Biomedical Engineering Reference
In-Depth Information
5.3
CLINICAL NEURO-CYBERNETICS: MOTOR
LEARNING IN NEURONAL SYSTEMS
Florian P. Kolb
Institute of Physiology, University of Munich, Munich, Germany
Dagmar Timmann
Department of Neurology, University of Duisburg-Essen, Essen, Germany
Although the understanding of complex cerebellar function is still a matter of discussion,
new imaging techniques have provided evidence that the human cerebellum is critically
involved in motor learning. For associative plastic motor processes this evidence has
been obtained by comparison of classically conditioned eyeblink results from cerebellar
patients with those from corresponding control subjects, the former showing typically re-
duced incidence levels of conditioned responses. Of particular interest was that
non-
associative
-motor-related processes such as habituation are also affected characteristi-
cally in cerebellar patients. Aside from these motor-related functions, we also review evi-
dence that the cerebellum may be involved as well in
non-motor
visuomotor associative
learning. The common denominator for impaired function may be an inadequate error-
detection or error-correction capability, a putative function of the olivo-cerebellar system.
In the final section of this chapter, we review computational models based on feedback-
error learning.
1.
INTRODUCTION
1.1. Types of Learning
Changes in the activity of the central nervous system are characteristic and
desirable in developing systems, as has been pointed out in previous chapter 5.2
by Tabak and Rinzel (Part III, this volume), which presents a model for spon-
Address correspondence to: Florian P. Kolb, Institute of Physiology, University of Munich, Pet-
tenkoferstrasse 12, 80336 Munich, Germany (F.Kolb@LMU.de).
507
