More complex brains are not always better: rats outperform humans in implicit category-based generalization by implementing a similarity-based strategy

Generalization from previous experiences to new situations is a hallmark of intelligent behavior and a prerequisite for category learning. It has been proposed that category learning in humans relies on multiple brain systems that compete with each other, including an explicit rule-based system and an implicit system. Given that humans are biased to follow rule-based strategies, a counter-intuitive prediction of this model is that other animals in which this rule-based system is less developed might generalize better to new stimuli in implicit category learning tasks that are not rule-based. To test this prediction, rats and humans were trained in rule-based and information integration category learning tasks with visual stimuli. Generalization performance of rats and humans was equal in rule-based categorization, but rats outperformed humans on generalization in the information-integration task. The performance of rats was consistent with a non-dimensional similarity-based categorization strategy. These findings illustrate through a comparative approach how the bias towards rule-based strategies can impede humans’ performance on generalization tasks.status: publishe