Context Effects on Reinforcement

点击翻译按钮获取中文摘要

9306736 Williams Research on animal learning has generally been segregated into the study of Pavlovian conditioning, which involves the association between stimuli and reinforcers, and instrumental (operant) conditioning, which involves the relation between responses and reinforcers. Dr. Williams' recent research has demonstrated striking parallels in how contingency effects operate for the two types of conditioning. He will now extend the analysis to determine in more precise detail whether contingency effects in the two domains depend on the same underlying associative processes. In particular, he asks whether standard measures of the strength of instrumental behavior, rate of reinforcement and choice between different response alternatives, can be explained in terms of the concept of stimulus value as predicted by Pavlovian association theory. This research may revolutionize the conceptual foundations of general theories of learning. The fundamental assumption of learning theory for most of this century has been that "associative strength" is composed of individual increases and decreases in associative value due to individual occurrences of reinforced and nonreinforced behavior. This assumption is challenged by recent findings showing that the animal seems sensitive to the overall rates of reward available in the situation independent of its own behavior. In other words, the animal uses its behavior to estimate these overall reward rates, rather than the reverse causal direction of the reward producing changes in the "strength" of the behavior. At the core of this issue is the viability of associative concepts to provide a reductionistic account of learning. This theoretical issue has ramifications for all of neuroscience, because recent emphases on identifying the neural mechanisms underlying learning and cognition must be predicated on the correct view of the learning processes in order to have any hope of success.