Decoding decision processes to improve metacognition about actions

People have been famously found to exhibit relatively poor access to higher-order cognitive processes, such as the determinants of own actions, demonstrated e.g. by choice blindness, where people confabulate reasons leading to their choices. It has been hypothesized, but not directly tested, that such metacognitive errors are partially caused by a lack of adequate feedback about the cognitive processes. In the present study, participants were often mistaken about whether their action was automatic or based on intentional deliberation. Uniquely, we were able to probabilistically determine this fact on a trial-by-trial basis and provide feedback to participants. Participants undergoing such feedback-based training improved their metacognitive sensitivity significantly more than control participants. We discuss possible interpretations of this improvement. Moreover, participants generally exhibited rather good meta-metacognition: awareness of their metacognitive sensitivity. We were able to quantify metacognitive sensitivity by comparing the reported and decoded cognitive processes leading to actions within the framework of “type 1” fuzzy signal detection theory. Although the results do not unequivocally support the hypothesis we set out to explore, its unique methodology can pave the way for further research: We extend the study of metacognition beyond the domains of perception and memory to the domain of action and beyond using the performance-confidence relation as a measure of metacognition.