The self-consistency model of subjective confidence.

How do people monitor the correctness of their answers? A self-consistency model is proposed for the process underlying confidence judgments and their accuracy. In answering a 2-alternative question, participants are assumed to retrieve a sample of representations of the question and base their confidence on the consistency with which the chosen answer is supported across representations. Confidence is modeled by analogy to the calculation of statistical level of confidence (SLC) in testing hypotheses about a population and represents the participant's assessment of the likelihood that a new sample will yield the same choice. Assuming that participants draw representations from a commonly shared item-specific population of representations, predictions were derived regarding the function relating confidence to inter-participant consensus and intra-participant consistency for the more preferred (majority) and the less preferred (minority) choices. The predicted pattern was confirmed for several different tasks. The confidence-accuracy relationship was shown to be a by-product of the consistency-correctness relationship: It is positive because the answers that are consistently chosen are generally correct, but negative when the wrong answers tend to be favored. The overconfidence bias stems from the reliability-validity discrepancy: Confidence monitors reliability (or self-consistency), but its accuracy is evaluated in calibration studies against correctness. Simulation and empirical results suggest that response speed is a frugal cue for self-consistency, and its validity depends on the validity of self-consistency in predicting performance. Another mnemonic cue-accessibility, which is the overall amount of information that comes to mind-makes an added, independent contribution. Self-consistency and accessibility may correspond to the 2 parameters that affect SLC: sample variance and sample size.