Feature representations and analytic/nonanalytic processing.

Abstract

Lee Brooks' original formulation of instance theory embedded the notion of an instance within the larger conception of a distinction between analytic and nonanalytic processing. Brooks has recently argued that features can be represented either in terms of their specific feature appearance, or in terms of the abstract information some particular instantiation embodies. This work reviews some recent studies that link reliance on different types of feature representations to different decision-making processes, and to different patterns of categorization behaviour. This in turn complicates the analytic/nonanalytic distinction, suggesting a more precise reformulation.