Abstract
We proposed a spatiotemporal model of simple cell based on resent physiological findings and indicated its capability of detecting velocity of moving image. The model consists of several simple cells mutually connected and four types of LGN cells (spatial type, on/off center and temporal type, nonlagged/lagged). One simple cell connects to four types of LGN cells which are spatially placed in a form of periodical band. The periodical band is shifted in position among simple cells. The spatial receptive-field profile of simple cell can be approximated to Gabor function and the shift correspond to phase difference of the function. We showed that the model has selectivities in spatial frequency, orientation and velocity of stimulus and the responces agree with the physiological experimental data. Then, we proposed a network consisting of simple cell models of different optimal spatial frequencies and indicated that the network can detect velocity of drifting grating stimuli.
