A comparison of disconfirmatory, confirmatory and control strategies on Wason's 2-4-6 task

Attempts to instruct subjects to falsify on inference tasks designed to model scientific problem-solving have indicated that, in most cases, falsification does not lead to improved performance. The only study in which instructions to falsify did improve performance was done with groups, using a novel inference task. The present study is an attempt to replicate the success of falsification in the group study with subjects run individually on an inference task that has been used extensively in the literature: Wason's 2-4-6 problem. Each subject was run in one of three strategy conditions: confirmatory, disconfirmatory or control (no-strategy).

Results indicated that subjects in the disconfirmatory condition solved the 2-4-6 problem significantly more often than subjects in other conditions. The superior performance of disconfirmatory subjects was explained by the fact that they deliberately sought and obtained significantly more falsificatory information than subjects in other conditions.

The critical difference between this study and previous attempts to instruct individual subjects to falsify on the 2-4-6 task is that subjects in the present study were given no feedback as to whether their guesses were right or wrong until the experiment was over. The implications of this result for scientific problem-solving are discussed.