Conditioning of simultaneous and serial feature-positive discriminations

Three experiments with rat subjects examined the development of simultaneous and serial feature-positive discriminations in appetitive conditioning. In Experiment 1, reinforced presentations of a simultaneous light-tone compound were intermixed with nonreinforced presentations of either the light or the tone. The compound stimulus acquired conditioned behaviors of a form characteristic of the predictive feature alone; the element common to reinforced and nonreinforced trials did not evoke conditioned behavior. In Experiment 2, reinforced presentations of a serial light-trace-tone compound were intermixed with nonreinforced tone-alone presentations. The light feature stimulus acquired conditioned behaviors characteristic of visual CSs. The tone stimulus, common to reinforced and nonreinforced trials, evoked conditioned behaviors characteristic of auditory CSs, but only when preceded by the light. In Experiment 3, variations in the interval between the light and tone on reinforced trials had little effect on responding to the light CS but substantially altered the pattern of responding to the tone CS. These results suggested that simultaneous and serial feature-positive discriminations may be solved differently. Performance in simultaneous feature-positive discriminations may be determined solely by associations between the feature stimulus and the reinforcer, but performance in serial discriminations may also involve the acquisition of a conditional cue function to the feature.     

Contextual control of pavlovian feature-positive and feature-negative discriminations

Three pigeons were trained with a Pavlovian serial feature-positive (F-P) discrimination task in a light context, in which the houselight was on, and with a Pavlovian serial feature-negative (F-N) discrimination task in a dark context, in which the houselight was off. Three other pigeons were trained with the F-P task in the dark context and the F-N task in the light context. These two contextual conditions were changed randomly trial by trial. The former birds learned the tasks within 60 sessions, by responding exclusively to the target keylight after the feature tone in the light context and by responding exclusively to the target not preceded by the feature in the dark context. Two of the latter birds required separate training of the F-P and the F-N tasks to acquire the discrimination: responding exclusively to the target after the feature in the dark context and responding exclusively to the target not preceded by the feature in the light context. The third bird, however, failed to learn the discrimination even with separate training. These results indicate that the four-term contingency (the context-feature-target-food relationship) controlled the birds’ behavior in the Pavlovian setting. The insertion of a temporal gap between the feature and the target impaired the F-N discrimination, although it had little effect on the F-P discrimination.     

Type of feature affects transfer in operant serial feature-positive discriminations

In two experiments, the effects of feature identity in operant serial feature-positive discriminations were examined with rats. Rats were trained with two serial feature-positive discriminations (F1 → T1+/ T1? and F2 → T2+/ T2?), in which different operants were reinforced with delivery of a sucrose solution during two auditory target cues (T1 and T2). The features (F1 and F2) were two visual cues, two flavored sucrose solutions, or one visual cue and one sucrose solution. Transfer of a feature’s control to the target of the other discrimination was observed only when the features were from the same modality. When observed, transfer responding was of the form originally trained to the target, rather than the feature, and was preserved after feature extinction. Control groups showed that the differential transfer was not solely the consequence of differential feature generalization. Implications for theories of occasion setting are discussed.     

Narrowing down the conditions for extinction of Pavlovian feature-positive discriminations in humans

The aim of this study was to delineate the minimal conditions for extinction of Pavlovian modulation in humans. Previous experiments at our lab showed that, after X ? A+/A?C acquisition training, X?C trials did not extinguish differential X ? A+/A?C responding, while X ? A?C trials did. Additionally, X ? A?C extinction training seemed only to extinguish differential X ? A+/A?C responding, while leaving differential responding on a concurrently trained Y ? B+/B?C discrimination intact. It thus seemed that the X ? A+/A?C discrimination can only be extinguished by X ? A?C extinction trials. (Rescorla, Journal of Experimental Psychology: Animal Behavior Processes 12, 16?C24, 1986), on the other hand, found that the minimal conditions for extinction were broader in pigeons: Namely, he found that an acquired X ? A+/A?C discrimination could be extinguished by presenting the original feature X in combination with a different target (B) that was minimally trained as an exciter. We thus wanted to examine whether this was also the case in humans. We found that nonreinforced X ? B?C presentations did not abolish discriminative X ? A/A responding when target B was a nonreinforced stimulus. Nonreinforced X ? B?C trials did extinguish the X ? A+/A?C discrimination when target B had previously been trained as a target for modulation (X ? B+/B?C or Y ? B+/B?C training) or as a reinforced exciter (B+). Our results thusf parallel and extend those in nonhuman animals (Rescorla, Journal of Experimental Psychology: Animal Behavior Processes 12, 16?C24, 1986).     

Higher order occasion setting

Higher order occasion setting with serially presented stimuli was investigated in an appetitively motivated, discrete-trial operant study with rats. Reinforcement of barpressing during an occasion-setting light (a discriminative stimulus) was contingent on immediately preceding second-order occasion setters (i.e., a click train or a buzzer served as a conditional discriminative stimulus). Moreover, the meanings of the clicks and buzzer were themselves indicated by a third-order occasion setter that preceded them (i.e., a white noise acted as a second-order conditional discriminative stimulus). Subjects responded more frequently and had shorter latencies to the first response in the presence of the light on trials during which barpressing was reinforced than on trials during which barpressing was not reinforced. The likelihood that the subjects solved the problem by responding to unique compound stimuli was minimized by the insertion of a 5-sec gap between the different controlling stimuli presented on each trial. Thus, these subjects appear to have mastered a second-order conditional discrimination, which is equivalent to third-order occasion setting if the discriminative stimulus (light) is viewed as a first-order occasion setter. Although the subjects learned to respond appropriately to each of the compound stimuli, differences in responding to specific stimuli were consistent with a higher order feature-positive effect. Some implications of higher order occasion setting are discussed, including the issue of independence between the different levels of occasion setting signaled by a single stimulus.     

Summation and transfer of negative occasion setting

Two experiments examined the summation and transfer of negative occasion setting in serial feature negative (X→A?/A+) and negative patterning discriminations (X→A?/X+/A+) in a discrete-trial operant task with rats. With both discrimination procedures, negative occasion setting transferred readily (but not perfectly) to cues trained as targets in another similar discrimination procedure, but not to cues that were separately trained. Furthermore, with both discrimination procedures, the negative occasion-setting powers of features from two individual discriminations summed, both when those features were trained with the same targets and when they were trained with different targets. After negative patterning discriminations, in which the feature (X) cues were separately reinforced, this summation of negative occasion setting occurred despite the concomitant summation of the features’ excitatory control in the absence of an explicit target cue. These data, which replicated and extended previously reported data from Pavlovian feature negative procedures, are discussed in the context of hierarchical, generalization, and network models of discrimination learning.     

Feature extinction enhances transfer of occasion setting

In three experiments, transfer effects in appetitive serial feature-positive discrimination procedures (X→A+,A?) were examined with rat subjects. Nonreinforced presentations of the feature (X) alone had little or no effect on that cue’s ability to modulate responding to the target cue with which it had originally been trained (A) but enhanced its ability to modulate responding to another cue (B), which had been the target within another serial feature-positive discrimination (Y→B+,B?). These effects were observed when the nonreinforcedX presentations occurred either concurrently withX→A+,A? training or afterward. Regardless of whether or not nonreinforcedX presentations were administered,X did not modulate responding evoked by a cue that had not been trained within a serial feature-positive discrimination. These data help resolve conflicts in published reports of transfer or its absence in this conditioning situation, and shed light on the nature of the modulatory or occasion-setting powers of feature cues in serial feature-positive discrimination learning.     

Negative priming and occasion setting in an appetitive Pavlovian procedure

Rats received training in which two auditory target stimuli, X and Y, were signaled by two visual stimuli, A and B, and followed by food (i.e., A-->X+, B-->Y+). The test consisted of presentations of X and Y preceded either by the same signal as during training (same trials: A-->X, B-->Y) or by the alternative signal (different trials: A-->Y, B-->X). After 8 training sessions, the animals responded less on same trials than on different trials; this effect was significantly reduced after 24 training sessions. In two additional experiments, animals that had also experienced presentations of A and B alone, either before or during training, showed the opposite pattern of results, responding more on same trials than on different trials. These results are interpreted as being due to the interaction between the effects of occasion setting and negative priming (see Wagner, 1981).     

Transfer of occasion setting: The role of generalization decrement

In Experiment 1, pigeons were trained on two feature-positive discriminations. A transfer test examined whether the feature from one discrimination enhanced responding to the target from the other. Transfer was obtained, but it was incomplete; the feature produced less responding to the transfer target than to its own. Experiments 2 and 3 examined whether this attenuation of responding was the product of generalization decrement induced by the novel combination of feature and target on transfer trials. Birds were trained on a pair of pseudo-occasion-setting discriminations in which each target was reinforced whether or not it was preceded by its feature. In a subsequent test, there was no loss of responding when novel combinations of features and targets were introduced; on the contrary, responding was, if anything, enhanced in this condition. This suggests that imperfect transfer is not due to generalization decrement but to the fact that an occasion setter is specific to its target stimulus.     

The ontogeny of the feature-positive effect in young chicks

The wide generality of the feature-positive effect (FPE) has caused speculation that the FPE may represent innate biases in the stimulus control of discriminative responding. There is little experimental evidence to date ragarding this possibility. In the present study, 1- or 4-day-old chicks were trained on a feature-positive (FP) or feature-negative (FN) discrimination with heat reinforcement. After the acquisition phase, these subjects received extinction training followed by a reacquisition phase. The FP performance was superior to the FN performance in both age groups. Extinction resulted in improved discrimination performance in both the FP and FN conditions. Unmasking of FN learning by the extinction treatment suggests that the FPE represents a deficit in performance, rather than an inability to learn the FN task. These data demonstrate that adult-like performance on feature discriminations is evident as early as the first day post-hatch.     

An exploration of the feature-positive effect in adult humans

Experiment 1 compared the acquisition of a feature-positive and a feature-negative discrimination in humans. In the former, an outcome was signaled by two stimuli together, but not by one of these stimuli alone. In the latter, the outcome was signaled by one stimulus alone, but not by two stimuli together. Using a within-group design, the experiment revealed that the feature-positive discrimination was acquired more readily than the feature-negative discrimination. Experiment 2 tested an explanation for these results, based on the Rescorla-Wagner theory, by examining how novel discriminations, based on a combination of a feature-positive and a feature-negative discrimination, were solved. The results did not accord with predictions from the theory. Alternative explanations for the results are considered.     

Stimulus control in multiple temporal discriminations

In multiple fixed interval (FI) schedules, rats are trained to discriminate different FIs that are signaled by different stimuli. After extensive training, the different stimuli often acquire control over performance, observed by an earlier increase in responding for stimuli that signal shorter FIs, as compared with stimuli that signal longer FIs. The order in which the different FIs are trained, either intermixed across cycles or in blocks of several cycles, may seem irrelevant given that average performance at asymptote may be similar. In this study, rats were trained in two procedures with multiple FIs presented intermixed within sessions or in blocks of one interval per session. Similar performance was observed at asymptote, but an inspection of early cycles in each session revealed that different stimuli acquired control over performance only when trained intermixed within each session. Although the stimuli reliably signaled the upcoming FI, when trained in successive blocks of 60 cycles, rats rapidly adjusted performance early in the sessions on the basis of the temporal aspects of the task, and not on the basis of the stimulus presented in the current cycle. These results are discussed in terms of overshadowing of the stimuli by temporal cues and in terms of conditions under which a stimulus acquires control over performance.     

Value transfer in concurrent-schedule discriminations by pigeons

When pigeons are trained on a discrete-trial simultaneous discrimination, some of the value associated with the positive stimulus appears to transfer to the negative stimulus (Zentall & Sherburne, 1994). Pigeons preferred a negative stimulus that had been discriminated from an always-positive stimulus (S+) over a negative stimulus that had been discriminated from a sometimes-positive stimulus (S±). A very different finding (suggestive of transitivity of preference or contrast) was reported by Belke (1992). On concurrent probe tests of stimuli associated with equal variable interval (VI) schedules but originally trained in alternative concurrent pairs (one with a richer schedule, the other with a poorer schedule—VI 20 sec vs. VI 40 sec and VI 40 sec vs. VI 80 sec), the stimulus originally paired with the poorer schedule was preferred. But Belke’s results may have been obtained because the pigeons had been trained to peck the VI 40 sec paired with the poorer schedule and they had been trained not to peck the VI 40 sec paired with the richer schedule. In the present experiment, we avoided this bias by training pigeons on two concurrent schedules in which the tested stimuli both had been associated with the poorer schedule of the pair [A(VI 20 sec) vs. B(VI 80 sec) and C(VI 40 sec) vs. D(VI 80 sec)]. Evidence for value transfer was demonstrated when on probe trials pigeons preferred B over D.     

Response modes in simultaneous and successive visual discriminations

Modes of responding in simultaneous and successive discrimination problems were investigated using a bidirectional transfer procedure. Rats receiving training on a simultaneous discrimination problem were transferred to successive discriminations. Rats trained on a successive problem were transferred to simultaneous problems. The transfer problems were either compatible or incompatible with the response mode present during training. The results suggest that a single response mode may underlie the solution of both simultaneous and successive discrimination problem.     

《辞源》释“怨刺”辨析

《汉书·礼乐志》所言《诗经》"怨利之诗",此"刺"字非个语"讽刺"之意,当依《毛传》《瞻印》释为"责"。验证《诗经》"怨刺"一类诗．其内容绝没有嘲讽朝政之意,而只是黄其想咎,欧其改过,重振朝纲;检索先秦两汉文献中所允之"刺"字,皆没有含"嘲讽"意味的义项;考查"讽刺"一词的由来及其意义演变,个语"讽刺"之意是"刺"字的后起义项,大约始于唐宋间。     

On the role of trial outcomes in delayed discriminations

Delayed simple discriminations are typically retained more accurately over longer delays by pigeons than are delayed conditional discriminations (e.g., Honig & Wasserman, 1981). In two experiments, we investigated the extent to which trial outcomes contribute to this difference by comparing performances when all trials ended with food reinforcement versus when only half of the trials did. Experiment 1 showed that when food was presented on all trials, contingent upon either pecking or not pecking the test stimulus, levels of retention and rates of forgetting were comparable for these two tasks. By contrast, Experiment 2 showed better retention of delayed simple than delayed conditional discriminations when half of the trials ended with food and the other half in extinction. Furthermore, delayed simple discriminations were retained more accurately with food versus no-food outcomes than with food at the end of every trial, whereas the reverse was true for delayed conditional discriminations. These findings indicate that retention differences between these tasks are another instance of the differential outcomes effect.     

The feature-positive effect and sign-tracking behavior during discrimination learning in the rat

Feature-positive (FP) and feature-negative (FN) successive discrimination learning was investigated in the rat. When a discrete, visual element serving to differentiate the discriminanda belonged to S+ (FP), rats acquired the discrimination more rapidly than when it formed a part of S? (FN). During the course of training, FP rats developed a tendency to direct their responding toward the differentiating feature in S+ while FN rats shifted their responding away from this element in S?. These findings were discussed in terms of the conceptions of stimulus-reinforcer relations and “sign-tracking” behavior.     

Intertrial interval effects in pavlovian serial feature positive discriminations

The effects of the intertrial interval (ITI) on learning and performance in Pavlovian appetitive serial feature positive (SFP) discriminations were examined in three experiments with rats. With longer ITIs, acquisition was more rapid, and there was less transfer of the feature’s behavioral control to a separately trained target cue, suggesting that longer ITIs encouraged the use of an occasion setting strategy. Behavior was also affected by discrimination-specific ITIs. Rats were trained with two SFP discriminations. The overall ITI was held constant, but the intervals between trials of one discrimination were varied by intermixing different numbers of trials from the other discrimination. Learning was more rapid when the intervals between trials of a single discrimination were longer. A sequential analyses showed that performance on a trial was impaired when it was preceded by a trial that included the same target cue but with the opposite trial outcome. The results are discussed in the frameworks of proactive interference effects and deletion-comparator processes (Cooper, Aronson, Balsam, & Gibbon, 1990.)