ABSTRACT
Traditional learning theory has developed models that can accurately predict and describe the course of learned behavior. These "psychological process" models rely on hypothetical constructs that are usually thought to be not directly measurable or manipulable. Recently, and mostly in parallel, the neural mechanisms underlying learning have been fairly well elucidated. The argument in this essay is that we can successfully uncover isomorphisms between process and mechanism and that this effort will help advance our theories about both processes and mechanisms. We start with a brief review of error-correction circuits as a successful example. Then we turn to the concept of stimulus elements, where the conditional stimulus is hypothesized to be constructed of a multitude of elements only some of which are sampled during any given experience. We discuss such elements with respect to how they explain acquisition of associative strength as an incremental process. Then we propose that for fear conditioning, stimulus elements and basolateral amygdala projection neurons are isomorphic and that the activational state of these "elements" can be monitored by the expression of the mRNA for activity-regulated cytoskeletal protein (ARC). Finally we apply these ideas to analyze recent data examining ARC expression during contextual fear conditioning and find that there are indeed many similarities between stimulus elements and amygdala neurons. The data also suggest some revisions in the conceptualization of how the population of stimulus elements is sampled from.
