tions. Without the crib sheet, subjects offered satisfactory explanations about
67% of the time but with the crib sheet 86%, a statistically significant con-
trast. However, arguably this was to a considerable extent an artifact of the
method, since the crib sheet reduced the interpretation of an underline to a
multiple choice problem with a 1-in–5 probability of getting it right by luck.
The pattern of findings provides further evidence against hypothesis 1, that
subjects lacked the knowledge, because providing knowledge via the crib
sheet did not enhance explanations much.
In summary, providing knowledge of what to look for did not help sub-
jects to detect shortfalls nor help subjects much to explain them. Saliency, on
the other hand, allowed most subjects to go on and explain the shortfalls. The
results favored the third hypothesis, that shortfalls were difficult to detect in
the midst of the stories despite appropriate knowledge and priming. This is in
keeping with the perceptual overtones of the notion of sensitivity, suggesting
a pattern recognition process that goes beyond simply knowing about short-
falls in principle and searching for them systematically.
Conclusion
In general, then, this series of studies provided strong support within the meth-
odology adopted for the importance of the dispositional side of thinking. Peo-
ple often do not perform nearly as well as they might in situations that call for
thinking principally because they miss the situations altogether and second-
arily because they fail to engage the situations thoughtfully. This challenges the
hegemony of abilities-centric accounts of thinking and indeed intelligence.
DISPOSITIONS AS TRAITS
The work previously outlined focused on the relative contribution of sensitiv-
ity, inclination, and ability to intellectual performance. Another question
concerns the extent to which sensitivity and inclination are trait-like con-
straints independent of ability. In particular, (a) are such candidate traits sta-
ble across time and task; (b) are they more domain general or domain spe-
cific; (c) are they statistically independent of ability measures; and (d) what
dispositional traits are there—one or many and which ones? Such questions
were not the central focus of this program of inquiry, but they were addressed
from time to time and research from other quarters speaks to them.
Stability Across Time and Task
Research on constructs such as need for cognition (Cacioppo et al., 1996) and
need for closure (Kruglanski, 1990) has demonstrated test-retest reliability.
The present program examined test-retest correlations on sensitivity scores for
366 PERKINS AND RITCHHART