The coding of velocity of movement in the human visual system.
A very simple model of velocity perception which requires only 17 channels is outlined. The important points of the model are that: (1) in each direction of movement just two temporal frequency channels are necessary at any spatial frequency, (2) at low temporal frequencies the spatial frequency domain is encoded by many channels, but only those at low spatial frequencies are direction-specific. Using a detection/discrimination technique the supposition that channels which detect high spatial, low temporal frequencies are not direction specific is investigated. Possible reasons for the apparent nondirectional behaviour of these channels are investigated: the notion that non-directionality reflects a failure of the stimulus to travel some threshold distance across the retina is rejected, but the proposal that a velocity threshold must be exceeded before the direction of a grating may be identified at detection threshold remains a possibility.