Abstract
Most animals including humans can judge whether an approaching object will hit them, but neural correlates for time-to-collision estimation in primates are still unknown. A previous model study suggested that time-to-collision estimation could be derived in MST using optic flow. In this study, we searched for neural correlates in the dorsal MST and caudal FEF. We examined discharge of single neurons using a Julesz pattern (20° × 20°) combined with a 0.5° single spot. For collision stimuli, a Julesz pattern of 5° × 5° visual angles was first presented on a computer display placed 50cm ahead of the monkeys' eyes. After 0.5-1s, the pattern expanded to simulate its approach to the monkeys at constant velocities of 13-120cm/s from a simulated distance of 75cm for MST neurons (100cm for FEF neurons) until the collision, while the monkeys fixated on a stationary spot on the screen center. We recorded 25 neurons in the dorsal MST that responded to collision stimuli. Of these, 13 neurons responded continuously during collision stimuli indicating that they simply responded to components of the optic flow (group A). The remaining 12 neurons exhibited peak discharge before the time of collision (group B). Time-to-collision of group B neurons depended on the approaching pattern velocity, whereas distance-to-collision was nearly constant, suggesting that group B neurons signaled distance-to-collision. In contrast, most caudal FEF neurons that responded to collision stimuli (8/11) exhibited peak discharge almost at the time of collision, and the peak discharge time remained constant irrespective of the approaching pattern velocity. Most of these neurons (7/8) were pursuit-related. These results suggest that dorsal MST and caudal FEF have different roles in estimating time-to-collision from visual motion.
