INTRODUCTION

One of the more celebrated conclusions in cognitive psychology refers to the limited computational capacity of controlled thought, as typically epitomized by Miller's (1956) estimate of a short-term-memory holding capacity of “seven-plus-or-minus-two” chunks. That people can only keep a limited amount of information active for controlled processing at any moment in time has inspired humbling conclusions in regard to problem solving (Newell & Simon, 1972), reasoning (Evans, Newstead, & Byrne, 1993), and, perhaps especially, judgment and decision making (Gilovich, Griffin, & Kahneman, 2002). This limitation is often raised as a main obstacle to people's attainment of classical rationality, suggesting that at best people can aspire to bounded rationality (Simon, 1990). In the context of this volume the implication is that at any moment in time controlled processes of thought can only access a small sample of observations.

The default interpretation seems to be to emphasize the liabilities of these limitations and to regard the current state in the evolutionary development as representing at best a local maximum. Organisms are thus restricted to limited samples of information, although there is agreement that on normative grounds as much information as possible is needed to optimize judgments and decisions. More rarely is the question raised of whether there can be a functional significance attached to apparent cognitive limitations.