Acquisition of a Free-Operant-Appetitive Response

LEARNING AND MOTIVATION 7, 394-405 (1976)

Acquisition of a Free-Operant-Appetitive Response in Pigeons as a Function of Prior Experience

with Response-Independent Food

ROBERT L. WELKER

Mount Holyoke College

Deficits in learning to escape from electric shock following exposure to response-independent preshocks have frequently been reported and have been referred to as learned helplessness. Experiments were conducted in order to determine whether a phenomenon similar to learned helplessness could be induced in appetitive free-operant procedures with pigeons. Subjects received preliminary training under one of the following conditions; protracted exposure to response- dependent grain presentations (key pecking), protracted exposure to response- independent grain deliveries, or short-term hopper training. Subjects were then tested for acquisition of a treadle-pressing response which was the only means of access to grain in the experimental chamber. The acquisition of the treadle-pressing response was retarded following protracted exposure to response-independent grain deliveries and the degree of this retardation was related to the complexity of the response-reinforcer contingency.

Administration of response-independent electric shocks to rats or dogs often retards their subsequent learning to escape from shock, relative to the performance of subjects which receive previous exposure to escapable shock or no shock (Maier, 1970; Maier, Albin, & Testa, 1973; Maier & Testa, 1975; Overmier & Seligman, 1967; Seligman & Beagley, 1975; Seligman & Maier, 1967; Seligman, Rosellini, & Kozak, 1975). This retardation in learning to escape from shock has been referred to as “learned helplessness,” a position which assumes that learning about the independence of shock presentations and behavior during pretraining transfers to the subsequent escape task (for reviews see Maier, Seligman, & Solomon, 1%9; Seligman, Maier, & Solomon, 1971).

Engberg, Hansen, Welker, and Thomas (1972) have presented data

This research is based upon a doctoral dissertation submitted to the University of Colorado in partial fulfillment of the requirements for a Ph. D. in Experimental Psychology. The author would like to thank his advisor, David R. Thomas, and the other members of his dissertation committee, G. Kimble, S. Maier, and R. Miles, for their helpful comments andcriticisms. The research was supported by National Institutes of Health Research Grant HD-0&486 and Training Grant MH-10427, administered by David R. Thomas, and a National Institute of Mental Health predoctoral research fellowship awarded to the author. Requests for reprints should be sent to Robert L. Welker, Department of Psychology, Mount Holyoke College, South Hadley, Mass. 01075.

394

Copyright 0 1976 by Academic Press. Inc. All rights of reproduction in any form reserved.

RESPONSE-INDEPENDENT FOOD 395

which could conceivably represent an appetitive parallel to learned helplessness. A group of pigeons that received protracted exposure to response-independent grain deliveries subsequently acquired an autoshaped key peck response more slowly than either a group that exercised operant control over grain deliveries in pretraining or a naive control group. Considering the functional similarities of these data and those discussed under the rubric of learned helplessness, Engbergetal. proposed a cognitive interpretation of their results which was viewed as an appetitive extension of learned helplessness, “learned laziness.”

A basic methodological problem inherent in the Engberg et al. experiment involves the dual nature of the contingencies that may be responsible for conditioning of the key peck in the autoshaping procedure. The procedure may be conceptualized as one that contains a stimulus- reinforcer contingency, and, following the occurrence of the first peck at the illuminated response key, an adventitious response-reinforcer contingency (Williams & Williams, 1969; Gamzu & Williams, 1973). Although Engberg et al. proposed that the retarded initiation of autoshaped key pecking was produced by previous learning about the independence of behavior and grain deliveries, it becomes theoretically cumbersome to speculate as to how this sort of learning would influence response initiation in the autoshaping procedure.

The following experiments were undertaken to further determine the plausibility of an appetitive parallel to learned helplessness by investigating the effects of exposing pigeons to response-independent grain deliveries upon their learning to behave in accordance with a response-reinforcer contingency.

EXPERIMENT 1

Experiment I was designed to determine whether results similar to those reported by Engberg et al. would be obtained with a testing procedure consisting only of a response-reinforcer contingency. Three groups of pigeons were pretrained under the following conditions: One group earned grain presentations by key pecking in accordance with a variable-interval schedule of reinforcement (VI), a second group received an equivalent number and pattern of response-independent grain presentations, and a third was merely trained to eat from the grain hopper. Following this training, a foot treadle was inserted along the back wall of the operant chamber and each grain delivery was contingent upon a treadle press. According to the learned laziness hypothesis, the group of pigeons exposed to the response-independent grain deliveries in pretraining should acquire treadling more slowly than both the control group and the group of pigeons pretrained with the key peck contingency. The latter group was included in the experimental design in order to control for the effects of experience

396 ROBERT L. WELKER

with and feeding in the experimental chamber per se relative to experience with response-independent grain presentations.

Method

Subjects. The subjects were 36 experimentally naive pigeons of mixed strains obtained from a local supplier and maintained at 75% of their ad lib weight. Throughout the experiment the pigeons were housed in individual cages in a colony room illuminated in accordance with a 12-hr light/dark cycle. Subjects were given free access to water.

Apparatus. Four operant chambers housed in sound-attenuating hulls served as the experimental apparatus. The inside dimensions of the chambers were 31 x 35 x 36 cm. A response key, grain hopper, and houselight were located on the front wall of each chamber. The response key (2.9 cm in diameter) was located in the middle of the wall approximately 2 1 cm above the floor, with the grain hopper (5.2 x 6.5 cm) situated 10 cm below. Grain deliveries were accompanied by illumination of the inside of the hopper chambers. The houselight was projected onto and diffused by a strip of translucent pearl-white acrylic plastic which covered the top 5.2 cm of the front wall. Intensity of the houselights averaged 47 ftL for the four chambers when measured from a position approximating that of subjects’ heads when key pecking. The operant chambers could be modified by insertion of a removable treadle (9 cm in length x 6 cm in width) which protruded from a weighted box (10 x 9 x 7 cm) that was placed along the back wall, opposite the magazine wall. Each treadle was positioned obliquely relative to the floor with the lowest side (front) approximately 2.5 cm above the floor. A piece of cardboard was taped diagonally from the top of the treadle housing to the wall of the operant chamber in order to prevent roosting on the housing. A Csec access to a hopper filled with mixed grain served as the reinforcing event throughout the experiment. Manipulation of all experimental variables was programmed through standard relay circuitry.

Procedure. The subjects were unsystematically assigned to one of three experimental groups (n = 12). These groups were designated, and will be referred to throughout the remainder of the text, as the Contingent, Noncontingent, and Control groups. On the first day of preliminary training, subjects in the Contingent group were trained to peck an unilluminated response key by the method of successive approximations. Immediately following the first response, a VI IO-set schedule of reinforcement was instituted. Subjects were allowed to earn 20 reinforcers under the VI lo-set schedule and 10 additional reinforcers under each of two VI schedules in which the mean temporal interval between reinforcers was successively increased to VI 30 set and VI 45 sec. Beginning with the second session and continuing throughout the next 15 sessions, subjects in


the Contingent group were allowed to earn 60 reinforcers per session for key pecking in accordance with a VI 45-set constant probability schedule (cf. Catania & Reynolds, 1968). Subjects in the Noncontingent group were given the same number of grain presentations as the Contingent subjects, but the grain deliveries were administered independently of the subjects’ behavior. Grain presentations were scheduled with mean interdelivery intervals progressing from 10 to 30 and then 45 set in a sequence congruent with that used for the Contingent group, but the subjects in the Noncontingent group were not yoked to subjects in the Contingent group. Subjects in the Control group were given one session of hopper training in which they received 40 grain presentations under the same temporal restrictions as those employed in the first session for the Contingent and Noncontingent subjects. The treadle was not present in the experimental chamber during this stage of training for any of the groups.

Twenty-four hours after termination of the above pretraining conditions, the treadle was inserted into the operant chamber and subjects could receive grain deliveries only by pressing the treadle. A continuous schedule of reinforcement was employed, and experimental sessions were 3 hr in duration. The unilluminated response key remained accessible during this stage of the experiment but key pecks had no effect upon grain deliveries. When necessary, subjects received supplemental feedings in order to maintain their 75% weights. These feedings occurred no sooner than 2 hr after the termination of individual experimental sessions. Individual subjects were terminated from the experiment when they had emitted 100 treadle presses, or, if they failed to emit 100 responses, when they had received a total of 24 hr of testing following their first treadle press response.

Acquisition of treadle pressing by the three experimental groups was compared, using the following measures: median time (minutes) to the first treadle press, median time elapsing between the first and fiftieth treadle press, and median time elapsing between the fiftieth and one-hundredth treadle press. These measures were devised in order to delineate more precisely the locus and permanence of the effects, if any, of the various types of pretraining upon learning to treadle press.

Results

The mean rate of key pecking for subjects in the Contingent group during the terminal session of preliminary training was 46.3 responseslmin, with a range of 11.5 to 79.9. All of the subjects were observed to be eating reliably from the hopper during the terminal grain presentations of preliminary training.

The median times elapsing prior to the initiation of the first response for subjects in the Contingent, Noncontingent, and Control groups were 58.5,


TIME IN MINUTES FROM RI-50 (Treadk-CRF)

FIG. 1. Individual scores for the time elapsing between the first and fiftieth responses for all subjects in Experiment I.

112, and 23 min, respectively. These measures do not represent reliable differences as indicated by the results of a Kruskal-Wahis analysis of variance, as individual scores were highly variable both within and across groups.

The three groups differed reliably in the median times elapsing between the first and fiftieth treadle press (H(2) = 6.495, p < .05). This measure was greatest for subjects in the Noncontingent group (29 min) and approximately equal for subjects in the Contingent and Control groups, 13 and 14 min, respectively. Supplemental analyses revealed that this measure for the Noncontingent group differed reliably from that of the Contingent and Control groups ( Us( n 12~112) = 32.5 and 34, respectively). The small difference between the Contingent and Control groups failed to reach statistical significance.

The median time elapsing during the emission of the final 50 treadle presses was considerably less than that elapsing during the first 50 responses for all three groups, and the groups did not differ reliably on this measure. Median times for this measure were 6.0, 7.5, and 9.0 min, respectively, for the Contingent, Noncontingent, and Control groups.

Figure 1 illustrates individual scores for the time from the first to the fiftieth response for all subjects. These data are presented in order to depict accurately the magnitude of the variability inherent in the measure. The range of variability was smallest for the Control group (8-64 min) and of approximately the same magnitude for the Contingent group (8- 138 min) and the Noncontingent group (9-373 min).

Median interresponse times for the first 10 treadle presses are compared graphically for the three groups in Fig. 2. The individual interresponse times were calculated in units of 0.5 min. The figure depicts rapid acquisition of the treadle-press response with appreciable differences among the groups observed only in the time elapsing between the first and second responses (H(2) = 7.405, p < .05). The results of multiple


0 - - Honcontingsnt

. -Control,

A----- -. Contlngsnt

RESPONSE N TO N+I (TREADLE CRF)

FIG. 2. Median interresponse time curves for the first 10 treadle presses in Experiment I.

Kruskal- Wallis analyses of variance indicated no reliable differences among the groups for the nine remaining interresponse times.

Discussion

The results of Experiment I demonstrated that protracted exposure to response-independent grain deliveries interfered with learning to behave in accordance with a response-reinforcer contingency. Subjects in the Noncontingent group exhibited retarded acquisition of treadle pressing relative to the Control group, whereas the birds which were previously trained to key peck acquired the treadle-press response in a manner similar to the controls. These findings may be viewed as a general corroboration and extension of those reported by Engberg ef al. (1972).

In comparing these data with those representing learned helplessness, however, a marked difference emerges. The escape behavior of subjects pretreated with response-independent shock typically fails to become more effecient with extended testing, and long escape latencies of inescapably preshocked subjects relative to controls are evident throughout at least 30 trials of testing (e.g., Maier, Albin, & Testa, 1973; Seligman, Rosellini, & Kozak, 1975). In the present experiment, interference with learning to treadle press following exposure to response-independent grain deliveries was overcome quite rapidly (i.e., within the emission of the first 10 responses).

Maier et al. (1973) have demonstrated that the degree of retardation in learning to escape from shock following response-independent preshocks is directly related to the complexity of the escape contingency. Rats


receiving response-independent preshock exhibit no deficit in acquiring a simple two-way shuttle response, whereas requiring the subjects to emit two responses to terminate shock (i.e., leave and return to the side of the shuttlebox in which shock is first encountered) yields a substantial retardation in learning the response relative to no-shock controls. Seligman and Beagley (1975) have demonstrated a similar relationship with rats, using bar pressing as the escape response.

Generalizing from the data of Maier et al. (1973) and Seligman and Beagley (1975), it was hypothesized that an increment in the complexity of the treadle-press contingency might increase the magnitude of retardation in the acquisition of treadling following response-independent feeding. If so, the results would bolster the apparent relationship between learned helplessness and laziness.

EXPERIMENT II

Two groups of pigeons received preliminary training under conditions identical to those described for the Noncontingent and Control groups of Experiment I. Following this training the treadle was inserted into the operant chamber, and grain was presented following every third response.

Method

Subjects. The subjects were 24 experimentally naive pigeons obtained from a local supplier and maintained at 75% of their ad lib weight. Subjects were housed under the same conditions as those described for Experiment I.

Apparatus. The apparatus was the same as described for Experiment I. Procedure. Subjects were unsystematically assigned to one of two

groups ( IZ = 12)) a Noncontingent and Control group. Training conditions and data analyses were carried out in a manner identical to that described for Experiment I with one exception. During the testing phase of the experiment, the first treadle press produced grain but subsequent reinforcers were programmed in accordance with a fixed-ratio three (FR3) schedule of reinforcement.

Results

All subjects were observed to be eating reliably from the hopper during the terminal grain presentations of preliminary training. The data for two subjects, a Control and a Noncontingent bird, were not included in the following analyses due to a malfunctioning of the programming apparatus.

As was observed in Experiment I, the median times to the first response for the Noncontingent group (75.5 min) and Control group (13.5 min) did not differ reliably. Again, individual scores for this measure were highly variable.


TIME IN MINIJTES mou RI- 50 (TREADLE R13)

FIG. 3. Individual scores for the time elapsing between the first and fiftieth responses for all subjects in Experiment II.

A large difference was observed between the groups in the median time elapsing between the first and fiftieth treadle press. This measure was 200 min for the Noncontingent group as opposed to 10 min for the Control group (U(nIl,nll) = 8, p < .05). Two subjects in the Noncontingent group were assigned a maximum score of 1440 min for this measure as they failed to complete 50 responses within 24 hr of their first treadle press. One subject emitted 12 responses and the other 18 during the 24-hr period.

Comparison of the median time elapsing between the first and fiftieth response for the control groups across Experiments I and II indicated no statistically significant difference. An equivalent comparison for the Noncontingent groups of Experiments I and II illustrated that the FR3 contingency produced a substantial increment in this measure (U(nll,nl2) = 28,~ < .05).

The median time elapsing during the emission of the final 50 treadle-presses was again considerably less than that elapsing during the first 50 responses for both the Control group (4 min) and Noncontingent group (4 mm).

Figure 3 illustrates individual scores for the time elapsing from the first to the fiftieth response for ah subjects in Experiment II. The range of variability observed in the Control group was 5.5-40.5 min, as opposed to 23- 1440 min for the Noncontingent group.

Median interresponse times for the first 10 responses in Experiment II are illustrated in Fig. 4. Notice that the median inter-response time curve for the Control group is indistinguishable from that of the Control group in Experiment I (Fig. 2). The curve for the Noncontingent group in Experiment II more closely approximates a gradual learning function than does that of the Noncontingent group of Experiment I. Multiple statistical analyses indicated that three of the ten interresponse times for the Noncontingent and Control groups of Experiment II differed at the .05 level. These included the time elapsing between the first and second,


RESPONSE N TO N+I (TREADLE FR3)

FIG. 4. Median interresponse time curves for the first 10 treadle presses in Experiment II.

second and third, and seventh and eighth responses (Us(n 11 ,n 11) = 26.5, 30, and 27.5, respectively).

Discussion

The results of Experiment II exhibited the same general trends as those obtained in Experiment I. The Noncontingent and Control groups differed reliably in their acquisition of treadle pressing only with regard to the time taken to emit the first 50 responses. In addition, the results of Experiment II illustrated that the degree of retardation of the acquisition of treadle pressing following protracted exposure to response-independent grain presentations was enhanced by increasing the complexity of the treadle-press contingency. Subjects in the Noncontingent group of Experiment II required a substantially greater amount of time to emit their first 50 responses than subjects in the Noncontingent group of Experiment I, whereas the increment in the complexity of the contingency produced no appreciable effect upon the acquisition of treadle pressing by subjects in the Control group. This finding, augmented by the failure of two subjects in the Noncontingent group of Experiment II to learn to treadle press within the allotted time, was viewed as support for the proposition that learned helplessness may be a phenomenon amenable to investigation with appetitive procedures.

Although interference with the acquisition of treadling for subjects in the Noncontingent group of Experiment II was enhanced by increasing the response requirement of the treadle-pressing contingency, this interfer-


ence was once again counteracted or overcome rather quickly for most subjects when compared with the degree of interference with learning demonstrated in learned helplessness experiments. It might be expected on the basis of the results obtained in Experiments I and II that the degree of interference with the acquisition of treadle pressing following response- independent food might be further magnified by additional increments in the complexity of the treadle-pressing contingency. This strategy has been employed in a sequel to the present experiments. Two groups of pigeons pretreated like the Noncontingent and Control groups of Experiments I and II acquired an FR6 treadle-press response at rates that did not differ statistically from those of the respective Noncontingent and Control groups of Experiment II. In other words, the results of Experiment II were replicated, but the difference between acquisition of treadle pressing for the Control and Noncontingent groups was no larger than that observed in Experiment II. These data suggest that there is not a simple linear relationship between the complexity of the contingency and the rate of learning to treadle press for subjects previously exposed to response- independent food.

Perhaps the differing degrees of relative permanence of interference with learning an operant response following response-independent food and shock is related to procedural asymmetries in the appetitive and aversive instances. When a dog or rat which has received response-independent shocks is placed into the shuttlebox or operant chamber it receives occasional shocks through the grid floor of the apparatus. These shocks presumably elicit the expectancy acquired during the pretraining: “Nothing I do matters” (Maier et at., 1969). Failure to respond and terminate shock might tend to confirm this expectancy,.as shock onset and offset would remain functionally independent of the subject’s behavior. Now, assume that the subjects in the Noncontingent groups of the present experiments developed the same expectancy with regard to their behavior and its outcomes. When they were exposed to the operant chamber during the treadle-pressing phase of the experiment the context of the chamber may have elicited this expectancy, but nothing occurred to confirm it. Quite to the contrary, grain presentations occurred only after the subjects emitted a treadle press. In other words, there was a forced exposure to the response-reinforcer contingency with each grain delivery. Failure to respond produced no immediate consequences.

Failures to escape from shock following response-independent shock may be eventually overcome by forced exposure to the escape contingency (Seligman, Maier, & Geer, 1968). In consideration of these data it might be concluded that the interfering effect of exposure to response-independent grain in the present experiments was rather quickly overcome by: (1) lack of confirmation of the expectancy that grain deliveries and behavior were


independent and (2) the forced exposure to the contingency accompanying each grain delivery during the treadle-pressing phase of the experiments.

Although the results of the present experiments have been interpreted throughout from a learned helplessness or laziness point of view, an alternative interpretation should be considered. Perhaps the retarded acquisition of treadling in the Noncontingent groups is attributable to the superstitious conditioning of responses during pretraining that were topographically incompatible with treadle pressing (cf. Gamzu, Williams, & Schwartz, 1973; Welker, Hansen, Engberg, & Thomas, 1973). The design of the present experiment does not provide a definitive test for differentiation between this alternative and learned laziness. Such a test would require, in the least, the fulfillment of two conditions: (1) a priori specification of the compatibility relations among response topographies in the pigeons’ behavioral repertoire and (2) observation of superstitious conditioning in pigeons during response-independent feeding with a variable-time 4%set schedule. It is ofinterest, however, that the subjects in the Contingent group of Experiment I were pretrained with a specific response (key pecking) which might have competed with treadle pressing, but did not. It is not obvious why response competition would occur only in the Noncontingent groups.

REFERENCES

Catania, C. A., & Reynolds, G. S. A quantitative analysis of the responding maintained by interval schedules of reinforcement. Journal ofthe Experimental Analysis ofsehavior, 1%8, 11, 327-383.

Engberg, L. A., Hansen, G., Welker, R. L., & Thomas, D. R. Acquisition of key-pecking via autoshaping as a function of prior experience: “Learned laziness?” Science, 1972,178, 1002- 1004.

Gamzu. E., & Williams, D. R. Associative factors underlying the pigeon’s key pecking in auto-shaping procedures. Journal of the Experimental Analysis of Behavior, 1973,19, 225-233.

Gamzu, E.. Williams, D. R., & Schwartz, B. Pitfalls of organismic concepts: “Learned laziness?” Science, 1973, 181, 367-368.

Maier, S. F. Failure to escape traumatic electric shock: Incompatible skeletal-motor responses or learned helplessness? Learning and Motivation, 1970, 1, 157- 169.

Maier, S. F., Albin, R. W., & Testa, T. J. Failure to learn to escape in rats previously exposed to inescapable shock depends upon the nature of the escape response. Journal of Comparative and Physiological Psychology, 1973, 85, 581-592.

Maier, S. F., Seligman, M. E. P., & Solomon, R. L. Pavlovian fear conditioning and learned helplessness. In B. A. Campbell & R. M. Church (Eds.), Punishment and aversive behavior. New York: Appleton-Century-Crofts, 1969. Pp. 299-342.

Maier, S. F., & Testa, T. J. Failure to learn to escape by rats previously exposed to inescapable shock is partly produced by associative interference. Journal of Comparative and Physiological Psychology, 1975, 88,554-564.

Overmier, J. B., & Seligman, M. E. P. Effects of inescapable shock upon subsequent escape and avoidance responding. Journal of Comparative and Physiological Psychology, 1967, 63, 28-33.

Set&man, M. E. P., & Beagley, G. Learned helplessness in the rat. Journal of Comparative and Physiological Psychology, 1975, 88, 534-541.


Seligman, M. E. P., 8~ Maier, S. F. Failure to escape traumatic shock. Journal of Experimental Psychology, 1967, 74, 1-9.

Seligman, M. E. P., Maier, S. F., & Geer, J. H. Alleviation of learned helplessness in the dog. Journal of Abnormal Psychology, 196&73, 256-262.

Seligman, M. E. P., Maier, S. F., & Solomon, R. L. Unpredictable and uncontrollable aversive events. In F. R. Brush (Ed.),Aversive conditioning andlearning. New York: Academic Press, 1971. Pp. 347-402.

Seligman, M. E. P., Rosellini, R. A., & Kozak, M. J. Learned helplessness in the rat: Immunization, time course and reversibility. Journal of Comparative and Physiologi- cal Psychology, 1975, 88, 542-547.

Welker, R. L., Hansen, G., Engberg, L. A., & Thomas, D. R. A. reply to Gamzu, Williams and Schwartz. Science, 1973, 181, 368-369.

Williams, D. R., & Williams, H. Auto-maintenance in the pigeon: Sustained pecking despite contingent non-reinforcement. Journal of the Experimental Analysis of Behavior, 1969, 12, 511-520.

Received March 26, 1975 Revised October 15, 1975

Documents

Acquisition of a Free-Operant-Appetitive Response