When does metacognition evolve in the opt-out paradigm?

Metacognition (awareness of one's own knowledge) is taken for granted in humans, but its evolution in non-human animals is not well understood. While there is experimental evidence of seemingly metacognitive judgements across species, studies rarely focus on why metacognition may have evolved. To address this, I present an evolutionary model of the opt-out paradigm, a common experiment used to assess animal's metacognition. Individuals are repeatedly presented with a task or problem and must decide between opting-out and receiving a fixed payoff or opting-in and receiving a larger reward if they successfully solve the task. Two evolving traits - bias and metacognition - jointly determine whether individuals opt-in. The task's reward, the mean probability of success and the variability in success across trials, and the cost of metacognition were varied. Results identify two scenarios where metacognition evolves: (1) environments where success variability is high; and (2) environments where mean success is low, but rewards are high. Overall, the results support predictions implicating uncertainty in the evolution of metacognition but suggest metacognition may also evolve in conditions where metacognition can be used to identify cases where an otherwise inaccessible high payoff is easy to acquire.