Understanding the neural basis of reinforcement learning

The project focuses on delineating neural mechanisms that drive reinforcement learning/value-based decision-making in healthy adults. In particular, we are interested in how the interplay between instrumental and Pavlovian valuation systems shapes choice behavior under various conditions.

We developed paradigms to systematically manipulate participants’ control over rewards/losses to understand how outcome controllability influences behavioral performance and its neural correlates (measured via electroencephalography or pupillometry). Moreover, we explore how non-invasive stimulation of regions within the prefrontal cortex affects task performance and neural activity. For this purpose, we have so far applied techniques such as transcranial direct current stimulation (tDCS), transcranial temporal interference stimulation (tTIS) and repetitive transcranial magnetic stimulation (rTMS). Finally, in two ongoing studies, we also investigate how experimentally induced heat pain interferes with choice behavior in healthy adults.

We investigated if non-invasive stimulation of the medial prefrontal cortex with transcranial direct current stimulation (tDCS) influenced how healthy adults respond to manipulation of outcome controllability in a reinforcement learning task. Our results indicate that, compared to a placebo condition, real stimulation enhanced approach tendencies to reward and withdrawal tendencies to losses, implying that this brain region is associated with the selection of choice strategies during value-based decision-making in a manner that resembles learned helplessness.

Sedlinská, T., Bolte, L., Melsæter, E., Mittner, M., & Csifcsák, G. (2023). Transcranial direct-current stimulation enhances Pavlovian tendencies during intermittent loss of control. Frontiers in Psychiatry, 14, 766. https://doi.org/10.3389/fpsyt.2023.1164208


We show that non-invasive stimulation of the medial prefrontal cortex via transcranial direct current stimulation (tDCS) prevents developing maladaptive choice strategies following reduced outcome (reward & loss) controllability in a reinforcement learning task. Thus, stimulating this region may be advantageous in conditions associated with impaired value-based decision-making that develops due to compromised control over environmental events.

Csifcsák, G., Bjørkøy, J., Kuyateh, S., Reithe, H., & Mittner, M. (2021). Transcranial direct current stimulation above the medial prefrontal cortex facilitates decision-making following periods of low outcome controllability. Eneuro, 8(5). https://doi.org/10.1523/ENEURO.0041-21.2021


This study showed that healthy adult participants' choice behavior during a reinforcement learning task is influenced by intermittent absence of outcome (reward & loss) controllability, relative to a reference group with retained control. While intermittent absence of control does not impair overall performance (choice accuracy), it leads to increased Pavlovian response tendencies (approach to potential reward & inhibition when facing potential loss) by the end of the task. Moreover, the effect of impaired outcome controllability is associated with ineffective top-down cognitive control over Pavlovian valuation, indexed by an electroencephalographic (EEG) signal arising from the medial prefrontal cortex, midline frontal theta power.

Csifcsák, G., Melsæter, E., & Mittner, M. (2020). Intermittent absence of control during reinforcement learning interferes with Pavlovian bias in action selection. Journal of cognitive neuroscience, 32(4), 646-663. https://doi.org/10.1162/jocn_a_01515