Rule abstraction, model-based choice, and cognitive reflection.

Area of Science:

• Cognitive Science
• Learning and Memory

Background:

• Cognitive tasks often reveal distinct response patterns suggesting dual processing systems.
• Causal and reinforcement learning tasks show both abstract and associative responding, but their relationship is unclear.

Purpose of the Study:

• To investigate the link between rule-based generalization (causal learning) and model-based/model-free responding (reinforcement learning).
• To examine cognitive reflection as a predictor of deliberative processing.

Main Methods:

• Assessed rule- and feature-based generalization in a causal learning task.
• Evaluated model-based and model-free responding in a reinforcement learning task.
• Measured individual differences in cognitive reflection.

Main Results:

• Rule-based generalization predicted model-based, but not model-free, reinforcement learning choices.
• Cognitive reflection correlated with performance in both tasks.
• Cognitive reflection did not independently predict model-based choice beyond rule-based generalization.

Conclusions:

• Evidence supports stable individual differences in higher-order cognitive processes across learning tasks.
• Task-specific mechanisms may underlie observed differences in higher-order processing.