Thinking and Deciding

LOGICAL ERRORS IN HYPOTHESIS TESTING 91
they were simply failing to use the very powerful evidence that was staring them in
the face. If they were to use it, it would cause them to change their original selection.
Dual processes and rationalization
If these examples show a bias in inference from evidence, another set of experiments
suggests a related bias in the search for evidence. In a series of studies carried
out by Evans and his collaborators (reviewed in Evans, 1982, ch. 9), it has been
found that when subjects are asked to give reasons for their choices of cards, the
reasons given often seem to be rationalizations that they thought of afterward to
explain their choices rather than true determinants of the choice. Evans has made
this point dramatically by arguing that reasoning involves “dual processes,” one that
actually draws conclusions and another that rationalizes the conclusions after they
are drawn. 5
The best evidence for this comes from experiments in which the actual determinants
of the choice of cards seem to have nothing to do with the reasons that subjects
give. Instead the choices of cards seem to be determined by an elementary process
of matching to the elements of the rule. When subjects given the following cards are
asked to test the rule “If there is a B on one side, then there will be a 3 on the other
side,” they seem to choose the B and 3 cards simply because they are both mentioned
in the rule itself.
B U 3 6
How do we know that this is the actual determinant of the choice? One very
clever way to tell is to change the rule so that it becomes “If there is a B on one side,
then there will not be a 3 on the other.” If subjects are simply matching, they will
still choose the B and 3 cards. Now, however, these are the right answers. This time,
the 3 card is relevant, because, if there is a B on the other side, the rule is false.
In fact, subjects tend to choose the B and 3 cards whether the rule is stated in
the original, affirmative form or in the new, negative form. When they are asked to
justify their answers, however, their justifications are correct only for the negative
form. One subject (from Wason and Evans, 1975) who chose the 3 card justified
this choice in the negative task by saying, “If there is a B on the other side, then the
statement is false.” This, of course, is perfectly correct. The same subject, however,
immediately after doing the negative task, chose the 3 card for the affirmative task as
well. Here, the justification was “If there is a B on the other side, then the statement
is true.” This, of course, is beside the point.
5 Surely this point may be made too strongly. After all, some people do succeed in solving even the
four-card problem, and many people do change their mind on the basis of reasons (evidence, goals, new
possibilities). The important aspect of Evans’s point is that the determinants of the conclusions we draw
are, all too often, not fully rational, in terms of being sensitive to the results of a search process.