Making Decisions does not Suffice for Minimal Cognition

In the early nineties, I stumbled upon Beer's Intelligence as adaptive behavior and I have been a fan of his work ever since. Beer has been one of the major players in developing a radical embodied, situated and dynamical perspective on cognition. In this radical perspective, the reciprocal relation between perception and action becomes the primary concern for cognitive theorizing, while the relevance of computation and representation for cognition are questioned. The radical perspective remains very controversial, and one of the major issues on which it is questioned is the cognitive status of basic perception-action couplings. This reaction can go as follows: Perception-action coupling is definitely important for on-line problem solving and action, but thinking, being the prime example of cognition, is something that we do best off-line, preferably reclining in a chair (e. Thus, the more traditional interpretation holds that the radical embodied perspective overstates the importance of perception-action coupling as the notion of cognition itself remains unaffected. How to proceed with this debate? Beer's paper aims to contribute to this debate by the careful study of " idealized models of minimally cognitive behavior " in order to " improve our intuitions, clarify the key issues and sharpen the debate " [p.4]. However, I doubt whether Beer's present study will significantly clarify the issues at stake in the debate about the radical embodied view. In my view, he leaves unclear a key issue in the debate: What do we consider cognition to be? Both parties answer this question differently. The radical embodied view stresses the primacy of physical organism-environment couplings by means of which problems—moving about, feeding, avoiding predators, and so on—can be solved in the actual world. In this view, any system capable of performing such feats of behavior is simply a cognitive system. In contrast, more traditional views stress a capacity for problem solving as the key ingredient of cognition. These problems can range from tasks derived from biology (e.g. sex-ratio decisions in parasitic wasps) to abstract reasoning (e.g. deciding on the next move in a chess match), as long as it involves making a decision that solves the problem in an intelligent (adaptive or rational) way. What counts here is the decision making process. In the traditional view, the physical execution of those decisions is not at the heart of cognition, but assumed to occur in addition to making the proper decisions. Beer's choice of categorical …