Adults with ADHD show reduced reliance on Bayesian inference strategies (lower alpha) than controls in complex learning settings¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? On the No-Explicit-Assist task, the ADHD group's best-fit alpha (0.12) is lower than controls' (0.23), indicating greater reliance on reinforcement learning over Bayesian learning. → Gabay 2025

Evidence¶

Gabay 2025¶

Gabay, Y., Jacob, L., Mansour, A. & Hertz, U. (2025). Computational markers show specific deficits for dyslexia and ADHD in complex learning settings. npj Science of Learning. https://doi.org/10.1038/s41539-025-00323-4

q2 · i? · associational · r2

Computational modeling of trial-by-trial choices with a hybrid reinforcement learning/Bayes model in the ADHD comparison; the alpha parameter was lower in the ADHD group on the No-Explicit-Assist task. The article reports the alpha contrast as significant at p = 0.008 in Table 1 (control 0.23 vs ADHD 0.12); no standardized effect size is printed, so no magnitude is asserted.

"On the Explicit-No-Assisttask, the difference between the two groups was in the best-fit value of the free parameter alpha, which was lower for the ADHD group. This parameter reflects the balance between the contributions of reinforcemen t learning and Bayesian learning processes, with lower values indic ating greater reliance on the rein forcement learning mechanism."

Discussion¶

Related Claims¶

• Adults with ADHD and adults with dyslexia show poorer reinforcement learning performance than matched controls across simple and complex multidimensional tasks — related
• The group-by-task interaction is not significant for either ADHD or dyslexia comparisons, though simple t-tests show task-specific group differences — related
• Adults with dyslexia show heightened decay rates than controls in complex learning settings, indicating quicker discounting of learned associations — related
• Performance is substantially better on the Explicit-Assist task than on the No-Explicit-Assist task in both the ADHD and dyslexia comparisons — related
• Children's reinforcement learning is best fit by a one-learning-rate model, whereas adolescents and adults are best fit by Bayesian models incorporating hidden-agent beliefs — related