This paper contributes to the emerging literature at the border between AI and cognitive neuroscience, analyzing the resource constraints that shape brain function. The brain is conceptualized as a set of ar-eas that collaborate to perform complex cognitive tasks. Both (1) computation within individual areas and (2) communication between collaborating ar-eas are viewed as resource-consuming activities. The efficient deployment of limited resources is formalized as a Linear Programming problem which the brain is hypothesized to solve on a mo-ment-by-moment basis. A model of language proc-essing is analyzed within this framework and found to exhibit resource utilization profiles consistent with those observed in functional neuroimaging studies of humans.
Subjects: 2. Architectures; 4. Cognitive Modeling
Submitted: Oct 16, 2006