Metacognition facilitates theory of mind through optimal weighting of trait inferences.

Abstract

The ability to represent and infer accurately others' mental states, known as Theory of Mind (ToM), has been theorised to be associated with metacognitive ability. Here, we considered the role of metacognition in mental state inference through the lens of a recent theoretical approach to explaining ToM, the 'Mind-space' framework. The Mind-space framework posits that trait inference, representation of the qualities of the mind giving rise to the mental state, is important in forming accurate mental state inferences. We tested a potential role for metacognition in facilitating optimal weighting of trait inferences, as well as several theoretical predictions regarding factors associated with the accuracy of trait inference and confidence in those trait inferences. Participants completed a judgement-of-confidence task in the trait inference domain alongside the Interview Task, a recently-developed task for assessing the accuracy of trait and mental state inferences. A simple relationship in which increased metacognitive sensitivity is associated with increased accuracy of mental states inferences was not found. However, when predicting trial-level performance, confidence in trait inference was shown to modulate the effect of trait inference accuracy on mental state inference accuracy. This effect was greater in magnitude with lower metacognitive sensitivity, i.e., when confidence is more likely to be misplaced. Furthermore, participants' trait inference ability was associated with the accuracy of their understanding of the average mind. In addition, the accuracy of specific trait inferences was predicted by the participant's similarity to the target, but this similarity benefit was reduced in participants whose self-perception was inaccurate. Reported confidence in a given trait inference was also predicted by participant-target similarity, such that participants showed greater overconfidence in judgements made about similar targets. This overconfidence effect was larger when self-perception was more erroneous. Results support several theoretical claims made by the Mind-space theory, and further elucidate the processes underlying accurate mental state inference.