Strategic and Graduated, Retrospective and Prospective Metacognitive Judgments by Monkeys

Studies of metacognition in nonhuman animals typically require subjects to make binary choices, e.g., skipping a trial on a cognitive task or placing a two-choice bet of confidence. Although animals can use that contingency to maximize reward, their performance is much less flexible than the graduated judgments that have been reported in the human literature. Here, we report three experiments in which monkeys wagered a small, medium, or large number of tokens on the accuracy of their performance (retrospective metacognition), or anticipated performance (prospective metacognition), on each trial of a matching-to-sample task. Subjects’ wagers tended to maximize reward and minimize risk. They were also influenced by the number of tokens needed to earn a reward on each trial. Alternative cues such as reaction time and test difficulty could not account for our results which show, for the first time, that monkeys can make graduated and strategic metacognitive judgments.