Superstitious learning of abstract order from random reinforcement.

Area of Science:

• Cognitive Science
• Animal Behavior
• Neuroscience

Background:

• Animals often learn spurious associations, typically explained by reinforcement learning.
• The capacity for complex cognitive structures independent of reinforcement remains an open question.

Purpose of the Study:

• To investigate if monkeys can infer abstract structures from random events.
• To explore cognitive mechanisms beyond simple stimulus-outcome learning.

Main Methods:

• Monkeys were presented with "learnable" (ordered) and "unlearnable" (random) pictorial sets.
• Choice preferences and learning were assessed, including under a preference-discouraging reward schedule.
• Simulations using a Q-learning model were conducted for comparison.

Main Results:

• Monkeys reliably learned the implicit order in the "learnable" set.
• Monkeys inferred ordering in the "unlearnable" set, showing consistent preferences.
• This inferred ordering persisted even under a preference-discouraging reward schedule, unlike Q-learning simulations.

Conclusions:

• Monkeys exhibit abstract structure inference from random events.
• Cognitive processes in monkeys extend beyond basic reinforcement learning.
• Model-based learning mechanisms likely contribute to understanding abstract structures.