Stimulus preexposure effects and associative learning

Psychologists want to understand individual adaptation / the process by which an individual organism interacts with its environment, is changed by the interaction, and is then equipped to behave differently in future interactions. This process (commonly referred to as learning) can be effectively studied in the laboratory with nonhuman animals whose environmental experience can be carefully controlled and whose behaviour can be rigorously recorded and measured. Such studies (using a procedure known as conditioning, introduced by the Russian physiologist Pavlov) have established that much learning can be explained in terms of the formation of links (or associations) between events (or more correctly, between the central representations of events) that occur together in the environment. Learning thus depends on the animal experiencing pairings of stimuli. The associative principle has dominated our view of learning for at least a hundred years. More recently, however, it has become clear that an important form of learning can occur without pairings. It has long been known that simple exposure to a single stimulus will produce a learned change in behaviour (the phenomenon known as habituation), which shows most clearly as the waning of the response initially evoked by the stimulus. But this procedure turns out to have more complex and interesting consequences. Preexposure to a stimulus not only reduces its ability to evoke its normal response, it also changes the ability of the stimulus to function normally when it comes to association formation / association formation proceeds slowly when either of the paired stimuli has been preexposed. In addition, the animal's ability to discriminate between similar stimuli appears to be enhanced when these have been preexposed. The aim of this project is to elucidate the nature of the learning that is responsible for these preexposure effects. The starting point for the experiments is the observation that the effects of preexposure on subsequent associative learning can be modulated by the presence of other events during preexposure / specifically, the effects produced by preexposure differ in the target stimulus is either preceded or followed by some other event during the preexposure phase. What the target stimulus predicts or how well it is itself predicted during preexposure determines the outcome of the procedure. What learning process is responsible for this? The experiments will investigate the hypothesis that in fact two learning processes are involved. One of them (that responsible for the simple habituation phenomenon) produces a reduction in the effective salience of the stimulus during preexposure (as if a loud noise gradually becomes effectively softer with repeated presentation). Our experiments will investigate the suggestion that this process operates best when the target stimulus is presented alone (or with a constant associate) during preexposure, but can be reversed when the target stimulus is predicted unreliably by some other event. The second process is one that changes some aspect of the attention paid to a stimulus (something distinct from salience / it is clearly possible to learn attend to a soft noise, and even, possibly, to learn to ignore a loud one). Our experiments will explore the suggestion that animals learn to attend to a stimulus the consequence of which are uncertain, but learn, eventually, to reduce the attention paid to one followed by inconsistent consequences. Our interest in these exposure-learning processes does not mean that we want to deny the central importance of associative learning itself. To the contrary, it is because we think that associative learning is a mechanism of fundamental psychological importance that we think it necessary to explore in detail the subsidiary learning processes that determine the operation of the associative mechanism.

It is proposed to conduct experiments using laboratory rats and standard classical conditioning procedures to explore the effects of prior exposure to the stimuli. The techniques used will be flavour aversion conditioning (with an LiCl injection as the unconditioned stimulus, US), and conditioned suppression with auditory and visual cues as the conditioned stimuli (CSs) and shock as the US. We will also use shock as a CS in appetitive conditioning procedures in which shock predicts the delivery of food. In addition to assessing the effects of preexposure on conditoning we will also look at its effects on the unconditioned response (UR) elicited by certain stimuli (the flavour quinine evokes a clear UR of neophobia; shock evokes unconditioned suppression). The retardation of associative learning routinely observed after preexposure to a stimulus may be a consequence either of a loss of associability by the stimulus, or of an habituation process that reduces its effective salience (or both). Measurement of the UR can help resolve this ambiguity. The planned experiments will investigate the effects of presenting another event along with the target stimulus during preexposure. There is already evidence available to suggest that both how well the target stimulus is predicted, and how well it predicts another event, can modulate the effects of preexposure. We propose to conduct a systematic survey of the full range of these effects, using our two, well investigated, training procedures. We will investigate the effects of these different forms of preexposure both on the ability of the target stimulus to function as a CS, and on its ability to function as a US. In addition, as noted above, we will monitor the effects on the ability of the stimulus to evoke its UR. It may be premature to speculate in advance of the results of these experiments, but our tentative hypothesis at this stage is that the explanation of preexposure effects is likely to require postulation of changes in (at least) two parameters, one reflecting the effective salience of the stimulus, the other its associability. We anticipate that different learning rules will apply in each case, with salience change depending on how well the stimulus is predicted during preexposure and associability change depending on how well the stimulus predicts its consequences. We acknowledge, however, that the picture may be substantially more complicated than this (that for instance the rules that determine the functioning of a stimulus as a CS may differ from those that determine its functioning as a US). The ultimate aim is to produce an account of the learning processes involved in exposure learning that can be 'bolted on' to our account of associative learning to produce a more comprehensive theory of learning. But the exposure learning processes will be of interest in their own right. In particular, an understanding of how the effective salience of the various components of a complex stimulus might change during preexposure would provide the basis for an account of perceptual learning and the process of representation formation more generally.