Metacognition in animals

Human decision-making often is guided by our certainty (or uncertainty) about the accuracy of our own thought processes. In essence, we make choices not based solely on information processing of stimuli in the external world but also on the basis of how confident we are that our information processing routines have provided enough information or the correct information for an accurate response. This ability, called metacognition, is sometimes defined as "thinking about thinking," but the term also refers to the monitoring and the cognitive control of "first-order" cognitive processes. Human metacognition may be intricately linked to important aspects of mind, including cognitive control, self-awareness, theory of mind, and consciousness. This might suggest that it would be a uniquely human capacity. However, comparative research indicates that other animals may be capable of metacognitive processes, particularly in the area of uncertainty monitoring. For example, when monkeys are trained to classify stimuli that lie along a psychophysical continuum (e.g., discriminating line length or circle size), their performance is poorest for the most difficult stimuli near the discrimination's breakpoint. When monkeys are given a third response that allows them to decline the current trial without receiving reward or punishment for a primary response, they selectively use this response on exactly those trials for which they are at greatest risk of making an incorrect primary response. The third response has come to be called the uncertainty response. The appropriate psychological or metacognitive interpretation of this response pattern is still debated. In this chapter, we will outline evidence for nonhuman metacognition, focusing on studies with monkeys. We will describe the methods that are needed to evoke uncertainty responses in animals. We will also describe the methods that let researchers distinguish true metacognitive capacities from responding based in associative learning and reinforcement signals. Finally, we will present evidence from studies that may show metacognitive capacities in nonhuman animals and that suggest that metacognition is a shared capacity across various species.