Decreased transfer of value to action in Tourette syndrome

Abstract

Objective

Tourette syndrome is a neurodevelopmental disorder putatively associated with a hyperdopaminergic state. Therefore, it seems plausible that excessive dopamine transmission in Tourette syndrome alters the ability to learn based on rewards and punishments. We tested whether Tourette syndrome patients exhibited altered reinforcement learning and corresponding feedback-related EEG deflections.

Methods

We used a reinforcement learning task providing factual and counterfactual feedback in a sample of 15 Tourette syndrome patients and matched healthy controls whilst recording EEG. The paradigm presented various reward probabilities to enforce adaptive adjustments. We employed a computational model to derive estimates of the prediction error, which we used for single-trial regression analysis of the EEG data.

Results

We found that Tourette syndrome patients showed increased choice stochasticity compared to controls. The feedback-related negativity represented an axiomatic prediction error for factual feedback and did not differ between groups. We observed attenuated P3a modulation specifically for factual feedback in Tourette syndrome patients, representing impaired coding of attention allocation.

Conclusion

Our findings indicate that cortical prediction error coding is unaffected by Tourette syndrome. Nonetheless, the transfer of learned values into choice formation is degraded, in line with a hyperdopaminergic state.

Introduction

Tourette syndrome (TS) is a childhood-onset hyperkinetic neurodevelopmental disorder characterized by the presence of motor and vocal tics. Tics share many commonalities with habitual behavior, as they are stereotyped and automatic sequences of actions that are triggered by specific internal or external stimuli (Leckman & Riddle, 2000). Common comorbid disorders in TS include obsessive-compulsive disorder (OCD), attention-deficit hyperactivity disorder (ADHD) and affective disorders (Eddy and Cavanna, 2014, Robertson, 2006, Simpson et al., 2011). The precise pathophysiology of TS remains unknown, but numerous findings point to alterations of cortico-basal ganglia-thalamo-cortical (CBGTC) loops (Mink, 2001). Although imbalances in various neurotransmitter systems have been reported for TS, the central role of dopaminergic transmission for this disorder is underlined by the effectiveness of neuroleptic medication in the treatment of TS (Huys et al., 2012, Leckman et al., 2010). The most parsimonious explanation of empirical findings might be an overall hyperdopaminergic state via increased dopaminergic innervation (Maia and Conceição, 2017, Maia and Conceição, 2018). Importantly, midbrain dopaminergic activity acts as a teaching signal (prediction error, PE) for reinforcement learning (RL) in the striatum (Balleine & O'Doherty, 2010) and probabilistic learning has been repeatedly shown to be impaired in TS (Kéri et al., 2002, Marsh et al., 2004). On the other hand, learning impairments in TS have been attributed to neuroleptic treatment, comorbid OCD or ADHD (Shephard et al., 2016a, Shephard et al., 2016b, Worbe et al., 2011).

Various theories postulate different predictions for RL alterations in a hyperdopaminergic state. One influential view hypothesizes increased impact of rewards alongside decreased impact of punishments (Palminteri et al., 2009). Alternative accounts state that increased tonic dopamine should diminish the impact of learned stimulus values on choices and thus increase choice stochasticity (Beeler, 2012, Hamid et al., 2015). Furthermore, current models propose that value computation in the striatum is context-sensitive; meaning that successful avoidance of a loss elicits a positive dopaminergic RL signal (Kishida et al., 2015, Palminteri et al., 2015). Importantly, learning from feedback is biased such that confirmatory feedback (i.e. obtained reward) is preferentially taken into account and this bias extends to counterfactual information (i.e. successfully avoided losses) (Palminteri, Lefebvre, Kilford, & Blakemore, 2017). This suggests that the hyperdopaminergic state in TS might increase learning from both rewards and successfully avoided losses (Palminteri et al., 2009, Palminteri et al., 2017) or increase choice stochasticity irrespective of factual and counterfactual information (Beeler, 2012, Hamid et al., 2015).

Cortical processes in probabilistic RL can be readily analyzed as event-related potentials (ERPs). The feedback-related negativity (FRN) is a fronto-central deflection from 200 to 300 ms following feedback presentation and encodes a PE signal (Fischer and Ullsperger, 2013, Sambrook and Goslin, 2015) which is supposed to stem from the medial frontal cortex (MFC). The P3a is a positive fronto-central ERP peaking between 300 and 500 ms following feedback and is thought to reflect allocation of attention toward relevant information (Polich, 2007). The P3b is a positive centro-parietal ERP from 400 to 600 ms which has been associated with updating of an internal prediction model (Fischer and Ullsperger, 2013, Polich, 2007). In healthy subjects, MFC activity has been linked to value updating for factual, but not counterfactual feedback while parietal activity predicted behavioral adaptation for both types of feedback (Fischer and Ullsperger, 2013, Jocham et al., 2014).

In this study, we aimed to further characterize probabilistic learning and its temporally resolved neural correlates in TS. We employed a probabilistic learning task while recording high-density EEG in TS patients and matched healthy controls. Using multiple variants of a standard RL model, we tested whether behavior is preferentially guided by confirmatory feedback. To assess the interrelation between behavior and neural activity, we employed single-trial regression analyses of model-derived predictors onto the EEG signal. We then compared the resulting regression weights between groups. We hypothesized impaired probabilistic learning in TS and expected these changes to be reflected in decreased regression weights. In an exploratory analysis we also evaluated whether the behavioral model parameters were related to clinical scores in the TS group.