Simple Inference Heuristics versus Complex Decision Machines

Deep thought has long been considered a virtue. For centuries, the philosophers of antiquity have been seen as paragons of their civilizations; today, we similarly revere our judges and justices who ponder present facts and past precedents to reach wise and fair conclusions. Aristotle named reason and the practical application of intelligence as virtues, and Descartes held thought to be the very definition of existence (Sahakian, 1968); recently, active thinking has been proposed to improve existence, actually increasing health and longevity (Holden, 1998). The ability to deliberate and ponder one’s actions is also often seen as a defining characteristic solely of human existence, distinguishing us from the more-reflexive animals. Even when this view is challenged, as it has been increasingly of late by primate research (e.g., Whiten and Byrne, 1998 — but see Heyes, 1998 for counter-arguments), the challenges serve primarily to increase our estimation and respect for the other lucky species shown to be capable of any intricate cogitation that rivals our own. We deep-thinking humans have also striven to emulate deep thought in the machines we build, creating artificial intelligences and decision tools that mimic or enhance the human mind. Expert systems, data-mining software, artificial theorem provers, and chess-playing computers all perform elaborate calculations or process great amounts of information in an attempt to approach and sometimes exceed human decision-making power. These complex decision machines make up one focus of this special issue of Minds and Machines. And because the human mind is often viewed as a kind of computer itself (Gigerenzer and Goldstein, 1996b), many would include it in this complex-cogitator category as well. But might there not be an alternative to all this heavy thinking? After all, humans are also renowned for making snap judgments, jumping to conclusions, and ignoring the evidence. Are we always wrong to do so, when we could ponder deeply instead? Recently, research has begun to suggest that simple inference mechanisms or heuristics can be surprisingly useful (e.g., Payne et al., 1993; Gigerenzer, Todd, and the ABC Research Group, 1999) — ‘fast and frugal’ thinking may be