The effects of context changes on the reinstatement of extinguished conditioned behavior in a conditioned suppression task with humans

Reinstatement refers to the return of previously extinguished conditioned responses to test trials of a conditional stimulus (CS) when presentations of the unconditional stimulus (US) alone are given following extinction. Four experiments were conducted to determine whether reinstatement could be found in a conditioned suppression task with humans and whether contextual changes can abolish it. Experiment 1 demonstrated the reinstatement of conditioned suppression when acquisition, extinction, US alone, and test trials were all given in the same context. Experiments 2 and 3 suggested that the reinstatement effect was still present when the US alone presentations were given in a different context to the subsequent test trials. Experiment 4 replicated this effect using additional controls over the amount of exposure to the various contexts. The results suggest that reinstatement can be robust across changing contexts. Aspects of the conditioned suppression task that promote the transfer of learning across contexts or the establishment of configural context-CS stimuli may underlie the apparent lack of contextual control over reinstatement.     

Contextual control of the extinction of conditioned fear

Rats received 15 pairings of a CS and shock in one context, and then a series of CS-alone trials in a second context. Even though this extinction procedure produced a complete loss of conditioned suppression, when the animals were returned to the site of original conditioning, suppression was renewed to a level comparable to that of animals that had not undergone extinction. Controls indicated that the renewed suppression was not due solely to pseudoconditioning, suggesting that the CS-US association had survived extinction. Renewed suppression was also demonstrated in a third context that was never associated with shock. Loss of suppression did not necessarily depend upon inhibitory conditioning of the extinction context. The data suggest that extinction of conditioned fear is specific to the context in which it occurs. They also suggest the possibility that animals might discriminate episodes in which a CS is reinforced and nonreinforced independently of the excitatory or inhibitory status of cues, like contextual stimuli, that are coincidentally present during those episodes.     

Contextual control over conditioned responding in an extinction paradigm

Four experiments studied contextual control over rats' freezing to conditioned stimuli (CSs) that had been paired with shock and were then extinguished. In Experiment 1, rats were exposed to a CS A-shock and a CS B-shock pairing in Context C. CS A was then extinguished in Context A, and CS B in Context B. Freezing was renewed when each CS was presented in the context where the other CS had been extinguished. In Experiments 2-4, rats were exposed to a CS A-shock pairing in A and a CS B-shock pairing in B. They were then exposed to Context C where one, both, or neither of the CSs were extinguished, or where both CSs continued to be reinforced. On test, the rats froze more to CS A than to CS B in Context A, and more to CS B than to CS A in Context B, but only if the CSs had been extinguished. Thus, after extinction, rats use contexts to regulate retrieval not only of their memory for extinction, but also of their memory for the original conditioning episode.     

Contextual conditioning and reinstatement of extinguished instrumental responding

In three experiments we studied the relationship between contextual conditioning and the reinstatement of extinguished lever pressing that occurs when noncontingent food is introduced following extinction. In all three experiments the non-contingent food was presented off-baseline (with the response levers not present). On subsequent tests, with the response levers present, animals that had been exposed to food showed more reinstatement of lever pressing than control animals. This finding rules out alternative mechanisms for the reinstated responding that rely on the interaction of non-contingent food and responding, such as superstitious reinforcement or the discriminative after-effects of food. In addition, in each experiment we demonstrated that manipulations known to affect contextual conditioning (signalling the food in Experiment 1, context extinction in Experiment 2, and switching contexts in Experiment 3) reduced the reinstatement. These results are consistent with the claim that contextual conditioning is important in controlling instrumental conditioning and closely parallel findings concerning the reinstatement of Pavlovian responsing following extinction.     

Contextual control over conditioned responding in a latent inhibition paradigm

We used 1-, 2-, and 3-context designs to study the control exerted by contexts over freezing in rats exposed to a conditioned stimulus (CS) in advance of its pairing with a shock unconditioned stimulus. The latent inhibition observed when preexposure, conditioning, and testing occurred in the same context was attenuated if preexposure occurred in a different context to conditioning and testing. Latent inhibition (i.e., attenuated performance) was restored in a CS-specific manner if preexposure and testing occurred in the same context and conditioning in a different one. Latent inhibition was also reduced by a long retention interval but remained specific for a particular context-CS relation. Finally, CS preexposure resulted in contextual control over the expression of excitatory conditioned performance. The results are discussed in terms of memory, associative, and associative-performance models of CS-preexposure effects.     

Reinstatement of extinguished conditioned responses and negative stimulus valence as a pathway to return of fear in humans

The present study investigated reinstatement of conditioned responses in humans by using a differential Pavlovian conditioning procedure. Evidence for reinstatement was established in a direct (fear rating) and in an indirect measure (secondary reaction time task) of conditioning. Moreover, the amount of reinstatement in the secondary reaction time task was significantly correlated with the difference in valence between the conditioned stimulus (CS)+ and the CS-after extinction. These data provide clear evidence for reinstatement and for the role of negative stimulus valence in the return of conditioned responding after extinction.     

Reinstatement of fear to an extinguished conditioned stimulus.

Four experiments are reported which demonstrate the ability of an unconditioned stimulus (UCS) presentation following extinction to partially reinstate the conditioned response. These experiments are interpreted in terms of the strengthening of an extinction-reduced UCS representation. The first two experiments address alternative interpretations in terms of sensitization, reinstating the stimulus conditions of acquisition, conditioning of background cues, and stimulus generalization. Experiment 3 suggests that reinstatement is possible with a UCS qualitatively different from that used in conditioning. Experiment 4 explores an alternative extinction procedure which especially preserves the conditioned stimulus-unconditioned stimulus association while encouraging modification of the UCS representation. The results are discussed both in terms of related empirical phenomena, such as spontaneous recovery and sensory preconditioning, and in relation to the general role of the UCS representation in conditioning.     

Role of conditioned contextual stimuli in reinstatement of extinguished fear

If the unconditioned stimulus (US) is presented independently of the conditioned stimulus (CS) following extinction, the conditioned response may be reinstated to the CS. Three experiments are reported that suggest that reinstatement is mediated by conditioning to contextual stimuli that are present during both US presentation and testing. Shocks presented to rats following the extinction of conditioned suppression reliably reinstated suppression to the CS, but only when they were presented in the context in which testing was later to occur. Reinstatement was also reversed by extinguishing fear to the context through nonreinforced exposure to the context between shock presentation and testing. Reinstatement was obtained in these experiments in spite of procedures that have been used in the past to minimize the influence of context conditioning. Moreover, fear of the context was never detected directly by depressed bar-press rates in the absence of the CS. The results do not support the hypothesis that reinstatement results from an increment in the strength of a memory of the US that has been weakened during extinction. Problems inherent in controlling and detecting levels of context conditioning that may influence behavior toward nominal CSs are discussed.     

Effects of multiple contexts and context similarity on the renewal of extinguished conditioned behaviour in an ABA design with humans

The ABA renewal procedure involves pairing a conditional stimulus (CS) and an unconditional stimulus (US) in one context (A), presenting extinction trials of the CS alone in a second context (B), and nonreinforced test trials of the CS in the acquisition context (A). The renewal of extinguished conditioned behaviour is observed during test. The current study tested the effects of multiple extinction contexts and context similarity in attenuating renewal. Participants (N = 99) took part in a fear conditioning ABA renewal procedure. Using a measure of self-reported expectancy of the US, ABA renewal was observed when a single extinction context that was dissimilar to the test context was used. Renewal was attenuated, though still present, when extinction occurred in multiple dissimilar extinction contexts or in a single extinction context that was similar to the test context. Renewal was completely abolished when multiple extinction contexts that were similar to the test context were combined. Multiple extinction contexts and context similarity act additively in their effect on attenuating renewal. The results are discussed in relation to the design of exposure therapy programs that seek to reduce relapse that can occur via renewal.     

Reinstatement of fear to an extinguished conditioned stimulus: two roles for context

The authors studied the role of context in reinstatement. Freezing was reinstated when the conditioned stimulus (CS) was extinguished in 1 context and rats moved to another context for reexposure to the shock unconditioned stimulus (US) and test. It was also reinstated (rather than renewed) when rats were shocked in the extinction context and moved to another context for test. This reinstatement was CS specific and reduced by nonreinforced exposures to the extinction context. Rats shocked in the context in which a stimulus had been preexposed froze when tested in another context. These findings suggest 2 roles for context in reinstatement: conditioning of the test context (M. E. Bouton, 1993) and mediated conditioning by the extinction context (P. C. Holland, 1990).     

Intertrial sources of stimulus control and delayed matching-to-sample performance in humans

Two experiments compared delayed matching-to-sample (DMTS) accuracy under 2 procedures in adults with mental retardation. In the trial-unique procedure, every trial in a session contained different stimuli. Thus, comparison stimuli that were correct on one trial were never incorrect on other trials in that session (or vice versa). In the 2-sample DMTS procedure, the same 2 comparison stimuli were presented on each trial, and their function changed quasi-randomly across trials conditional upon the sample stimulus. Across 2 experiments, 7 of 8 subjects showed the highest overall accuracy under the trial-unique procedure, and no subject showed consistently higher accuracy under the 2-sample procedure. Negative, exponential decay functions fit to logit p values showed that this difference was due largely to the steeper delay-mediated decline in sample control for the 2-sample procedure. Stimulus-control analyses indicated that, under the 2-sample procedure, the selection of the comparison stimulus on Trial N was often controlled by the comparison stimulus selection on Trial N-1 rather than the Trial-N sample stimulus. This source of competing stimulus control is not present in trial-unique procedures. Experiment 2 manipulated intertrial interval duration. There was a small but consistent increase in accuracy as a function of intertrial interval duration under the 2-sample procedure, but not under the trial-unique procedure.     

Expectancy for airpuff and conditioned eyeblinks in humans

Three experiments were designed to test the hypothesis that expectancy for an unconditioned stimulus (US) serves as an intervening variable in the well-documented empirical relationship between the awareness of stimulus contingencies and the occurrence of conditioned reactions (CR) in human classical conditioning. Experiment 1 showed that, when subjects are instructed to respond by keypressing to US in a conditioning-like procedure, reaction (RT) provides the same kind of information as a direct rating of expectancy. In experiment 2, the RT task was superimposed on an otherwise standard eyeblink conditioning procedure. Reliable discriminative conditioning failed to occur, although changes in RT give evidence of increasing expectancy for US. Experiment 3 aimed to repeat experiment 2 with several modifications in procedure, intended to facilitate conditioning. The between-subjects correlation between RT to US and frequency of eyeblink CRs was reasonably high (-0.52). However, when consecutive trials were considered, a shift was apparent between the two variables: RT changes occurred earlier than CR development. The implications of these results with respect to the expectancy theory of conditioning are discussed.     

Contextual control of letter perception

This study examined the hypothesis that the type of list context in which a given item is embedded may affect the way in which it is perceptually processed. In the first experiment, subjects matched four types of test items inserted in four types of lists. The types of test items and the types of lists in which they were embedded were two-letter words, two-letter spelling clusters, unrelated letter pairs, and pairs of novel letters. The first experiment indicated that only unrelated letter pairs were significantly affected by type of list in which they were inserted. The pattern of the effect was confirmed in the latency data of the second experiment and the interpretation was that only unrelated letter and related letter-group contexts were operating in the first two experiments. The third and fourth experiments were concerned with contrasting contexts at the feature and letter levels of processing. By using lists in which type of catch trial either emphasized local features or global aspects of a pattern, the processing times of both familiar and unfamiliar letters were affected. The results of these experiments suggest that information processing models should incorporate structures which represent the way that context selectively facilitates processing of items at particular levels.     

Contextual control of biconditional task performance: Evidence for cue and response competition in rats

A novel paradigm is presented that was designed to mimic aspects of cue and response competition seen in humans in conflict procedures such as the Stroop task. Rats were trained simultaneously on two biconditional discrimination tasks, one auditory and one visual, in two different contexts: C1, in which A1:LP1 → R, A2:LP2 → R; and C2, in which V1:LP1 → R, V2:LP2 → R, where C1/C2 represent different training contexts (produced by different operant chambers), A1/A2 are different auditory cues, V1/V2 are different visual cues, LP1/LP2 are discrete operant responses, and R is reward. At test, rats received presentations of audiovisual compounds of these training stimuli in extinction. These compounds had dictated either the same (A1V1 or A2V2) or different (A1V2 or A2V1) responses during training: termed congruent and incongruent trials, respectively. Experiment 1 showed that following equal training on the two biconditional tasks, the contextual cues came to control responding to conflicting information provided by incongruent stimulus compounds such that animals responded according to the stimulus element previously trained in that test context. Experiment 2 demonstrated that differential training on the biconditional discriminations (with rats receiving training on the two discriminations in the ratio 3∶1) resulted in greater interference from the overtrained task when animals were tested in the undertrained context. This finding is similar to the classic Stroop asymmetry seen in human performance whereby dominant word reading interferes with colour naming for incongruent colour–word compounds. Further analysis also revealed some evidence for a reverse Stroop effect in which the undertrained stimulus element interfered with performance on the overtrained task.     

Contextual control in discrimination reversal learning

In 3 human predictive learning experiments, the authors examined contextual control of responding in discrimination reversal learning. In Phase 1, a discrimination between 2 stimuli (A+, B-) was trained in Context 1. During Phase 2, participants received discrimination reversal training (A-, B+) in Context 2. Testing occurred in Context 1 and Context 2 (Experiments 1A and 1B) or in Context 1 and Context 3 (Experiment 2). During the test phase, performance in Context 1 and Context 2 reflected the contingencies trained during Phase 1 and Phase 2, respectively. When testing occurred in Context 3, there was no discriminative responding between A and B. In addition, the experiments demonstrated that discriminating stimuli with a consistent reinforcement history were also affected by contextual manipulations. Results indicate that each training context acquires the ability to control performance. Unique-cue and configural approaches account for a major part of the results.     

Effects of endorphin blocking on conditioned SCR in humans

In order to test the hypothesis that low levels of endogenous opioids (endorphins) predispose to strong conditioning effects, female Ss (N = 36) were assigned to a placebo group, a low-dose naltrexone group, or a high-dose naltrexone group and then underwent a classical conditioning procedure. This procedure consisted of an acquisition phase in which all Ss received 5 pairings of a CS+ (neutral picture) and a UCS (100 dB white noise). The CS− (neutral picture) was never followed by a UCS. During extinction, Ss received 4 unreinforced presentations of CS+ and CS−. Throughout the experiment, skin conductance responses (SCRs) to the CSs and UCSs were recorded. Acquisition was successful in that CS+ slides elicited stronger SCRs than CS − slides. However, during acquisition, there was no interaction between drug and differential response (CS+ vs CS−). During extinction, there was no overall remaining effect of conditioning. Again, no evidence was found to suggest that (remaining) effects of conditioning were stronger in the naltrexone treated Ss than in the placebo S's. If anything, the opposite seemed to be true with especially high-dose naltrexone Ss showing relatively weak conditioning effects.     

Contextual control of long-term habituation in rats

This study examined contextual control of long-term habituation and whether such effects are dependent on the habituating response system. Habituation of the acoustic startle response transferred from the home cage to the testing context, whereas habituation of lick suppression was context specific (Experiments 1 and 2). Contextual control of habituation was demonstrated between 2 experimental contexts for lick suppression to a tone (Experiment 3) and bar-press suppression to a light (Experiment 4). Experiment 5 extinguished habituation of lick suppression and the orienting response to a tone with 27 exposures to the habituation context. Context specificity of both responses also was found. Previous failures to demonstrate contextual control of habituation may be due to the choice of response system and to less sensitive procedures to detect response recovery. The habituation mechanism for startle is independent from the process or processes that underlie habituation in other response systems, but the nature of these mechanisms is not yet known.     

Contextual control over task-set retrieval

Contextual cues signaling task likelihood or the likelihood of task repetition are known to modulate the size of switch costs. We follow up on the finding by Leboe, Wong, Crump, and Stobbe (2008) that location cues predictive of the proportion of switch or repeat trials modulate switch costs. Their design employed one cue per task, whereas our experiment employed two cues per task, which allowed separate assessment of modulations to the cue-repetition benefit, a measure of lower level cue-encoding processes, and to the task-alternation cost, a measure of higher level processes representing task-set information. We demonstrate that location information predictive of switch proportion modulates performance at the level of task-set representations. Furthermore, we demonstrate that contextual control occurs even when subjects are unaware of the associations between context and switch likelihood. We discuss the notion that contextual information provides rapid, unconscious control over the extent to which prior task-set representations are retrieved in the service of guiding online performance.     

Flavour liking and preference conditioned by caffeine in humans

When caffeine consumers repeatedly experience a novel flavoured drink containing caffeine, the rated pleasantness of the drink flavour increases progressively. These results could be interpreted in terms of the flavour acting as a Pavlovian conditioned stimulus (CS) predicting the consequences of caffeine ingestion. However, all studies of this phenomenon to date have used between-subjects designs, and one criticism of this is that changes in pleasantness might have arisen from nonspecific effects. A more rigorous test is to examine changes in pleasantness for two drinks, a CS+ flavour paired with caffeine and CS- paired with placebo. Accordingly, 20 moderate caffeine consumers consumed both CS+ and CS- drinks in counterbalanced order over eight conditioning trials at breakfast, with hedonic and sensory characteristics evaluated on each trial. As predicted, the rated pleasantness of the CS+ drink increased whereas pleasantness of the CS- drink did not change. Despite this, participants did not have an overall preference for the CS+ flavour posttraining. However, both those who chose the CS+ and those who chose the CS- at the end showed the same direction and rate of change in pleasantness for the two drinks during training, but spurious differences in baseline preference obscured this effect in terms of an overall change in preference. Overall these data suggest that changes in pleasantness of drinks paired with caffeine delivery are best explained in terms of Pavlovian associations between drink flavour and the postingestive effects of caffeine.