Abstract

We discuss the nature of cognitive representations and present a scheme for the encoding of information which accounts for both categorical and graded aspects of cognitive events. Accordingly, object categories are mapped onto the identity of cell populations via the process of categorical perception, while graded aspects such as vividness and con dence level are mapped onto the rate and degree of synchrony of neural populations. We show that the process of categorical perception can solve the problem of reference of perceptual-cognitive states, and we demonstrate by computer simulation models that synchrony can encode for grouping and binding of coherent information. In these models, recurrent connections play a major role in generating synchronised activity in cell populations which receive consistent sensory stimulation, and in generating temporal uctuations in their discharge rate. We suggest that such uctuations underlie autonomous computations re ecting uctuations in the certainty (con dence level) of perceptual and cognitive hypothesis.

Citations