Neural Substrates of Cognitive Processing Architectures

Abstract

To best understand the biological underpinnings of psychological phenomena, one must understand two distinct but related dimensions: (1) the response properties of the underlying neurons and (2) the underlying cognitive processes and their organization (i.e., the cognitive architecture used). Many neurobiological models of behavior are defined in terms of evidence accumulation processes. However, these models often do not consider heterogeneity of response profiles of the neurons, nor do they consider that the behavior could be produced by multiple operations whose architecture must be appreciated. To this end, I recorded single neuron activity in macaque frontal eye field (FEF) understand the system from both directions. First, I developed an algorithm that identifies categories of functional neuron types in a more refined, more assumption-free manner than has been done previously. Then, I developed a behavioral task that explicitly manipulated two factors with two levels each, the first such approach in non-human primates, which allows the inference of cognitive architecture. Finally, I recorded neural activity during this and a related complex search task and identified neural correlates of multiple cognitive operations within the same task. I found that visually responsive neurons in FEF identify behaviorally relevant stimuli but do not participate in the decision to make or withhold eye movements whereas movement-related neurons in FEF integrate the identification of a relevant object with the decision to make an eye movement, ultimately driving oculomotor responses. In sum, I have tested a research approach by which heterogeneity of response profiles and complexity of cognitive architectures are mutually informative, resulting in a more thorough mechanistic understanding of directed behavior.