Effects of response elimination procedures upon the subsequent reacquisition of autoshaping

In Experiment 1, three groups of pigeons were autoshaped and then administered one of three different response-elimination procedures. Group TRC (truly random control) and Group Backward (backward conditioning) ceased responding more rapidly than Group CS-only. In a subsequent reacquisition test with a novel CS, Groups TRC and Backward were retarded relative to Group CS-only. In Experiment 2, the CS-only and TRC response-elimination treatments were not differentially effective in extinguishing the response. The treatments were followed either by five sessions of unpredictable US presentations (US-only) or by five sessions of remaining in their home cages (hold) prior to the reacquisition test. The TRC treatment retarded reacquisition relative to the CS-only treatment; the US-only posttreatment manipulation failed to reliably retard reacquisition relative to Hold, although retarding effects of US-only are discernible in a block-by-block analysis. Applications and limitations of a context-blocking account of these results are discussed.     

Tests of two explanations of response elimination by noncontingent reinforcement

Two experiments are reported on the elimination of autoshaped keypecking in pigeons by introducing added feedings that have the effect of removing the contingency between keylight (CS) and feedings. In Experiment 1, added feedings were signaled, by an already conditioned stimulus, in one group but not in another. Contrary to the Rescorla-Wagner theory, but consistent with scalar expectancy theory, responding in these groups declined at equal rates. In Experiment 2, following acquisition, some groups were exposed to repeated sessions in which they received feedings alone prior to receiving both CS and feedings noncontingently. Although prior exposure to feedings did reduce the total amount of responding in subsequent noncontingent sessions, it did not, contrary to scalar expectancy theory, reduce the initial level of responding to the CS. It is suggested that a differential between reinforcing conditions in the CS as compared with neighboring non-CS periods is more fundamental to conditioning than is acknowledged in the Rescorla-Wagner theory or in scalar expectancy theory.     

The effects of activity conditioned in random CS/US training on performance in autoshaping

Random presentations of keylights and food retarded acquisition and suppressed asymptotic rates of keypecking in autoshaping. Sequences of 10 sessions of random training alternated with 10 sessions of autoshaping resulted in poor performance (in terms of both the acquisition and asymptotic indices) relative to a group that received sequences of CS-only (rather than random) training alternating with autoshaping. When the birds that previously were trained with the random sequence were exposed to CS-alone extinction, retardation of acquisition was alleviated but the asymptotic suppression was not (Experiment 1). Pigeons with a history of autoshaping without prior random training showed no asymptotic suppression when exposed to the random procedure. These birds, when switched between two-session sequences of random training alternating with two-session sequences of autoshaping, were able to (1) surpass pigeons that received CS-only rather than random treatment in terms of asymptotic levels of responding in autoshaping, and (2) showed improvement in extinction performance with repeated random/autoshaping sequences (Experiment 2). Detailed observations of pigeons in random training showed that stereotypic activity behaviors were acquired, and these generally persisted in autoshaping; the degree of change in these behaviors was related to asymptotic rates of keypecking in autoshaping (Experiment 3). The same kinds of behaviors were observed when pigeons initially were autoshaped, and these persisted in subsequent random and fixed-interval 10-sec training. We suggest that the suppression effect is reflected in activity, conditioned in random training, which persists in autoshaping (especially if the activity is compatible with the kinds of behaviors elicited by the autoshaping contingency itself), and that the effect may be a deficit due to performance factors rather than to associative learning.     

Reacquisition following extinction in appetitive conditioning

In four experiments utilizing an appetitive conditioning preparation, reacquisition of conditioned responding was found to occur both rapidly and slowly following extinction. In Experiment 1, acquisition of responding to a tone that had been conditioned and extinguished occurred more rapidly than acquisition in either a group that received equivalent exposure to the food unconditioned stimulus or a “rest” control group that received only exposure to the apparatus in the first two phases. However, reacquisition was impaired relative to acquisition in a “learning-experienced” group that had previously received conditioning and extinction with a different stimulus. Experiments 2 and 3 produced similar results, but also found that high responding during reacquisition was confined to trials that followed reinforced, rather than nonreinforced, trials. Experiment 4, in which very few initial conditioning trials were used, produced reacquisition that was slow compared with both learning-experienced and rest controls. The results are consistent with a role for sequential learning: Reacquisition is rapid when animals have learned that reinforced trials signal other reinforced trials.     

Temporal control during maintenance and extinction of conditioned keypecking in ring doves

Timing of conditioned keypecking was investigated using a delay autoshaping procedure. In two experiments, the CS-US interval (ISI) was varied and the temporal pattern of responding during unreinforced probe trials extending beyond the reinforced ISI was recorded. Both experiments showed temporal control of keypecking at ISIs of 4, 8, and 16 sec, regardless of whether the probe duration was held constant (Experiment 1) or covaried with the ISI (Experiment 2). Analyses showed that the timing of keypecking was scalar. Increased responding near the end of the probe was due possibly to independent timing of the probe duration. In a third experiment, a group of subjects trained at an ISI of 8 sec were extinguished by presentation of only probe trials. Although the maximal response rate declined progressively, timing of keypecking did not change.     

Second-order conditioning of the pigeon’s keypeck

Pigeons learned to peck a keylight (S2) when it was paired with a stimulus (S1) that already evoked keypecking. Control procedures showed that S2 acquired control over responding because it was paired with S1 and because S1 had a conditioning history, thereby supporting the claim that S2 was a second-order conditioned stimulus. Second-order conditioning occurred as rapidly when S1 was a keylight as when it was a tone. Test procedures showed that after second-order conditioning, responding to S2 was markedly debilitated by the extinction of responding to S1, indicating that the ability of S2 to evoke a response importantly depends upon the continued ability of S1 to do so. Our demonstration that directed motor action in the pigeon is susceptible to second-order conditioning suggests a new interpretation of conditioned reinforcement in instrumental learning. Our demonstration that the effectiveness of S2 depends upon the continued effectiveness of S1 indicates that S-S associations are formed in this version of the second-order conditioning experiment.     

Signaling functions of the second-order CS: Partial reinforcement during second-order conditioning of the pigeon’s keypeck

The signaling function of the second-order CS (S2) was manipulated in second-order autoshaping by arranging a partial reinforcement schedule. S2 was paired with a well-conditioned first-order CS (SI) on a continuous reinforcement or a 25% reinforcement schedule in different groups. Schedule of reinforcement did not influence the number of S2-S1 pairings required to establish keypecking to S2. However, in the postacquisition sessions, responding to S2 was initially weaker but persisted for many more sessions on the 25% schedule than on the 100% schedule. The data indicate that S2-S1 pairings are responsible both for the acquisition of second-order keypecking to S2 and for the subsequent conversion of S2 into an inhibitory stimulus.     

The effects of Pavlovian CSs on two food-reinforced baselines— with and without noncontingent shock

Two experiments explored the effects of Pavlovian (tone-shock) CS+, CS?, and truly random control (TRC) contingencies on two different food-reinforced instrumental baselines. One food-reinforced baseline contained noncontingent shock, while the other did not. In the first experiment, a TRC contingency was shown to produce suppression of food-reinforced responding, while a CS? contingency did not. When noncontingent shock was added to the baseline, however, the TRC stimuli failed to produce suppression, and the CS? contingency increased response rates over baseline level. In a second experiment, the effects of TRC and CS+ contingencies were compared on these same two baselines. While the CS+ produced suppression on both shock and no-shock baselines, the TRC contingency again produced suppression on only the no-shock baseline.     

Sign and goal tracking during delay and trace autoshaping in pigeons

In two experiments, pigeons were exposed to an autoshaping procedure in which a keylight was followed by food under delay or trace conditions, while measures were taken of keypecking and time spent near the key. In Experiment 1, the birds in the trace group spent less time near the key than did the delay birds. Moreover, the trace birds exhibited a pattern of withdrawal from the key during the trial. In Experiment 2, visual observations of the birds’ location and latencies to eat both indicated that the trace birds’ withdrawal from the conditioned stimulus was accompanied by goal tracking. The difference in performance between the delay and trace conditions was taken as evidence that a trace stimulus may exert control over behavior as an occasion setter.     

The effect of number and minimum duration of post-response-prevention escapes on the resistance to extinction of an avoidance response

After one-way avoidance training, rats were exposed, during avoidance response prevention, to light (CS-only) presentations or to light-shock (CS-US) pairings. Subgroups were then given 1, 5, or 10 trials during which they could escape immediately (unrestricted) or after 5 sec (restricted) by means of the previously conditioned avoidance response from a simultaneous light-shock compound. All animals were then exposed to avoidance extinction. The number of unrestricted escapes increased responding for CS-only animals, but had no significant effect on the performance of CS-US animals. Nevertheless, resistance to extinction was considerably less for CS-only animals given 10 unrestricted escapes than for CS-US animals given one unrestricted escape. One restricted escape had no more effect than one unrestricted escape for either response-prevention group. However, 5 restricted escapes elevated responding for CS-only animals to the level of CS-US animals. Extinction responding for CS-US animals increased significantly only after 10 restricted escapes. Since CS-only animals showed no further increase, resistance to extinction once more was greater for CS-US animals. These results, together with the very brief unrestricted escape latencies of CS-only animals, support a greater role for Pavlovian extinction than for response competition in the facilitation of avoidance extinction by CS-only response prevention. The fact that 10 restricted escapes were required to elevate resistance to extinction for CS-US animals over that obtained with one unrestricted escape attests to the effectiveness of Pavlovian conditioning during avoidance response prevention in elevating CS aversiveness to a near ceiling level.     

Classical-classical transfer: Effects of prior appetitive conditioning upon aversive conditioning in rabbits

An appetitive-ayersive transfer experiment with rabbits determined that prior paired and unpaired tone CS and water US presentations, given in jaw-movement (JM) conditioning, respectively facilitated and retarded the acquisition of the nictitating membrane (NM) CR when the tone was subsequently paired with a shock US. In addition, the unpaired tone and water deliveries reduced the level of JM conditioning that was undertaken following the completion of NM CR acquisition. Finally, the reacquisition of the NM CR was accompanied by a large savings effect in contrast to the failure of the JM CR to display reacquisition savings. When the present findings are compared to the results of previous work addressing the influence of prior NM conditioning procedures upon subsequent JM CR acquisition, an asymmetry in appetitive-aversive interactions is indicated. This asymmetry encourages a reinterpretation of the opponent-process explanation of appetitive-aversive interactions. Moreover, the observed effects of the unpaired CS suggest the operation of a salience factor.     

Specification of the stimulus-reinforcer relation in multiple schedules: Delay and probability of reinforcement

Control of pigeons’ keypecking by a stimulus-reinforcer contingency was investigated in the context of a four-component multiple schedule. In each of three experiments, pigeons were exposed to a schedule consisting of two two-component sequences. Discriminative stimuli identifying each sequence were present only in Component 1, which was 4, 6, or 8 sec in duration, while reinforcers could be earned only in Component 2 (30 sec in duration). Control by a stimulus-reinforcer contingency was sought during Component 1 by arranging a differential relation between Component 1 cues and schedule of reinforcement in Component 2. In Experiment 1, rate of keypecking during Component 1 varied with the presence and absence of a stimulus-reinforcer contingency. When a contingency was introduced, rate of keypecking increased during the Component 1 cue associated with the availability of reinforcement in Component 2. In Experiment 2, the stimulus-reinforcer contingency was manipulated parametrically by varying the correlation between Component 1 cues and Component 2 schedules of reinforcement. Responding in Component 1 varied as a function of strength of the stimulus-reinforcer contingency. The relatively high rates of Component 1 responding observed in Experiments 1 and 2 pose difficulties for conceptions of stimulus-reinforcer control based on probability of reinforcement. In these two experiments, the stimulus-associated probabilities of reinforcement in Component 1 were invariant at zero. An alternate dimension of stimulus-reinforcer control was explored in Experiment 3, in which Component 1 cues were differentially associated with delay to reinforcement in Component 2, while probability of reinforcement was held constant across components. When the stimulus-reinforcer contingency was in force, rate of responding in Component 1 varied inversely with delay to reinforcement in Component 2. In a quantitative analysis of data from Experiments 2 and 3, relative rate of responding during Component 1 was strongly correlated with two measures of relative delay to reinforcement.     

The form of timing distributions in pigeons under penalties for responding early

The peak procedure was used in two experiments. Pigeons in the penalty group in Experiment 1 were rewarded with food in the first phase for the first peck after 12.5 sec had elapsed since the onset of a keylight. In the second phase, reward was withheld if the pigeons pecked within 6.25 sec after keylight onset. Responses in time were tabulated on occasional unrewarded tests in which the keylight was left on for 37.5 sec. Under the penalty contingencies, the response distribution in time remained nearly symmetric about the peak, while the spread of the distribution narrowed, and the time of peak responding came slightly earlier. The yoke group underwent a schedule of rewards similar to that for the penalty group, but without the penalty contingencies. Their response distributions remained similar throughout. The results of Experiment 1 were replicated in Experiment 2, which showed further that the changes due to the penalty contingencies did not generalize to the use of another key on which the penalty contingencies were not in effect. The narrower spread under the penalty contingencies is explained in terms of a change in threshold for when to start responding, and more weight being given to timing versus responding in the presence of the signal per se. No explanation was found for the change in the time of peak responding.     

The role of signaled periods of nonreinforcement in responding on a random schedule in autoshaping

Little responding develops to a conditioned stimulus (CS) that is placed in a random relation to an unconditioned stimulus (US). However, if the USs not preceded by that CS are themselves signaled by another stimulus, then the CS does come to elicit responding. This result has been attributed (e.g., by Durlach, 1983) to the signal’s blocking of conditioning to background cues that otherwise would prevent conditioning of the CS. However, Goddard and Jenkins (1987) have suggested the alternative that signaling the USs promotes responding due to the adventitious creation of periods of signaled nonreinforcement. Two experiments were conducted to assess this alternative, involving an autoshaping preparation in pigeons. In Experiment 1, little responding to a keylight CS presented in a random relation to a food US occurred, despite the explicit presentation of a discrete noise signaling periods of no food in the intertrial interval (ITI). Experiment 2 was designed to replicate the procedure of Goddard and Jenkins, in which an auditory stimulus extended throughout the ITI of a random schedule, terminating only prior to extra USs and during the CS. Contrary to their findings, little responding developed to the target CS. However, responding did develop when the sound-free period occurred only prior to the extra USs. These results offer little support for the hypothesis that signaled periods of nonreinforcement promote responding on random schedules. However, they are consistent with the view that signaling of ITI USs acts by preventing conditioning of potentially competitive background cues.     

Savings in classical conditioning in the rabbit as a function of extended extinction

In the present experiments, savings phenomena following a limited amount of initial acquisition and extended extinction were examined. Experiments 1 and 2 compared rates of reacquisition following brief acquisition and various amounts of extinction in conditioning of the rabbit’s nictitating membrane and heart rate response, respectively. Experiment 3 compared rates of acquisition to a novel stimulus (e.g., light) following brief acquisition and various amounts of extinction to another stimulus (e.g., tone). In addition, in Experiment 3 recovery of responding to the extinguished stimulus during acquisition to the novel, cross-modal stimulus was examined. Experiments 1, 2, and 3 demonstrated that with a limited number of acquisition trials (1) there was a graded reduction in the rate of reacquisition as a function of the number of extinction trials in both conditioning preparations, (2) there was a graded reduction in the rate of cross-modal acquisition as a function of the number of extinction trials, but (3), in Experiment 3, recovery of responding to the extinguished stimulus during cross-modal training of the novel stimulus appeared uniformly robust even in the face of extended extinction.     

Retardation tests of inhibition following discriminative autoshaping

Four experiments explored various applications of the retardation test of inhibition following discriminative autoshaping. In all studies, discriminative autoshaping consisted of food-reinforced presentations of a green key-light CS+ and nonreinforced presentations of a vertical white line on a green background CS?. In Experiment 1, the inhibitory properties of the vertical Une CS? were assessed by comparing the acquisition of keypecking to the now-reinforced vertical line CS in groups previously receiving discriminative autoshaping (as above), discriminative autoshaping with a red key CS?, and nondiscriminative autoshaping. The results indicated that the former CS? was retarded in being transformed into a CS+. In Experiment 2, the discriminative autoshaping procedure was followed by testing for acquisition with independent groups receiving one of four different CSs differentiated by line orientation. The results indicated that the orientation attribute of the CS? was inhibitory, in that a between-subjects resistance-to-reinforcement gradient of inhibition was obtained. In Experiment 3, discriminative autoshaping was followed by a within-subjects resistance-to-reinforcement generalization test based on line orientation, which failed to yield an orderly gradient. The implementation of a DRO contingency in the within-subjects test (Experiment 4) was ineffective in generating an incremental gradient. Implications for inhibitory assessment methodology are discussed.     

Temporal encoding in trace conditioning

Conditioned lick suppression in rats was used to explore the role of timing in trace conditioning. In Experiment 1, two groups of rats were exposed to pairings of a CS (CS1) with a US, under conditions in which the interstimulus interval (ISI) that separated CS1 offset and US onset was either 0 or 5 sec. Two additional groups were also exposed to the same CS1→US pairings with either a 0 or a 5-sec ISI, and then received “backward” second-order conditioning in which CS1 was immediately followed by a novel CS2 (i.e., CS1→CS2). A trace conditioning deficit was observed in that the CS1 conditioned with the 5-sec gap supported less excitatory responding than the CS1 conditioned with the 0-sec gap. However, CS2 elicited more conditioned responding in the group trained with the 5-sec CS1-US gap than in the group trained with the 0-sec CS1-US gap. Thus, the CS1-US interval had inverse effects on first- and second-order conditioned responding. Experiment 2 was conducted as a sensory preconditioning analogue to Experiment 1. In Experiment 2, rats received the CS1?CS2 pairings prior to the CS1→US pairings (in which CS1 was again conditioned with either a 0 or a 5-sec ISI). Experiment 2 showed a dissociation between first- and second-order conditioned responding similar to that observed in Experiment 1. These outcomes are not compatible with the view that differences in responding to CSs conditioned with different ISIs are mediated exclusively by differences in associative value. The results are discussed in the framework of the temporal coding hypothesis, according to which temporal relationships between events are encoded in elementary associations.     

Effects of signaled retention intervals on pigeon short-term memory

In three delayed matching-to-sample experiments, pigeons were given distinctive stimuli that were either correlated or uncorrelated with the scheduled retention intervals. Experiment 1 employed a single-key, go/no-go matching procedure with colors as the sample and test stimuli; lines of differing orientations signaled short or long delays for one group, whereas the lines and the delays were uncorrelated for the other group. The function relating discriminative test performance to delay length was steeper in the correlated group than in the uncorrelated group. In addition, the line orientation stimuli controlled differential rates of sample responding in the correlated group, but not in the uncorrelated group. In Experiment 2, subjects extensively trained with correlated line orientations were exposed to reversed cues on probe trials. Miscuing decreased discriminative test responding at the short delay, but enhanced it at the long delay. As in the correlated group of the first experiment, rates of sample keypecking were higher in the presence of the “short” time tag than in the presence of the ”long” time tag. Experiment 3 used a three-key choice-matching procedure and a within-subjects design, and equated reinforcement rate at the short and long delays. When auditory stimuli were correlated with delay length, the function relating choice accuracy to delay was steeper than when the stimuli and the delays were uncorrelated. The consistent effects of signaled retention intervals on memory performance may be understood in terms of differential attention to the sample stimuli.     

Short-term retention of “surprising” events following different training conditions

Terry and Wagner (1975) have suggested that short-term retention of information about an event is enhanced if the occurrence of the event is surprising. To investigate this idea, we trained two groups of pigeons in a preparatory-releaser procedure in which half the trials started with the presentation of food (the preparatory event). The preparatory food presensation was signaled by an 8-sec white keylight in the signaled, but not in the unsignaled, group. After a retention interval, varying between 2 and 32 sec, the releaser stimulus (CS_R), a red keylight, was presented for 8 sec in the absence of any reinforcement. The remaining trials were initiated by the presentation of CS_R, and the first peck occurring 8 sec after the onset of CS_R was reinforced by food. The preparatory event controlled responding to CS_R at the short retention interval, with the level of control declining systematically with increasing retention intervals. On probe test trials, the presentation of the preparatory food event was preceded by a stimulus that had previously been paired (CS+) or unpaired with food (CS?). Discriminative responding to CSr was better following CS? than following CS+ in the unsignaled, but not the signaled, group. These results suggest that the enhanced retention following surprising preparatory events reflects a generalization decrement induced by changing the signaling conditions between training and testing.     

Noncontingent and contingent no-choice intervals and concurrent performance

Pigeons responded to changeover-key concurrent variable-interval variable-interval reinforcement schedules while there were intervals during which the changeover key was inoperative (no-choice intervals). In Experiment 1, a multiple schedule on the changeover key signaled choice and no-choice intervals. All subjects showed near-perfect discrimination during initial discrimination training and rapid reacquisition of discrimination following contingency reversals. In Experiment 2, the onset of no-choice intervals was unsignaled and contingent on interchangeover time. The temporal distribution of changeover-key responses conformed to the temporal distribution of choice intervals. The results of both experiments suggest that changeover responding is modifiable as a function of its immediate consequences. The results of Experiment 2, in particular, suggest that time or some correlate of time since the last changeover response can determine subsequent changeover behavior.