Abstract
After adaptation to transparent stimuli moving in different directions, the motion aftereffect (MAE) is perceived in a single direction. This phenomenon has been explained by postulating that direction-selective neurons in an initial stage (identified with V1) excite neurons in a second stage (MT) which have similar direction tuning, and inhibit those which have opposite tuning (Grunewald 1995 NIPS 837; Grunewald & Lankheet, Nature 384:358). The excitation from stage 1 to stage 2 is assumed to be narrowly tuned, allowing two separate peaks of activity to emerge at stage 2, corresponding to the perception of transparent motion during adaptation. However, since the inhibition is assumed to be much more broadly tuned, its distribution — and hence the rebound activity corresponding to the MAE — has only a single peak, so only a single direction of motion is perceived during the MAE. This model, while successfully explaining several aspects of the MAE, takes as input only the directions of motion present in the stimulus. We develop a biologically plausible model of the MAE, aiming to make it compatible with a wide range of psychophysics, electrophysiology and imaging results. Initial processing is based on the known properties of V1 neurons, including tuning to temporal and spatial frequency (Simoncelli & Heeger Nat Neurosci 4:461). Tuning to speed is achieved by combining the outputs of many V1 units. Connectivity following the intersection of constraints rule (Adelson& Movshon Nature 300:523) leads to pattern-selectivity in MT. This enables us to probe the model's behavior with a variety of stimuli. For both transparent moving gratings and random dot patterns, the model successfully predicts the direction(s) of motion perceived during the adaptation period and during the MAE. We are currently extending the model in order to include non-Fourier motion detectors, to better describe transparency changes during plaid adaptation (Wilson & Kim, Vis Neurosci 11:1205–20).
This research was supported by FCT SFRH/BD/13344/2003 and the Programa Gulbenkian de Estimulo a Investigacao.