Assessing conditionals: An experimental comparison of suppositional and inferential accounts

Two main normative accounts have been proposed to explain the acceptability of a conditional. The suppositional account equates acceptability with the probability of the consequent given the antecedent, whereas the inferential account maintains that it depends on the support the antecedent provides to the consequent. Previous empirical work has addressed the descriptive accuracy of normative and non-normative accounts, but the results have been inconclusive. In the present study, we introduce a novel experimental procedure designed to dissociate the predictions of competing accounts and employ it in three experiments. Overall, the results indicate that most individuals endorse a normative interpretation of conditionals; however, the specific account that best captures their judgments depends on the type of conditional evaluated. For conditionals in which the antecedent and the consequent stand in an arbitrary relation (e.g., “If the toy is a triangle, then it will be yellow”), judgments align more closely with the suppositional account. In contrast, for conditionals where the antecedent and the consequent are directly causally related (e.g., “If the laser is misaligned, then the prosthesis will have cracks” or “If the prosthesis has cracks, then the laser was misaligned”), judgments are better captured by the inferential account. Possible extensions of current models aimed at achieving a comprehensive characterization of conditionals are briefly discussed.