Heuristics and Biases (Kahneman & Tversky)

The Heuristics and Biases Approach

Inspired by the examples of biased real-world judgments of the sort identified by Meehl and his peers, and guided by the clear normative theory explicated by Edwards and others, Kahneman and Tversky developed their own perspective on bounded rationality. Although acknowledging the role of task complexity and limited processing capacity in erroneous judgment, Kahneman and Tversky were convinced that the processes of intuitive judgment were not merely simpler than rational models demanded, but were categorically different in kind. Kahneman and Tversky described three general-purpose heuristics – availability, representativeness, and anchoring and adjustment – that underlie many intuitive judgments under uncertainty. These heuristics, it was suggested, were simple and efficient because they piggybacked on basic computations that the mind had evolved to make. Thus, when asked to evaluate the relative frequency of cocaine use in Hollywood actors, one may assess how easy it is to retrieve examples of celebrity drug-users – the availability heuristic piggybacks on highly efficient memory retrieval processes. When evaluating the likelihood that a given comic actor is a cocaine user, one may assess the similarity between that actor and the prototypical cocaine user (the representativeness heuristic piggybacks on automatic pattern-matching processes). Either question may also be answered by starting with a salient initial value (say, 50%) and adjusting downward to reach a final answer (the anchoring and adjustment heuristic, whose underlying mechanisms are debated in Chapters 6 and 7).

In the early experiments that defined this work, each heuristic was associated with a set of biases: departures from the normative rational theory that served as markers or signatures of the underlying heuristics. Use of the availability heuristic, for example, leads to error whenever memory retrieval is a biased cue to actual frequency because of an individual’s tendency to seek out and remember dramatic cases or because of the broader world’s tendency to call attention to examples of a particular (restricted) type. Some of these biases were defined as deviations from some “true” or objective value, but most by violations of basic laws of probability. (For elegant examples, see Chapter 1 by Tversky and Kahneman).

Several aspects of this program are important to note at the outset because they set the stage for a discussion of the criticisms it aroused. First, although the heuristics are distinguished from normative reasoning processes by patterns of biased judgments, the heuristics themselves are sensible estimation procedures that are by no measure “irrational.” Second, although heuristics yield “quick and dirty” solutions, they draw on underlying processes (e.g., feature matching, memory retrieval) that are highly sophisticated. Finally, note that these heuristic processes are not exceptional responses to problems of excessive complexity or an overload of information, but normal intuitive responses to even the simplest questions about likelihood, frequency, and prediction.

The Positive and Negative Agendas. As the preceding discussion implies, Kahneman and Tversky distinguished between two messages or agendas for the heuristics and biases program, one “positive” and one “negative.” The positive agenda is to elucidate the processes through which people make a variety of important and difficult real world judgments. Is a corporation’s explosive growth likely to continue? Is a coup more likely in Ecuador or Indonesia? What is a reasonable estimate of next year’s GNP? Thus, representativeness, availability, and anchoring and adjustment were proposed as a set of highly efficient mental shortcuts that provide subjectively compelling and often quite serviceable solutions to such judgmental problems.

But, the solutions were just that – serviceable, not exact or perfectly accurate. Thus the second, negative, agenda of the heuristics and biases program was to specify the conditions under which intuitive judgments were likely to depart from the rules of probability. When, in other words, are everyday judgments likely to be biased? Kahneman and Tversky’s experience teaching statistics and their observations of predictions made in applied settings led them to conclude that people often fail to anticipate regression to the mean, fail to give adequate weight to sample size in assessing the import of evidence, and fail to take full advantage of base rates when making predictions. Their three (now familiar) heuristics were offered as an explanation of the when and why of such errors. Thus the two agendas blend together: Identifying particular biases is important in its own right, but doing so also sheds light on the underlying processes of judgment. (Kahneman & Tversky, 1982b, also offered positive and negative approaches to judgment errors, a perspective that is taken up by Kahneman & Frederick, Chapter 2).

Automatic or Deliberate? There is another dichotomous aspect of the heuristics and biases approach that warrants discussion. Heuristics have often been described as something akin to strategies that people use deliberately in order to simplify judgmental tasks that would otherwise be too difficult for the typical human mind to solve. This use of the term fits with the “cognitive miser” metaphor that proved popular in the field of social cognition (Fiske & Taylor, 1991). The metaphor suggests, perhaps unfortunately and unwisely, that the biases documented in the heuristics and biases tradition are the product of lazy and inattentive minds. The implication is unfortunate and potentially misleading because the biases identified in this tradition have not been appreciably reduced by incentives for participants to sit straight, pay attention, and devote their full cognitive resources to the task (Camerer & Hogarth, 1999; Grether & Plott, 1979; Wilson, Houston, Etling, & Brekke, 1996; see Chapters 37 through 42 for descriptions of real-world judgments characterized by a high level of domain-specific expertise and motivation that nonetheless fit the patterns described by the heuristics and biases program). After reviewing 74 studies, Camerer and Hogarth (1999) concluded that incentives can reduce self-presentation effects, increase attention and effort, and reduce thoughtless responding, but noted that “no replicated study has made rationality violations disappear purely by raising incentives” (p. 7).

Imperviousness to incentives is just what one would expect from considering the other half of the dichotomy, or the other way that heuristics have been described. In particular, Tversky and Kahneman (1983; see Chapter 1) tied heuristics to “natural assessments” elicited by the task at hand that can influence judgment without being used deliberately or strategically. When deciding whether an elegantly-dressed lawyer is more likely to be a public defender or a member of a large corporate firm, for example, one cannot help computing the similarity between the individual and the prototype of each professional niche. This assessment then informs the judgment of likelihood in the absence of deliberative intent.

It seems to us that both uses of the term are valid and have their place. When deciding whether there are more coups in Ecuador or Indonesia, for example, one automatically searches for known instances of each (availability). Yet one can also deliberately recruit such instances and use the ease with which they come to mind as an explicit strategy – as when deciding to bet on one team over another after explicitly considering the number of star players on each. Similarly, existing research on anchoring makes it clear that many anchoring effects occur in the absence of any explicit adjustment (Mussweiler & Strack, 1999; see Chapman & Johnson, Chapter 6). Often people’s estimates are automatically contaminated by previously mentioned values. Sometimes, however, the anchoring and adjustment heuristic is deliberately employed. If asked when George Washington was first elected president, most people do not know the answer; but they do know it was after 1776 and they adjust from that year (Epley & Gilovich, Chapter 7). Anchoring and adjustment is thus sometimes used as an explicit strategy of judgment.

At the time of writing, the authors considered the “cognitive miser” view more pervasive than the “natural assessments” view. They expected the rise of “two system” models of reasoning, described in the source’s Chapter 2 and Chapters 22 through 24, to change that balance. Such models allow for rapid, automatic assessments that deliberate reasoning may or may not override. Reasserting this account of heuristics was an aim of the collection, including its opening chapter by Tversky and Kahneman and Chapter 2 by Kahneman and Frederick.