Objectives/Aims Learning from experience and making decisions based on integrated environmental feedback is crucial for human functioning and wellbeing. Difficulties in learning and decision-making have been found in several psychiatric conditions. Pavlovian bias, a tendency to approach reward and remain passive in the face of punishment, can be advantageous in some situations, while in others, it can lead to maladaptive decisions and needs to be overcome by cognitive control. It has been suggested that healthy humans rely more heavily on Pavlovian bias when instrumental control over environmental reinforcers is compromised. In our study, we were focusing on the influence of transcranial direct current stimulation (tDCS) on Pavlovian bias during and after an intermittent loss of control over rewards and losses.

Methods In our pilot study, 19 adults underwent three blocks of an orthogonalized go-nogo reinforcement learning task. Blocks 1 and 3 had a response-feedback contingency of 70–30%, enabling learning via trial-and-error. In the second block, the outcome was independent of the participants’ responses (50%-50% contingency level). Cortical responses of all participants were recorded via EEG. Multi-electrode tDCS targeting the medial prefrontal cortex was administered in a randomised, double-blind placebo-controlled manner.

Results We conducted a repeated-measures ANOVA with ‘session’ (PRE x tDCS x POST), ‘valence’ (Win x Avoid) and ‘PB-congruency’ (Pavlovian bias congruent x incongruent) as within-factors, ‘group’ (Active stimulation x Sham) as the between- factor and ‘accuracy’ as the dependent variable. The interaction of ‘session’ and ‘PB- congruency’ with F(2,16)=2.62, p=0.09 were marginally significant, pointing towards slightly enhanced Pavlovian bias in the second block. However, the interaction of ‘session’, ‘PB-congruency’ and ‘group’ was not significant (F(2,16)=0.46, p=0.63). The evaluation of the feedback-related negativity (FRN) in the EEG revealed gradually increasing amplitudes in reward trials in the sham group, whereas we found a trend towards reduced FRN amplitude in the active group with F(2,13)=2.83, p=0.08.

Conclusions Our preliminary data show that a loss of control over feedbacks might increase the effect of Pavlovian bias on the choices of all participants. Although active tDCS seems to attenuate cortical responses during feedback evaluation, this effect is not accompanied by alterations in choice behaviour. Our data collection is still ongoing. The results from the full sample of 50 participants will be analysed by February 2020.