Movement and thought: identical control mechanisms by the cerebellum

摘要: During the past three decades, knowledge of cerebellum has increased markedly, and research yielded important new concepts such as micro- zones cerebellar cortex, synaptic plasticity motor learning. Furthermore, involvement in a certain category mental functions become apparent based on anatomical pathophysiological findings proposed keynote article by Leiner, Leiner Dow this issue. I support proposal from viewpoint control systems: acquires 'dynamics' or 'in- verse dynamics' object through repeated exercise, using 'dynamics memory' it automates quick, precise smooth learned movement. This system principle applies primarily to movement, but is general enough apply any type neural including that functions, thought. Here, trace development theories up problems. The fundamental functional module cer- ebellum corticonuclear microcomplex (CNMC), consisting small area cortex (microzone), group vestibular nuclear neurons inferior olive 1. These structures are interconnected way major signal flow mossy fiber pathway (m Fig. 1A) modulated its sidepath microzone, modulation modified according error signals (c mediated neurons. As basis for error- driven adaptation CNMCs, long-term depression (LTD) occurs synapses Purkinje cell, which active at time when arrive 2. An argument been raised whether not additional accounts memory storage within CNMC 3'4. However, does affect basic idea cerebellum, LTD plays an essential role learning mechanism. most form reflex, can be either autonomic motor. vestibulo-ocular reflex (VOR) generates eye movements compensate head move- ments, so minimize retinal errors during VOR adaptively artificially amplified means prism lens goggles rotation visual surrounds various combinations with rotation. adaptability abolished lesions flocculus. In our model s, CNMC, flocculus, organs relaying VOR, constitutes controller acting upon includes brainstern oculomotor system, muscles ball (Fig. 1B). viewed modifiable controller. Its performance perfect output velocity, equal input, sign reversed: achieved attains dynamics (llg) inversely (g). A also developed voluntary touching one's nose finger. fact normal subject bring his finger quickly accurately eyes closed, i.e. without feedback (finger-to-nose test), suggests process converts mode feedforward mode. Since patients injury fail finger-to-nose test, appears situated cerebellum. Based loop connection between paravermal part cerebral thought while (CX 1C) receiving instruction movement acts skeletomuscular referring feed- back (eft), develops, practice, (g') equivalent g Thence, possible instead actual external replaced internal other areas ebellar cortices connected parallel fashion third step modeling thus involves assumption controller, same 6 1 D). When feeclforward reproduces, arm trajectory in- structed trajectory, should bear (l/g) limb instructed fed into inverse forms identical given instruction. It interesting note here primitive shown lB. what learnt these models dynamics, individual actually practiced. simu- lation study Kawato et a/. demonstrated after practice particular combination two 1C D, robot will trajectories directions smoothly. propose term learning' expressing manner circuitry retains (either not) 448