Abstract
Neurons in the MST area are sensitive to a continuum of combinations of rotational and translational motions. However, the receptive field structure of these neurons is still not well understood. Here, we investigated how local motions are integrated to compensate the noise in complex motions. We estimated the sensitivity to the rotational and to the radial components when the individual trajectories in a spiral pattern were randomly perturbed. Two conditions were tested _ in the first each element had a fixed trajectory, while in the second the amount of trajectory perturbation varied on every frame. The width of the noise distribution added to the individual trajectories varied between 2° and 25°. We compared the discrimination thresholds for identifying the radial component in a spiral pattern as inward or outward and the rotational component as clockwise or counter-clockwise. The results indicate that when noise in successive frames is uncorrelated, the sensitivity to motion direction was higher than when each dot moves on a fixed trajectory. When the noise level was not too high, the discrimination thresholds were constant. Taken together, these results indicate that the integration of local motion in space and time reduces the effect of noise. However, the rotational discrimination thresholds exceeded significantly the thresholds for the radial motion and were less affected by the perturbation of the individual trajectories. This suggests that the integration of local motion is not uniform in all directions and/or that the local sensors are not independent. The reduced sensitivity to the rotational component of motion may be explained if motion sensors with opposite directional preferences integrate the flow information along circles centered at the preferred heading direction.
This study was funded by grant TK01-200 of National Science Fund, Bulgaria.