Signal detection theory (SDT) is one of the most popular frameworks for analysing data from studies of human behaviour, including investigations of confidence. SDT based analyses of confidence deliver both standard estimates of sensitivity (d prime), and a second estimate based only on high confidence decisions, meta d prime. The extent to which meta d prime estimates fall short of d prime estimates is regarded as a measure of metacognitive inefficiency, quantifying the contamination of confidence by additional noise. These analyses rely on a key but questionable assumption, that repeated exposures to an input will evoke a normally shaped distribution of perceptual experiences (the normality assumption). Here we show, via analyses inspired by an experiment and modelling, that when distributions of experiences do not conform with the normality assumption, meta d prime can be systematically underestimated relative to d prime. Our data highlight that SDT based analyses of confidence do not provide a ground truth measure of human metacognitive inefficiency.

No abstract