The idiosyncratic nature of confidence

Abstract

Confidence is the feeling of knowing that accompanies decision making and guides processes such as learning, error detection, and inter-personal communication. Bayesian theory proposes that confidence is a function of the probability that a decision is correct given the evidence. Empirical research has shown, however, that humans tend to report confidence in very different ways. This idiosyncratic behaviour suggests that different individuals may perform different computations to estimate confidence from uncertain evidence. We tested this hypothesis by collecting confidence reports from healthy adults making decisions under either visual or numerical uncertainty. We found that for most individuals, confidence did indeed reflect the perceived probability of being correct. However, in approximately half of them, confidence also reflected a different probabilistic quantity: the observed Fisher information. We isolated the influence of each of these two quantities on confidence, and found that this decomposition is stable across weeks, and consistent across tasks involving uncertainty in both perceptual and cognitive domains. Our findings provide, for the first time, a mechanistic interpretation of individual differences in the human sense of confidence.