Affordance-Based Control of Visually Guided Action

The purpose of this essay is to compare and contrast existing theoretical approaches to understanding the visual guidance of action and to introduce a new approach. The focus is on tasks, such as steering, braking, and intercepting, that are (more or less) continuously guided on the basis of visual information. The prominent approach, information-based control, captures important aspects of behavior but is incompatible with the theory of affordances, a core principle of the ecological approach. Information-based control also fails to capture how actors behave in ways that take the limits of their action capabilities into account. I attempt to resolve these problems by introducing a new approach, affordance-based control, which asserts that a primary function of vision is to allow actors to see the world in terms of what they can and cannot do. Affordance-based control captures the tight coupling between information in optic flow and movement that is characteristic of visually guided action but also provides a parsimonious explanation of how actors take into account the dynamic properties of their body and the environment. Notes 1More recently, CitationMcBeath, Shaffer, and Kaiser (1995) derived an alternative model of fly ball catching, which predicts that fielders change both speed and direction so as to cancel the optical curvature of the ball. The strategy that fielders actually use is the focus of ongoing debate (CitationMcLeod, Reed, & Dienes, 2001; CitationShaffer, McBeath, Roy, & Krauchunas, 2003; CitationZaal & Michaels, 2003). 2Stoffregen (2002) made an analogous argument about time-to-contact and τ. He pointed out that time-to-contact refers not to an affordance but rather to the kinematics of an event involving impending collision. It is only when time-to-contact is taken with reference to an action-relevant property of the animal that it could refer to an affordance. This has led to ambiguity regarding the role of affordance perception in interceptive actions (such as catching and hitting). 3This does not necessarily mean that stationary and moving fielders will be equally accurate in their perception of catchableness. Because information about ideal running speed must be calibrated, and because calibration can drift over time, fielders who are restricted from moving for long periods of time may be less accurate. Such inaccuracy would result in a systematic bias to overestimate or underestimate catchableness that could change over time.