Motion selectivity in macaque visual cortex. II. Spatiotemporal range of directional interactions in MT and V1.

Abstract

We measured the spatial and temporal limits of directional interactions for 105 directionally selective middle temporal (MT) neurons and 26 directionally selective striate (V1) neurons. Directional interactions were measured using sequentially flashed stimuli in which the spatial and temporal intervals between stimuli were systematically varied over a broad range. A direction index was employed to determine the strength of directional interactions for each combination of spatial and temporal intervals tested. The maximum spatial interval for which directional interactions occurred in a particular neuron was positively correlated with receptive-field size and with retinal eccentricity in both MT and V1. The maximum spatial interval was, on average, three times as large in MT as in V1. The maximum temporal interval for which we obtained directional interactions was similar in MT and V1 and did not vary with receptive-field size or eccentricity. The maximum spatial interval for directional interactions as measured with flashed stimuli was positively correlated with the maximum speed of smooth motion that yielded directional responses. MT neurons were directionally selective for higher speeds than were V1 neurons. These observations indicate that the large receptive fields found in MT permit directional interactions over longer distances than do the more limited receptive fields of V1 neurons. A functional advantage is thereby conferred on MT neurons because they detect directional differences for higher speeds than do V1 neurons. Recent psychophysical studies have measured the spatial and temporal limits for the perception of apparent motion in sequentially flashed visual displays. A comparison of the psychophysical results with our physiological data indicates that the spatiotemporal limits for perception are similar to the limits for direction selectivity in MT neurons but differ markedly from those for V1 neurons. These observations suggest a correspondence between neuronal responses in MT and the short-range process of apparent motion.