On the Modelisation of a Cognitive Process: A Viability Approach

Aubin, J.-P. (1988). On the Modelisation of a Cognitive Process: A Viability Approach. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-88-014

Preview

Text WP-88-014.pdf Download (337kB) | Preview

Abstract

A dynamical description of an abstract cognitive system which should be closer to some cognitive considerations than pure automata or networks of automata is proposed. The system operates "sensory-motor" states, whose components are the state of the environment, its variation and the cerebral motor activity.

The main addition is the introduction of a "conceptual control" that is postulated in order to define a "learning process", which is a set-valued map associating conceptual controls with sensory-motor states.

A learning process must be consistent with a 'recognition mechanism" which determines at each instant the set of possible metaphors, linking the perception of the environment and its variations with conceptual controls, as well as with "viability constraints" describing the consumption of the cognitive system, associating with each state of the environment the set of viable motor activities.

It also has to be consistent with an 'action law", describing the evolution of the state of the environment in terms of the cerebral motor activity, and a "motor activity" law, describing the evolution of the motor activity in terms of the perception of the environment and the conceptual controls. It suggests also that the evolution of a learning process obeys an "inertia principle" which allows to select specific choices of learning procedures.

This paper justifies this approach, which can used to prove mathematically the existence of a largest learning process and of specific "heavy evolutions" obeying an "inertia principle".

Actions (login required)

View Item