24.04.2013 Views

Confirmation Bias: A Ubiquitous Phenomenon in Many Guises

Confirmation Bias: A Ubiquitous Phenomenon in Many Guises

Confirmation Bias: A Ubiquitous Phenomenon in Many Guises

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

184 RAYMOND S. NICKERSON<br />

fit what one sees <strong>in</strong>to the taxonomic b<strong>in</strong>s at<br />

hand. In accordance with the confirmation bias,<br />

people are more likely to look for, and f<strong>in</strong>d,<br />

confirmation of the adequacy of a taxonomy<br />

than to seek and discover evidence of its<br />

limitations.<br />

Formal Reason<strong>in</strong>g and the Selection Task<br />

I have already considered a task <strong>in</strong>vented by<br />

Wason and much used <strong>in</strong> rule-discovery experiments<br />

that revealed a tendency for people to test<br />

a hypothesized rule primarily by consider<strong>in</strong>g<br />

<strong>in</strong>stances that are consistent with it. Wason<br />

(1966, 1968) <strong>in</strong>vented another task that also has<br />

been widely used to study formal reason<strong>in</strong>g. In a<br />

well-known version of this task, participants see<br />

an array of cards and are told that each card has a<br />

letter on one side and a number on the other.<br />

Each of the cards they see shows either a vowel,<br />

a consonant, an even number, or an odd number,<br />

and participants are asked to <strong>in</strong>dicate which<br />

cards one would have to turn over <strong>in</strong> order to<br />

determ<strong>in</strong>e the truth or falsity of the follow<strong>in</strong>g<br />

statement: If a card has a vowel on one side then<br />

it has an even number on the other side.<br />

Suppose the array that people perform<strong>in</strong>g this<br />

task see is as follows:<br />

Given this set of cards, experimenters have<br />

generally considered selection of those show<strong>in</strong>g<br />

A and 7 to be correct, because f<strong>in</strong>d<strong>in</strong>g an odd<br />

number on the other side of the A or f<strong>in</strong>d<strong>in</strong>g a<br />

vowel beh<strong>in</strong>d the 7 would reveal the statement<br />

to be false. The cards show<strong>in</strong>g B and 4 have<br />

been considered <strong>in</strong>correct selections, because<br />

whatever is on their other sides is consistent<br />

with the statement. In short, one can determ<strong>in</strong>e<br />

the claim to be false by f<strong>in</strong>d<strong>in</strong>g either the card<br />

show<strong>in</strong>g the A or the card show<strong>in</strong>g the 7 to be<br />

<strong>in</strong>consistent with it, or one can determ<strong>in</strong>e the<br />

claim to be true by f<strong>in</strong>d<strong>in</strong>g both of these cards to<br />

be consistent with it. Wason found that people<br />

perform<strong>in</strong>g this task are most likely to select<br />

only the card show<strong>in</strong>g a vowel or the card<br />

show<strong>in</strong>g a vowel and the one show<strong>in</strong>g an even<br />

number; people seldom select either the card<br />

show<strong>in</strong>g a consonant or the one show<strong>in</strong>g an odd<br />

number. Numerous <strong>in</strong>vestigators have obta<strong>in</strong>ed<br />

essentially the same result. Experiments with<br />

this task and variants of it have been<br />

reviewed many times (Cosmides, 1989; Evans,<br />

1982; Evans, Newstead, & Byrne, 1993; Tweney<br />

& Doherty, 1983; Wason & Johnson-Laird,<br />

1972).<br />

The logic of Wason's selection task is that of<br />

the conditional: if P then Q. In the case of the<br />

above example, P is there is a vowel on one side,<br />

and Q is there is an even number on the other.<br />

Select<strong>in</strong>g the card show<strong>in</strong>g the 7 is analogous to<br />

check<strong>in</strong>g to see if the not-Q case is accompanied<br />

by not-P, as it must be if the conditional is true.<br />

The basic f<strong>in</strong>d<strong>in</strong>g of experiments with the task<br />

lends support to the hypothesis—which is<br />

supported also by other research (Evans, Newstead,<br />

& Byrne, 1993; Hollis, 1970)—that<br />

people f<strong>in</strong>d the modus tollens argument (not-Q,<br />

therefore not-P) to be less natural than the<br />

modus ponens form (P, therefore Q). And it is<br />

consistent with the idea that, given the objective<br />

of assess<strong>in</strong>g the credibility of a conditional<br />

assertion, people are more likely to look for the<br />

presence of the consequent given the presence<br />

of the antecedent than for the absence of the<br />

antecedent given the absence of the consequent.<br />

Several experiments have shown that performance<br />

of the selection task tends to be<br />

considerably better when the problem is couched<br />

<strong>in</strong> familiar situational terms rather than abstractly<br />

(Johnson-Laird, Legrenzi, & Legrenzi,<br />

1972; Wason & Shapiro, 1971), although it is by<br />

no means always perfect <strong>in</strong> the former case<br />

(E<strong>in</strong>horn & Hogarth, 1978). People also generally<br />

do better when the task is couched <strong>in</strong> terms<br />

that require deontic reason<strong>in</strong>g (decid<strong>in</strong>g whether<br />

a rule of behavior—e.g., permission, obligation,<br />

promise—has been violated) rather than <strong>in</strong>dicative<br />

reason<strong>in</strong>g (determ<strong>in</strong><strong>in</strong>g whether a hypothesis<br />

is true or false; Cheng & Holyoak, 1985;<br />

Cosmides, 1989; Cosmides & Tooby, 1992;<br />

Gigerenzer & Hug, 1992; Griggs & Cox, 1993;<br />

Kroger, Cheng, & Holyoak, 1993; Manktelow &<br />

Over, 1990, 1991; Markovits & Savary, 1992;<br />

Valent<strong>in</strong>e, 1985; Yachan<strong>in</strong> & Tweney, 1982).<br />

In relat<strong>in</strong>g confirmation bias to Wason's<br />

selection task, it is important to make three<br />

dist<strong>in</strong>ctions. The first is the dist<strong>in</strong>ction between<br />

the objective of specify<strong>in</strong>g which of four cards<br />

<strong>in</strong> view must be turned over <strong>in</strong> order to<br />

determ<strong>in</strong>e the truth or falsity of the assertion<br />

with respect to those four cards and the<br />

objective of say<strong>in</strong>g which of the four types of<br />

cards represented should be turned over <strong>in</strong> order<br />

to determ<strong>in</strong>e the plausibility of the assertion

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!