T

Fig. 23 Types of cues in an experiment of visual learning in the monkey (Zhang and Barash 2000). (a) The monkey was trained to make a brief, rapid eye movement (a saccade) either toward the stimulus (prosaccade) or in the opposite direction (antisaccade). This was done in the course of the analysis of neuronal correlates of visual decision making and performance in the posterioparietal *cortex, but for our purpose, only the behavioural part of the experiment will be mentioned. (b) The monkey was placed in front of a computer screen, and trained in successive stages ('shaped') to fixate on a stimulus at the centre of the visual field.A coloured stimulus, either red (full circle) or green (open circle), was then presented for 250 ms away from the fixating point. The colour was the 'target cue', instructing the monkey which type of saccade should be performed at the later part of the task (definitions 1 and 2). After an additional working memory interval of 1 s, the fixation stimulus disappeared. This signalled to the monkey to make the response (the 'GO' cue, definitions 1 and 3): prosaccade if the target stimulus was red, antisaccade if it was green. A proper response was rewarded (*reinforced) with juice. (Adapted from Zhang and Barash 2000.)

control by a conditioned stimulus (CS) due to the presence in training of another conditioned stimulus, that is more salient or is a better predictor of the unconditioned stimulus (US) or the "reinforcement. Two notable types of 'cue competition' are overshadowing and blocking.

Overshadowing is an increase in the associability of the more salient or more predictive of two or more stimuli present at the same time, at the expense of the other stimuli. Pavlov (1927) noted that when a conditioned salivary reflex was established in a dog to a simultaneous application of two stimuli, the more salient stimulus came to evoke salivation when presented alone, whereas the less salient stimulus was ineffective by itself. His conclusion was that 'When the stimuli making up the compound act upon different analysers, the effect of one of them when tested singly was found very commonly to overshadow the effect of the others' (ibid.). An example of overshadowing in instrumental conditioning was provided by Pearce and Hall (1978): two groups of rats were trained to press a lever for food. In one group, the delivery of the food was always preceded by a brief flash of light. In the other group, the light was uncorrelated with the food. In the test, the rats that had experienced the light-food correlated schedule pressed the lever at a slower rate than those trained under the uncorrelated light-food schedule. The authors concluded that pairing the light with food retarded the development of normal instrumental conditioning, because the light was a better predictor of the food, and therefore the stimulus-reinforcer association overshadowed the formation of the response-reinforcer association. The literature contains many other examples of overshadowing in classical and instrumental conditioning.

Blocking is inhibition of the conditioning to a stimulus, CS1, in a compound CS1+ CS2 stimulus, by previous pairing of CS2 with the US (Kamin 1968). In a classical experiment, Kamin (ibid.) used a conditional emotional response procedure ("fear conditioning). Rats were trained to press a bar for food, and upon mastering the operant task, were further conditioned to associate a CS with a fear-inducing US (electric shock). The fear response resulted in suppression of bar pressing in response to the CS. This was observed regardless of whether the CS was light, white noise, or a compound stimulus of both. But in the consecutive phases of the experiment, an interesting phenomenon emerged. When the rats were first conditioned to fear the noise, then conditioned to the compound stimulus, and then tested on the light, they showed essentially no fear response to the light. A similar phenomenon was noted when the animals were first conditioned to the light, then to the compound stimulus, and then tested on the noise. Thus prior training to an element of the compound stimulus blocked conditioning to the new, superimposed element. The interpretation is that in order for an association between a CS and a US to be formed, the US must "surprise the animal. If the animal already knows that CS2 predicts the US, addition of CS1 does not involve much additional surprise, and therefore association of CS1 with the US is weak. (On the role of the unexpected in learning, see also "algorithm; on additional aspects of blocking phenomena; see Miller and Matute 1996; Holland 1988.)

The analysis of 'cue competition' has contributed to the compelling evidence that even apparently 'simple' forms of classical conditioning are actually manifestations of rather complex information processing in the brain, which involves interactions among the "internal representations of the associated stimuli. Hence in this respect, the cues have provided a valuable cue to the mechanisms underlying conditioning.

Selected associations: Attention, Clever Hans, Context, Recall, Stimulus

0 0

Post a comment