Can increased processing noise induce better decisions? Evidence polarization through exponential weighting.

We examined how people adapt to processing noise in decision making under uncertainty. We assume that processing noise increases as the number of mental operations required for executing a decision increase. We gave participants repeated choices between “Safer” and “Riskier” slot-machines that paid out points with different long-run averages. To manipulate processing noise, we varied the location of the slot-machines such that they were either in the same position throughout trials (congruent) or different positions (incongruent). Incongruent trials thus required additional mental operations to map choices of a particular machine to the outcome it produced. Counterintuitively, we found that participants chose the superior option (higher long-run average) more often under conditions of higher processing noise (incongruent) than lower processing noise (congruent). A modelling analysis suggested that this pattern can be explained via a convex exponential weighting function that assigns greater weight to larger outcomes (higher numbers of points) which has the effect of polarizing evidence and thus choice towards the superior option. We conclude that prioritizing larger values can improve performance under processing noise.