MODELING AND EXPERIMENTAL STUDY OF CONFIDENCE IN SENSORY JUDGMENTS
A mathematical model of decision making and confidence in sensory tasks is developed on the base of Signal Detection Theory. The model argues that confidence may be represented as a sum of a sensory evidence and non-sensory (frequency and motivational) ones (determined by stimuli a priori probabilities and costsâ€”payoffs for responses). This representation of confidence permits to convert confidence obtained in a single observation into expected decision correctness probability and utility of a chosen response. The model predicts better accuracy of confident responses regarding to all responses. It was confirmed in the both main kinds of sensory discrimination tasks: â€œGreaterâ€”Lesserâ€ and â€œSameâ€”Differentâ€ for visual stimuli: spatial (circles diameters) and temporal (durations) ones presented simultaneously and successively correspondingly. Thus, the fact found seems to be rather general one.