# Kamin blocking

 John W Moore and Nestor A. Schmajuk (2008), Scholarpedia, 3(5):3542. doi:10.4249/scholarpedia.3542 revision #89027 [link to/cite this article]
Post-publication activity

Curator: John W Moore

In psychology, the term blocking refers broadly to failures to express knowledge or skill because of failures of learning or memory, as in the everyday experience of “blocking” of the name of a familiar face or object. Kamin blocking refers to failures of learning and/or the expression of classically conditioned responses (CRs) when a target conditioned stimulus (CS) is presented to an animal as part of a compound that includes another CS that had been used previously to establish the target CR.

By convention, we shall refer to the blocked CS as B (for blocked) and the previously established CS as A (for antedating). In a series of experiments reported in three chapters published in the late 1960s, Kamin showed that the prior conditioning response training with CS A interferes with the acquisition of the CR to CS B when they are presented together as compound stimulus AB. This conditioning paradigm, which can be abbreviated A+ $$\rightarrow$$ AB+, takes its name from psychologist Leon J. Kamin, who first reported the phenomenon in a 1968 chapter (Kamin, 1968). A related chapter by Kamin appeared in 1969 (Kamin, 1969a). A third chapter on blocking also appeared in 1969 (Kamin, 1969b).

The plus signs “+” in A+ $$\rightarrow$$ AB+ indicate that CS A and compound CS AB are followed by the contiguous presentation of an unconditioned stimulus (US). This two-stage training is usually followed by presentations of B alone and without reinforcement by the US. This is often abbreviated as B-, where the minus sign “-” denotes the withholding of the US. In this paradigm, blocking refers to a low level of conditioned responding to B alone when compared to various control procedures. The most commonly employed control procedure is compound conditioning (AB+) but without the antedating A+ stage. Because B is followed by the US equally often in the blocking and compound-only conditions, the observation of reduced conditioning to B only in the blocking conditioning is hard to reconcile with theories of classical conditioning asserting that contiguity of a CS with a US suffices for the establishment of a CR. Kamin reasoned that the traditional theories were incomplete. He suggested that blocking implies that some higher-order “attentional” processes are involved in conditioning, processes that had been regarded with suspicion (if not disdain) by most psychologists concerned with animal learning and behavior. The strength of Kamin’s evidence from his blocking experiments fueled the then nascent “cognitive” perspective, which in the ensuing decades became a dominant feature of modern learning theory and computational models of classical conditioning. Kamin interpreted blocking in terms of surprise, attention, and predictability, and he credited Egger and Miller (1962) for anticipating many of his ideas. For animal learning theory, the cognitive revolution began in earnest with Kamin blocking.

This essay summarizes Kamin’s basic findings and his ideas about their implications for associative learning and classical Pavlovian conditioning. The experiments described in the three chapters cited above employed Estes and Skinner’s Conditioned Emotional Response (CER) procedure with rats (Estes & Skinner, 1941). The CER procedure using rats as subjects has been a popular model system for quantitative studies of classical conditioning. Kamin choose the CER because of its sensitivity to variables important for classical conditioning, a training procedure described originally by Pavlov (1927). The CER is often referred to as conditioned suppression because the learned response is a CS-elicited reduction in the rate of free-operant response rates established by food reward using various schedules of reinforcement. Conditioned suppression occurs when a CS paired with a foot shock US results in a reduction in the baseline operant rate of responding for food reinforcement in the presence of the CS. Withdrawal of the CS results in baseline recovery. A CER is measured as the reduction in the rate of responding in the presence of the CS expressed as a suppression ratio, R1/(R1 + R2), where R1 is the number of operant responses during the CS and R2 is the number of operant responses during a comparable interval of time antedating the CS.

In Kamin’s experiments, the operant behavior was bar pressing for food by motivated (food restricted) rats. The CSs, lights and noise, were 3 minutes in duration and foot shook employed to instill classical conditioning were 1-milliampere foot shocks of 0.5-sec duration delivered via metal grids comprising the floor of the operant chamber. The foot shock is timed to overlap with the last 0.5 seconds of the CS. When first presented, the CS produces a shallow and transient suppression of bar pressing attributable to an orienting response. However, a few pairings of the CS with foot shock results in reduced bar pressing as indexed by suppression ratios below 0.5. Estes and Skinner (1941) suggested that this suppression reflects conditioned fear, and this interpretation is consistent with other indices of fear such as freezing behavior or suppression of licking in the presence of the CS by water-motivated rats.

Kamin (1969b) summarizes the initial blocking experiments with rats in the CER procedure in general terms: “First, condition an animal to respond to a simple CS, consisting of Element A. Then, condition the animal to respond to a compound, consisting of Element A plus a superimposed Element B. Finally, test the animal with Element B alone (p 42).” Blocking occurs when this prior conditioning to Element A interferes with conditioned responding to Element B during the test. “To conclude that the prior conditioning to Element A was responsible for the failure to respond to Element B we must, of course, show that animals conditioned to the compound without prior conditioning to A do respond when tested with B. [Furthermore,] we ought also show that if compound conditioning is followed by conditioning to A alone, the animal will respond when tested with B (p 42).”

Thus, initial demonstrations of blocking employed an experimental design involving three groups: a blocking group for which conditioning to A preceded compound conditioning to AB, a group that received compound conditioning to AB only, and a group that received compound to AB prior to subsequent conditioning to A alone. Any demonstration of blocking requires that conditioned suppression of responding to B on test trials be significantly impaired in Group Blocking in comparison with the other two groups. Subsequent demonstrations of blocking employed similar control procedures.

## Other demonstrations of Kamin blocking

In their initial demonstration of blocking in the rabbit eyeblink conditioning procedure, Marchant and Moore (1973) employed five groups of subjects. Using the A+ $$\rightarrow$$ AB+ paradigm, Stimulus A was a tone (T) and Stimulus B was a light (L).

Rabbits in the blocking group (T+ $$\rightarrow$$ TL +), were trained for 3-4 daily sessions of 100 trials to T+. Stage-2 compound training to TL+ proceeded for five additional days, followed by two days of extinction testing involving 60 L- test trials. Only one CR occurred to L- among the four subjects; a CR rate of 0.00, which is virtually complete blocking.

One control group (sit $$\rightarrow$$ TL+) did not receive Stage-1 training to T+. Instead, it was confined in the conditioning chambers under restraint for a period of time equal in duration to first-stage conditioning to T+ in the blocking group., i.e., for 3-4 daily sessions of 100 trials each. Conditioning to tone-light compound (TL+) began midway through the fifth session and proceeded for five daily sessions, followed by the extinction test to L-. Marchant and Moore (1973) referred to this procedure as a “sit”-control, because the rabbits were confined under restraint with blink-recording paraphernalia attached for the duration of Stage 1, but received no conditioning to Element A presented alone. This group is equivalent to Kamin’s group that received compound conditioning to A and B (AB+) without Stage-1 training to A+. Tests revealed that conditioning to L was unimpaired, with a CR rate of 0.32. The relatively low rate of CRs to L- reflects overshadowing, a phenomenon thought to be related to blocking.

A second control group (sit $$\rightarrow$$ L+) received L+ training in Stage-2 instead of TL+. This group controlled for any proactive interference of conditioning to L+ in Stage 2 that might result from confinement in the training chambers prior to L+ training. The group received the same extinction-test as the other groups. The CR rate for this group of for rabbits was 0.67. The relatively high CR rate reflects the absence of blocking and overshadowing.

A third control group (T+ $$\rightarrow$$ L+) resembled the blocking group, except that Stage-1 training to T (T+) was followed by training to L alone in Stage 2 (L+). The CR rate to L- on extinction-test trials was 0.43. Like group sit $$\rightarrow$$ L+, this CR reflects the absence of blocking and overshadowing, but may reflect proactive interference from T+ training in Stage 1.

A fourth control group resembled the order-reversal control employed by Kamin, with TL+ training in Stage 1 preceding L+ training in Stage 2. The CR rate for this group was 0.275. Stage-1 training to TL+ resulted in fewer CRs during the test phase than in the other control groups, but significantly more than in the blocking group. The low CR rate may reflect a combination of overshadowing and retention or retrieval deficit or perhaps even a weak backward-blocking effect (see below).

### Conditioned odor aversion in Limax

Sahley, Rudy, and Gelperin (1981) demonstrated Kamin blocking of learned odor aversion in the terrestrial mollusk Limax using three control groups. The blocking group received trials on which carrot odor was paired with bitter tasting quinidine sulfate in Stage 1. In Stage-2, this group experienced compound conditioning in which carrot and potato odor were combined together and paired with quinidine. Because of the prior association between carrot odor and the quinidine, animals did not form an aversion to the odor of potato. The pretrained carrot odor blocked the development of an aversion to potato.

One of Sahley et al’s control groups did not receive any training in Stage-1 but did receive the compound conditioning to the carrot-potato compound in Stage-2. Test trials showed that the animals did develop an aversion to potato odor, which implies no blocking. Aversive conditioning to potato odor in this group was just as great as that to carrot odor, indicating an absence of overshadowing of one odor by the other. A second control group received aversion conditioning to carrot odor in Stage-1 and potato odor in Stage-2. Aversion conditioning to both odors was robust. These two control groups matched two blocking controls employed by Kamin and Marchant and Moore (1973).

Sahley et al’s third control group differed: Stage-1 training employed a backward conditioning procedure in which quinidine preceded exposure to carrot odor, with Stage-2 the same as in the blocking group. There was no evidence of blocking of the potato odor aversion. The backward conditioning control is important because this procedure equates the number of exposures to carrot odor and quinidine in this control with that of the blocking group. However, backward conditioning procedures often result in conditioned inhibition. Theoretically, this training would convert the carrot odor into a “safety signal” with respect to quinidine. Combining this safety signal with potato in Stage-2 could theoretically enhance aversion conditioning to potato odor following Stage-2 compound training, according to the Rescorla-Wagner (1972) model of associative learning (e.g., Rescorla, 1971). There was no evidence for this effect in the Sahley et al study (Wagner & Rescorla, 1972).

Another possible blocking control involves presenting the US alone in Stage-1 of the two-stage paradigm (e.g., foot shocks in the CER procedure, air puffs or periorbital eye-shocks in eyeblink conditioning procedures, and quinidine in Limax odor aversion learning). Unlike backward conditioning, US-alone control groups would likely not produce conditioned inhibition of Stimulus A. Instead, it could result in a US pre-exposure effect, whereby the experimental context predicts the US such that the learned association between the context and the US interferes with subsequent conditioning to the AB compound CS in Stage-2, thereby acting more as an alternative blocking procedure than as a control for blocking. A better alternative blocking control would train animals to a third CS, Stimulus C (C+) in Stage-1. This procedure is designed to reduce context blocking because conditioning to Stimulus C would overshadow any conditioning to the context. One risk is that features shared by C and B could promote stimulus generalization between the two. Generalization from C and B would lead to an overestimation of blocking by increasing conditioned responding to B.

### Conditioned proboscis extension in honey bees

However, the C+ control was used in conditioned-proboscis extension to odorant stimuli in the honeybee (Smith & Cobey, 1994). Variations of US intensity in the honey bee system produce many of the same effects on blocking as observed in vertebrates, e.g., unblocking when US intensity increases from stage 1 to stage 2 in the 2-stage blocking paradigm (Smith, 1997), suggesting shared mechanisms of blocking among vertebrates (rats, rabbits, pigeons) and invertebrates (snails and bees).

## Blocking of conditioned inhibition

The forgoing account of blocking control groups pertain to excitatory conditioning procedures, i.e., procedures such as pairing a CS and US in order to increase the frequency and magnitude of CRs such as conditioned suppression, conditioned eyeblinks, and conditioned odor aversions. Suiter and LoLordo (1971) were among the first to demonstrate Kamin blocking of inhibitory conditioning. Inhibitory conditioning (also called conditioned inhibition) describes learning in which the CS signals the omission of an otherwise expected US. Conditioned inhibition to the inhibitory CS is expressed by its ability to reduce the CR elicited by an excitatory CS and by the slower rate at which the inhibitory CS can acquire an excitatory CR through pairings with a US. Note that blocking of inhibition is not to be confused with inhibition of blocking. The former implies that a CS predicts the absence of the US and an absence of the CR. The latter implies a procedure whereby a CS interferes with the development of the expression of blocking, e.g., extinction of conditioning to A in a paradigm represented by the string A+ $$\rightarrow$$ A$$-\rightarrow$$ AB+, in which A loses its potential to block CR acquisition to B.

Suiter and LoLordo (1971) employed the CER procedure with rat subjects in a three-stage experimental design with groups of rats trained to bar press for food reinforcement. For the blocking-of-inhibition group, foot shocks were presented in Stage-1 only when L was not present. That is, L and foot shock were explicitly unpaired. As with the backward conditioning procedure discussed above, uncorrelated presentations of L and foot shock cause L to become a “safety signal” with respect to conditioned suppression of bar pressing. In Stage-2, the blocking-of-inhibition group received presentations of the compound CS consisting of L and T presented together but never in the presence of foot shock. Stage-3 was a test for any learned inhibition to T. The test consisted of a series of trials on which T was paired repeatedly with foot shocks (T+). Any blocking of conditioned inhibition acquired in Stage-2 through negatively correlated presentations of the LT and foot shock would retard acquisition of the CER to T in Stage-3.

The Suiter and LoLordo (1971) study employed four control groups. One control group received unsignaled shocks in Stage-1 without L. This was followed by the same LT-uncorrelated-shock procedure employed in the blocking group in Stage 2 and followed in turn by the Stage-3 retardation of conditioning test to T. Blocking of inhibitory conditioning emerged as more rapid acquisition of conditioned suppression in the blocking-of-inhibition group than in the free-shock control-group. The unsignaled shocks in Stage-1 would have promoted excitatory conditioning of the context, which should have enhanced the inhibitory conditioning of the LT- compound and to T as an element of that compound, and this was born out by the retarded excitatory condition to T in this control group when contrasted with the blocking-of-inhibition group. A second control group received L and foot shocks negatively correlated in both Stages 1 and 2. T was withheld until Stage-3, where it developed conditioned suppression just as rapidly as the blocking-of-inhibition group. In the third and fourth control groups, Stage-1 consisted of inhibitory conditioning to the compound (LT negatively correlated with foot shocks) followed in Stage-2 by either further inhibitory conditioning to L or unsignaled shocks. Conditioned Inhibition of the CER was equally robust in these two groups in Stage-3 excitatory conditioning tests.

Although the Rescorla-Wagner model can predict Kamin blocking of conditioned inhibition in the CER procedure, Suiter and LoLordo’s findings are also consistent with Kamin’s “surprise” interpretation of blocking, which assumes that blocking is a consequence of failure to attend to the to-be-blocked stimulus because of its redundancy in predicting the absence of foot shock. In the Suiter and LoLordo study, rats failed to notice T because it was redundant with respect to predicting the absence of foot shock.

Support for Kamin’s interpretation of blocking in both excitatory and inhibitory CER conditioning procedures comes from experiments by Mackintosh and Turner (1971). Mackintosh and Turner (1971) report unblocking of the effects of Stage-1 excitatory or inhibitory conditioning if the intensity of the foot shocks either increased or decreased at the start of Stage 2, replicating finding reported by Kamin (1969a). No unblocking occurred, however, when the switch in shock intensity was introduced four trials into Stage-2. Mackintosh and Turner concluded that blocking is not merely a failure to attend to the added CS introduced in Stage-2, because if this “novel” alteration in training sufficed to cause unblocking, than some unblocking should be evident in tests. Mackintosh and Turner (1971) argued that by the fourth Stage-2 presentation of the compound CS, animals had learned that the added element was redundant with respect foot-shock intensity. That is, they had learned to ignore the added element, as opposed to failing to notice its presence. These ideas became important in Mackintosh’s (1975) attention theory of conditioning. Consistent with Mackintosh’s (1975) theory, Kamin (1968, p 17) noted that transient reductions of conditioned suppression that occur when a new CS is combined with a previously trained CS dissipate quickly with additional compound conditioning.

### Kamin blocking in an appetitive reinforcement task

Demonstration of Kamin blocking reviewed thus far involved aversive conditioning procedures such as Estes and Skinner’s CER procedure with rats. These demonstrations involved a number of control procedures. Are there similarly rigorous demonstrations of blocking with an appetitive conditioning procedure? vom Saal and Jenkins (1970) reported blocking of stimulus control of an auditory go/no-go discrimination in the pigeon key pecking procedure developed by Skinner and his students (e.g., Ferster & Skinner (1957). The target behavior in the vom Saal and Jenkins (1970) demonstration of blocking was the stimulus control of key pecking exerted by differential reinforcement of exemplars from visual dimensions such as color (wavelength, saturation, and shape) and acoustical stimuli varying in tonal frequency and intensity. Stimulus control refers to the extent to which animals respond to specific clusters of stimulus features, as reflected in the accuracy of discrimination performance and the steepness of stimulus generalization gradients (Terrace, 1966).

In the vom Saal and Jenkins (1970) study, pigeons in the blocking group were trained in Stage-1 on a red versus green go/no-go visual discrimination task and then switched in Stage 2 to a compound-cue discrimination task in which red and green were combined with a tone and a noise. One control group received no Stage-1 training. Another control group received partial reinforcement training in Stage-1, such that neither red nor green were reliable predictors of differential reinforcement. A third control group received reinforcement for key pecks to both red and green. When tested after Stage-2, the blocking group showed less auditory control than the other groups. The vom Saal and Jenkins (1970) study is noteworthy not only because it demonstrate Kamin blocking in an appetitive learning procedure but also because it supports attentional interpretations of blocking such as that proposed by Mackintosh (1975), in which animals learn to ignore stimuli that are not predictive of differential reinforcement.

### Noteworthy observations from Kamin’s studies of blocking

Kamin (1968) reported that the light (L) and noise (N) CSs employed in the initial demonstrations of blocking were not equally salient, because suppression to L was greater than the corresponding suppression to N. Nevertheless, initial training to light CS (L) was just as effective in blocking conditioning to the noise CS (N) as the noise CS is in blocking conditioning to the light CS, when the latter is the added CS element to the LN compound. Salience refers to the rate of CR acquisition or development when the stimulus is paired with reinforcement. However, it is usually easier to block conditioning to a CS of weaker salience with a CS of stronger salience than the other way around (Marchant & Moore, 1973, with the notable exception of taste-potentiated odor aversion conditioning, Rusiniak, Hankins, Garcia, & Brett, 1979).

Kamin (1968) noted that the first presentation of a stimulus (typically L) in the CER procedure produced a transient suppression of ongoing bar pressing, which he referred to as a “mildly disruptive effect” akin to Pavlovian external inhibition. Transient disruptions of bar pressing are not conditioned responses, as they typically dissipate after the initial stimulus presentation. Conditioned suppression of bar pressing to a stimulus only dissipates after a series of non-reinforced extinction trials.

Kamin (1968, p 14) reported that prior conditioning to L did not attenuate conditioning to N-alone and vice versa. This finding validates one of the controls for proactive interference mentioned above.

Kamin (1968, p 16-18) reported that blocking in the two-stage paradigm was not overcome by extended compound-CS training in Stage-2. By contrast, reducing the number of Stage-1 training trials to N attenuated the blocking effect observed on L test trials. Eight noise trials in Stage-1 was enough to produce complete blocking of L. Complete blocking was also observed with CSs of 1-minute duration instead of the usual 3-minute CS durations and with 3-milliampere foot shocks instead of the usual 1-milliampere foot shocks. Furthermore, L was blocked just as effectively by a noise-off CS as a noise-on CS. In short, blocking in the CER procedure with rats is robust and easy to demonstrate, according to Kamin’s experiments.

Kamin (1968, p 19) reported that blocking could be prevented by non-reinforced presentation of N (extinction training) before compound-CS training. This observation supports Kamin’s hypothesis that the amount of learning about the blocked CS (L) depends on to the depth of conditioned suppression to N at the start of Stage-2. That is, learning about L depends on the discrepancy between the level of conditioned responding to N at the start of Stage-2 and the asymptotic level of conditioning achievable with the foot-shock reinforcement.

### Savings of conditioning to a blocked CS

Kamin employed savings tests as a measure of blocking with greater sensitivity than the post-Stage-2 extinction test to L (L-). Savings tests typically involved a series of four foot-shock reinforced trials (L+) followed the extinction test. They were important in Kamin’s efforts to pinpoint Stage-2 trials responsible for any attenuation of blocking effects. Using savings tests, Kamin concluded that any “unblocking” effects (i.e., conditioning to L despite its compounding with N) could be detected by savings tests. These test indicated that conditioning to L in compound with N occurred on the initial Stage-2 trial.

Kamin (1968, p 20) noted that combining L and N into a compound CS in Stage-2 of the two-stage paradigm was always noticed by the animal, because the addition of the new element to the original CS resulted in an immediate decrease in conditioned suppression relative to the final Stage-1 trial. Later, Kamin argued that the magnitude of this decrease in suppression on this initial compound trial predicted the number of reinforced trials following the extinction-test needed to fully condition the blocked CS element. The larger the initial disruption of conditioned suppression on the first Stage-2 trial, the fewer the number of reinforced trials needed to undo the blocking effect of compound conditioning. On its face, this result may appear counterintuitive because the added element initially results in an attenuation of conditioned suppression. The magnitude of this initial attenuation of conditioned suppression predicts greater savings of learning (fewer trials) than the need for more trials to overcome the blocking effect. Thus, greater CER disruption predicts faster CER learning. This observation was central to Kamin’s “surprise” theory of conditioning.

## CS-US contiguity

Kamin (1968, p 21) discussed the role of contiguity of CSs with the US in producing blocking. Blocking only occurs when N (the blocker) occurs contiguously with L (the blocked stimulus) at the time of the US. That is, blocking only occurred when the blocking stimulus and the blocked stimulus were contiguous with the US, even though the duration of the blocking stimulus (N) was only 5 seconds. Kamin referred to the requirement that N overlap the US as temporally specific blocking. The finding that CS-overlap yields greater blocking than non-overlap is supported by a study of rabbit eyeblink conditioning by Kehoe, Schreurs, and Amodei (1981). This study showed that blocking of the second component of a serial compound CS was greater when the two CSs overlapped than when they did not overlap.

Kamin maintained that overshadowing and blocking are instances of the same underlying processes, whereby training of an element prior to compound training has the same effect on conditioning to the added element as if the added element had been trained de novo as the weaker element of a compound CS (p 29). Kamin (1969b) reported that overshadowing of one CS element by another can be reduced or eliminated by increasing the intensity of the US during compound conditioning. From this evidence, Kamin argued that blocking and overshadowing represent the same basic processes, namely, the advantage of rapid or previously established acquisition of a CER to the more salient component of the compound CS.

Kamin suggested that blocking and overshadowing are different phenomena, the former being due to learning while the latter is due to perceptual processes. The two phenomena are reconciled in the Rescorla-Wagner model of associative learning. Indeed, Kamin suggested that stimulus intensity (salience) and the amount of pre-superimposition training are two ways to modulate blocking effects. Kamin’s CER experiments support this view. Marchant and Moore (1973) demonstrated the functional equivalence of amount of pre-training and salience of the CS, using the classically conditioned eyeblink.

## Kamin’s “Blotting Out” hypothesis and error correction

In describing his earliest impressions of blocking, Kamin observed that “It is indeed as if the animal does not see the light [blocked CS] when a previously trained noise-training response is acting (Kamin, 1968, p 22).” He rejected this “blotting out” theory because of an experiment (Kamin, 1968, p29) showing that increasing the intensity of the foot-shock US after stage-1 training to N eliminated blocking to L in the stage-2 compound training stage. Increasing the intensity of the US effectively raised the potential asymptotic level of CER conditioning. In short, blocking is inversely related to the difference between the amount of conditioning attained by the pre-trained element (N) with a comparatively weak US and the greater amount of conditioning attainable with a stronger US.

These ideas lie at the heart of the Rescorla-Wagner model. For his part, Kamin noted that learning requires that the US be to some extent surprising. Switching from a 1-milliampere foot shock to a 4-milliampere foot shock would constitute a surprise in that the stage-1 CS (N) predicts a relatively weak US. A sudden increase the intensity of the US would not only increase the asymptotic potential for the conditioned response, it would constitute an unanticipated change of circumstances. The resulting surprise activates learning and the superimposed CS (L) acquires the CER. This is the core idea of error-correction learning, as represented by the Rescorla-Wagner (1972) model and other modern theories, such as the time-derivative models of Sutton and Barto (1991).

## Kamin’s surprise hypothesis of learning

Kamin (1969b) attributed surprise to the animal’s recognition of the added CS component (B). This recognition results in external inhibition of the previously established CER component (A). This external inhibition results in a transient reduction of the CER (less suppression). Any conditioning to B occurs only during bouts of distraction from A. Once B loses its capacity as a distraction from A, conditioning to B effectively ceases no matter how much compound conditioning follows this transition phase between single-component conditioning to A and compound conditioning to B.

Kamin provided corroborating evidence for this scenario by examining the savings in CER acquisition to B following tests for blocking. As outlined previously, blocking tests consisted of presenting the blocked component alone for one non-reinforced trial after the compound-conditioning phase. Blocking occurred to the extent that median suppression ratios equal 0.5 in value. Kamin demonstrated that median suppression ratios to component B failed to reflect the learning that occurs on transiting from single component conditioning to A and conditioning to the AB compound. A savings test showed some conditioning to B during the compound conditioning phase. That is, conditioning to B was faster after the blocking test than to a novel cue.

Kamin established that the amount of savings of CER acquisition to B depended on the degree to which conditioned suppression decreased on the first compound conditioning trial. Thus, any conditioning to B during the compound conditioning phase resulted from the transient reduction of CER magnitude on the first compound trial. Kamin suggested that this learning on the initial compound conditioning trial came from the surprise evoked by component B. This surprise dissipates with recognition that B is redundant predictor of foot shock.

Kamin suggested that all conditioning depends on the surprise value of a CS’s predictive relationship regarding foot shocks. Learning ceases when the CS’s predictive relationship is established. The introduction of a salient additional CS component, when noticed, triggers a reassessment of associative relationships involving foot shocks. This reassessment involves backward scanning of the short-term memory. Kamin’s assertions about backward scanning and learning are confused because it is suppose to mediate conditioning to B on transition trials, on the one hand, and to mediate learning caused by unpredicted modulations of the foot shock, on the other.

The backward scanning idea says that animals engage in “retrospective processing” of events. Thus, the associative connection between a CS and US, which controls the CER, can be modified only when the qualities of the US and their predictability are altered unexpectedly. Such alterations are noticed immediately and “alert” the animal to pay greater heed to events on subsequent trials. As with backward scanning, this “prospective processing” can result in modifications of the CS-US associative relationships. Either scenario can account for the new learning and possible unblocking of B when combined with A to form a compound CS. Kamin reported frustration in being able to concoct experiments that would clearly choose between the retrospective backward scanning account of conditioning and the prospective “alerting” account of learning and unblocking.

Kamin maintained that either account could be correct but that in either case the animal “is still left with a selective role in determining those occasions when stimulus inputs do enter into learned associations (Kamin, 1969b, p 63).” This statement is at odds with purely mechanistic accounts of learning because it is up to the animal to decide to alter a particular associative link or not. Misgivings about the mentalistic aspects of Kamin’s accounts of associative learning in general and blocking/unblocking in particular triggered efforts to development mechanistic theories that could encompass the evidence of an active role for the animal in choosing (selecting) stimulus events that best meets its needs. Such a mechanistic theory would have to reconcile traditional contiguity theories of learning (especially conditioning), which assert that contiguity is necessary and sufficient for learning an associative relationship, with the overwhelming evidence from Kamin’s blocking experiments that contiguity is not sufficient. The dilemma for mechanists was resolved with the emergence of the Rescorla-Wagner model.

## Experimental tests of surprise and unblocking

Gray and Appignanesi (1973) reported that novel stimuli presented in Stage-2 disrupted the blocking effect in the CER procedure. The novel stimulus consisted of a brief (300 msec) “flash” of the compound CS after each Stage-2 trial. Savings tests revealed unblocking expressed as more rapid acquisition to Stimulus B than in the normal A+ $$\rightarrow$$ AB+ paradigm. By this test, compound CS “flashes” only disrupted blocking when they followed reinforced compound CS trials by 3 and 5 but not 10 sec. Gray and Appignanesi (1973) suggested that this aspect of their unblocking results favored Kamin’s backward scanning idea that surprise triggers a retrospective assessment of the events preceding the surprising event. This backward assessment presumably caused an association between B and the US to be formed “in short-term memory,” i.e., after the event. An alternative interpretation is that the unexpected flashes introduced in Stage-2 simply caused an “alerting” reaction that promoted attention to B on subsequent trials. The two processes are compatible in principle, but the timing data favors the former.

A similar conclusion emerged from a CER study by Dickinson, Hall, and Mackintosh (1976), which employed three different methods to induce post-trial surprise during Stage-2 compound conditioning. One method consisted of adding an extra shock at the termination of the AB compound stimulus, i.e., 2 shocks instead of only 1, as predicted by Stimulus A because of Stage-1 training. A second method of inducing post-trial surprise was to use two shocks as a US in Stage-1 training to Stimulus A but to postpone the second shock in Stage-2. The third method of inducing surprise employed a two-shock US in Stage-1 but omitted the second shock in Stage-2. All of these methods produced unblocking. The authors favored the retrospective processing explanation of unblocking, while rejecting the prospective “alerting” idea. However, their interpretation of unblocking differed from that of Kamin and Gray and Appignanesi (1973). Instead of suggesting that retrospective processing causes an associative connection between B and the US, they suggested that the surprising alteration of the US causes a retrospective increase in the associability of B. This up-tick in associability results in formation of a B $$\rightarrow$$ US association, but the increase in the predictive relationship between B and the US does not occur until the next reinforced Stage-2 trial. How this happens is unclear. An alternative interpretation of the Dickinson et al. (1976) data would be that adding a second shock in Stage-2 raises the asymptotic level of conditioning, which promotes conditioning to B. Postponing or eliminating second shocks could result in some extinction of conditioning to A, and this, too, would promote conditioning to B, according to the Rescorla-Wagner (1972) model.

A subsequent CER study by Donegan, Whitlow, and Wagner (1977) did not consistently replicate the unblocking effect caused by post-trial “flashes” of the AB compound employed by Gray and Appignasesi (1973). In attempting to replicate this study, however, Donegan at al (1977) increased the duration of “flashes” from 300 to 500 msec, thereby making it more a “reinstatement” than a “flash.” It is not clear whether this difference was responsible for the discrepancies between the two studies.

Kohler and Ayres (1979) approached the question of surprise and unblocking using a lick suppression task instead of the CER procedure employed by Kamin. Their approach to introducing surprise was to vary the duration of Stage-1 and/or Stage-2 trials and to compare the post-Stage-2 acquisition of conditioned suppression to Stimulus B with that of groups for which the duration of the CS did not vary. Blocking of B occurred with both variable and fixed CS durations. Maleske and Frey (1979) approached the question of surprise in blocking in rabbit eyeblink conditioning by changing the CS-US interval from Stage-1 to Stage-2, and they too failed to observe decrements in blocking. Thus, there was no evidence that any “surprise” attendant upon a varying CS-US interval caused unblocking. This finding appears to undermine Kamin’s claim that blocking is a temporally specific phenomenon, but it does not gainsay the importance of CS-US contiguity in promoting blocking noted previously.

In sum, efforts to produce unblocking through surprise-triggered learning to Stimulus B in the two-stage paradigm have not been universally successful, and favorable evidence has been subject to other interpretations, particularly in terms of the Rescorla-Wagner (1972) model. Nevertheless, debates about the role of predictability, surprise, and attention in conditioning gave impetus to a number of contemporary theories and computational models of conditioning that incorporate these ideas.

## Kamin blocking as learning versus performance deficit

Whether blocking is the result of retrospective or prospective processing, Kamin believed that it is due to the failure in acquiring a CS-US association in the presence of a CS that had been trained previously or possessed greater salience with respect to the specific US. Miller and his associates proposed that blocking results from a failure of performance, not acquisition of learning. Briefly, blocking of a CER in the A+ $$\rightarrow$$ AB+ paradigm occurs because the pre-trained blocking CS (A) prevents retrieval or expression of the memory of the B $$\rightarrow$$ US association presumably established during the AB+ stage (Miller & Schachtman, 1985). Attempts to uncover this association involved a number of procedures, such as extinction of the A $$\rightarrow$$ US association (Arcediano, Escobar, & Matute, 2001; Blaisdell, Gunther, & Miller, 1999), reminder treatments that involve presentations of the blocked CS or US (Balaz, Gustin, Cacheiro, & Miller, 1982; Schachtman, Gee, Kasprow, & Miller, 1983), and spontaneous recovery (Pineno, Urushihara, & Miller, 2005). However, extinction does not always cause unblocking, and the notion that blocking is a retrieval or performance deficit remains controversial (McPhee, Rauhut, & Ayres, 2001; Rauhut, McPhee, DiPietro, & Ayres, 2000).

## Recent developments

A number of recent findings have changed our view of the blocking phenomenon and the mechanisms assumed to participate in it. As mentioned above, Blaisdell et al. (1999) indicated that responding to the blocked stimulus can recover if the blocking CS is extinguished. Shanks (1985) and Miller and Matute (1996) demonstrated that blocking can be obtained by first presenting the AB reinforced compound followed by the reinforced A element (backward blocking). Furthermore, it has been reported that spontaneous recovery from backward blocking is present following a retention interval, Pineno et al. (2006) showed that backward blocking was attenuated when the blocking association was extensively trained and that forward blocking was attenuated by extensive training of the blocking association. More recently, Beckers et al. (2005) showed in a human causal learning experiment that the strength of blocking and backward blocking increases when the subject knows that (a) A produces a result smaller than the maximal possible outcome (maximality), and (b) the effects of A and B can be added (or results in the maximal possible result (additivity). Beckers et al. (2006) found similar results in rats.

## Recovery from blocking

### Backward blocking and recovery from backward blocking

Whereas in blocking, A-US training precedes AB-US training, in backward blocking AB-US trials precede A-US trials. First reported in causal judgments in humans (Shanks, 1985), Miller and Matute (1996, Experiment 2) demonstrated that, by using a sensory preconditioning procedure in the two first phases of their experimental design, backward blocking could be also obtained in animals. The backward blocking group received presentations of stimuli A and B followed by outcome O, presentations of A followed by O, and presentations of O followed by a US. Control groups received presentations of either C followed by O or just remained in the training cage during the second phase, while the other phases were identical to those in the experimental group. Responding to B was stronger in the controls than in the experimental group. Also using sensory preconditioning, Pineño et al. (2005) demonstrated spontaneous recovery from backward blocking following a retention interval.

## Blocking is attenuated with extended training

Pineno et al. (2006) studied in rats the effect of extending the blocking A-O association following compound training of AB–O, on responding to B. O was a flashing light, which was in turn associated with a footshock US. In Experiment 1, backward blocking was attenuated when the blocking association was extensively trained. Experiment 2 showed that forward blocking was also attenuated by extensive further training of the blocking association following the AB–O trials, replicating an earlier report of this phenomenon (Azorlosa & Cicala, 1988).

## Theoretical approaches

Rescorla and Wagner (1972), Mackintosh (1975), Pearce and Hall (1980), and Wagner (1981) proposed models that are able to describe the original blocking results. Rescorla and Wagner (1972) introduced a rule that assumes that CSs compete to gain association with the US, when the prediction of the US matches the actual value of the US then CS-US associations do not change. Mackintosh (1975) suggested that attention to a given CS increases when that CS is the best predictor of the US, and decreases otherwise. Pearce and Hall (1980) proposed that attention to a given CS decreases when the US is accurately predicted. According to Wagner’s (1981) theory, a stimulus representation can be in one of three states, A1 (active), A2 (active), or I (inactive). An excitatory association between a CS and a US increases when their representations are both in the A1 state. After training, presentation of the CS by itself activates a representation of the US (initially in the I state) into the A2 state. An inhibitory association between a CS and a US increases when the CS representation is in the A1 state and the US representation is in the A2 state, that is, the US is not present but evoked by another CS. Modifications to those above-mentioned models (e.g., Van Hamme and Wasserman, 1994; Dickinson and Burke, 1996), or newer models (Miller and Schachtman, 1985; Schmajuk, Lam, and Gray, 1986) were needed to explain the recent results of retrospective revaluation. Van Hamme and Wasserman (1994) offered a modified version of the Rescorla and Wagner (1972) model that is able to explain some of the new results. They proposed that the association of a CS with the US decreases when the CS is absent and the US is absent but expected, instead of staying constant as in the original model. The Van Hamme and Wasserman (1994) version of the Rescorla-Wagner (1972) rule is able to describe that (a) extinction of the blocking CS results in the recovery of the response to the blocked CS (Blaisdell et al., 1999), and (b) backward blocking (Miller and Matute, 1996). Dickinson and Burke (1996) proposed a revised version of Wagner’s (1981) Sometimes Opponent Process (SOP) theory. Whereas Wagner (1981) suggested that if the representations of two stimuli are in the A2 state no learning occurs, Dickinson and Burke (1996) postulated that in this situation an excitatory association is formed. This association is weaker, however, than that formed when both stimuli are in the A1 state. In addition, whereas Wagner (1981) suggested that if the CS is represented in the A2 state and the US in the A1 state no learning occurs, Dickinson and Burke (1996) postulated that in this situation an inhibitory association is formed between the CS and the US. The Dickinson and Burke (1996) modified SOP model can describe (a) that extinction of the blocking CS results in the recovery of the response to the blocked CS, and (b) backward blocking. According to the “comparator hypothesis” (Miller and Schachtman, 1985; Miller and Matzel, 1988; Denniston et al., 2001; Stout and Miller, 2007), during testing, the target CS generates two representations of the unconditioned stimulus (US): a direct one through its own CS-US association, and an indirect, serial one through CS-Comparator CS and Comparator CS-US associations. The Comparator CS (i.e., another CS, the Context (CX), or both) is the one with which the target CS was trained. When the strength of the direct representation is greater than the indirect one, the potential for excitatory responding is larger than that for inhibitory responding. Conversely, when the strength of the indirect representation is greater than the direct one, the potential for inhibitory responding is larger than that for excitatory responding. The comparator hypothesis has been successfully applied to describing (a) that extinction of the blocking CS results in the recovery of the response to the blocked CS, and (b) backward blocking. Schmajuk, Lam, and Gray (SLG, 1996) proposed a model of classical conditioning that incorporates and extends the properties of several previous models. The SLG model includes (1) a recurrent system that stores CS-CS and CS-US associations and permits the generation of inferences (Schmajuk and Moore, 1988), (2) a real-time attentional variable regulated not only by the novelty of the US, as in the Pearce and Hall (1980) model, but also by the novelty of the CSs and the context (CX), and (3) an extended, real-time, modified version of the Rescorla and Wagner (1972) rule that describes, not only CS-US associations, but also CS-CS associations. As shown by Schmajuk and Larrauri (2006), the model can describe (a) that extinction of the blocking CS results in the recovery of the response to the blocked CS, (b) the reinforced presentations of A results in the attenuation of blocking AND backward blocking, (c) backward blocking, (d) unblocking by decreasing the US intensity, as well as (e) the maximality and additivity properties of blocking and backward blocking. In addition, the model also predicts that the associability of (attention to) the CSs decreases during compound AB presentations, a result supported by Mackintosh and Turner’s (1971) data mentioned before, and also by Holland’s (1985) more recent data.

## References

• Arcediano, F., Escobar, M., & Matute, H. (2001). Reversal from blocking in humans as a result of posttraining extinction of the blocking stimulus. Animal Learning & Behavior, 29, 354-366.
• Azorlosa, J.L., & Cicala, G.A. (1988). Increased conditioning in rats to a blocked CS after the first compound trial. Bulletin of the Psychonomic Society, 26, 254-257.
• Balaz, M.A., Gutsin, P., Cacheiro, H., & Miller R.R. (1982). Blocking as a retrieval failure: Reactivation of associations to a blocked stimulus. The Quarterly Journal of Experimental Psychology B: Comparative and Physiological Psychology, 34B, 99-113.
• Beckers, T., De Houwer, J., Pineno, O. & Miller, R.R. (2005). Outcome additivity and outcome maximality influence cue competition in human causal learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 238-49.
• Beckers, T., Miller, R.R., De Houwer, J., Urushihara, K. (2006). Reasoning rats: forward blocking in Pavlovian animal conditioning is sensitive to constraints of causal inference. Journal of Experimental Psychology: General, 135, 92-102.
• Blaisdell, A., Gunther, L. & Miller, R. (1999). Recovery from blocking achieved by extinguishing the blocking CS. Animal Learning & Behavior, 27, 63-76.
• Denniston, J., Savastano, H. & Miller R. (2001). The extended comparator hypothesis: Learning by contiguity, responding by relative strength. In R.R. Mowrer & S.B. Klein (Eds.), Handbook of contemporary learning (pp. 65-117). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
• Dickinson, A., Hall, G. & Mackintosh, N.J. (1976). Surprise and the attenuation of blocking. Journal of Experimental Psychology: Animal Behavior Processes, 2, 313-322.
• Dickinson, A. & Burke, J. (1996). Within-compound associations mediate the retrospective revaluation of causality judgments. Quarterly Journal of Experimental Psychology, 49B, 60-80.
• Donegan, N.H., Whitlow, J.W. Jr. & Wagner, A.R. (1977). Post-trial reinstatement of the CS in Pavlovian conditioning: facilitation or impairment of acquisition as a function of individual differences in responsiveness to the CS. Journal of Experimental Psychology: Animal Behavior Processes, 3, 162-177.
• Egger, M.D. & Miller, N.E. (1962). Secondary reinforcement in rats as a function of information value and reliability of the stimulus. Journal of Experimental Psychology, 64, 97-104.
• Estes, W.K. & Skinner, B.F. (1941). Some quantitative properties of anxiety. Journal of Experimental Psychology, 29, 390-400.
• Ferster, C.B. & Skinner, B.F. (1957). Schedules of Reinforcement. New York: Appleton-Century-Crofts.
• Gray, T. & Appignanesi, A.A. (1973). Compound conditioning: Elimination of the blocking effect. Learning and Motivation, 4, 374-380.
• Holland, P.C. (1985). Element pretraining influences the content of appetitive serial compound conditioning in rats. Journal of Experimental Psychology: Animal Behavior Processes, 11, 367–387.
• Holland, P. (1999). Overshadowing and blocking as acquisition deficits: No recovery after extinction of overshadowing or blocking cues. The Quarterly Journal of Experimental Psychology, 52B, 307-333.
• Kamin, L.J. (1968). “Attention-like” processes in classical conditioning. In M.R. Jones (Ed.). Miami Symposium on the Prediction of Behavior, 1967: Aversive Stimulation. Coral Gables, Florida: University of Miami Press (Pages 9-31).
• Kamin, L.J. (1969a). Predictability, surprise, attention, and conditioning. In B. A. Campbell and R. M. Church (Eds.) Punishment and Aversive Behavior. New York: Appleton-Century-Crofts (Pages 279-296).
• Kamin, L.J. (1969b). Selective association and conditioning. In N.J. Mackintosh and W.K. Honig (Eds.) Fundamental issues in associative learning. Proceedings of a symposium held at Dalhousie University, Halifax, June 1968. Halifax: Dalhousie University Press, 1969 (Pages 42-64).
• Kehoe, E.J., Schreurs, B.G. & Amodei, N. (1981). Blocking acquisition of the rabbit’s nictitating membrane response to serial compound stimuli. Learning and Motivation, 12, 92-108.
• Kohler, E.A. & Ayres, J.J.B. (1979). The Kamin blocking effect with variable-duration CSs. Animal Learning & Behavior, 7, 347-350.
• Mackintosh, N. (1975). A theory of attention: Variations in the associability of stimuli with reinforcement. Psychological Review, 82, 276-298.
• Mackintosh, N.J. & Turner, C. (1971) Blocking as a function of novelty of CS and predictability of UCS. The Quarterly Journal of Experimental Psychology, 23, 359-366.
• Mackintosh, N.J. (1975). A theory of attention: Variations in the associability of stimuli with reinforcement. Psychological Review, 82, 276-298.
• Maleske, R.T. & Frey, P.W. (1979). Blocking in eyelid conditioning: Effect of changing the CS-US interval and introducing an intertrial stimulus. Animal Learning & Behavior, 7, 452-456.
• Marchant, III, H.G. & Moore, J.W. (1973). Blocking of the rabbit’s conditioned nictitating membrane response in Kamin’s two-stage paradigm. Journal of Experimental Psychology, 101, 155-158.
• McPhee, J.E., Rauhut, A.S. & Ayres, J.J.B. (2001). Evidence for learning deficit versus performance deficit theories of latent inhibition in Pavlovian fear conditioning. Learning and Motivation, 32, 274-305.
• Miller, R.R. & Schachtman, T.R. (1985). The several roles of context at the time of retrieval. In P.D. Balsam & A. Tomie (Eds.), Context and learning. Hillsdale, N.J.: Erlbaum (pp 167-194).
• Miller, R.R. & Schachtman, T.R. (1985). Conditioning context as an associative baseline: Implications for response generation and the nature of conditioned inhibition. In R.R.
• Miller, R. & Matzel, L. (1988). The comparator hypothesis: A response rule for the expression of associations. In G.H. Bower (Ed.), The psychology of learning and motivation (Vol. 22, pp. 51-92). Orlando, FL: Academic Press.]
• Miller, R., and Matute, H. (1996). Biological significance in forward and backward blocking: Resolution of a discrepancy between animal conditioning and human causal judgement. Journal of Experimental Psychology: General, 125, 370-386.
• Pearce, J. & Hall, G. (1980). A model for Pavlovian conditioning: Variations in the effectiveness of conditioned but not unconditioned stimuli. Psychological Review, 87, 332-352.
• Pineño, O., Urushihara, K. & Miller, R.R. (2005). Spontaneous recovery from forward and backward blocking. Journal of Experimental Psychology: Animal Behavior Processes, 31, 172-183.
• Pineño, O., Urushihara, K., Stout, S., Fuss, J. & Miller, R. (2006) When more is less: Extending training of the blocking association following compound training attenuates the blocking effect. Learning and Behavior, 34, 21–36.
• Rauhut, A.S., McPhee, J.E., DiPietro, N.T. & Ayres, J.J.B. (2000). Conditioned inhibition training of the competing cue after compound conditioning does not reduce cue competition. Animal Learning & Behavior, 28, 92-108.
• Rescorla, R.A. (1971). Variations in the effectiveness of reinforcement and nonreinforcement following prior inhibitory conditioning. Learning and Motivation, 2, 113-123.
• Rescorla, R.A. & Wagner, A.R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A.H. Black and W.F. Prokasy (Eds.), Classical Conditioning II: Recent Research and Theory. New York: Appleton-Century Crofts (Pages 64-99).
• Sahley, C., Rudy, J.W. & Gelperin, A. (1981). An analysis of associative learning in a terrestrial mollusc: 1. Higher-order conditioning , blocking, and a transient US pre-exposure effect. Journal of Comparative Physiology, 144, 1-8.
• Schachtman, T.R; Gee, J.; Kasprow, W.J.; & Miller, R.R. (1983). Reminder-induced recovery from blocking as a function of the number of compound trials. Learning and Motivation, 14, 154-164.
• Schmajuk, N., Lam, Y. & Gray, J. (1996). Latent inhibition: A neural network approach. Journal of Experimental Psychology: Animal Behavior Processes, 22, 321-349.
• Schmajuk, N.A., and Larrauri, J.A. (2006). Experimental challenges to theories of classical conditioning: Application of an attentional model of storage and retrieval. Journal of Experimental Psychology: Animal Behavior Processes, 32, 1–20.]
• Shanks,D.R. (1985). Forward and backward blocking in human contingency judgment. Quarterly Journal of Experimental, 37B, 1–21.]
• Smith, B.H. & Cobey, S. (1994). The olfactory memory of honey bee, Apis mellifera: II. Blocking between odorants in binary mixtures. Journal of Experimental Biology, 195, 91-108.
• Smith, B.H. (1997). An analysis of blocking in odorant mixtures: An increase but not a decrease in intensity of reinforcement produces unblocking. Behavioral Neuroscience, 111, 57-60.
• Stout, S.C. & Miller, R.R. (2007). Sometimes-competing retrieval (SOCR): A formalization of the comparator hypothesis. Psychological Review, 114, 759-783.
• Suiter, R.D. & LoLordo, V.M. (1971). Blocking of inhibitory Pavlovian conditioning in the conditioned emotional response procedure. Journal of Comparative and Physiological Psychology, 76, 137-144.
• Sutton, R.S. & Barto, A.G. (1990). Time-derivative models of Pavlovian reinforcement. In M. Gabriel and J. Moore (Eds.) Learning and computational neuroscience: Foundations of adaptive networks. Cambridge, MA: MIT Press (Pages 497-537).
• Terrace, H.S. (1966). Stimulus control. In W.K.Honig (Ed.) Operant Behavior: Areas of Research and Application. New York: Appleton-Century-Crofts (Page 271-344).
• Van Hamme, L. & Wasserman, E. (1994). Cue competition in causality judgments: The role of nonpresentation of compound stimulus elements. Learning and Motivation, 25, 127-151.
• vom Saal, W. & Jenkins, H.M. (1970). Blocking the development of stimulus control. Learning and Motivation, 1, 52-64.
• Wagner, A.R. & Rescorla, R.A. (1972). Inhibition of Pavlovian conditioning: Application of a theory. In R.A. Boakes and M.S.Halliday (Eds.), Inhibition and Learning. London: Academic Press (pp 301-336).
• Wagner, A. (1981). SOP: A model of automatic memory processing in animal behavior. In N.E. Spear & R.R. Miller (Eds.), Information processing in animals: Memory mechanisms (pp. 5-47). Hillsdale, NJ: Erlbaum.

Internal references

• Howard Eichenbaum (2008) Memory. Scholarpedia, 3(3):1747.
• Wolfram Schultz (2007) Reward. Scholarpedia, 2(3):1652.