Doi:10.1016/j.neubiorev.2004.09.00

Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 Memory processes in classical conditioning Department of Psychology, University of Vermont, Burlington, VT 05405, USA Classical conditioning provides a rich and powerful method for studying basic learning, memory, and emotion processes in animals.
However, it is important to recognize that an animal’s performance in a conditioning experiment provides only an indirect indication of whatit has learned. Various remembering and forgetting processes, in addition to other psychological processes, may intervene and complicatewhat investigators can infer about learning from performance. This article reviews the role of context, interference, and retrieval in a numberof classical conditioning phenomena (e.g. extinction), and provides an overview of how long-term and short-term memory processesinfluence behavior as it is studied in classical conditioning.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Learning vs. performance; Context; Extinction; Short-term memory; Long-term memory 1. Extinction, context, and interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664 2. Long-term memory processes in animal conditioning and learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665 3. Short-term memory processes in conditioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668 4. Conditioned stimuli as retrieval cues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669 5. Summary and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671 Classical conditioning in animals provides a powerful Given proper control groups for alternative nonassociative tool for studying the biological processes underlying processes such as sensitization and pseudoconditioning, the learning, memory, and emotion. In conditioning, once evocation of the CR is a reasonably good index of learning.
a conditional stimulus (CS) is associated with an uncondi- But it is important to realize that what an animal does in a tional stimulus (US), a constellation of conditioned conditioning experiment is not the same as what it knows.
responses (CRs) comes to be elicited by the CS.
Researchers in behavioral aspects of learning and memoryhave long separated learning, the hypothetical psychologi-cal and physical changes in the brain, from performance, * Corresponding author. Tel.: C1 802 656 4164; fax: C1 802 656 8783.
the manifestation of that change in behavior. The present E-mail addresses: [email protected] (M.E. Bouton), mbouton@ article selectively reviews the kinds of memory processes 0149-7634/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neubiorev.2004.09.001 M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 that separate learning and performance, and thus compli- presenting CS without the US on a number of trials cate simple inferences about psychological processes eliminates (‘extinguishes’) that responding. The study of (learning, memory, emotion) from behavior.
extinction is interesting in its own right, because something The distinction between learning and performance was like it must be available to animals in order to allow them to an important part of the thinking of early learning theorists.
update and modify their behavior in a changing environ- For instance, Edward Tolman, one of the most important ment. In addition, it has been explicitly linked to various theorists of the 20th century, made a convincing case for it.
therapies designed to eliminate unwanted cognitions, For example, in the well-known latent learning experiment, emotions, and behaviors that themselves depend on learning Tolman and Honzik allowed rats to explore a 14-unit It is tempting to suppose that the loss of responding in T-maze on a series of trials. One group received reward each extinction merely reflects the destruction of the original time it reached the end of the maze, whereas another group learning that led to it. But we know this is not the case; the did not. Not surprisingly, the rewarded group moved original association is retained, perhaps fully intact. The through the maze more efficiently, making fewer errors extinction phenomenon is therefore another place where it is (entering fewer dead ends) over the first 11 trials. But when important to understand the distinction between learning the non-rewarded group was then rewarded, they began to (what the animal knows) and performance (what the animal move efficiently through the maze beginning on the next trial. According to Tolman and Honzik, the nonrewarded There are several ways to show that the original group had been learning about the maze the whole time, association is intact after extinction. First, there is even though that learning or knowledge had not been ‘reinstatement.’ In reinstatement, if the US is now presented evident in their behavior. The function of reward was not to a few times after extinction is complete, responding will stamp behavior in, but instead to motivate the animal to return to the CS when the CS is presented again One of perform. A motivational function of reward was widely the main reasons reinstatement occurs is that the animal accepted in subsequent theory . Learning is not the associates the US with the context (background stimuli same as performance. Motivation is required for the typically defined as emanating from the box in which the experiment is conducted) when the US is presented after Modern thinking has followed another of Tolman’s extinction. When the animal is subsequently tested with the ideas. He argued, at a time when it was not fashionable to do CS in the context, the contextual conditioning triggers so, that learning was not the simple attachment of a responding to the CS. One of the most important types of behavioral response to an environmental stimulus (so-called evidence supporting this view is that the US must be S-R learning). Instead, the animal represented its experience presented in the context in which the CS will be tested. If the in some sort of cognitive way. For instance, Tolman animal is presented with the US in an irrelevant context, it claimed that rats learned cognitive maps of the environment.
does not produce reinstatement Although the Although the idea that animals learn a literal map is contextual conditioning that causes reinstatement is not debatable, the cognitive view of what is learned in classical always evident in behavior directly elicited by the context, conditioning has become dominant This view assumes the strength of reinstatement correlates with the strength of that there is a distinction between what is learned and what contextual conditioning when it is measured with sensitive is manifest in behavior. For example, theorists now suppose context-preference tests . Reinstatement indicates that learning involves some sort of encoding of information, that extinction is not the same as unlearning. Responding to storage of that information in memory, and then the retrieval the extinguished CS can return depending on what the of it. Learning processes thus involve the encoding and storage of information. Performance, on the other hand, A second phenomenon indicating that extinction is not depends at least in part on successful retrieval. The current unlearning, and that the CS has become dependent on the article is mainly interested in considering how memory and context, is the ‘renewal effect’ In renewal, an memory retrieval processes operate in classical condition- animal might receive conditioning trials with the CS in one ing. An appreciation of these processes is essential for a context (Context A) and then extinction trials in another complete understanding of this deceptively simple learning (Context B). When the animal is then returned to the process. They are methodologically important because they original context and presented with the CS, responding to introduce a layer of complexity when inferring an animal’s the CS recovers (is ‘renewed’). Although most studies of learning or knowledge from behavior or performance.
renewal have used the ABA design (in which conditioning,extinction, and testing are conducted in Contexts A, B, andA, respectively), other work indicates that renewal can also occur if the contexts are ABC or even AAB Suchresults suggest that extinction is especially dependent on the Consider extinction, a learning phenomenon that has context. It would be a mistake to think, based on the been investigated in our laboratory for many years. Just as a animal’s lack of responding in the extinction context, that CS–US pairing comes to evoke responding, subsequently M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 One of the interesting things about research on the words. (The word ‘fire’ can mean different things, and evoke renewal effect is that it suggests that extinction depends different behaviors, depending on whether it is shouted in more on the context than conditioning does. Although it the movie theater or the shooting gallery.) Interestingly, seems relatively easy to find that extinction performance is occasion setters are not generally assumed to elicit lost after a context switch, conditioning performance is performance by themselves, as ordinary CSs might; rarely lost when the CS is switched to a familiar context, instead, they mainly operate by modulating performance which would minimize possible external inhibition effects created by novel contexts used in some experiments The occasion setting mechanism is similar to another . A change of context after conditioning does not possible role of context, namely, to enable retrieval of the attenuate conditioned fear to a CS as measured by CS’s association with the US. As mentioned above, when the animal is outside the extinction context, it might fail to appetitive conditioning or taste aversion condition- retrieve extinction, which can be taken as a CS–no US ing The fact that extinction is more context- association Consistent with this view, retrieval cues dependent than conditioning is consistent with the idea that that remind the animal of extinction can abolish the renewal the animal codes extinction as a kind of conditional exception to the rule—one that depends on the current this view, the context and memory retrieval processes are context When the context is changed, extinction necessary in the translation of knowledge into performance performance disappears and conditioning performance A growing literature on the brain processes involved in A related recovery-after-extinction phenomenon is extinction is broadly consistent with the behavioral research ‘spontaneous recovery’. In this phenomenon, if the just summarized (see for one review). For instance, experimenter merely allows time to pass after extinction, extinction appears to be linked to new brain plasticity the extinguished response can recover . Just as extinction is more sensitive to context than conditioning, voltage-gated calcium channels in the shorter term .
so it is more sensitive to the effects of the passage of time.
One implication is that facilitation of these synaptic In fact, we have argued that the passage of time affects processes should help facilitate extinction, which has been extinction precisely because it is a kind of context Just shown with administration of an NMDA partial agonist, as the renewal effect indicates that extinction is sensitive to D-cycloserine It is not known whether facilitated the physical context, spontaneous recovery suggests that it extinction results from deeper extinction learning that is less is sensitive to the context provided by time.
context-dependent or merely learning that is easier to Research thus clearly indicates that the current perform- retrieve in the right context. At the systems level, fear ance elicited by a CS can underestimate what the animal extinction may be linked in part to activity in the infralimbic actually ‘knows’ about the CS. The major factor that region of the medial prefrontal cortex or to influences performance after extinction—besides the latent possible GABAergic interneurons in the lateral amygdala CS–US association—appears to be the current context. How A role for hippocampus, an area long thought to be does the context operate? Conditioning theorists have had involved in context learning is also suggested by data much to say about this. First, several influential conditioning indicating that the ABC renewal effect is suppressed by models assume that the context enters into direct associ- inactivation of the hippocampus although the ABA ations with the US, just as a CS might . These renewal effect is not affected by lesions . The pattern context-US associations would be expected to summate may be consistent with the idea that negative occasion with the CS–US association to generate performance.
setting by Context B, which is presumably a major source of Unfortunately, an emphasis on this idea cannot explain the ABC effect, is especially dependent on the hippocampus details of the results just reviewed . For example, it is . Reinstatement, which depends on direct associations not consistent with the fact that an extinguished CS is between the context and the US, also depends on an intact especially sensitive to context-US associations and hippocampal system in fear conditioning , though measurable context-US associations do not seem to be not in appetitive conditioning The various brain processes are likely to contribute in different ways to the A second possibility is that the contexts have the acquisition of new learning in extinction and its suppression properties of ‘occasion setters’ (see for reviews). On this view, the context is not merely associatedwith the CS (or the absence of the CS), but instead selects oractivates the CS’s own current association with the US 2. Long-term memory processes in animal conditioning Thus, the extinction context activates something like the animal’s CS–no US association It is as if the contextdetermines the current meaning of the ambiguous CS, much It is interesting to note that the extensive literature on as verbal contexts disambiguate the meaning of ambiguous long-term memory in animals also reinforces the view that it M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 is important to distinguish between learning and perform- It is worth observing that the weakness of the context ance. The word ‘forgetting’ describes a behavioral switch effect in humans has a parallel in the animal research.
phenomenon in which performance assessed at Time 2 is As we noted above, a change of context after simple shown to be inferior to that shown at Time 1, typically the excitatory conditioning usually produces little effect on time when the task is originally learned. Although it is responding to the CS . This sort of result can perhaps tempting to assume that the behavioral phenomenon is be seen as a case of outshining: The animal is responding to explained by the decay or erosion of the original memory a highly salient cue for the US (the CS) that overcomes the trace, we know that this often is not true. For example, it is absence of contextual support in the changed context.
often possible to reverse forgetting (and apparently recover Conditioning theories essentially argue that the CS is a more a forgotten memory) by exposing the animal to a retrieval informative cue than the context Other experiments in cue This sort of result has been taken to mean that which context switches have been said to impair memory forgetting often results from an inability to access or retrieve retrieval often have not included an overt CS; for example, the target trace (see for some of the relevant investigators may test passive avoidance , in which the animal refrains from entering a compartment where it Research on remembering and forgetting is often had previously been shocked, and thus responds directly to consistent with Tulving’s Encoding Specificity Principle, contextual (apparatus) cues. With few cues that can which emphasizes the importance of the similarity between potentially outshine them, it may not be surprising to see background contextual cues present during learning and an effect of switching the apparatus. Interestingly, in many testing . The general idea, which has been implicitly cases an extant context switch effect can still be attenuated if accepted in the extinction research presented above, is that the animal is given an extra retrieval cue —a retrieval depends on a match between the conditions present potential parallel to mental reinstatement in humans. The during learning and the conditions present during testing. In parallel between the animal and human literatures is even the original research with human participants, the specificity stronger than this. The fact that the context is especially was defined by subtle semantic shadings of target, to-be- important after extinction is consistent with other parts of remembered words. Tulving and Thomson showed that the human memory literature, which indicate that context the memory for words on a list was affected by weakly- switch effects are similarly easier to detect in interference associated words that were present at input, which designs in which the participant learns a conflicting word apparently influenced the semantic meaning encoded for list in a second phase . Apparently analogous to the target. Their experiments dramatically showed that these extinction, the second word list produces context-specific weakly-associated input words were better able to retrieve interference with memory for the first.
the target words than were more strongly-associated words One important implication of this work, of course, is that that were not present at input. Even though the presence of forgetting is not necessarily due to the erasure of the learned the strongly-associated words at the test almost certainly information. Instead, forgetting may be the result of a generated the target word through free association, the retrieval failure caused by a change of context .
target words were not recognized. Although the initial work Forgetting might also be caused by interference emanating manipulated input cues that could profoundly affect the from conflicting information learned at an earlier or later meaning of target words, subsequent work has shown that point. Interference has a long and distinguished history memory depends on a variety of contextual stimuli, in human learning theory When first-learned information interferes with memory for second-learned Consistent with this idea, memory for verbal material in information, we have ‘proactive interference.’ When humans is often inferior when the room is switched second-learned information interferes with first-learned between learning and testing However, it is often information, we have ‘retroactive interference’. In the disappointingly difficult to detect this effect In a domain of animal learning, retroactive interference is recent meta-analysis, Smith and Vela found overall represented by several possible paradigms, including evidence of such context-dependent memory, but extinction or counterconditioning (where Phase 2 interferes suggested that it is sometimes difficult to find owing to a with Phase 1). Proactive interference is perhaps represented number of other processes that can help support recall at by phenomena like latent inhibition, in which initial the time of the test. For example, ‘outshining’ refers to the exposure to the CS without the US can interfere with idea that if other cues besides the context can support conditioning when the CS and US are subsequently paired.
memory retrieval, and if they are present in both the Interestingly, it has been common in the animal learning training and testing context, then they will diminish any tradition to assume that proactive and retroactive inter- context switch effect. Contexts that are not particularly ference effects occur because of a failure at the level of salient also would not support a context switch effect. And learning. As we have already seen, extinction is sometimes humans can also defeat a context switch effect by thinking thought to result from a destruction of the original learning.
about (mentally reinstating) the original context when they Similarly, latent inhibition is often attributed to a failure to learn during the second (conditioning) phase owing, for M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 example, to the habituation of attention to the CS , and deprivation state Mood effects have been Nonetheless, extinction and latent inhibition may both reported in humans, but, perhaps as we have seen with follow from performance processes Consistent with exteroceptive context effects they can be influenced by this idea, these and other proactive and retroactive a number of other factors that make their effects small and interference phenomena are dependent on context (e.g.
difficult to demonstrate . In animals, there is evidence renewal effects have been obtained in both domains) and that stress hormones that could be correlates of emotion can time (e.g. spontaneous recovery effects have been reported play the role of context in a renewal design. Specifically, in both domains) (see for a review). For example, in Ahlers and Richardson showed that administration of latent inhibition, if Phase 2 conditioning is assessed after a ACTH after passive avoidance extinction renewed avoid- delay in the latent inhibition paradigm, the conditioned ance behavior. The idea that ACTH provided a context is response becomes stronger, as if conditioning had been consistent with the fact that dexamethasone (which learned but performance was obstructed by a memory of suppresses ACTH production) delivered during condition- latent inhibition, which was forgotten over the delay ing nullified this effect; ACTH needs to be part of the . In addition, if preexposure and conditioning are conditioning context to recover behavior after extinction.
conducted in different contexts, a return to the preexposure We have argued that the passage of time itself provides a context after conditioning is complete can renew latent gradually changing context discussed above, inhibition performance . Once again, it would be a spontaneous recovery can be understood this way .
mistake to assume that the performance evident at a Certain forgetting phenomena have been suggested as particular point in time is a simple product of the CS’s posing a challenge for this type of view , although the challenge has been addressed and arguably resolved .
Other research supports the idea that memory also Ultimately, it is almost universally held that ‘context’ can be depends on internal contexts provided by drugs . In provided by many kinds of cues and all should ‘state-dependent retention’, memory is inferior when there is a mismatch between interoceptive drug state at the time of Our understanding of what kinds of events reactivate learning and testing. For example, when fear extinction is memories is not complete. As implied above, one of the conducted while the rat is under the influence of a main ideas is that presenting aspects of the original training benzodiazepine (chlordiazepoxide or diazepam), fear is situation is best for reactivating a target memory. Often, an renewed when the animal is subsequently tested in the sober effective cue can be thought of as part of the original state Fear extinction can thus be dependent on the drug context. Thus, the extinction cue in the experiments context. State-dependent retention has important methodo- mentioned earlier worked to retrieve extinction logical implications for studies of the biological basis of because it was coded as part of the extinction context; learning and memory, where it is common to give a drug presentation of even a part of a context can presumably during or soon after training and then test memory in the trigger completion of the entire pattern of the context that it absence of the drug. Although the drug may often be of has been associated with As we will discuss in a later interest because it potentially disrupts a biological con- section, a CS is often thought to evoke behavior because it solidation process, if memory is being tested without a drug retrieves or activates a representation of the US. This that was otherwise present during learning and storage, then implies that the laws of associative learning as developed in retrieval failure is also a candidate explanation for poor research on classical conditioning may be relevant to performance during testing. For example, chemicals that are understanding how retrieval cues are learned and estab- supposed to disable learning and consolidation processes, lished However, the effectiveness of different cues at such as the protein synthesis inhibitor anisomycin and the time of testing might depend further on poorly- several NMDA antagonists (e.g. MK-801, ketamine, understood details about the timing of their presentation phencyclidine, and CGS 19155) can also generate state- with respect to the memory test , the duration of their dependent retention effects under some conditions (aniso- presentation , and what other cues are presented along with them In addition, the effectiveness of means that performance tests in the absence of the chemical different cues may wax and wane as the retention interval might overestimate its impact on learning as opposed to increases The area would benefit from more retrieval. Such a possibility highlights the importance of systematic, theory-driven research on the specific factors using control conditions in which learning is tested in the presence of the same drugged state; if a drug affects learning Another complication is that activation of a forgotten or consolidation when given at the time of conditioning, it memory is not a neutral event, but itself enables further should later affect performance in either the drugged or the learning. For instance, when forgotten memories are reactivated in a new context, the new context becomes Other research suggests that context effects can be effective at retrieving the memory, as if it has been added created by a variety of contextual cues, including hormones to the original training memory . (The word (particularly in non-physiological doses) , time of day ‘reactivation’ is sometimes meant to imply susceptibility M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 to further modification.) In addition, events that disrupt themselves, until all the baits have been removed .
memory when they are administered soon after learning can These literatures provide insights into the nature of short- also disrupt memory when they are administered soon after term memory processes in animals. For example, in both reactivation. For example, Misanin et al. found that delayed match-to-sample and the radial maze, animals seem exposure to electroconvulsive shock soon after a reactiva- able to code short-term information either retrospectively tion treatment made the memory difficult to retrieve. Similar (they remember the previous sample or previous arms that evidence has been produced more recently by Nader et al.
have been visited) or prospectively (they remember the who found that introduction of anisomycin likewise correct upcoming comparison stimulus or arms that are abolished a reactivated memory (see also for a related yet to be visited) The implications for studies effect of MK-801). The findings suggest that reactivation of classical conditioning have mainly been discussed in can make a memory susceptible to disruption again, and that studies of occasion setting, where a feature CS informs the similar neurobiological processes are engaged after initial subject of whether or not a subsequent target CS will be learning and reactivation, although they say little about the mechanisms of retrieval itself One danger is to assume There is also evidence that short-term memory processes that the amnestic effect of any agent (e.g. ECS, anisomycin, are important even in simple cases of classical conditioning.
or MK-801) is necessarily to abolish reconsolidation. For Consistent with thinking about human memory processing, example, although the delivery of ECS soon after learning current theory assumes that conditioning requires some was originally thought to abolish consolidation processing of the CS and US together in short-term memory subsequent work suggested that it merely made the memory after individual conditioning trials for the association to be more difficult to retrieve Thus, the trace had been stored in long-term memory. In a classic experiment, encoded after all. We have already seen that anisomycin and Wagner et al. showed that learning of a particular MK-801 might affect retrievability of a memory rather than CS–US relation was damaged in rabbits if a surprising merely disrupting consolidation . The point is that if episode was presented within 300 s after each learning trial.
reactivation treatments initiate new learning, an amnestic The role of short-term memory in conditioning has agent may either interfere with reconsolidation or make figured importantly in the influential theories of Wagner the reconsolidated memory more difficult to retrieve.
The original formulation of his model, which The distinction between learning and performance is paralleled models of human information processing explicitly proposed that storage of the CS–US It is also interesting to note that reactivation treatments association in long-term memory depended on the CS and that involve exposure to a CS operationally resemble US being processed (‘rehearsed’) in short-term memory extinction trials. Typically, however, CS exposures that after each conditioning trial. Importantly, surprising CSs reactivate and engage reconsolidation do not create extinc- and USs—specifically, those that are not already ‘primed,’ tion (the CS improves performance, rather than weakens it, or currently represented in short-term memory, when they compared to control groups that receive no CS). It is possible are presented in a trial—were assumed to command more that extinction learning, which involves the learning of new rehearsal than those that are not surprising. Priming can conflicting information, simply depends on more exposure to occur either through recent presentation of the event itself or the CS. The relationship between these processes is through recent presentation of a retrieval cue associated interesting and important, but poorly understood at present.
with the event (which would retrieve the item from long-term memory and put it into short-term memory).
Furthermore, consistent with an important characteristic of 3. Short-term memory processes in conditioning working memory in humans, the capacity of short-termmemory was assumed to be limited. These ideas and Short-term memory processes have also been important assumptions allowed the model to explain am impressive in studies of animal learning. For example, there is a amount of new data (see for review).
substantial amount of interest in variations of the delayed In a more recent version of the model, Wagner matching to sample procedure, in which animals are given a put the ideas in a connectionist framework. In this sample stimulus at one point in time and then given that model, known as the ‘sometimes opponent process’ model, stimulus and another a few seconds later, when they are or ‘SOP,’ CSs and USs are assumed to have corresponding required to respond to the stimulus that matches the sample ‘nodes’ in memory. The presentation of a CS or US is in order to acquire reinforcement With typical assumed to activate the node. Initially, the node is activated methods, the pigeon’s choice returns to chance when the to a highly active state (‘A1’), but this quickly decays to a interval between the presentation of the sample and then the less active state (‘A2’), where it stays a bit longer before choice stimuli is in the order of many seconds Short returning to the inactive (I) state. (An activated node is term memory for recent choices similarly influences essentially one that is represented in short-term memory.) behavior on the radial maze, where rats choose among The animal learns about the CS only when it has been put a number of baited arms, almost without repeating into its maximally active state (A1). If the US node is M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 simultaneously activated to the same A1 state, then the presented on trials that are more spaced in time, there may animal will learn a connection or association between the be a significant increase in conditioned responding on the two. As a consequence of conditioning, the CS is able to spaced trials On the other hand, the increase in activate the US node into the A2 (not the A1) state; this responding does not reach the level in animals that have activation ultimately produces a conditioned response.
received spaced trials throughout training . This Notice that if time were to elapse between presentation result might suggest that massed trials also have an enduring of the CS and US, the CS could decay from A1 to the less- effect on learning, although it is also possible that the active A2 state and therefore be unavailable for learning.
animals trained with massed trials and tested with spaced The theory actually supposes that nodes are composed of ones might not generalize perfectly between trials presented many elements that move from A1 to A2 and then on the two schedules. The evidence suggests that recent Inactivity in a probabilistic fashion. Given these dynamics, presentations of the conditioning events can generate a the theory explains the results of ‘trace conditioning’ short-term suppressive effect on performance and might experiments, in which one observes increasingly poor conditioning as the temporal gap between CS offset and US Of course, priming in short-term memory can only onset increases 1 The US node similarly decays from account for the effects of trial massing within the range of A1 to A2 to Inactivity after each US presentation, although short intertrial intervals. In our appetitive conditioning here the consequences are somewhat different. If the US is experiments in which rats are given pairings of 10- or 30-s in A2 at the time the CS is in A1, an inhibitory CS–US CSs with food pellet USs, priming effects are clearly evident association is learned. This is one account of inhibitory when the trials are separated by 60 s, but not when they are conditioning that can occur in ‘backward conditioning,’ separated by 240 s It is worth noting that when the CS is presented relatively soon after the offset of additional spacing of trials beyond 240 s can have additional positive effects on conditioning This fact implies that The dynamics of short-term memory in SOP can go some additional mechanisms do contribute to trial-spacing effects distance in explaining a number of other interesting effects (see for review). Our own research suggests that long in conditioning. For example, recent work in our own intertrial intervals may create better conditioning because laboratory has investigated the well-known trial-spacing they allow extinction of contextual cues that receive effect, in which trials that are spaced relatively widely in conditioning during CS–US pairings These otherwise time produce better conditioning than trials that are massed compete with (or block conditioning of the CS.
in time Although a number of explanations of Quantitative conditioning models suppose exactly this this effect have been proposed recent research process . SOP gives the process psychological suggests that massing trials closely together in time may flesh by arguing that conditioned contextual cues wouldblock CS conditioning because they activate the US node make conditioning inferior because the presentation of CS to A2, and thus reduce the surprisingness of the US when it and US on one trial primes their representations into short- term memory (the A2 state) and makes them less surprising intertrial intervals weakens the context’s ability to put the on the next trial The results were not consistent To summarize the trial-distribution findings, very short Interestingly, although it is once again tempting to intertrial intervals can hurt conditioning because they prime assume that priming reduces learning, massed training may the CS and US in short-term memory, making them less have its clearest effect on performance. For example, likely to command performance and/or learning on the next presentation of the CS a few seconds before it is presented trial. At longer intertrial intervals, the context becomes less again can reduce responding on the second CS presentation likely to retrieve the representation of the US (and perhaps after conditioning has already occurred . When the CS) and therefore allows better conditioning. Several conditioning occurs in a within-subject procedure that psychological mechanisms thus play a role in trial-spacing intermixes short and long intertrial intervals, there may be effects. But a productive way to conceptualize them is to less responding in the short intertrial intervals a emphasize the influence priming effects in short-term learning deficit caused by the short intervals should be manifest in behavior at all intervals. And when a CS that hasreceived conditioning in a massed-trial procedure is 1 SOP also allows other mechanisms to contribute to the trace conditioning deficit. If the US occurs alone at the end of a very long gap Our discussion to this point has accepted the idea that (‘trace interval’), it could be associated with contextual cues that are present once conditioning has occurred, the CS functions as a at the time Contextual conditioning would reduce any possible retrieval cue that activates a US node or representation. The conditioning of the CS, because it would allow the context to activate idea has received direct empirical support. For example, (prime) the US node to the A2 state, reducing its surprisingness andtherefore causing blocking Rescorla conditioned fear in rats by pairing a light CS M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 with a loud noise (created by a klaxon). In a subsequent Also importantly, how the animal currently values the phase, he presented the klaxon repeatedly by itself to representation of the outcome in instrumental learning habituate fear to it. During a third phase, he presented the CS depends upon a subtle process known as ‘incentive alone. At this point, the CS elicited less fear than it did in a control group that did not receive habituation to the noise.
trained rats to lever-press for food pellets while they were Rescorla argued that habituation had modified the rat’s satiated and then tested lever pressing in extinction. At memory representation of the noise; when the CS was then this time, the rats were either hungry or not. Hunger had presented again, it activated a less frightening US represen- no effect on the level of responding. But if the rats had tation. Other experiments revealed a corresponding inflation been given a separate opportunity to eat the pellets while effect rats that received fear conditioning with a they were hungry, hunger strongly increased responding in weak-intensity shock US showed augmented fear of the CS the extinction test. The animal needed to learn about the if they were then exposed to more intense shocks. The effect of food on hunger. Hunger thus motivates behavior argument is thus that the response evoked by a CS depends in a subtle way; the organism will perform an action while in part on the status of the animal’s US representation. A CS hungry if the behavior has been associated with a may generate behavior at least in part because it serves as a particular reinforcer, and if the reinforcer has been retrieval cue for that representation.
associated with something like amelioration of the hunger These findings have been extended in a number of ways.
For instance, Holland has presented striking evidence The evidence thus suggests that animals form relatively that the rat forms a rich representation of the US, and that rich representations as a result of conditioning. One effect of responding to the CS depends on the status of that a CS is to retrieve a representation of the US. Consistent representation. In the basic experiment, rats receive pairings with this, conditioned responding (or performance) is a of different tone CSs with differently-flavored sucrose product of the animal’s knowledge of a CS–US or response- solutions. When one of these is then separately paired reinforcer association and how the animal currently values with illness (which creates a conditioned aversion to the the US. The assignment of value, especially in the operant flavor), the rat exhibits less appetitive responding to the situation, involves motivational as well as memory paired tone during final tests. Remarkably, as if the tone processes, which brings us full circle to the ideas of Tolman evoked an almost palpable image of the flavor it was that we mentioned at the start of this article.
associated with, rats reacted to novel combinations of tonesin a way that paralleled what they had separately learnedabout the corresponding combinations of flavors. In addition, the effect of taste-aversion revaluation of the USappears to be less noticeable when conditioning involves a This short review has only scratched the surface of the more extended number of CS–US pairings, as if conditioned complexity of classical conditioning. In addition to responding becomes more ‘automatic,’ rather than rep- describing some of the many roles for memory in resentation-mediated, with extended exposure to the conditioning, we have repeatedly noted that any study of conditioning must always distinguish learning from per- Related findings have been reported in operant con- formance. We have seen that extinction does not depend on ditioning, where animals learn to associate a behavior (such unlearning; erased performance at the end of extinction does as lever pressing) with a reinforcer instead of a CS with a not reflect erased knowledge. Instead, performance after US. In this case, there is a similar literature indicating that extinction is a product of the animal’s knowledge of two the animal associates one event (the response) with a conflicting associations, and how the context selects representation of the outcome. In a simple experiment, if a between them. Similarly, forgetting in studies of long- rat is trained to lever-press for a food pellet, separate term memory does not necessarily result from erasure or pairings of the pellet and illness will condition an aversion decay, but instead often results from reduced access, either to the pellet—and will also cause the rat to suppress its due to retrieval failure or interference. Once again, the performance of the operant when it is tested in extinction context plays a role. Short-term memory effects are also As in classical conditioning, the current amount evident in classical conditioning; in this case, recent of responding depends on the strength of a hypothetical presentations of events can temporarily suppress perform- association and the extent to which the animal currently ance and learning. Finally, the idea that a CS functions as a values the outcome. Interestingly, extended training in retrieval cue provides another link between conditioning which the rat has many response-outcome pairings can and memory, as well as a further separation between what decrease the sensitivity of the response to the devaluation Although we have focused primarily on memory behavior has become ‘automatized,’ i.e. more habit-like and processes, a number of other factors also matter in the less dependent on the current status of the representation of translation of learning into performance. We have barely mentioned motivational factors (see for one recent M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 review). In addition, the qualitative nature of the CS can [12] Bouton ME, Bolles RC. Role of conditioned contextual stimuli in influence the type of conditioned response one observes reinstatement of extinguished fear. J Exp Psychol Anim BehavProcess 1979;5:368–78.
So can the duration of the CS: CSs that end in the [13] Bouton ME, King DA. Contextual control of the extinction of US after a relatively brief interval can evoke conditioned conditioned fear: tests for the associative value of the context. J Exp responses that are qualitatively different from those evoked Psychol Anim Behav Process 1983;9:248–65.
by CSs that end in the US after a longer interval [14] Frohardt RJ, Guarraci FA, Bouton ME. The effects of neurotoxic In any of these cases, it would be a mistake to make hippocampal lesions on two effects of context after fear extinction.
Behav Neurosci 2000;114:227–40.
inferences about learning from measurements of a single [15] Wilson A, Brooks DC, Bouton ME. The role of the rat hippocampal response, a point that has been made in other animal system in several effects of context in extinction. Behav Neurosci learning paradigms as well The behavioral product of classical conditioning is not a single, monolithic [16] Bouton ME, Bolles RC. Contextual control of the extinction of conditioned response. Conditioning instead engages whole conditioned fear. Learn Motiv 1979;10:445–66.
[17] Bouton ME, Brooks DC. Time and context effects on performance in behavior systems, or sets or constellations of behaviors that a Pavlovian discrimination reversal. J Exp Psychol Anim Behav are evolutionarily adapted to help optimize the organism’s interaction with the US Although the topic of [18] Bouton ME, Ricker ST. Renewal of extinguished responding in a behavior systems is beyond the scope of the present article, second context. Anim Learn Behav 1994;22:317–24.
it further reminds us that classical conditioning involves a [19] Penick K, Solomon PR. Hippocampus, context and conditioning.
sophisticated and surprisingly varied set of behavioral [20] Devenport L. Sampling behavior and contextual change. Learn processes that will always come into play whenever conditioning is used as a tool to study the brain.
[21] Gisquet-Verrier P, Alexinsky T. Does contextual change determine long-term forgetting? Anim Learn Behav 1986;14:349–58.
[22] Hall G, Honey RC. Contextual effects in conditioning, latent inhibition, and habituation: associative and retrieval functions of contextual cues. J Exp Psychol Anim Behav Process 1989;15:232–41.
[23] Kaye H, Mackintosh NJ. A change of context can enhance Preparation of this manuscript was supported by Grant performance of an aversive but not of an appetitive conditioned RO1 MH64847 from the National Institute of Mental Health response. Q J Exp Psychol B 1990;42:113–34.
to MEB. We thank Jaylyn Waddell and Ceyhun Sunsay for [24] Harris JA, Jones ML, Bailey GK, Westbrook RF. Contextual control their discussion and comments. Send correspondence to over conditioned responding in an extinction paradigm. J Exp Mark E. Bouton, Department of Psychology, University of Psychol Anim Behav Process 2000;26:174–85.
[25] Bouton ME, Peck CA. Context effects on conditioning, extinction, and reinstatement in an appetitive conditioning preparation. AnimLearn Behav 1989;17:188–98.
[26] Rosas JM, Bouton ME. Renewal of a conditioned taste aversion upon return to the conditioning context after extinction in another one.
[27] Rosas JM, Bouton ME. Context change and retention interval can [1] Tolman EC, Honzik CH. Introduction and removal of reward, and have additive, rather than interactive, effects after taste aversion maze performance in rats. U Cal Pub Psych 1930;4:257–75.
extinction. Psychon Bull Rev 1998;5:79–83.
[2] Hull CL. A behavior system; an introduction to behavior theory [28] Bouton ME. Conditioning, remembering, and forgetting. J Exp concerning the individual organism. New Haven: Yale University Psychol Anim Behav Process 1994;20:219–31.
[29] Brooks DC, Bouton ME. A retrieval cue for extinction attenuates [3] Spence KW. Behavior theory and conditioning. New Haven: Yale spontaneous recovery. J Exp Psychol Anim Behav Process 1993;19: [4] Mowrer OH. Learning theory and behavior. New York: Wiley; 1960.
[30] Pavlov IP. Conditioned reflexes. Oxford, UK: Oxford University [5] Rescorla RA, Solomon RL. Two-process learning theory: relation- ships between Pavlovian conditioning and instrumental learning.
[31] Robbins SJ. Mechanisms underlying spontaneous recovery in autoshaping. J Exp Psychol Anim Behav Process 1990;16:235–49.
[6] Tolman EC. Cognitive maps in rats and men. Psychol Rev 1948;55: [32] Bouton ME. Context, time, and memory retrieval in the interference paradigms of Pavlovian learning. Psychol Bull 1993;114:80–99.
[7] Hulse SH, Fowler H, Honig WK. Cognitive processes in animal [33] Mackintosh NJ. A theory of attention: variations in the associability behavior. Hillsdale, NJ: Erlbaum; 1978.
of stimuli with reinforcement. Psychol Rev 1975;82:276–98.
[8] Rescorla RA. Pavlovian conditioning: it’s not what you think it is.
[34] Pearce JM, Hall G. A model for Pavlovian conditioning: variations in the effectiveness of conditioned but not unconditioned stimuli.
[9] Barlow DH. Anxiety and its disorders: the nature and treatment of anxiety and panic, 2nd ed. New York: Guilford Press; 2002.
[35] Rescorla RA, Wagner AR. A theory of Pavlovian conditioning: [10] Rescorla RA, Heth CD. Reinstatement of fear to an extinguished variations in the effectiveness of reinforcement and nonreinforce- conditioned stimulus. J Exp Psychol Anim Behav Process 1975;1: ment. In: Black AH, Prokasy WK, editors. Classical conditioning II: current research and theory. New York: Appleton–Century–Crofts; [11] Bouton ME. Differential control by context in the inflation and reinstatement paradigms. J Exp Psychol Anim Behav Process 1984; [36] Wagner AR. A model of automatic memory processing in animal behavior. In: Spear NE, Miller RR, editors. Information processing M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 in animals: memory mechanisms. Hillsdale, NJ: Erlbaum; 1981. p.
[57] Harris JA, Westbrook RF. Evidence that GABA transmission mediates context-specific extinction of learned fear. Psychopharma- [37] Wagner AR, Brandon SE. Evolution of a structured connectionist model of Pavlovian conditioning (AESOP). In: Klein SB, [58] Holland PC, Bouton ME. Hippocampus and context in classical Mowrer RR, editors. Contemporary learning theories: Pavlovian conditioning. Curr Opin Neurobiol 1999;9:195–202.
conditioning and the status of traditional learning theory. Hillsdale, [59] Corcoran KA, Maren S. Hippocampal inactivation disrupts con- textual retrieval of fear memory after extinction. J Neurosci 2001;21: [38] Bouton ME. Context and retrieval in extinction and in other examples of interference in simple associative learning. In: [60] Frohardt RJ, Guarraci FA, Bouton ME. The effects of neurotoxic Dachowski L, Flaherty CF, editors. Current topics in animal hippocampal lesions on two effects of context after fear extinction.
learning: brain, emotion, and cognition. Hillsdale, NJ: Erlbaum; [61] Wilson A, Brooks DC, Bouton ME. The role of the rat hippocampal [39] Pearce JM, Bouton ME. Theories of associative learning in animals.
system in several effects of context in extinction. Behav Neurosci [40] Bouton ME, King DA. Effect of context on performance to [62] Holland PC, Lamoureux JA, Han JS, Gallagher M. Hippocampal conditioned stimuli with mixed histories of reinforcement and lesions interfere with Pavlovian negative occasion setting. Hippo- nonreinforcement. J Exp Psychol Anim Behav Process 1986;12: [63] Fox GD, Holland PC. Neurotoxic hippocampal lesions fail to impair [41] Bouton ME, Swartzentruber D. Analysis of the associative and reinstatement of an appetitively conditioned response. Behav occasion-setting properties of contexts participating in a Pavlovian discrimination. J Exp Psychol Anim Behav Process 1986;12:333–50.
[64] Spear NE. The processing of memories: forgetting and retention.
[42] Holland PC. Occasion setting in Pavlovian conditioning. In: Bower G, editor. The psychology of learning and motivation, vol.
[65] Sara SJ. Retrieval and reconsolidation: toward a neurobiology of 28. Orlando, FL: Academic Press; 1992. p. 69–125.
remembering. Learn Mem 2000;7:73–84.
[43] Swartzentruber D. Modulatory mechanisms in Pavlovian condition- [66] Tulving E, Thomson DM. Encoding specificity and retrieval ing. Anim Learn Behav 1995;23:123–43.
processes in episodic memory. Psychol Rev 1973;80:352–73.
[44] Bouton ME, Nelson JB. Mechanisms of feature-positive and feature- [67] Godden DR, Baddeley AD. Context-dependent memory in two negative discrimination learning in an appetitive conditioning natural environments: on land and underwater. Br J Psychol 1975;66: paradigm. In: Schmajuk NA, Holland PC, editors. Occasion setting: associative learning and cognition in animals. Washington, DC: [68] Smith SM. Remembering in and out of context. J Exp Psychol Hum American Psychological Association; 1998. p. 69–112.
[45] Brooks DC, Bouton ME. A retrieval cue for extinction attenuates response recovery (renewal) caused by a return to the conditioning Davies GM, Thomson DM, editors. Memory in context: context in context. J Exp Psychol Anim Behav Process 1994;20:366–79.
memory. New York: Wiley; 1988. p. 13–34.
[46] Brooks DC. Recent and remote extinction cues reduce spontaneous [70] Smith SM, Vela E. Environmental context-dependent memory: a recovery. Q J Exp Psychol 2000;53B:25–58.
review and meta-analysis. Psychon Bull Rev 2001;8:203–20.
[47] Brooks DC, Palmatier MI, Garcia EO, Johnson JL. An extinction cue [71] Land CL, Riccio DC. Nonmonotonic changes in the context shift reduces spontaneous recovery of a conditioned taste aversion. Anim effect over time. Learn Motiv 1998;29:280–7.
[72] Zhou Y, Riccio DC. Manipulation of components of context: the [48] Myers KM, Davis M. Behavioral and neural analysis of extinction.
context shift effect and forgetting of stimulus attributes. Learn Motiv [49] Falls WA, Miserendo MJD, Davis M. Extinction of fear-potentiated [73] Gordon WC, McCracken KM, Dess-Beech N, Mowrer RR. Mech- startle: blockade by infusion of an NMDA antagonist into the anisms for the cueing phenomenon: the addition of the cuing context amygdala. J Neurosci 1992;12:853–63.
[50] Santini E, Muller RU, Quirk GJ. Consolidation of extinction learning to the training memory. Learn Motiv 1981;12:196–211.
involves transfer from NMDA-independent to NMDA-dependent [74] Mowrer RR, Gordon WC. Effects of cuing in an ‘irrelevant’ context.
memory. J Neurosci 2001;21:9009–17.
[51] Cain C, Blouin A, Barad MG. L-type voltage-gated calcium channels [75] Wittrup M, Gordon WC. Alteration of training memory through are required for extinction, but not for acquisition or expression, of cueing. Am J Psychol 1982;95:497–507.
conditioned fear in mice. J Neurosci 2002;22:9113–21.
[76] McGeoch JA. Forgetting and the law of disuse. Psychol Rev 1932; [52] Ledgerwood L, Richardson R, Cranney J. Effects of extinction of conditioned freezing. Behav Neurosci 2003;117:341–9.
[77] Postman L, Underwood BJ. Critical issues in interference theory.
[53] Walker DL, Ressler KJ, Lu KT, Davis M. Facilitation of conditioned fear extinction by systemic administration or intra-amygdala [78] Kraemer PJ, Spear NE. The effect of nonreinforced stimulus infusions of D-cycloserine as assessed by fear potentiated startle in exposure on the strength of a conditioned taste aversion as a function of retention interval: Do latent inhibition and extinction [54] Millad MR, Quirk GJ. Neurons in medial prefrontal cortex signal involve a shared process? Anim Learn Behav 1992;20:1–7.
memory for fear extinction. Nature 2002;420:70–4.
[79] Aguado L, Symonds M, Hall G. Interval between preexposure and [55] Gewirtz JC, Falls WA, Davis M. Normal conditioning inhibition and test determines the magnitude of latent inhibition: implications for extinction of freezing and fear-potentiated startle following electro- an interference account. Anim Learn Behav 1994;22:188–94.
lytic lesions of medial prefrontal cortex in rats. Behav Neurosci [80] Bouton ME, Swartzentruber D. Slow reacquisition following extinction: context, encoding, and retrieval mechanisms. J Exp [56] Goosens KA, Maren S. Pretraining NMDA receptor blockade in the Psychol Anim Behav Process 1989;15:43–53.
basolateral complex, but not the central nucleus, of the amygdala [81] Maren S, Holt W. The hippocampus and contextual memory prevents savings of the conditional fear. Behav Neurosci 2003;117: retrieval in Pavlovian conditioning. Behav Brain Res 2000;110: M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 [82] Westbrook RF, Jones ML, Bailey GK, Harris JA. Contextual control [107] Miller RR, Springer AD. Retrieval failure induced by electrocon- over conditioned responding in a latent inhibition paradigm. J Exp vulsive shock: reversal with dissimilar training and recovery agents.
Psychol Anim Behav Process 2000;26:157–73.
[83] Overton DA. Contextual stimulus effects of drugs and internal states.
[108] Blough DS. Delayed matching in the pigeon. J Exp Anal Behav In: Balsam PD, Tomie A, editors. Context and learning. Hillsdale, [109] Roberts WA, Grant DS. Studies of short-term memory in the pigeon [84] Bouton ME, Kenny FA, Rosengard C. State-dependent fear using delayed matching to sample procedure. In: Medin DL, extinction with two benzodiazepine tranquilizers. Behav Neurosci Roberts WA, Davis RT, editors. Processes of animal memory.
Hillsdale NJ: Erlbaum; 1976. p. 79–112.
[85] Radyushkin KA, Anokhin KV. Recovery of memory in chicks after [110] Olton DS, Samuelson RJ. Remembrance of places passed: spatial disruption during learning: the reversibility of amnesia induced by memory in rats. J Exp Psychol Anim Behav Process 1976;2:97–116.
protein synthesis inhibitors. Neurosci Behav Physiol 1999;29:31–6.
[111] Cook RG, Brown MF, Riley DA. Flexible memory processing by [86] Harrod SB, Flint RW, Riccio DC. MK-801 induced retrieval, but not rats: use of prospective and retrospective information in the radial acquisition, deficits for passive avoidance conditioning. Pharmacol maze. J Exp Psychol Anim Behav Process 1985;11:453–69.
[112] Zentall TR, Urcuioli PJ, Jackson-Smith P, Steirn JN. Memory [87] Jackson A, Koek W, Colpaert FC. NMDA antagonists make learning strategies in pigeons. In: Dachowski L, Flaherty CF, editors. Current and recall state-dependent. Behav Pharmacol 1992;3:415–21.
topics in animal learning: brain, emotion, and cognition. Hillsdale, [88] Costanzo DJ, Riccio DC, Kissinger S. State-dependent retention produced with estrus in rats. Physiol Behav 1995;57:1009–11.
[113] Swartzentruber D. Perspectives on modulation: modulator- and [89] Holloway FA, Wansley RA. Multiple retention deficits at periodic target-focused views. In: Schmajuk NA, Holland PC, editors.
intervals after active and passive avoidance learning. Behav Biol Occasion setting: associative learning and cognition in animals.
Washington, DC: American Psychological Association; 1998. p.
[90] Davidson TL. The nature and function of interoceptive signals to feed: toward integration of physiological and learning perspectives.
[114] Wagner AR, Rudy JW, Whitlow JW. Rehearsal in animal conditioning. J Exp Psychol 1973;97:407–26.
[115] Wagner AR. Expectancies and the priming of STM. In: Hulse SH, [91] Eich E. Searching for mood dependent memory. Psychol Sci 1995;6: Fowler H, Hoing WK, editors. Cognitive aspects of animal behavior.
Hillsdale, NJ: Erlbaum; 1978. p. 177–210.
[92] Ahlers ST, Richardson R. Administration of dexamethasone prior to [116] Atkinson RC, Shiffrin RM. Human memory: a proposed system and training blocks ACTH-induced recovery of an extinguished its control processes. In: Spence KW, Spence JT, editors. The avoidance response. Behav Neurosci 1985;99:760–4.
psychology of learning and motivation, vol. 2. New York: Academic [93] Riccio DC, Richardson R, Ebner DL. Memory retrieval deficits based upon altered contextual cues: a paradox. Psychol Bull 1984; [117] Wagner AR. Priming in STM: an information processing mechanism for self-generated or retrieval-generated depression in performance.
[94] Bouton ME, Nelson JB, Rosas JM. Stimulus generalization, context In: Tighe TJ, Leaton RN, editors. Habituation: perspectives from change, and forgetting. Psychol Bull 1999;125:171–86.
child development, animal behavior and neurophysiology. Hillsdale, [95] Bouton ME. Context, ambiguity, and unlearning: sources of relapse after behavioral extinction. Biol Psychiatry 2002;52:976–86.
[118] Wagner AR, Brandon SE. A componential theory of Pavlovian [96] Spear NE, Riccio DC. Memory: phenomena and principles. Boston, conditioning. In: Mowrer RR, Klein SB, editors. Handbook of contemporary learning theories. Hillsdale, NJ: Erlbaum; 2001.
[97] Rudy JW, O’Reilly RC. Contextual fear conditioning, conjunctive representations, pattern completion, and the hippocampus. Behav [119] Smith MC, Coleman SR, Gormezano I. Classical conditioning of the rabbit’s nicatating membrane response at backward, simultaneous [98] Gordon WC, Smith GJ, Katz DS. Dual effects of response blocking and forward CS–US intervals. J Comp Physiol Psychol 1969;69: following avoidance learning. Behav Res Ther 1979;17:479–87.
[99] Miller JS, Jagielo JA, Spear NE. Differential effectiveness of various [120] Maier SF, Rapaport P, Wheatly KL. Conditioned inhibition and the prior-cuing treatments in the reactivation and maintenance of UCS–CS interval. Anim Learn Behav 1976;4:217–20.
memory. J Exp Psychol Anim Behav Process 1991;17:249–58.
[121] McNish KA, Betts SL, Brandon SE, Wagner AR. Divergence of [100] Arnold HM, Spear NE. Order and duration of stimuli are important conditioned eyeblink and conditioned fear in backward Pavlovian determinants of reactivation. Anim Learn Behav 1993;(2):391–8.
training. Anim Learn Behav 1997;25:43–52.
[101] Gisquet-Verrier P. Coherence of retrieval cues, rather than [122] Barela PB. Theoretical mechanisms underlying the trial-spacing additivity, determines prior cuing effectiveness in the rat. Anim effect in Pavlovian fear conditioning. J Exp Psychol Anim Behav [102] Gisquet-Verrier P, Dekeyne A, Alexinsky T. Differential effects of [123] Holland PC. Trial and intertrial durations in appetitive conditioning several retrieval cues over time: evidence for time dependent in rats. Anim Learn Behav 2000;28:121–35.
reorganization of memory. Anim Learn Behav 1989;17:394–408.
[124] Terrace HS, Gibbon J, Farrell L, Baldock MD. Temporal factors [103] Misanin JR, Miller RR, Lewis DJ. Retrograde amnesia produced be influencing the acquisition and maintenance of an autoshaped electroconvulsive shock after reactivation of a consolidated memory keypeck. Anim Learn Behav 1975;3:53–62.
[125] Gallistel CR, Gibbon J. Time, rate, and conditioning. Psychol Rev [104] Nader K, Schafe GE, Le Doux JE. Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature [126] Bouton ME, Sunsay C. Importance of trials versus accumulating time across trials in partially reinforced appetitive conditioning.
[105] Przybyslawski J, Sara SJ. Reconsolidation of memory after its J Exp Psychol Anim Behav Process 2003;29:62–77.
reactivation. Behav Brain Res 1997;84:241–6.
[127] Sunsay C, Stetson L, Bouton ME. Memory priming and trial [106] Duncan CP. The retroactive effect of electroshock on learning.
spacing effects in appetitive Pavlovian learning. Learn Behav 2004; J Comp Physiol Psychol 1949;42:32–44.
M.E. Bouton, E.W. Moody / Neuroscience and Biobehavioral Reviews 28 (2004) 663–674 [128] Pfautz PL, Wagner AR. Transient variations in responding to [142] Balleine BW. Instrumental performance following a shift in primary Pavlovian conditioned stimuli have implications for the mechanisms motivation depends on incentive learning. J Exp Psychol Anim of ‘priming’. Anim Learn Behav 1976;4:107–12.
[129] Lattal KM. Trial and intertrial durations in Pavlovian conditioning: [143] Dickinson A, Balleine B. The role of learning in the operation of issues of learning and performance. J Exp Psychol Anim Behav motivational systems. In: Pashle H, Gallistel R, editors. Steven’s handbook of experimental psychology. Learning, motivation, and [130] Barnet RC, Grahame NJ, Miller RR. Trial spacing effects in emotion, vol. 3. New York: Wiley; 2002. p. 497–533.
Pavlovian conditioning: a role for local context. Anim Learn Behav [144] Holland PC. Conditioned stimulus as a determinant of the form of the Pavlovian conditioned response. J Exp Psychol Anim Behav Process [131] Sunsay C, Bouton ME. A theoretical investigation of the trial spacing effect in appetitive Pavlovian conditioning; In Preparation.
[145] Timberlake W, Grant DL. Auto-shaping in rats to the presentation of [132] Kamin LJ. Predictability, surprise, attention, and conditioning. In: another rat predicting food. Science 1975;190:690–2.
Campbell BA, Church RM, editors. Punishment and aversive [146] Akins CK. Effects of species-specific cues and the CS–US interval behavior. New York: Appleton–Century–Crofts; 1969. p. 279–96.
on the topography of the sexually conditioned response. Learn Motiv2000;31:211–35.
[133] Rescorla RA. Effect of US habituation following conditioning.
[147] Akins CK, Domjan M, Gutie´rrez G. Topography of sexually J Comp Physiol Psychol 1973;82:137–43.
conditioned behavior in male Japanese quail (Coturnix japonica) [134] Rescorla RA. Effect of inflation of the unconditioned stimulus value depends on the CS–US interval. J Exp Psychol Anim Behav Process following conditioning. J Comp Physiol Psychol 1974;86:101–6.
[135] Holland PC. Event representation in Pavlovian conditioning: image [148] Holland PC. CS–US interval as a determinant of the form of and action. Cognition 1990;37:105–31.
Pavlovian appetitive conditioned responses. J Exp Psychol Anim [136] Colwill RM, Rescorla RA. Associative structures in instrumental learning. In: Bower GH, editor. The psychology of learning and [149] Vandercar DH, Schneiderman N. Interstimulus interval functions in motivation, vol. 20. New York: Academic Press; 1986. p. 55–104.
different response systems during classical discrimination condition- [137] Dickinson A. Expectancy theory in animal conditioning. In: ing in rabbits. Psychon Sci 1967;9:9–10.
Klein SB, Mowrer RR, editors. Contemporary learning theories: [150] Wilkie DM, Spetch ML. Pigeon’s delayed matching to sample errors Pavlovian conditioning and the status of traditional learning theory.
are not always due to forgetting. Behav Anal Lett 1981;1:317–23.
Hillsdale, NJ: Erlbaum; 1989. p. 279–308.
[151] Roberts WA, Grant DS. Interaction of sample and comparison [138] Adams CD. Variations in the sensitivity of instrumental responding stimuli in delayed matching to sample with the pigeon. J Exp to reinforcer devaluation. Q J Exp Psychol B 1982;34:77–98.
Psychol Anim Behav Process 1978;4:68–82.
[139] Colwill RM, Rescorla RA. The role of response reinforcer [152] Roitblat HL, Harley HE. Spatial delayed matching-to-sample associations increases throughout extended instrumental training.
performance by rats: learning, memory, and proactive interference.
J Exp Psychol Anim Behav Process 1988;14:71–82.
[140] Balleine BW. Incentive processes in instrumental conditioning. In: [153] Timberlake W. Motivational modes in behavior systems. In: Mowrer RR, Klein SB, editors. Handbook of contemporary learning Mowrer RR, Klein SB, editors. Handbook of contemporary learning theories. Hillsdale, NJ: Erlbaum; 2001. p. 307–66.
theories. Hillsdale, NJ: Erlbaum; 2001. p. 155–209.
[141] Dickinson A, Balleine B. Motivational control of goal-directed [154] Odling-Smee FJ. Background stimuli and the inter-stimulus interval action. Anim Learn Behav 1994;22:1–18.
during Pavlovian conditioning. Q J Exp Psychol A 1975;27:387–92.

Source: http://paardengedrag.be/wp-content/uploads/2012/02/Memory-processes-in-classical-conditioning.pdf