The Evolution of Accuracy and Bias in Social Judgment

자료: http://www.sscnet.ucla.edu/comm/haselton/webdocs/HaseltonFunder.pdf

Martie G. Haselton (UCLA), David C. Funder (UC Riverside)

... Chapter in preparation for M. Schaller, J. A. Simpson, & D. T. Kenrick (eds.) Evolution and Social Psychology, Psychology Press. 

... Draft: December 1, 2004

※ Thanks to the authors and the related people for for providing the article on the Internet. Followings are just an excerpt of some parts therein for reference.

1. The Evolution of Accuracy and Bias in Social Judgment

Humans are an intensely social species and therefore it is essential for our interpersonal judgments to be valid enough to help us to avoid enemies, form useful alliances and find suitable mates; flawed judgments can literally be fatal. An evolutionary perspective implies that humans ought to have developed sufficient skills at solving problems of interpersonal judgment, including gauging the personalities of others, to be useful for the basic tasks of survival and reproduction. Yet, the view to be derived from the large and influential bias-and-error literature of social psychology is decidedly different—the social mind seems riddled with fundamental design flaws. We will argue in this paper that flawed design is probably the least plausible explanation for the existence of so many errors. We present an evolutionarily-based taxonomy of known bias effects that distinguishes between biases that are trivial or even artifactual and lead virtually nowhere, and those that have interesting implications and deserve further study. Finally, we present an evolutionary perspective that suggests that the ubiquity, automaticity, and success of interpersonal judgment, among other considerations, presents the possibility of a universal Personality Judgment Instinct.

2. ADAPTATIONS FOR SOCIAL LIFE

Archeological evidence and behavioral patterns observed in extant hunter-gatherer groups indicate that the human species has been intensely social for a long time (e.g., Chagnon, 1983, Tooby & Devore, 1987). Human offspring have a remarkably extended period of juvenile dependency, which both requires and provides the skills for surviving in a complex social world (Hrdy, 1999). Humans evolved language and universal emotional expressions which serve the social purpose of discerning and influencing the thoughts of others (e.g., Darwin, 1872; Ekman, 1973; Pinker, 1994), and humans will infer social intentions on the basis of minimal cues, as Heider and Simmel (1944) demonstrated in their classic experiment involving chasing triangles and evading circles. Recent work has shown that children above age 4 and adults in disparate cultures (Germans and Amazonian Indians) can categorize intentions—chasing, fighting, following, playing, and courting (for adults)— from no more than the motion patterns of computerized v-shaped arrowheads (Barrett, Todd, Miller, & Blythe, in press).

Most notably, humans have a deeply-felt need for social inclusion. Deprivation of social contact produces anxiety, loneliness, and depression (Baumeister & Leary, 1995); indeed, as William James (1890) observed: “Solitary confinement is by many regarded as a mode of torture too cruel and unnatural for civilised countries to adopt.” Participants in laboratory studies who are left out of a face-to-face triadic ball toss respond with depressed mood and decreased self esteem (Williams & Sommer, 1997). These effects can even be produced by a computerized version of the game in which participants use key presses to “toss” the ball back and forth to human-like figures on a screen (Williams, Cheung, & Choi, 2000), and persist when participants are told that the other players have been scripted or are mere computer programs (Zadro & Williams 2003, cited in Williams, Case, & Govan, 2003). Neuroscience evidence suggests that being ostracized activates the same brain regions involved in the sensation of physical pain (Eisenberger, Lieberman, & Williams, 2003). Rejection hurts, literally.

This acute social sensitivity makes sense in the light of the many problems of social adaptation that have long faced members of our species: the formation of cooperative alliances for hunting and protection (e.g., Tooby & Devore, 1987; Tooby & Cosmides, 1988), hierarchy negotiation (Kyl-Heku & Buss, 1996), mate choice (Buss, 2003; Miller, 2000; Symons, 1979; choice of allies and friends (Tooby & Cosmides, 1996), and social exchange (Cosmides 1989), to name a few. Given the importance of these problems, we should expect finely honed adaptations for forming social judgments and making social decisions that are, at the very least, good enough to promote survival and reproduction. We would certainly not expect thousands of years of social evolution to yield a psychological apparatus fundamentally prone to social misperception, judgmental flaws, and maladaptive interpersonal behavior. Yet, this is the picture one gets from a good deal of conventional research in social psychology.

3. IS THE SOCIAL MIND DEEPLY FLAWED?

A large part of social psychology – including some of its most famous and influential research programs – consists of a loosely-connected set of non-intuitive and curious effects, each of which demonstrates a context in which humans can be led to make incorrect judgments according to one or more standards of logic, statistics, or even morality (Krueger & Funder, in press). An especially famous error is the putative tendency for people to infer that dispositions (enduring aspects of personality) have stronger effects on the behavior of others than do situations, coined the fundamental attribution error (Ross, 1977; but see Funder, 1982). Humans also have been accused of false consensus, confirmation bias, overconfidence bias (as well as pessimistic bias), hindsight bias, and the sinister attribution error. And, experimenters have caught humans in the act of committing the planning fallacy, the external agency illusion, and the transparency illusion. These are just a few examples from a very long list (for reviews see Gilovich, Griffin, & Kahneman, 2002; Fiske & Taylor, 1991), a list that grows longer all the time. One recent example is the dud-alternative effect in which adding an implausible alternative (e.g., a “dud” in a horserace) increases the judged likelihood that a good alternative will win, when in fact the inclusion of more alternatives must reduce the probably of success for any given candidate (Windschitl & Chambers, 2004). An especially terrifying recent example is the bias blind spot (Pronin, Gilovich, & Ross, 2004), which is the bias to not know you’re biased! The cumulative effect of this ever-growing list is a view of the human social mind as fraught with shortcomings, a view that is almost always detectable implicitly and often is expressed explicitly as well (Lopes, 1991).

The emphasis on bias and error is understandable to some degree. Bias effects tend to be counterintuitive (Lopes, 1991; Funder, 2003), funny—they make for good anecdotes and amusing classroom demonstrations (Crandall, 1984)—and to the degree they really do afflict cognition and associated life outcomes, they call out for study so that they can be fixed. However, the view of human judgment as dominated by error is both implausible and theoretically impoverished.

When a putative error of human judgment is discovered, there are three possible explanations. First, the error might not be an error at all. The experimental situation or instructions to subjects or the standards by which error has been defined might be misleading or incorrect, so that the putative error is better considered an experimental artifact. Second, the error might be one that, on balance, leads in realistic situations to adaptive decisions more often than not. The error might be produced by a usually-adaptive heuristic, or be the result of a tendency to favor less costly errors over more costly ones (see below). For example, to the extent that behavior really is predictable from stable traits and attitudes, the fundamental attribution error (to the extent it is itself not an artifact) will tend to produce correct decisions in realistic circumstances. Third, the error might reveal a flaw in psychological design such that the mind is fundamentally prone to get a broad class of decisions wrong. This explanation is the most frequently offered of the three but is, we submit, the least plausible one.

Furthermore, these loosely connected findings, despite their number, do not add up to a broad, coherent theory of human social thought and behavior. The reason is that demonstrations of error characteristically begin by assuming that human judgment to be perfect, and attain their news value from the conclusion that it is not. But this conclusion does not provide even the beginnings of an explanation of how judgments are ever made correctly; the initial assumption of perfection bypasses any possibility of a broader account. Instead, the long lists of errors powerfully convey the usually implicit, sometimes explicit and surely misleading message that good judgment is rarely achieved.

WHERE DO BIASES COME FROM? (continued)