Adaptive curiosity about metacognitive ability

Metacognition provides control and oversight to the process of acquiring and using knowledge. Across three experiments, we found a specific form of curiosity in humans about the quality of their own metacognition, using a novel approach that dissociates perceptual from metacognitive information searches. Observers displayed a strategic balance in their curiosity, alternating between a focus on perceptual accuracy and metacognitive performance. Depending on the context, this curiosity was modulated by an internal evaluation of metacognition, leading to increased feedback requests when metacognition was likely to be inaccurate. Using an observer model, we describe how this curiosity trade-off can arise naturally from a recursive evaluation and transformation of decisions’ evidence. These results show that individuals are inherently curious about their metacognitive abilities and can compare perceptual and metacognitive precision to fine-tune performance monitoring. We propose that this form of curiosity may reflect humans’ drive to refine their self-model.