Introduction
Hindsight bias, the tendency to believe that one "knew it all along" once an outcome is revealed, is a pervasive cognitive distortion. It makes events seem more predictable than they actually were, influencing probability judgments, causal reasoning, and responsibility attribution. In accident contexts, such distortions may foster unfair blame. As with other cognitive biases, however, individuals find it difficult to recognize their own hindsight bias.
Purpose
This study aimed to develop a performance-based task to measure individual tendencies toward hindsight bias, and to validate this method using both neuroimaging (fMRI) and confidence ratings.
Method
Participants completed a computer task consisting of two stages. In the first stage, they provided both an accident probability judgment for each image and a confidence rating for that judgment. In the second stage, the same images were re-presented together with an expert's probability assessment, and participants recalled their own initial probability judgment and confidence rating. Hindsight bias was quantified as the reduction in the difference between participants' probability judgments and the expert assessments when comparing initial versus recalled values (larger reductions indicating stronger hindsight bias). Brain activity during task performance was measured using fMRI, and confidence ratings were further analyzed to clarify whether the observed shifts reflected hindsight bias rather than anchoring.
Results
Preliminary fMRI analyses suggested activity in the medial prefrontal and orbitofrontal cortices during recall, consistent with hindsight-related processing. These neural tendencies appeared dissociable from anchoring, as confidence ratings showed that shifts toward expert evaluations could not explain the observed bias.
Conclusion
These preliminary findings suggest that the task provides a promising approach for measuring hindsight bias. While further validation is needed, the method shows potential applications in accident investigation and bias-aware safety education. Future work will extend this approach to the development of training programs aimed at reducing hindsight bias and supporting safer decision-making in occupational and industrial settings.