Affordance-based perception-action dynamics: A model of visually guided braking.

Abstract

Behavioral dynamics is a framework for understanding adaptive behavior as arising from the self-organizing interaction between animal and environment. The methods of nonlinear dynamics provide a language for describing behavior that is both stable and flexible. Behavioral dynamics has been criticized for ignoring the animal's sensitivity to its own capabilities, leading to the development of an alternative framework: affordance-based control. Although it is theoretically sound and empirically motivated, affordance-based control has resisted characterization in terms of nonlinear dynamics. Here, we provide a dynamical description of affordance-based control, extending behavioral dynamics to meet its criticisms. We propose a general modeling strategy consistent with both theories. We use visually guided braking as a representative behavior and construct a novel dynamical model. This model demonstrates the possibility of understanding visually guided action as respecting the limits of the actor's capabilities, while still being guided by informational variables associated with desired states of affairs. In addition to such "hard" constraints on behavior, our framework allows for the influence of "soft" constraints such as preference and comfort, opening a new area of inquiry in perception-action dynamics.