Modelling behaviour: the need for a computational approach

Abstract

The principal objective of this paper is to argue the thesis that a science concerned with the study of behaviour requires the computational approach in a serious way for its theoretical advancement. It is pointed out that modelling behaviour requires the articulation of explanations at three levels. The methodology of computational simulations is indispensable to articulating explanations at the first level which underlie explanations at the other two levels. The paper contrasts the currently fashionable approaches in artificial intelligence studies to the kind of constraints viable behavioural models must satisfy. Teachability and open-endedness are two of the essential characteristics of organisms that any satisfactory model must have. It is argued that analogy-based computational techniques and paradigmatic learning/teaching techniques are two modelling aspects that require imaginative study.