In each of three experiments animals received blocking, Aϩ AXϩ, in which food was always presented after one stimulus, A, that was occasionally accompanied by another stimulus, X. They also received a simple discrimination, AXϩ BXϪ, in which the presence and absence of food was signaled by two compounds that contained one unique cue, A or B, and one common cue, X. In each of these designs, X can be said to be redundant relative to A as a signal for food. Test trials at the end of training revealed that responding during X was stronger after blocking than after the simple discrimination. These results contradict predictions from theories of learning that assume changes in associative strength of a stimulus are determined by a global error term based on the outcome predicted by all the stimuli that are present for a conditioning trial. The results are interpreted, instead, by assuming either that animals store a memory of every trial to which they have been exposed, or that learning is governed by an error term based on the significance of individual stimuli.Keywords: blocking, relative validity, appetitive conditioning, redundant cuesIn any conditioning task some conditioned stimuli (conditional stimulus, CS) will be more informative about whether the unconditioned stimulus (US) will be delivered than others. Moreover, it appears that informative stimuli gain more control over behavior than those that are uninformative, or redundant. A classic demonstration of this effect was provided by Wagner, Logan, Haberlandt & Price (1968( , see also Wagner, 1969 in an experiment with two groups. One group received what we shall refer to as a simple discrimination in which the US was presented after one compound CS but not another, with each compound containing a unique and a common cue, AXϩ BXϪ. In this discrimination, A can be regarded as informative because it signals reliably whether the US will be delivered, whereas X can be regarded as redundant as it is less informative than A about the delivery of the US. Subsequent test trials revealed a strong conditioned response (CR) in the presence of A, and a weaker CR in the presence of X. The second group received the same two compounds but they were each paired with the US on half of the trials (AXϩ/Ϫ BXϩ/Ϫ). Here, X and A are equally informative about the delivery of the US, and subsequent test trials revealed a stronger CR during X than after the simple discrimination (see also: Pearce, Esber, George & Haselgrove, 2008). Thus even though X was paired with the US according to the same schedule in both groups, it gained less influence over behavior when it was less informative than equally informative about the trial outcome, relative to the stimuli that accompanied it. This outcome of the experiment is referred to as the relative validity effect.Another demonstration of how the informativeness of a stimulus determines the influence it has over behavior is blocking. In the classic demonstration of this effect by Kamin (1969), a blocking group of rats first received trials in...

Pigeons in two experiments were first trained with a set of simple discriminations of the form AX+ CX-, BY+ DY- where A, B, C, and D were relevant, and belonged to one dimension, and X and Y were irrelevant and belonged to a different dimension. They were then tested with a discrimination of the form AX+ AY- BX-. The experiments revealed that the discrimination between AX+ and BX- was acquired more readily than between AX+ and AY-, which indicates that the original training resulted in the associability of the relevant stimuli being greater than that of the irrelevant stimuli. Experiment 2 revealed that the status of other stimuli from the two dimensions influenced these changes in associability. The associability of X and Y was enhanced by making other stimuli from the same dimension relevant, and the associability of A and B was reduced by making other stimuli from the same dimension irrelevant. The associability of the stimuli is attributed to the attention they are paid.

In the blocking paradigm, subjects receive reinforced presentations of a compound, AX, after reinforced presentations of A alone. Following this training, responding to X is often diminished relative to a control group, which did not receive the prior training with A. Standard associative theories of learning such as the Rescorla-Wagner model (Rescorla & Wagner, 1972) explain this effect by assuming that A and X compete for control over behavior. In contrast, theories such as the comparator hypothesis assume that learning about X is unaffected by the properties of A, but it is the expression of this learning at test that is affected by the blocking manipulation. The aim of the 3 reported experiments was to distinguish between these 2 accounts. According to the comparator hypothesis, devaluing A following blocking should increase subsequent responding to X. In all 3 experiments the blocking effect was found to persist following devaluation of A, providing support for standard associative theories.Keywords: conditioning, blocking, comparator hypothesis Effects such as blocking (Kamin, 1969) have been taken as evidence that cues compete for the control they acquire over behavior. In a typical blocking experiment, pairings of a compound comprising two stimuli, A and X, with an unconditioned stimulus (US) are preceded by pairings of just A with the US (Aϩ then AXϩ). Subsequent responding to X is then found to be weaker than if the original training with A is omitted. According to one influential class of theories, blocking is a consequence of cues competing for a limited pool of associative strength (e.g., Pearce, 1994;Rescorla & Wagner, 1972). The Rescorla-Wagner theory, for example, explains blocking by assuming that the initial training with A allows this stimulus to acquire considerable associative strength, leaving little to be gained by X during compound conditioning. This idea is expressed formally by Equation 1, in which the rate of learning about X, ⌬V X , is determined by the discrepancy between the maximum associative strength that the US is able to support on a given trial, , and the aggregate associative strength of all cues present on that trial, ⌺V. The learning rate parameters ␣ and ␤ represent the salience of X and the US respectively. If pretraining with A results in it gaining high associative strength, then the value of ⌺V will be close to when X is introduced and according to Equation 1 there will be little scope for conditioning with X.An implication of the foregoing explanation concerns the effect on X of presenting A by itself after the blocking treatment. Standard theories of associative learning, including the RescorlaWagner (1972) theory, assume that the associative strength of a stimulus can be modified only if it is physically present. Thus presentations of A, either by itself or paired with the US, after blocking should have no impact on responding to X. In contrast, other theories allow that performance to X can be affected by subsequent manipulations of the associative strength of A.Th...

It has long been acknowledged that discrimination training of the kind AX BX results not only in the stimuli involved acquiring different levels of associative strength, but also in changes in the attention they are paid. Over 50 years ago, Lawrence (1949) proposed that this arrangement results in stimuli that are relevant to the discrimination being paid more attention than are irrelevant stimuli. The discrimination above, where A is relevant because it consistently signals reinforcement (S ) and B is relevant because it consistently signals nonreinforcement (S ), raises the question of whether attention to the former, to the latter, or to both stimuli will be enhanced by this training.At a theoretical level, the answer to this question would appear to be that attention to both cues will be enhanced. Sutherland and Mackintosh (1971) proposed that if a stimulus successfully predicts the outcome of a trial, attention to that stimulus and all other exemplars from the same dimension will increase. Since both S and S meet this requirement, there should be an increase in attention to both. Mackintosh (1975) proposed that attention will increase to a stimulus that is the best predictor of the trial outcome and will generalize to other, similar stimuli. Although the trial outcome was generally regarded as being the presence, rather than the absence, of reinforcement, Mackintosh (1975, p. 288; see also Le Pelley, 2004) acknowledged that his proposals could equally well apply to stimuli that are the best predictors of nonreinforcement.Although there is a growing body of evidence to suggest that attention will be high to stimuli that consistently signal reinforcement (Pearce & Mackintosh, 2010, provide a recent review), there is less evidence showing that attention to stimuli that consistently signal nonreinforcement will also be high. The purpose of the two experiments reported here, therefore, was to determine whether more attention is paid to a stimulus that is a reliable signal for nonreinforcement than to an irrelevant stimulus.If changes in attention occur during discrimination learning, they should be reflected in new learning when the stimuli are used in a different task. However, it is necessary to show that any apparent differences in the associability of these stimuli (the ease with which they are subsequently learned about) are not simply the result of associations acquired during the original training. This problem has been tackled in two main ways. The first involves training subjects, after an initial discrimination has been acquired, with the same stimuli but a different response requirement. In an early experiment of this kind, Lawrence (1949) trained rats to choose between two arms of a maze, which differed in size, brightness, and texture. Only one of these dimensions was relevant, such that approaching one cue (S ) from this dimension was reinforced but approaching the other (S ) was nonreinforced. Subjects were then transferred to a new task where turning left in the presence of one of the cues from t...

Rats were trained in two experiments to find a submerged platform that was situated in one of two of the four corners of a rectangular pool with a curved long wall. Different landmarks occupied two of the corners on every trial and the platform was always situated near a landmark. For the place group in each experiment, the location of the platform was indicated by the shape of the pool and stimuli outside the pool (place cues), but not the landmarks within the pool. For the landmark groups, the landmarks indicated where the platform could be found, but not the place cues. During Stage 2, two of the place cues were relevant, and two of the landmarks were irrelevant, for a new discrimination. The place cues gained better control over searching for the platform in the place than the landmark group when the place cues had initially been relevant by signalling the presence (Experiment 1) or the absence (Experiment 2) of the platform. The results are said to show that animals pay more attention to relevant than irrelevant cues.Many of the discriminations made by animals involve some stimuli that are relevant to the solution, and some that are irrelevant. If the same stimuli are used for a new discrimination, then learning often progresses more readily with the previously relevant than irrelevant stimuli, (e.g. Lawrence, 1949;1950). Furthermore, if relevant and irrelevant stimuli are from different dimensions, and the new discrimination involves novel stimuli from these dimensions, learning often progresses more readily with the novel stimuli that belong to the previously relevant rather than the previously irrelevant dimensions (e.g. Mackintosh & Little, 1969). These differences in conditionability, or associability, have been attributed to animals paying more attention to relevant than irrelevant stimuli during discrimination learning (e.g. Sutherland & Mackintosh, 1971;Mackintosh, 1975). The purpose of the present paper is to explore the generality of the circumstances that promote these changes in attention.One source of support for the claim that animals pay more attention to relevant than to irrelevant stimuli is the intradimensional shift -extradimensional shift (IDS-EDS) effect. Typically, subjects receive a number of discriminations of the form AX+ BX−, AY+ BY−, where A and B represent values from one dimension, say colour, which is relevant to the solution of the discriminations and X and Y represent values from another, irrelevant dimension, say orientation. Once the initial discriminations have been mastered, subjects are transferred to a set of new discriminations involving novel values from the two dimensions.