Answers are grounded in this paper and cite their sources.
A growing body of empirical evidence suggests that confidence judgments are modulated by action-related cues, yet the underlying mechanisms remain unclear. Recent decision-making models propose that the evidence accumulation variable is continuously transmitted to the motor system, allowing motor plans to evolve in parallel with decision formation. At the motor level, a secondary leaky integration process acts as a filter that regulates the speed-accuracy tradeoff: the stronger the filter, the slower and more accurate the responses. Here, we tested whether action-related effects on confidence are mediated by this secondary motor integration process. We manipulated sensory noise using a doubly stochastic random dot motion task and varied the motor-level filter via speed-accuracy instructions. We reasoned that the positive effect of sensory noise on confidence observed in previous studies should be attenuated by stronger filtering in the accuracy condition compared to the speed condition. However, we found strong empirical evidence against this qualitative prediction. Computational modeling further revealed that the predicted attenuation was extremely small and that confidence predictions based on secondary motor integration offered no advantage over those based on evidence accumulation. Together, these results show that confidence need not be computed from secondary motor integration to account for the observed pattern.
No takes yet. Share an insight, caveat, or question.
Garnier-Allain et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: