The developmental trajectories of generalization and discrimination in reinforcement learning reflect multiple underlying cognitive processes

• DOI: 10.31234/osf.io/kvu7c_v2
• Publication Date: 2025-01-29T01:02:48Z
• Authors: Wei Chen; Aaron T. Nakamura; Jialing Ding; Naohiro Okada; Shinsuke Koike; Sho Yagishita; Shin Ishii; Haruo Kasai; Yuko Yotsumoto; Ming Bo Cai
• Keywords: Behavioral Neuroscience; Judgment and Decision Making; Cognitive Development; Cognitive Neuroscience; Developmental Psychology; Cognitive Psychology; Learning; Social and Behavioral Sciences; Neuroscience

A hallmark of human intelligence is the ability to appropriately generalize past experiences new scenarios. Organizing concepts into categories allows brain learned stimulus-reward contingencies stimuli within same category. However, when category membership does not fully predict associated rewards, needs balance generalizing values and discriminating value differences among individual items. How do generalization discrimination in reward learning develop throughout lifespan as semantic knowledge gradually acquired? Using a probabilistic task with hierarchical structure organized by categories, we found children can based on at least early three four years old. Furthermore, age-related performance improvement explained combination increased decision certainty, decreased initial exploration, change memory retention, reduced categorical rate.