Perception, Categories, and Possibilities for Action

Beer’s approach to addressing questions about embodiment, situatedness, and dynamics is to investigate an evolved model agent capable of exhibiting the most basic form of behavior that one might call “cognitive.” One “minimally cognitive behavior” performed by Beer’s agent is the discrimination between circles to catch and diamonds to avoid. Indeed, it is often assumed that the ability to perceive the categories to which objects in the world belong is a basic form of cognition. We wholeheartedly agree that the perception of boundaries separating categories plays a fundamental role in adaptive behavior, and that this topic provides an appropriate focal point for fruitful investigations of minimally cognitive behavior. But we argue that the entities composing the most important categories are possibilities for action rather than objects, and that perception detects these boundaries rather than creates them (Michaels, Prindle, & Turvey, 1985). From an ecological perspective (Gibson, 1979), successful behavior depends on the ability to perceive which actions are possible and which actions are not possible (Turvey, 1992). The Outfielder Problem (Oudejans, Michaels, Bakker, & Dolne, 1996)—that is, the problem of how an outfielder runs to catch a fly ball—provides a convenient example to illustrate this point, and is not unlike the behavior performed by Beer’s agent when it moves to catch a falling circle. When an outfielder runs to catch a fly ball, she must know whether or not it is possible to run quickly enough to reach the landing location before the ball does. If the ball is hit too hard or the fielder is too slow, she may slow down and catch it on a bounce rather than on the fly. The other behavior performed by Beer’s agent (i.e., moving to avoid a falling diamond) is akin to avoiding traffic when crossing the street. When a pedestrian crosses the street, he must know whether or not it is possible to walk quickly enough to pass in front of an approaching car. If the car is approaching too quickly or the pedestrian is too slow, he may wait for the car to pass. In both cases, the situation invites a qualitatively different kind of behavior depending on whether or not the action is possible. This is the sense in which actions are categorical—we either catch the ball on a fly or let it bounce; we either cross in front of the car or wait for it to pass. But the boundaries separating categories are not invented by the organism and imposed on continuous stimuli. Rather, they are defined by the fit of organism and environment such that possibilities for action are real and out there to be discovered. If behavior is to be appropriate for the situation (e.g., if the outfielder is to know whether to catch the ball on the fly or let it bounce), the behavioral possibilities of the environment must be perceived.