A self-organizing neural network model of spatio-temporal visual receptive fields is proposed. It consists of a one-layer linear learning network with multiple temporal input channels, and each temporal channel has different impulse response. Every weight of the learning network is modified according to a Hebb-type learning algorithm proposed by Sanger. It is shown by simulation studies that various types of spatio-temporal receptive fields are self-organized by the network with random noise inputs. Some of them have similar response characteristics to X- and Y-type cells found in mammalian retina. The properties of receptive fields obtained by the network are analyzed theoretically. It is shown that only circularly symmetric receptive fields change their spatio-temporal characteristics depending on the bias of inputs. In particular, when the inputs are non-zero mean, the temporal properties of center-surround type receptive fields become heterogeneous and alter depending on the positions in the receptive fields.

HASP, a model of Human Associative Processor, has been proposed in a previous paper. Among other associative memory models hitherto proposed, HASP is unique in its multiple match resolution. By this function several items associated with one particular key can be retrieved one by one, and set-theoretically defined multiple key search operations can be performed. However, there was a problem in the performance of HASP. It was the occurance of unknown patterns in a recollection sequence of multiply matched items. It would become a serious defect in appling HASP to such a task as serial association, in which an output became a key input for the next association. In this paper, the origin of the problem is identified and HASP is improved not to produce unknown patterns. The idea is to weaken the competitive powers of the unknown patterns by modifying the strength of inhibitory recurrent paths. A series of simulation studies has been carried out and it is confirmed that the improved HASP can perform multiple key search operations without being contaminated by unknown patterns. Several psychological implications of the performance of HASP are also discussed.

This paper describes a series of experiments on spatial frequency adaptation. The adapting stimulus was a vertical square-wave grating which was moved right and left with constant velocity at the back of a stationary vertical slit. By setting the slit width narrower than a half cycle of the grating, adapting stimuli can be confined to only single edges and bars which repeatedly appear in the slit. Even under this slit vision condition, when adapting frequency was low (0.3, 0.5 and 0.75 c/deg), threshold elevations occurred in the same way as without the slit, where several cycles of the adapting grating could be seen. The position of the largest peak in the elevation profile corresponded to the fundamental frequency of the adapting grating, and the second peak always appeared at a frequency higher than 2 c/deg. The second peak corresponds to a higher harmonic frequency in the adapting stimulus and 2 c/deg corresponds to the lowest adaptable frequency of a sustained system.