Multiple demonstrations of metacognition in nonhumans: Converging evidence or multiple mechanisms?

Abstract

Metacognition allows one to monitor and adaptively control cognitive processes. Reports from the last 15 years show that when given the opportunity, nonhuman animals selectively avoid taking difficult tests of memory or perception, collect more information if needed before taking tests, or "gamble" more food reward on correct than on incorrect responses in tests of memory and perception. I review representative examples from this literature, considering the sufficiency of four classes of mechanism to account for the metacognitive performance observed. This analysis suggests that many of the demonstrations of metacognition in nonhumans can be explained in terms of associative learning or other mechanisms that do not require invoking introspection or access to private mental states. Consideration of these accounts may prompt greater appreciation of the diversity of metacognitive phenomena and may inform theoretical positions about the nature of the mental representations underlying metacognition.

Figure 1

Cartoons of three candidate processes of introspective memory assessment. The column to the left of the dashed line represents the contents of cognitive processing on trials on which monkeys choose to take the memory test. The right column depicts the same on trials on which the test is declined. The memory flag hypothesis posits an indicator for the presence of memory. Monkeys use the metacognitive response contingent on the indicator, but are not aware of the content of the memory. In the case of the image retrieval model, the decision to take the memory test is based on the vividness of the memory retrieved. The episodic retrieval hypothesis proposes that monkeys take the test when they can remember the context of the study episode and decline the test when this information cannot be recovered.

Figure 2

Left, a rhesus monkey, ignorant of the food’s location, collects more information before making a choice. Right, an informed monkey makes a choice without going to the effort of confirming the location of the food. Such selective information seeking suggests that the monkey knows when he knows, and only seeks more information as needed.