Computer experiments in motor learning

Abstract

Motor learning can be defined as the information processing adaptations necessary for a system to achieve advantageous control over its relationship to its physical environment. For convenience, such learning is divided into two major processes: (1) perfection of the subsystem organization necessary to convert highly coded command sequences into the multitudinous signals necessary to the efficient utilization of musculature or other effectors, etc., and (2) the development of the control capabilities necessary for the timely generation of effective coded command sequences. This paper considers only the latter process as it might possibly be achieved in so-called "artificially intelligent" systems such as those using electronic data processing system control, as exemplified by a program described in detail in reference 1. Because of time and space limitations, this program will only be briefly summarized here to the extent necessary to explain the significance of the results obtained with various simple systematic creative means of response generation or synthesis.