作者: Dana H. Ballard
DOI:
关键词: Function (mathematics) 、 Neural substrate 、 Parallel processing (DSP implementation) 、 Connectionism 、 Encoding (memory) 、 Theoretical computer science 、 Representation (mathematics) 、 Cortex (anatomy) 、 Computer science 、 Data structure
摘要: The cerebral cortex is a rich and diverse structure that the basis of intelligent behavior. One deepest mysteries function neural processing times are only about one hundred as fast fastest response for complex At very least, this would seem to indicate does massive amounts parallel computation.This paper explores hypothesis an important part can be modeled connectionist computer especially suited problem solving. uses special representation, termed value unit encoding, represents small subsets parameters in way allows access many different parameter values. This thought computing hierarchies sensorimotor invariants. substrate interpreted commitment data structures algorithms compute invariants enough explain behavioral times. A detailed consideration model has several implications underlying anatomy physiology.