五味裕章
中枢神経系における運動の適応・学習メカニズムモデルに関する研究
Abstract:In biological systems such as human beings and monkeys, many kinds of sophisticated limb, finger and eye movements are performed by complex neural processing in the central nervous system (CNS). The purpose of this study is to reveal an basic mechanism of such information processing for several kinds of movements, and to make computational models for this processing.
This paper examines the "internal model control hypothesis" as a fundamental motion control mechanism in the brain from behavioral, computational and physiological view points. Experimental results of measuring arm-stiffness during point to point movements are first shown to criticize the "virtual trajectory control hypothesis" which is considered an alternative to the "internal model control hypothesis". Then physiological, clinical and neuroanatomical data are joined to emphasize the hypothesis that internal models of controlled objects (e.g., eye and limbs) should be acquired in the cerebellum so as to smoothly and desirably control these objects. To support this hypothesis by experimental results, the relationships between eye movements and cerebellum Purkinje cell activities were analyzed. Moreover, several computational schemes for acquiring the internal models of controlled objects (e.g., inverse dynamics models) and mechanisms for utilizing these internal models are examined by using artificial neural networks, and some simulation results are shown.