When the monkey was trained repeatedly in a delayed match-to-sample task with a fixed sequence of visual stimuli, responses of inferior temporal neurons to adjacent stimuli in the sequence are mutually correlated although the monkey was not required to associate the stimuli with each other. There exist two kinds of models accounting for this correlation, but neither model is sufficiently supported by physiological evidence; rather, they seem incompatible with some findings on the perirhinal cortex. We present a different model consisting of two networks corresponding to area TE and the perirhinal cortex that explains the above phenomenon based on a more plausible mechanism, and show that the plasticity in the perirhinal cortex may play a key role in implicit association learning.
Characteristics of motion aftereffect (MAE) following adaptation to two dot-populations moving in orthogonal directions at different speeds are reported. The results of experiments indicate that: (1) when the contrast of test stimuli (randomly moving dots presented after adapting stimuli) is high, the perceived direction of MAE inclines toward the opposite direction of the population which is closer in speed to the test stimuli than the other; (2) when the test contrast is low, the MAE direction is not so much depend on test speed and inclines toward the opposite direction of the population whose speed is more dominant in human sensitivity than that of the other. Simple assumptions based on physiological findings enable us to account for the complex results.
We propose a novel rule extraction algorithm adopting the mathematical model called Basis Pursuit (Chen, et al. 1998), where it is represented by a linear combination of kernel functions (similar to MLP or Support Vector Machines) but gives sparse function representation compared to those models. In this algorithm, a number of logical rules are set to the kernel functions in advance. Applying a linear programming method, we obtain classification rules as a small subset of them. If less logical rules are discovered, they will be the core knowledge in a database. We applied our algorithm to several known benchmark problems and its effectualness is verified.