Metacognition in human decision-making: confidence and error monitoring

Abstract

People are capable of robust evaluations of their decisions: they are often aware of their mistakes even without explicit feedback, and report levels of confidence in their decisions that correlate with objective performance. These metacognitive abilities help people to avoid making the same mistakes twice, and to avoid overcommitting time or resources to decisions that are based on unreliable evidence. In this review, we consider progress in characterizing the neural and mechanistic basis of these related aspects of metacognition-confidence judgements and error monitoring-and identify crucial points of convergence between methods and theories in the two fields. This convergence suggests that common principles govern metacognitive judgements of confidence and accuracy; in particular, a shared reliance on post-decisional processing within the systems responsible for the initial decision. However, research in both fields has focused rather narrowly on simple, discrete decisions-reflecting the correspondingly restricted focus of current models of the decision process itself-raising doubts about the degree to which discovered principles will scale up to explain metacognitive evaluation of real-world decisions and actions that are fluid, temporally extended, and embedded in the broader context of evolving behavioural goals.

Figure 1.

The drift-diffusion model. Accumulating evidence (the decision variable, y-axis) over time (x-axis) is shown for two illustrative trials (marked a and b, grey and black lines), one on which the choice θ is made and the other on which the choice –θ is made. A decision is triggered when evidence reaches θ or –θ. Grey line, trial 1; black line, trial 2.

Figure 2.

Theories of error detection within the DDM framework. The drift-diffusion process is illustrated schematically for two trials, one in which decision θ is the correct response and one trial in which this decision is incorrect. Both decisions occur at the same time point (a). Following the correct response (grey line), post-decision processing continues to accumulate in favour of the decision just made. Following errors (black line), the drift rate regresses to its true mean, causing the DV to re-cross the decision bound (b), subsequently cross a change-of-mind bound (c), and finally cross the originally correct decision bound, –θ (d). The grey shaded area indicates a period of uncertainty, or conflict, between the re-crossing of the θ bound (b) and later crossing of the –θ bound (d).

Figure 3.

Schematic of a model in which both the mean and variance of information in an array are estimated in a serial sampling framework. (a) The left panel shows the posterior probability distribution p (H | data) over a continuous space of possible perceptual hypotheses (e.g. these dots are moving to the right with 30% coherence; this signal is 40% visible; etc.) at a given time, t. This distribution reflects the evidence sampled from the stimulus thus far, i.e. between onset and time t (grey dots). The new sample received at time t is shown in red. Right panel: at time t + 1, this distribution (light grey) is updated in the light of the newly sampled information, giving rise to a new probability distribution. In this model, confidence is reflected in the precision of the posterior distribution, i.e. the reciprocal of its standard deviation. (b) The evolving posterior probability distribution over perceptual hypotheses (y-axis) for each successive time point (x-axis; blue–red colourmap; red values indicate higher probabilities). The posterior distribution is updated following the arrival of successive samples with low variance (left panel) or high variance (right panel). Precision of the probabilistic representation of evidence strength increases more rapidly for the low-variability samples.