Learning about multiple attributes of reward in Pavlovian conditioning

Abstract

The nature of the reward representation in Pavlovian conditioning has been of perennial interest to students of associative learning theory. We consider the view that it consists of a range of different attributes, each of which may be governed by different learning rules. We investigated this issue through a series of experiments using a time-sensitive Pavlovian-to-instrumental transfer procedure, aiming to dissociate learning about temporal and specific sensory features of a reward. Our results successfully demonstrated that learning about these different features appears to be dissociable, with learning about the specific sensory features of a Pavlovian unconditioned stimulus (US) occurring very rapidly across a wide range of experimental procedures, while learning about the temporal features of the US occurred slightly less quickly and was more sensitive to parametric disruption. These results are discussed with regard to the potential independence or interdependence of the relevant learning processes, and to some recent neurophysiological recording and brain lesion work, which provide additional means to investigate these dissociations.