Modelling mental imagery in the ACT-R cognitive architecture

AbstractI present a novel approach to modelling spatial mental imagery within the ACT-R cognitive architecture. The proposed method augments ACT-R’s representation of visual objects to enable the processing of spatial extent and incorporates a set of linear and affine transformation functions to allow the manipulation of internal spatial representations. The assumptions of the modified architecture are then tested by using it to develop models of two classic mental imagery phenomena: the mental scanning study of Kosslyn, Ball, and Reiser (1978) and mental rotation (Shepard & Metzler, 1971). Both models provide very close fits to human response time data.

Return to previous page