Anne Fleur Kortekaas-Rijlaarsdam

97 REINFORCEMENT LEARNING IN ADHD: EFFECTS OF MPH 5 I N T RODUC T I ON The ability to learn the association between actions and consequences is a central feature of adaptive behavior (Rushworth & Behrens, 2008; Wolfram Schultz, 2000). This type of learning is called instrumental learning or reinforcement learning and requires the formation of a stable representation of action-consequence associations. It has been argued that a number of different mental disorders are associated with impairments in such fundamental learning processes (Sonuga-Barke, Cortese, Fairchild, & Stringaris, 2016). The everyday impairments seen in individuals with attention deficit/hyperactivity disorder (ADHD) have been hypothesized to be caused by deficits in instrumental learning(Luman, Tripp, & Scheres, 2010; Sonuga-Barke et al., 2016). Overall, studies of reinforcement learning in ADHD are scarce and the results are inconsistent (for an overview, see (Sonuga-Barke et al., 2016)). While individuals with ADHD show impaired accuracy levels in instrumental and probabilistic learning tasks (Frank, Santamaria, O’Reilly, & Willcutt, 2007; Groen et al., 2008; Hauser et al., 2014; Luman et al., 2009), the ability to improve in accuracy over the course of the task (learning rate), seems normal in ADHD (Luman, Goos, & Oosterlaan, 2015). Besides impairments in learning novel stimulus-response relationships, deficits (i.e. lower accuracy levels) have been observed in applying learned associations to novel stimulus- pair combinations (Frank et al., 2007). This indicates that successful learning of stimulus-response associations does not guarantee successful generalization of this knowledge to novel contexts. This is important, as, in daily life, children are constantly required to apply associations learnt in one context (e.g. during structured classroom tasks such as independently completing a task) to another (e.g. during less structured tasks in teams). Generalization of knowledge is thus an essential part of successful learning in daily life (Gershman & Niv, 2015; Tamminen, Davis, & Rastle, 2015). Reversal learning is the ability to change behavior in response to alterations in reinforcement contingencies. Children with ADHD are found to show more errors than controls when contingencies were reversed without warning (Itami & Uno, 2002). A recent study by Alsop et al. (2016) showed that children with ADHD are impaired in adjusting their responses to changes in reinforcement schedules compared to typically developing children on a signal detection task: When the response alternative associated with a higher rate of reinforcement was switched, the children’s response allocation (bias) shifted towards the newly reinforced alternative, but this effect was significantly smaller for children with ADHD. In contrast, Hauser et al. (2014) showed comparable behavioral performance (RT and accuracy) for adolescents with ADHD