Metacognitive monkeys or associative animals? Simple reinforcement learning explains uncertainty in nonhuman animals

Area of Science:

• Animal Cognition
• Learning Theory
• Behavioral Neuroscience

Background:

• Animals often adapt their behavior to avoid challenging tasks.
• Previous research suggested this avoidance learning requires immediate feedback.
• Metacognitive processes were proposed to explain avoidance learning without immediate feedback.

Purpose of the Study:

• To investigate the learning mechanisms underlying avoidance of difficult trials in monkeys.
• To test whether associative reinforcement learning can account for adaptive avoidance behavior.
• To challenge the necessity of metacognitive explanations for this phenomenon.

Main Methods:

• Analysis of existing empirical data on monkey perceptual discrimination learning.
• Modeling of animal behavior using a simple associative reinforcement learning framework.
• Comparison of model predictions with experimental findings, including those with deferred feedback.

Main Results:

• The associative reinforcement learning model successfully predicted monkeys' adaptive avoidance of difficult trials.
• The model accounted for findings where avoidance occurred even with deferred feedback.
• The model also explained other previously challenging findings for associative theories.

Conclusions:

• A simple associative reinforcement learning model adequately explains adaptive avoidance of difficult trials in monkeys.
• Metacognitive explanations are not necessary to account for these findings.
• Low-level associative processes are sufficient to explain complex adaptive learning behaviors.