Grant: Discovery Grants Program - Individual

Association formation contributes to the survival and fitness of both humans and nonhumans. It refers to the learning of relationships about different features of the environment when they occur closely together in either time or space. When an infant learns to discriminate the end of one word from the beginning of another word, or when an animal learns to reject a particular food because it is associated with past illness, an inter-event connection has been formed. Our research is directed at revealing the conditions under which associations are formed in order to create precise mathematical models of the processes involved. Past experiments from many laboratories, including our own, suggest that association formation is a complex process which does not always produce a result consistent with the actual statistical association between the events in question. We study these departures from statistical normality in conditioning experiments with rats and in timing and categorization tasks with human volunteers. Our research has shown that, under special circumstances, some events may become strong associates despite the relatively weak statistical relationship between them. These observations have encouraged the development of neuron-like mathematical accounts of the associative learning process, called error-correction theories. Further tests of the predictions of these models should reveal new phenomenon and lead to the development of even better theoretical accounts.