Adaptive Regulation of Confidence Following Metacognitive Conflict: Evidence from Behavior and Computational Modeling

It is well established that external feedback is effective not only for enhancing cognitive performance but also for regulating metacognitive monitoring. However, external feedback could be detrimental when it conflicts with one’s internal feedback signal (i.e. the level of confidence we have in our decisions). Currently, however, it is unknown how discrepancies between internal and external feedback (“metacognitive conflict”) influence cognitive and metacognitive performance. Here, we examined this question in a series of perceptual decision-making experiments in which participants reported both their level of confidence and received performance feedback. Metacognitive conflict occurs if participants have high confidence in their decision, but external feedback indicates that the decision was erroneous, or vice versa. Results revealed that metacognitive conflict triggers next-trial behavioral adjustments: participants exhibit significantly lower confidence following metacognitive conflict. Critically, this adaptation effect was confined to the trial level, with no sustained modulation observed when manipulating metacognitive conflict at the block level. To better characterize the computational mechanisms underlying this effect, we deployed an evidence accumulation model with additional post-decision accumulation to quantify confidence. Model fits revealed that participants become more conservative in reporting confidence after experiencing metacognitive conflict, as evidenced by biased post-decision starting points. Jointly, our findings provide much needed insight into the metacognitive consequences of metacognitive conflict.