Computational markers show specific deficits for dyslexia and ADHD in complex learning settings.

The current study examined how individuals with neurodevelopmental disorders navigate the complexities of learning within multidimensional environments marked by uncertain dimension values and without explicit guidance. Participants engaged in a game-like complex reinforcement learning task in which the stimuli dimension determining reward remained undisclosed, necessitating that participants discover which dimension should be prioritized for detecting the maximum reward. For comparison, a control condition featuring a simple reinforcement learning task was included in which the predictive dimension was explicitly revealed. The findings showed that individuals with ADHD and dyslexia exhibited reduced performance across both tasks compared to their controls. Computational modeling revealed that relative to controls, participants with ADHD exhibited a markedly decreased ability to utilize demanding yet more optimal Bayesian inference strategies, whereas participants with dyslexia demonstrated heightened decay rates, indicating quicker discounting of recently learned associations. These findings illuminate different computational markers of neurodevelopmental disorders in naturalistic learning contexts.