ABSTRACT
We are developing a computer system, using a Xerox LISP machine linked to an Encore computer, for representing our idea of a movement schema. We recount the motivation for this work, describe work in progress to develop this computer system, and sketch a broader context toward which we shall direct this research.
Our current approach to simulating actions springs from a desire (Greene, 1959) to study the structure of perception, motor skills, and thinking in persons and future computers. We have tried to model on a digital computer some features of a baby's sensorimotor development. The baby becomes able to recognize and manipulate objects, taking into account their movements and other spatial relationships, and performs purposeful actions. These actions are more naturally described in terms of their effects on the world than in terms of muscle movements because totally different combinations of muscle contractions might produce the same desired effect. By working between the physiological level (which is not cognizant of task-level considerations) and the usual symbolic programming level (which often manipulates structures that have already been formed elsewhere), we hoped to learn how the sensorimotor structures themselves come into existence.
