Abstract
We routinely make decisions that combine independent representations, such as combining vision and audition to decipher speech (McGurk & MacDonald, 1976). Recent work has argued that confidence representations exist in a domain-general format that could facilitate this integration: adult observers can compare their confidence across independent perceptual dimensions (orientation and frequency; de Gardelle & Mamassian, 2014), between auditory and visual stimuli (de Gardelle et al., 2016), and some work hints that this domain-generality might extend to non-perceptual tasks like memory and executive functioning (Mazancieux et al., under review). Here, we test a domain-general confidence hypothesis throughout development, investigating whether confidence is by nature domain-general, or if this emerges with experience. In Experiment 1, 6-7-year-olds compared their confidence in two decisions from the same visual dimension (e.g., number and number) and from two distinct visual dimensions (e.g., number and emotion) with equivalent ability (Fig. 1 and 2), supporting the hypothesis that visual confidence is domain-general in childhood. In Experiment 2, we similarly found that individual differences in certainty comparison are strongly correlated in 6-9-year-olds across otherwise uncorrelated visual dimensions (number, area, and emotion, Fig. 3). In Experiment 3, we extend this work to examine whether this domain-generality also exists between perception and memory in both children and adults. Finally, in Experiment 4, we attempt to identify what the common currency of confidence might be that allows for these cross-domain comparisons, examining whether response times, probability of accuracy, or objective difficulty (the ratio of two quantities) underlie these decisions. Together, our findings support the idea that confidence, particularly perceptual confidence, is represented in a common format even in childhood, providing one account for how independent representations could be compared.