Measures of confirmation and the inverse conjunction fallacy

Abstract

The conjunction fallacy is frequently explained by the hypothesis that individuals evaluate confirmation or justification rather than mathematical probability. Under this view, fallacious judgments occur because a conjunction can provide a higher degree of justification than its components, characterized by a balance between minimizing potential risk and maximizing potential information gain. However, this explanatory model fails to account for the inverse conjunction fallacy, in which individuals judge a property as more likely to be true of a general set than of a specific intersection or subset. Analysis of multiple probabilistic formulations demonstrates that Shogenji’s measure of justification—and confirmation measures generally—cannot successfully predict the inverse conjunction fallacy while remaining appropriate to natural language interpretation and the requirements of probabilistic consistency. The inability of these measures to generalize to related fallacies concerning set membership suggests that confirmation theory does not provide a robust account of why individuals deviate from the laws of probability. Consequently, measures of confirmation are insufficient to reconcile human judgment with the normative rules of probability in these contexts. – AI-generated abstract.