INCENTIVE MOTIVATION AND APPROACH-AVOIDANCE TENDENCIES

Transcription

1 INCENTIVE MOTIVATION AND APPROACH-AVOIDANCE TENDENCIES by Robert W. Rice, B.A. A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfilment of the requirements for the degree of Master of Arts. Department of Psychology McGill University Montreal August, Robert W. Ri ce 1967

2 ACKNOWLEDGMENTS The research reported in this thesis was supported by a grant (No. APT-74) from the National Research Council of Canada to Dr. Dalbir Bindra.

3 TABLE OF CONTENTS Page INTRODUCTION METHOD RESULTS DISCUSSION SUMMARY REFERENCES

4 INTRODUCTION Mowrer's (1947) and Spence's (1956) suggestion that classical conditioning may play a significant role in the acquisition and performance of instrumental responses bas fostered two broad lines of research. One of these bas been concerned with the role of classcally conditioned responses in instrumental behavior (Wynne & Solomon, 1955; Solomon & Turner, 1962; Miller & DeBold, 1965; Williams, 1965) and bas produced mainly negative results. The proposition that the occurrence of certain classically conditioned responses (e.g., autonomie "anxiety" reactions, anticipatory goal responses, or rg) is an essential step in instrumental responding seems no longer tenable. The second line of research bas examined the role of classically conditioned stimuli in instrumental behavior and is based on the assumption that in the course of instrumental training, certain stimuli, apart from whatever eue function they might serve in the control of the instrumental response, serve a motivational function by being classically paired with the reinforcement. A typical experimental design for studying incentive motivation involves, first, pairing classically a certain stimulus with a positive (CS+) or negative (CS-) reinforcer and then determining the effect of the non-contingent presentation of this reinforcement-linked or incentive-motivational stimulus on an instrumental response that bas been separately trained. Note that these reinforcement-linked CSs can also be used as secondary reinforcers when they are contingent upon the occurrence of the instrumental response.

5 - 2 - The effect of an incentive-motivational stimulus, CS+ or CS-, on instrumental behavior may be studied in several ways. One may study (1) the effect of a CS+ on an appetitively trained instrumental response, (2) the effect of a CS+ on an aversively trained instrumental response (e.g., an avoidance response), (3) the effect of a CS- on an appetitively trained instrumental response (the CER or conditioned emotional response paradigm), and (4) the effect of a CS- on an aversively trained instrumental response. Within each of these classes the effects of incentive-motivational stimuli on both acquisition and performance of the instrumental response may be studied. In general, the presentation of a CS+ in an appetitive learning situation facilitates the acquisition of an instrumental response (Bower & Grusec, 1964; Trapold, 1966; Bacon & Bindra, in press; Trapo1d & Winokur, in press) and enhances performance of tasks a1ready learned (Walker, 1942; Estes, 1943, 1947). The presentation of a CS+ can a1so faci1itate the acquisition of an aversive1y-reinforced (avoidance) response (Bacon & Bindra, in press), but the effect of CS+ on the performance of an already acquired aversively trained instrumental response is not known. The presentation of a CS- in an aversive learning situation is known to enhance the performance of an avoidance response (Overmier & Leaf, 1965; Krieckhaus & Chi, 1966; Rescor1a, 1966), but its effect on the acquisition of an avoidance response is not known. A CS- retards the acquisition and inhibits the performance of an appetitively trained instrumental response; this bas been well documented in a variety of studies on conditioned emotional response (Estes & Skinner,

6 ; Hunt & Brady, 1951). The general impression created by these studies is that CS+ facilitates appetitive instrumental responding, while cs- facilitates aversive instrumental responding, and it is also generally supposed that CS+ and CS- would have reverse effects on, respectively, aversive and appetitive instrumental responses. However, this simple, intuitively-sound summary is contradicted by the fact that CS+ may under certain circumstances facilitate the performance of an avoidance response (Bacon & Bindra, in press). There is also reason to believe that sorne studies with negative results may not have found their way into print; for example, an unpublished study done in this laboratory revealed no effects of CS+ on the performance of an appetitive instrumental response. It seems reasonable to suppose, therefore, that the effects of incentivemotivational stimuli on instrumental responses may be determined by a complex set of factors. The present investigation is a preliminary attempt to discover sorne of these factors. A recent paper by Bindra & Palfai (in press) suggests that specifie changes in the components of general activity produced by an incentivemotivational stimulus may partly determine the effects that that incentive-motivational stimulus would have on an instrumental response. They studied the effects of a CS+ and a CS- on the various components of general activity of the rat, and found that CS+ facilitated exploratory behavior, so that the frequency of occurrence of acts of sniffing, rearing, and walking was greater during the presentation of CS+ than during its absence. Similarly they found that the presentation of the CS- increased the incidence of defensive reactions so that the frequency of freezing

7 - 4 - and crouching was much greater in the presence of the CS- than in its absence. The present investigation started with the assumption that these changes in the components of general activity may be related in sorne way to the effects that incentive-motivational stimuli have on instrumental responding. 1 examined the effects of a CS+ and a CSon the performance of a straight alley running response. METHOD Subjects Ss were 60 naive male hooded rats of the Royal Victoria Hospital strain purchased from the Quebec Breeding Farm, St. Eustache, Quebec. Ss were housed six per cage and were maintained on a 23-hr. water deprivation schedule throughout the experiment. Food was available in the home cages at all times. Apparatus Classical conditioning apparatus. ~s were classically conditioned in four identical sound-deadened chambers. Each chamber was 10 in. long, 10 in. wide, and 9 in, high, with a metal grid floor. The grid floors of two of the chambers were connected to an electric shock generator-scrambler. The other two chambers were equipped with a dipper deviee which could be operated to deliver 0.5 cc of water through a hole in one wall. A click generator, connected to a loudspeaker placed in each chamber was used to present the CS. Relays and tape-timers controlled the CS and US (shock or water) presentations.

8 - 5 - Activity measuring apparatus. Activity was measured in a 12-in. long, 8-in. wide, and 10-in. high Plexiglas box with grey cardboard walls and a metal grid floor. A mirror was arranged above this box to enab1e E to observe S from his chair. Three microswitches, each connected to a 0.1-sec. running time meter, were used to record the duration of three categories of acts disp1ayed by ~ Instrumental training apparatus. A straight alley, made of unpainted p1ywood, was used for instrumental training. It consisted of a 6-in. long, 6-in. wide and 18-in. high start box, which was connected to a 36-in. long alley by a manua11y operated door. The alley served as a goal box. During approach training, the goal box contained a water spout; during avoidance training a piece of plywood placed over the goal box floor insulated the S from shock. Response latency and running time were measured automatically by a system of switches, relays and a photocell. A mirror was used to enable E to observe S. Procedure The Ss were divided into two experimental groups, CS+ and CS-, and one control group, C. Each group contained 20 rats. Al1 Ss were maintained on the 23-hr. water deprivation schedu1e for at least four days prior to training. The training and testing procedures will be described in the arder that they were conducted in the experiment. Classical conditioning. On Days 1 and 2 ~s were àdapted to the conditioning chambers. Water was delivered intermittent1y to the CS+ rats during the adaptation sessions. On Days 3-10, each ~ received one of

9 - 6 - three classical conditioning treatments. In the case of the CS+ group, the termination of a 15-sec. train of 1/sec. clicks {CS) was accompanied by the delivery of water {US). In the case of the CS- group, the termination of the same CS was accompanied by the onset of a 1-sec.,.5 ma shock delivered through the grid floor. Ralf the animals in Group C received unpaired presentations of the CS and water, the other half received unpaired presentations of the CS and shock. Nine CS-US pairings {or, in the case of the C group, nine unpaired presentations of CS and US) were given each day; the interval between presentations ranged from 5 min. to 15 min. Activity. The ~s were given a 1-hr. adaptation session in the activity box one day prior to the activity testing session. The test session consisted of three consecutive 3-min. observation periods, pre-cs, CS, and post-cs. The CS was presented throughout the second period, but not in the first (pre-cs) or the third (post-cs) period. For each 3-min. period,! recorded the amount of time spent in the following activities: grooming {washing, scratching, etc.), sitting (crouching, freezing) and exploration (general moving about, rearing, sniffing, etc.). Activity measures were obtained from 20 CS+ animais, 17 CS- animais and 10 control animais. Problems of scheduling prevented testing 13 rats. Instrumental training. Training was preceded by three 45-min. adaptation sessions in the straight ailey apparatus. For the purpose of instrumental training, ~s were divided into two sets of 30 animals each {10 CS-, 10 CS+, 10 C). One set of animals was taught a running response reinforced with water (approach) and the ether was taught a running response reinforced by shock (avoidance).

10 - 7 - In the water-approach condition, the training procedures were as follows. S was placed in the start box; after a 20-sec. delay, the start-box door was opened. S was allowed unlimited time to reach the goal box, where it was given 10 sec. access to the water spout. Each ~ received five trials on Day 1 spaced at approximately 5-min. intervals; 10 trials were given on Days 2, 3, and 4. Training was continued until a ~ had performed 8 out of 10 trials with both response latencies and running times under 2 sec. Most Ss performed at this level by Day 5 or 6. Three C group animais failed to reach criterion by Day 8 and were therefore discarded. Two C group rats died before they were tested, reducing the number in this group to five. The training procedures for the shock-avoidance condition were parallel to those of the water-approach condition. S was placed in the start box and after a 20 sec. delay the start-box door was opened. After an additional 5 sec. delay, the grid floors of the start box and the alley were electrified with a.5 ma of current. terminated until after ~ had escaped in the goal box. The shock was not Each S received five training trials spaced at approximately 5-min. intervals on Day 1 and ten trials similarly spaced on Days 2, 3, and 4. Training continued until response latencies and running times were bath less than 2 sec. on 8 out of 10 trials (the same criterion as in the water-approach condition). On the day following training all ~s were given an additional "refresher" classical conditioning session. On the day following the conditioning session ~s were tested under extinction conditions. The same testing procedure was used for bath the water-approach and the shock-

11 - 8 - avoidance conditions. On Day 1 of testing all ~s received three reinforced trials (approximately 1 min. between trials) followed by nine extinction, or test trials. Each S received nine extinction trials a day (three pre-cs, followed by three CS, followed by three post-cs trials) for four days (a total of 36 extinction trials). On Trials 4, 5, and 6 (CS trials), the CS was presented for 15 sec. while S was confined in the start box and the CS was continued for another 15 sec. after the door was opened. The CS was not present during Trials 1-3, or 7-9 (pre- and post-cs trials). On all trials, ~ was removed from the apparatus 10 sec. after entering the goal box or after 30 sec. if it did not reach the goal box. Response latency and running time were recorded for each trial. RESULTS Activity The means of the time scores obtained in the test session for each category of acts (exploration, sitting and grooming) are presented in Figure 1. Two types of comparisons can be made with the data for each group. The three periods (pre-cs, CS, and post-cs) may be compared within a group, and the three groups may be compared with respect to each period. In the CS+ group comparisons between the CS period and the pre-cs period showed that the introduction of CS increased exploration (t = 10.29, p <.001), and decreased bath sitting (t = 5.49, p <.001) and grooming (t = 12.24, p <.001). In the CS- group, comparisons between the CS period and the pre-cs period showed that the introduction of CS decreased exploration (t = 15.69, p <.001) increased sitting

12 - Sa ~--~ ~ i Grooming w 100 Cl:: 50 0 u V) le : -..; cs+,.t / " ' ---- )"'":,.(..., "- tl"... "" cs- w ~ 0~--~ ~Â~------~----;., /. 1- /, Sitting 100 /. /.. f' / CS-'. /. ' & 50 --, t._ ~""' A c\ "'"' >- -- A"-CS+ 0~----~ ~ ~~--~ M 1 N. TEST PERIOD Figure 1. Mean number of seconds spent in exploration, grooming and sitting for Groups CS+, CS- and C for the three 3-min. observation periods (pre-cs, CS, post-cs).

13 - 9 - increased sitting (t = 18.08, p <.001) and decreased grooming (t = 7.52, p <.001). In the C group, these same comparisons showed that the introduction of the CS had no effect. DurLng the CS period, the CS+ group spent more time than the C group in exploration (t = 18.42, p <.001), spent less t me than the C group in sitting (t = 4.42, p <.001), and 1ess time than the c group in grooming (t = 5.83, p <.001). The csgroup, in the same period, spent less time than the C group in exploration (t = 16.34, p <.001). Note a1so that the CS+ group was more active (explored more, sat less) than the C and CS- group even in the pre-cs period; but there were still marked activity changes in response to the CS. Instrumental responses The small number of animals in each group, especially the control group (as noted earlier), and the large number of variables under investigation pose serious problems for the statistical analysis of the data. The major problem arises from the fact that Ss were tested duruing extinction. Performance measures which were dependent upon the completion of the instrumental response (response latency, running time) become progressively invalid after Day 1 because an increasing number of Ss stopped responding. Latency scores after Day 1 are skewed toward the arbitrary maximum of 30 sec.; and running times after Day 1 are too few to be treated statistica11y. Under these conditions any over-all analysis of variance would no doubt be meaningless. Therefore, the statistical treatment of the data, when appropriate, consisted of se1ected comparisons by nonparametric methods between the scores of the various groups (CS+, C, CS-) in the same periods (pre-cs, CS, post-cs),

14 and comparisons within a given group of scores in different periods. The results of such comparisons can, of course, be considered only as suggestive and not necessarily reliable. Water-approach. Median response latencies and running times for each ~ were calculated separately for the three test periods (pre-cs, CS, post-cs) of each extinction session. Ss which did not leave the start box received no running time scores; hence the number of observations for each group varied from day to day. For reasons noted in the above paragraph, the data are presented only for the first extinction session. The means of the median latencies and running times are displayed in Figure 2. During the CS period, the CS+ group had longer latencies than the C group (p =<.02, Mann-Whitney u test), and slightly longer latencies than the cs- group (p =<.10). During the same period the latencies for the CS- group also tended to be slightly longer than those for the C group (p =<.10). There were no significant differences in running time scores in the CS period. There were also no significant differences between the groups in latencies and running times during both the pre-cs and the post-cs periods. The latencies and the running times for all groups showed a significant increase (extinction effect) from the pre-cs period to the post-cs period (p =<.05). The mean number of completed responses ~ touching the dry water spout) in all three periods of each of the four extinction sessions were calculated for each group. These data are displayed in Figure 3. There

15 - loa - Day 1 Latenc y u w 1./) z 0 w ~ Day " (N =10) 1 //' CS+ (N =10) "7 c\ ""'... "'... 1 ''7,_.-t: 'I tl Running Time cs 0~------~ ~ ~ pre-cs post-cs 3-TRIAL PERIODS Figure 2. Mean latencies and running times for Groups CS+, CS- and C in the water-approach condition on Day 1.

16 - lob V) w ~ z 1- w x loc...! <( z z <( w ~ 3.0 ol _.2...~3~ ~-----' EXTINCTION DAYS Figure 3. Mean number of goal box entries for Groups CS+, CS- and C in the water-approach condition for Days 1-4.

17 is a significant extinction effect between Day 1 and Day 4 for a11 the groups (p =<.01, Chi Square). There were no differences between periods in any group. Shock-avoidance. The means of the median latencies and running times for Day 1 are presented in Figure 4. No differences were found for any of the comparisons between periods or between groups except those between pre-cs and post-cs periods (extinction effect). Both latencies and running times for a11 groups showed an increase from the pre-cs period to the post-cs period (p =<.01, Mann-Whitney U test). The mean number of comp1eted responses for each day of extinction are presented in Figure 5. There were no noticeable differences between periods in any group on any of the days. On Day 1 there were no differences between groups. On Day 2, however, the CS- group made more complete avoidances than the CS+ group (p =<.01, Chi Square). On Day 3 and Day 4, the CS- group avoided more than the CS+ group (p =<.01) and the C group (p =<.01). The CS+ group did not differ from the C group on these days. DISCUSSION The changes in general activity produced by the presentation of the positive and negative incentive-motivationa1 CS s corroborate the findings of Bindra and Palfai (in press), except in one detail. In the present study, animals in the CS+ group were, prior to CS onset, more active initially than the C group rats; the CS- group was less active than the C group. Bindra and Palfai reported no activity differences

18 - lla - Day 1 latency 4.0 (N=10) (N=lO} / 1 A CS+ / # /,/ / 2.0 cs- (N= 10) u w VJ z 0 w ~ 1- Day 1 Running Time c (N:;::lQ) '/ (N=9) h/1. CS+ ff ~. ~ 7 ~\ (N=lO).~ èsol ~~~ ~ ~---~--~ pre-cs CS post-cs 3-TRIAL PERIODS Figure 4. Mean latencies and running times for Groups CS+, CS- and C in the shock-avoidance condition for Day 1.

19 - llb (/) w 0::: 1- z w x 0 c:t:l <( 0 C) o z z <( w ~ 3.0 c ol-----._------~ ~3~--~--~4~--~ EX Tl NCTION DAYS Figure 5. Mean number of goal box entries for Groups CS+, CS- and C in the shock-avoidance condition for Days 1-4.

20 among the groups prior to the onset of the CS. These activity differences are probably attributable to similarities between the stimulus qualities of the test situation and those of the classical conditioning chambers (e.g., both apparatuses had grey walls and gr\d floors). Evidently the single adaptation session used in the present study was not sufficient to extinguffihall the affective associations that the Ss might have brought to the test situation. Perhaps the most noteworthy result of the present investigation is that the incentive-motivational stimuli had a marked effect on general activity, but very little on instrumentally trained behavior. In the presence of the CS, ~ in the CS+ group showed a marked increase in the amount of time spent in exploration and ~ in the cs- group showed an increase in sitting. But the effects of CSs on the performance of both the avoidance and the approach instrumental tasks were slight when compared to their effects on activity. A review of the data of other investigators indicates that the magnitude of the effects on instrumental responses is invariably small and the effects are shortlived CWalker, 1942; Estes, 1943; Overmier & Leaf, 1966; Rescorla, 1966; Bacon & Bindra, in press). The adverse effect of CS+ in the approach situation, as reflected in the increase response latency, is contrary to the general finding that presentation of a CS+ facilitates the performance of instrumental appetitive responses. The studies reporting facilitation have generally tested the effects of a CS+ on a lever-pressing task rather than on a running response (e.g., Walker, 1942; Estes, 1943). These two results

21 suggest that the direction of the CS+ effect on instrumental performance is partly a function of the characteristics of the instrumental task. It may be useful here to consider the factors that might be determining whether the CS+ effect is an incrementai or a decrementai one. Response stability appears to be one of these factors. It may be hypothesized that the CS+ would be more likely to lead to exploratory acts in a situation involving no specifie training (e.g., the activity situation) than in a situation involving a stable instrumentally trained response. If the CS+ is presented during the performance of a weak or unstable response then it is possible that exploratory acts may appear sporadically. Generally, the occurrence of exploratory acts would interfere with performance of the instrumental task. If we further assume that a lever-pressing response, as used in previous studies, is more stable than the running response used here, then the discrepancy between the two sets of results could be attributed to the response stability factor. That, during extinction sessions, the lever pressing response used in previous studies might have been more stable than the present running response is suggested by differences in training procedures. In the present experiment all completed runway responses were reinforced; in the lever-pressing experiments, e.g., Walker's and Estes', the lever-pressing responses were reinforced only at variable intervals. The onset of the extinction or test trials would therefore be expected to produce a decrementai effect sooner in the runway situation than in the lever-pressing situation.

22 Factors other than response stability, may have increased the probability that in the present experiment exploratory responses rather than the alley-running response would occur during the presentation of the CS+. For example, habituation to the testing apparatus may be one of these factors. Ss lever-pressing in a Skinner box typically have more time to habituate to their surroundings prior to the onset of the CS than do S in the straight alley. In addition, the presentation of the CS+ during the 15-sec. period before opening the start-box door may have also influenced the direction of the results, since behavior, notably exploration, was not controlled in the start box. In the studies using the lever-pressing response (e.g., Walker, 1942; Estes, 1943) the CS+ was presented during the ongoing operant responding. The slight increase in response latencies produced by cs- in the approach task may be attributed to the same factors involved in the conditioned emotional response (CER) paradigm. When a CS- is presented in an appetitively reinforced lever-pressing situation, rats typically cease responding throughout the CS period (Estes & Skinner, 1941; Hunt & Brady, 1951). There is evidence that the effect of a CS- on a running response is less disruptive (Strouthes, 1964), presumably because running is to sorne extent associated with escape. This would be consistent with the fact that in the present experiment the animals did not completely cease responding. The shock-avoidance data indicate that the present design was inadequate to isolate the incentive-motivational effects of the cs- from its effects as a secondary reinforcer. If the resistance to extinction

23 of the cs- group was enhanced solely by the incentive-motivational qualities of the CS, then the effect should be specifie to the CS trials and should not appear in the pre-cs and post-cs trials. The data indicated no such differences: on Days 2, 3, and 4 there were no differences between the CS, pre-cs and post-cs conditions. There is evidence that the offset of a CS- may serve to reinforce an avoidance response (Kamin, 1956). During the extinction trials the CS was terminated 10 sec. after a ~ entered the goal box. Therefore those ~s in the CS- group that reached the goal box would receive sorne secondary reinforcement for the response, but ~s in the C group and CS+ group would receive none. This confounding intermittent secondary reinforcement alone could account for the prolonged responding of the cs- group during extinction. Broadly speaking, the results of this preliminary investigation suggest that the effects of incentive-motivational stimuli on instrumental performance are determined not only by the valance of the CS or the type of motivation (appetitive vs. aversive), but also by such variables as response topography and response stability. Systematic, formal examination of the interactions of these variables with positive and negative incentive motivation is therefore indicated. SUMMARY The incentive-motivational effects of stimuli classically paired with water (CS+) or shock (CS-) on general activity, as well as on the performance of an instrumentally trained running response reinforced

24 either by water (approach) or by shock (avoidance), were examined in 55 rats. Presentation of CS+ in an activity measuring apparatus resulted in an increase in exploration, while presentation of CS- resulted in an increase in sitting. Both CS+ and CS- increased the latency of the water-approach response in the initial test (extinction) trials. Rats presented CS- during extinction of the shock-avoidance response showed a decrease in rate of extinction. No other effecmof the CSs on response latency, running time, or rate of extinction were found. The marked effects on general activity compared to the relatively slight effects on instrumental responding reported here and in other studies suggest that incentive-motivation interacts with certain response variables, such as response topography and stability.

25 REFERENCES Bacon, W.E., and Bindra, D. The generality of the incentive-motivational effects of c1assically conditioned stimuli in instrumental learning. Acta Biol. Exper., (in press). Bindra, D., and Pa1fai, T. The nature of positive and negative incentive-motivational effects on general activity. J. camp. physiol. Psychol., (in press). Bower, J.S., and Grusec, T. Effect of prior Pavlovian discrimination training upon learning an operant discrimination. J. exp. anal. Behav., 1964, 1, Estes, W.K. Discriminative conditioning: I. A discriminative property of conditioned anticipation. J. exp. Psycho1., 1943, 32, Estes, W.K. Discriminative conditioning: II. Effects of a Pavlovian conditioned stimulus upon a subsequently established operant response. J. exp. Psychol., 1947, 38, Estes, W.K., and Skinner, B.F. Sorne quantitative properties of anxiety. J. exp. Psycho1., 1941, 12, Hunt, H.F., and Brady, J.V. Sorne effects of electro-convulsive shock on a conditioned emotiona1 response ("anxiety"). J. camp. physiol. Psychol., 1951, 44, Kamin, L.J. The effects of termination of the CS and avoidance of the US on avoidance 1earning. J. camp. physiol. Psychol., 1956, 49,

26 Krieckhaus, E.E., and Chi, c.e. Role of freezing and fear in avoidance decrements following mammi1lothalamic tractotomy in the cat: I. Two-way avoidance behavior. Psychonom. Sei., 1966, ~ Miller, N.E., and DeBold, R.C. Classical1y conditioned tongue-1icking and operant bar pressing recorded simultaneously in the rat. J. comp. physio1. Pszchol., 1965, 59, Mowrer, O.H. On the dual nature of learning--a re-interpretation of "conditioning" and "problem solving." Harvard Education Review, 1947,!Z Overmier, J.B., and Leaf, R.C. Effects of discriminative Pavlovian fear conditioning upon previously or subsequent1y acquired avoidance responding. J. comp. physiol. Psycho1., 1965, 60, Rescorla, R.A. Predictabi1ity and number of pairings in Pavlovian fear conditioning. Psychonom. Sei., 1966, ~ Spence, K.W. Behavior theory and conditioning. New Haven: Yale University Press, Salomon, R.L., and Turner, Lucille H. Discriminative c1assica1 conditioning in dogs paralyzed by curare can later control discriminative avoidance responses in the normal state. Psychol. Rev., 1962, 69, Strouthes, A. Effect of CS-onset UCS-termination delay, UCS duration, CS-onset, UCS-onset interval, and number of CS-UCS pairings. J. exp. Psychol., 1965, 69, Trapold, M.A. Reversal of an operant discrimination by non-contingent discrimination reversai training. Psychonom. Sei., 1966, ~'

27 Trapold, M.A., and Winokur, S. Transfer from classical conditioning and extinction to acquisition, extinction and stimulus generalization of a positively reinforced instrumental response. (in press). Walker, K.C. The effect of a discriminative stimulus transferred to a previously unassociated response. J. exp. Psychol., 1942, ]1, Williams, D.R. Classical conditioning and incentive motivation. In W.F. Prokasy (Ed.), Classical conditioning: A symposium. New York: Appleton-Century-Crofts, Wynne, L.C., and Salomon, R.L. Traumatic avoidance learning: acquisition and extinction in dogs deprived of normal peripheral autonomie function. Genet. Psychol. Monogr., 1955, 52,