| Two experiments investigated matching bias in two conditional reasoning tasks. Participants performed either the Wason selection task, the conditional truth table evaluation task (to check cases falsifying a conditional rule), or both. The conditionals used in the experiments were all indicative. Results, comparing selection patterns between single-negation and double-negation sentences such as "if not not-B, then 2", showed that the matching bias account (Evans and Lynch, 1973) was supported at least in the conditions where the consequent was negated in both tasks. However, in the conditions where the antecedent was negated, the selection proportion data slightly more fitted a model which postulates that participants deleted one negative component in the conditional and consider the case corresponding to the deleted conditional as falsifying the original conditional than the matching bias model. This strategy is more rational than the process influenced by the matching bias. |
