The Effects of Rewards on Trial-and-Error Learning in School-Aged Children

Abstract

Humans’ ability to rapidly identify appropriate actions in new situations is critical for survival and functional behavior. This skill develops through trial and error where humans get rewards for their actions, and through few attempts, learn the optimal subsequent action. Here, we used computerised reasoning games to test developmental changes in whether and how rewards affect trial-and-error learning. School-aged children (5- to 15-year-olds) played online games in which they had to access their physical knowledge and predict physical events to succeed. We systematically changed the rewards so children received either high positive rewards for success, high negative rewards for failure, or no reward. We evaluated differences between the groups in terms of performance (success rate, attempt and attempt duration) and strategy (which tools were selected and where they were positioned). Children in the Negative-Rewards group showed significant improvement with age—they succeeded more, used fewer attempts, and applied more efficient strategies. In contrast, we did not find developmental changes in the Positive-Rewards or No Reward groups. Our findings demonstrate a developmental shift in how rewards affect trial-and-error, and suggest that negative rewards emerge as a powerful cognitive reinforcer during late childhood.