Tested a motivated processing model of the hindsight bias or "knew-it-all-along" effect. Hindsight occurs when people exaggerate the extent to which they could have predicted the occurrence of an event. Although typically thought to be a "cold" cognitive process which automatically focuses attention on outcome-consistent information, some research suggests that, with sufficient motivation, subjects can ignore outcome information. The present study examined whether outcomes having negative connotations for the self might be less susceptible to the hindsight bias (e.g., an outcome that contradicts a decision to which one is highly committed).
One of two decision scenarios with strong moral components (gay job applicant and pregnant teen) was provided to each subject. A few subjects who were insufficiently committed to their choice were dropped from the analyses. Some subjects were given information indicating how their choice would have turned out--either favorably or unfavorably. Others were given no information. All subjects then provided likelihood estimates for a number of events that either did or did not occur in the scenario. One week later subjects were telephoned and asked to recall these outcome likelihood estimates. Hindsight was measured by comparisons between the outcome and no-outcome (control) conditions.
Results for the gay job applicant scenario indicated virtually no hindsight effects, while results for the pregnant teenager scenario indicated hindsight for roughly half of the likelihood measures. In both scenarios, however, outcomes that contradicted one's decision were no more or less susceptible to hindsight effects than were outcomes that affirmed one's decision. The study, then fails to support the motivated processing hypothesis, but provides additional evidence that hindsight effects are not necessarily automatic. Alternative hypotheses are addressed to explain the results.
Additional analyses indicated that, irrespective of choice, high need for cognition (NFC) subjects exhibited an unusual "reverse" hindsight bias in both scenarios; recalled likelihood estimates were smaller than controls' for outcomes that occurred and greater for outcomes that did not occur. Although the elaborate processing typical of high NFC subjects should increase the likelihood that they consider more alternative outcomes, no measure was taken to verify this.