ACTING AS INTUITIVE SCIENTISTS:

Abstract— In judging the efficacy of multiple causes of an effect, human performance has been found to deviate from the "normative"Δ P contingency rule However, in cases of multiple causes, that rule might not be normative, scientists and philosophers, for example, know that when judging a potential cause, one must control for all other potential causes. In an experiment in which they were shown trial-by-trial effects of two potential causes (which sometimes covaried), subjects used conditional rather than unconditional contingencies to rate the efficacy of the causes. A conditional contingency analysis may explain various "nonnormative" cue-integration effects (e.g., discounting) found in the literature and is relevant to how people unravel Simpson's paradox.