A mechanistic model of ADHD as resulting from dopamine phasic/tonic imbalance during reinforcement learning

作     者：Veronneau-Veilleux, Florence Robaey, Philippe Ursino, Mauro Nekka, Fahima 

作者机构：Univ Montreal Fac Pharm Montreal PQ Canada Univ Ottawa Childrens Hosp Eastern Ontario Ottawa ON Canada Univ Bologna Dept Elect Elect & Informat Engn Guglielmo Marconi Bologna Italy Univ Montreal Ctr Rech Math Montreal PQ Canada McGill Univ Ctr Appl Math Biosci & Med Montreal PQ Canada 

出 版 物：《FRONTIERS IN COMPUTATIONAL NEUROSCIENCE》 (Front. Comput. Neurosci.)

年 卷 期：2022年第16卷

页      面：849323页

核心收录：

基　　金：Natural Sciences and Engineering Research Council (NSERC) , Canada through the PGS-D program NSERC-Industrial Chair in Pharmacometrics - Novartis Pfizer Syneos FRQNT Projet d'equipe (FN) 

主　　题：attention deficit hyperactivity disorder tonic and phasic dopamine neurocomputational model basal ganglia reinforcement learning 

摘      要：Attention deficit hyperactivity disorder (ADHD) is the most common neurodevelopmental disorder in children. Although the involvement of dopamine in this disorder seems to be established, the nature of dopaminergic dysfunction remains controversial. The purpose of this study was to test whether the key response characteristics of ADHD could be simulated by a mechanistic model that combines a decrease in tonic dopaminergic activity with an increase in phasic responses in cortical-striatal loops during learning reinforcement. To this end, we combined a dynamic model of dopamine with a neurocomputational model of the basal ganglia with multiple action channels. We also included a dynamic model of tonic and phasic dopamine release and control, and a learning procedure driven by tonic and phasic dopamine levels. In the model, the dopamine imbalance is the result of impaired presynaptic regulation of dopamine at the terminal level. Using this model, virtual individuals from a dopamine imbalance group and a control group were trained to associate four stimuli with four actions with fully informative reinforcement feedback. In a second phase, they were tested without feedback. Subjects in the dopamine imbalance group showed poorer performance with more variable reaction times due to the presence of fast and very slow responses, difficulty in choosing between stimuli even when they were of high intensity, and greater sensitivity to noise. Learning history was also significantly more variable in the dopamine imbalance group, explaining 75% of the variability in reaction time using quadratic regression. The response profile of the virtual subjects varied as a function of the learning history variability index to produce increasingly severe impairment, beginning with an increase in response variability alone, then accumulating a decrease in performance and finally a learning deficit. Although ADHD is certainly a heterogeneous disorder, these results suggest that typical features