 983-986. Weisman, R. G., Denny, M. R., Platt, S. A., & Zerbolio, D. J., Jr. The facilitation of extinction by a stimulus associated with long nonshock confinement periods. J. comp. physiol. Psychol., 1966, pg, 26-30. Winer, B. J. Statistical principles 2; ex erimental design. New York: McGraw-Hill, 196 . m R m L V. H S R E W N U F. T A'- 1 "1|“!me