Longitudinal Effects of Brain Stimulation on Decision-Making and Reward Learning

Longitudinal Effects of Brain Stimulation Techniques (ECT, TMS, VNS, taVNS) on Decision-Making and Reward Learning

This study investigates the longitudinal effects of brain stimulation treatments on decision-making and reward learning. Here, we use daily assessments of reinforcement learning behavior in patients with depression who receive neuromodulation treatment to study potential associations between changes in decision-making and reward learning and changes in clinical symptoms.

Study Overview

• Status: Not yet recruiting
• Conditions: Depression / Major Depressive Disorder

Detailed Description

Brain stimulation techniques are increasingly applied to treat mental disorders, especially when psychotherapy and psychopharmacotherapy fail (Conroy & Holtzheimer, 2021). The efficacy of treating treatment-resistant depression has been demonstrated using electroconvulsive therapy (ECT; Espinoza & Kellern, 2022; Subramanian et al., 2022), transcranial magnetic stimulation (TMS, Trapp et al., 2025), and vagus nerve stimulation (VNS, Kamel et al., 2022). With regard to transcutaneous auricular vagus nerve stimulation (taVNS), a motivation-enhancing effect has been demonstrated in individuals with depression (Ferstl et al., 2024).

It is unclear to what extent these stimulation techniques also affect cognitive processes, such as learning and decision-making behavior, and whether such changes can explain treatment success. Temporary cognitive side effects, such as memory impairments, are known to occur with ECT treatments (Landry et al., 2021). Initial evidence regarding learning and decision-making behavior is available for taVNS, in which the vagal auricular branch is noninvasively stimulated at the ear. Acute taVNS stimulation at the left ear showed a reduced learning rate particularly during punishment (Kühnel et al., 2020), as well as increased reward sensitivity (Weber et al., 2021). Additionally, TMS over the left dorsolateral prefrontal cortex was shown to alter the ratio of reward-to-punishment learning rates (Biernacki et al., 2023). However, these initial findings on learning and decision-making behavior and the underlying computational parameters relate to acute stimulation. Long-term effects of brain stimulation techniques on theses processes, as used in routine clinical practice (e.g., eight TMS sessions or daily stimulation of the vagus nerve) remain unclear to date. Likewise, the influence of fluctuations in mental state, such as mood, on learning and decision-making behavior over the course of a treatment, as well as potential changes in this influence after successful treatment, remain mostly unexplored. However, a recent study found, that fluctuations in metabolic state affected reward sensitivity and punishment learning rates in obesity (Kühnel et al., 2025). In the context of effort-based decision-making, it has been shown that fluctuations in motivation influence reward sensitivity (Hewitt et al., 2025).

It is also not yet possible to say to what extent stimulation-induced changes in clinical symptoms are reflected in possible changes in decision-making and reward learning. However, increasing evidence of altered computational mechanisms in patients points towards this potential connection. For example, higher punishment learning rates were observed in individuals with depression (Pike & Robinson, 2022) and anhedonia was linked to reduced reward sensitivity (Huys & Browning, 2025). Also, greater temporal discounting of future rewards was observed in individuals with depression compared to those without a diagnosis (Amlung et al., 2019). An open question remains, that is whether, as symptoms improve, changes in cognitive processes also diminish.

Based on the above, the following objectives have been established: First, the acute and mid-term effects of the mentioned neurostimulation therapies on computational processes of decision-making and reward learning will be investigated over the course of a clinical application. We expect, that brain stimulation lowers punishment learning rates (Hypothesis 1). Subsequently, the effects of the various stimulation methods on punishment learning rates will be compared (acutely after stimulation as well as over the course of the treatment). In addition, the influence of state fluctuations, such as mood, on computational parameters of decision-making and reward learning, such as learning rates, will be modeled. We expect, that fluctuations in mood, motivation, and metabolic state are associated with changes in punishment learning rates (Hypothesis 2). Next, we investigate the effect of brain stimulation on working memory. We expect, that ECT impairs working memory compared to the other treatments (Hypothesis 3). Finally, we expect that changes in punishment learning rates predict changes in symptoms of depression, general well-being, and somatic symptoms (Hypothesis 4), as well as cognitive side effects on working memory (Hypothesis 5).

Here, we use daily assessments of reinforcement learning with a gamified online task (Neuser et al., 2023) over the course of 8 weeks to track longitudinal effects of brain stimulation on decision-making and reward learning in patients with depression who receive neuromodulation treatment (N = 100). Daily assessments are accompanied by an ecological momentary assessment (EMA) of mood and metabolic states. At least five runs of the reinforcement learning task should be played before the first stimulation as a baseline. Additionally, working memory (backward digit span task), depressive symptoms (BDI-II), well-being (WHO-5), and somatic symptoms (PHQ-15) are measured online at the start of the study before the first treatment, after 4 weeks, and after 8 weeks at the end of the study. Patients will be recruited directly from the clinic after brain stimulation treatment is indicated for them..

• Study Type: Observational
• Enrollment (Estimated): 100

Contacts and Locations

• Study Contact: Name: Nils B Kroemer, Prof. Dr.; Phone Number: +49 228 287 11151; Email: nkroemer@uni-bonn.de

Study Locations

• Germany
  • Bonn, Germany, 53127
    • Section of Medical Psychology, Department of Psychiatry & Psychotherapy, Faculty of Medicine, University of Bonn
    • Contact: Prof. Dr. rer. nat. Nils B Kroemer; Phone Number: +49 228 28719123; Email: nkroemer@uni-bonn.de

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

The study population consists of patients in the Department of Psychiatry and Psychotherapy at the University Hospital Bonn, who receive an indication for a brain stimulation treatment (ECT, TMS, VNS, taVNS).

Description

Inclusion Criteria:

• Indication for a brain stimulation treatment (ECT, TMS, VNS, taVNS) in the Department of Psychiatry and Psychotherapy at the University Hospital Bonn
• Be able and willing to provide informed consent.

Exclusion Criteria:

• Non-German speakers
• Unclear ability to give consent to the study participation

Study Plan

How is the study designed?

Number of groups / cohorts

5

Cohorts and Interventions

Group / Cohort
Electroconvulsive therapy (ECT); Participants receive ECT as a treatment within the clinic. ECT intentionally causes a generalized seizure by passing electrical currents through the brain under anesthesia. The treatment protocol (e.g., number of treatments) is independent of study participation and follows the doctor's orders. Participants are enrolled after the treatment indication is confirmed.
Transcranial magnetic stimulation (TMS); Participants receive TMS as a treatment within the clinic. TMS is a non-invasive procedure in which a magnetic coil is used to induce electrical currents in the brain and thus influence cortex activity. The treatment protocol (e.g., number of treatments, type of TMS) is independent of study participation and follows the doctor's orders. Participants are enrolled after the treatment indication is confirmed.
invasive vagus nerve stimulation (VNS); Participants receive VNS as a treatment within the clinic. VNS is a surgical treatment in which a stimulation device is implanted that sends electrical impulses to the vagus nerve. The treatment protocol (e.g., stimulation settings) is independent of study participation and follows the doctor's orders. Participants are enrolled after the treatment indication is confirmed.
transcutaneous vagus nerve stimulation (tVNS); Participants receive tVNS as a treatment from the clinic for at home stimulation. To stimulate vagal afferents, the electrode will be placed at the cymba conchae of the right ear using a previously established conventional stimulation protocol (30s ON, 30s OFF; tVNS E device, tVNS Technologies GmbH, Erlangen, Germany). The stimulation can be applied for up to 4h per day and participants self-select stimulation time and duration based on the doctor's orders. Participants are enrolled after the treatment indication is confirmed.
Control; Participants receive an indication for a stimulation treatment within the clinic, but do not make use of it. Participants are included after treatment indication.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Punishment learning rates	The primary outcome are punishment learning rate estimates from a computational reinforcement learning model. Learning rates will be compared within and between stimulation conditions. Reinforcement learning is repeatedly measured with a bandit task with fluctuating reward probabilities (reward learning task). Reward learning behavior will be collected online over up to 60 runs, each including 150 trials.	Assessed online up to 5 times before the first stimulation treatment and daily throughout the study (8 weeks).

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Stimulation-induced mid-term changes in the backward digit span	The primary outcome is the change in the backward digit span. Changes over time will be compared within conditions and between ECT and the other conditions. The backward digit span is the maximum sequence length that could be repeated correctly backwards in the digit span task (at least one out of two correct). The task runs two sequences per sequence length, starting at two digits and increasing by one digit up to eight digits. The task is terminated if both sequences of a sequence length could not be repeated correctly backwards. Scores range from 0 to 16 and are the number of correctly repeated sequences.	Assessed before the first stimulation treatment and after 4 and 8 weeks (10 minutes).
Correct choices in the reward learning task	The outcome describes the percentage of correct choices (wins) collected during the reward learning task. Choices are collected in each of the 150 trials per run.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Reward sensitivity	The primary outcome is reward sensitivity estimates from a computational reinforcement learning model reflected by the inverse temperature parameters. Changes over time will be compared within and between stimulation conditions. Reinforcement learning is repeatedly measured with an bandit task with fluctuating reward probabilities (reward learning task). Reward learning behavior will be collected online over up to 60 runs, each including 150 trials.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Weighting of learned values and rewards at stake (lambda)	The primary outcome is the weighting of win probability to reward magnitude weighting (lambda) from a computational reinforcement learning model. Changes over time will be compared within and between stimulation conditions. Reinforcement learning is repeatedly measured with an bandit task with fluctuating reward probabilities (reward learning task). Reward learning behavior will be collected online over up to 60 runs, each including 150 trials.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Stimulation-induced mid-term changes in reward learning rates	The primary outcome is the change over time in reward learning rate estimates from a computational reinforcement learning model. Changes over time will be compared within and between stimulation conditions. Reinforcement learning is repeatedly measured with an bandit task with fluctuating reward probabilities (reward learning task). Reward learning behavior will be collected online over up to 60 runs, each including 150 trials.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between mood state and punishment learning rates	The primary outcome is the association (regression coefficient) quantifying the relationship between mood state and punishment learning rates. Punishment learning rates are measured in the reward learning task. Mood state is calculated from two 0-100 visual analog scales (state happiness minus state sadness), which are asked before every run of the reward learning task.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between metabolic state and punishment learning rates	The primary outcome is the association (regression coefficient) quantifying the relationship between metabolic state and punishment learning rates. Punishment learning rates are measured in the reward learning task and metabolic state is asked before every run of the reward learning task on a 0-100 visual analog scale (sated - hungry).	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between changes in punishment learning rates and changes in depressive symptoms	The primary outcome is the association (regression coefficient) quantifying the relationship between changes in punishment learning rates and changes in depressive symptoms. The regression model predicts changes in depressive symptoms from changes in punishment learning rates. Depressive symptoms are measured with the BDI-II (total scores, ranging from 0 to 63). Punishment learning rates are measured in the reward learning task.	Depressive symptoms are assessed before the first stimulation treatment and after 4 and 8 weeks. Reward learning is assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between motivational state and punishment learning rates	The primary outcome is the association (regression coefficient) quantifying the relationship between motivational state and punishment learning rates. Punishment learning rates are measured in the reward learning task and motivational state is asked before every run of the reward learning task on a 0-100 visual analog scale (not motivated - motivated).	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between changes in punishment learning rates and changes in mental well-being	The primary outcome is the association (regression coefficient) quantifying the relationship between changes in punishment learning rates and changes in mental well-being. The regression model predicts changes in mental well-being from changes in punishment learning rates. Mental well being is measured with the WHO-5 (total scores, range from 0 to 30). Punishment learning rates are measured in the reward learning task.	Mental well being is assessed before the first stimulation treatment and after 4 and 8 weeks. Reward learning is assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between changes in punishment learning rates and changes in somatic symptoms	The primary outcome is the association (regression coefficient) quantifying the relationship between changes in punishment learning rates and changes in somatic symptoms. The regression model predicts changes in somatic symptoms from changes in punishment learning rates. Somatic symptoms are measured with the PHQ-15 (total scores, range from 0 to 30). Punishment learning rates are measured in the reward learning task.	Somatic symptoms are assessed before the first stimulation treatment and after 4 and 8 weeks. Reward learning is assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Association between changes in the backward digit span and changes in punishment learning rates	The primary outcome is the association (regression coefficient) quantifying the relationship between changes in the backward digit span and changes in punishment learning rates. The backward digit span is measured in the backward digit span task. Punishment learning rates are measured in the reward learning task.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Response times in the reward learning task	The outcome describes response times collected during the reward learning task. Response times (milliseconds) are collected in each of the 150 trials per run.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Backward digit span	The primary outcome is the maximum sequence length that could be repeated correctly backwards (at least one out of two correct). The task runs two sequences per sequence length, starting at two digits and increasing by one digit up to eight digits. The task is terminated if both sequences of a sequence length could not be repeated correctly backwards. Scores range from 2 to 8.	Assessed before the first stimulation treatment and after 4 and 8 weeks (10 minutes).
BDI-II (Beck Depression Inventory-II)	Questionnaire assessing depressive symptoms. Will be associated with behavioral outcomes. Total scores range from 0 to 63, with higher scores indicating greater depressive symptom severity.	Assessed before the first stimulation treatment and after 4 and 8 weeks (10 minutes).
WHO-5 (World Health Organization-Five Well-Being Index)	Questionnaire assessing mental well being. Will be associated with behavioral outcomes. Total scores range from 0 to 30, with higher scores indicating better mental well being.	Assessed before the first stimulation treatment and after 4 and 8 weeks (3 minutes).
PHQ-15 (Patient Health Questionnaire-15)	Questionnaire assessing the severity of somatic symptoms. Will be associated with behavioral outcomes. Total scores range from 0 to 30, with higher scores indicating higher severity of somatic symptoms.	Assessed before the first stimulation treatment and after 4 and 8 weeks (5 minutes).

Other Outcome Measures

Outcome Measure	Measure Description	Time Frame
Mood state	Ecological Momentary Assessment (EMA) using two 0-100 visual analog scales (mood state is calculated as state happiness minus state sadness), administered repeatedly across the reward learning task and lab sessions to capture mood state. We will assess changes in mood and associations with learning.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Metabolic state	Ecological Momentary Assessment (EMA) using a 0-100 visual analog scale (sated - hungry), administered repeatedly across the reward learning task and lab sessions to capture metabolic state. We will assess changes in metabolic state and associations with learning.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).
Motivational state	Ecological Momentary Assessment (EMA) using a 0-100 visual analog scale (unmotivated - motivated), administered repeatedly across the reward learning task and lab sessions to capture motivation. We will assess changes in motivation and associations with learning.	Assessed online at least 5 times before the first stimulation treatment and daily throughout the study (8 weeks).

Collaborators and Investigators

• Sponsor: Dr. Nils B. Kroemer
• Collaborators: Department of Psychiatry University of Bonn
• Investigators: Principal Investigator: Nils B Kroemer, Prof. Dr., Section of Medical Psychology, Department of Psychiatry & Psychotherapy, Faculty of Medicine, University of Bonn

Publications and helpful links

General Publications

• Espinoza RT, Kellner CH. Electroconvulsive Therapy. N Engl J Med. 2022 Feb 17;386(7):667-672. doi: 10.1056/NEJMra2034954. No abstract available.
• Neuser MP, Kuhnel A, Krautlein F, Teckentrup V, Svaldi J, Kroemer NB. Reliability of gamified reinforcement learning in densely sampled longitudinal assessments. PLOS Digit Health. 2023 Sep 6;2(9):e0000330. doi: 10.1371/journal.pdig.0000330. eCollection 2023 Sep.
• Huys QJM, Browning M. A Computational View on the Nature of Reward and Value in Anhedonia. Curr Top Behav Neurosci. 2022;58:421-441. doi: 10.1007/7854_2021_290.
• Kuehnel A, Zietz J, Theuer JK, Neuser MP, Kroemer NB; Dense sampling of choices links high learning rates to obesity and low reward sensitivity to binge eating; medRxiv; 2025 Jun 20
• Weber I, Niehaus H, Krause K, Molitor L, Peper M, Schmidt L, Hakel L, Timmermann L, Menzler K, Knake S, Oehrn CR. Trust your gut: vagal nerve stimulation in humans improves reinforcement learning. Brain Commun. 2021 Mar 14;3(2):fcab039. doi: 10.1093/braincomms/fcab039. eCollection 2021.
• Trapp NT, Purgianto A, Taylor JJ, Singh MK, Oberman LM, Mickey BJ, Youssef NA, Solzbacher D, Zebley B, Cabrera LY, Conroy S, Cristancho M, Richards JR, Flood MJ, Barbour T, Blumberger DM, Taylor SF, Feifel D, Reti IM, McClintock SM, Lisanby SH, Husain MM; National Network of Depression Centers Neuromodulation Task Group. Consensus review and considerations on TMS to treat depression: A comprehensive update endorsed by the National Network of Depression Centers, the Clinical TMS Society, and the International Federation of Clinical Neurophysiology. Clin Neurophysiol. 2025 Feb;170:206-233. doi: 10.1016/j.clinph.2024.12.015. Epub 2024 Dec 19.
• Subramanian S, Lopez R, Zorumski CF, Cristancho P. Electroconvulsive therapy in treatment resistant depression. J Neurol Sci. 2022 Mar 15;434:120095. doi: 10.1016/j.jns.2021.120095. Epub 2021 Dec 18.
• Pike AC, Robinson OJ. Reinforcement Learning in Patients With Mood and Anxiety Disorders vs Control Individuals: A Systematic Review and Meta-analysis. JAMA Psychiatry. 2022 Apr 1;79(4):313-322. doi: 10.1001/jamapsychiatry.2022.0051.
• Landry M, Moreno A, Patry S, Potvin S, Lemasson M. Current Practices of Electroconvulsive Therapy in Mental Disorders: A Systematic Review and Meta-Analysis of Short and Long-Term Cognitive Effects. J ECT. 2021 Jun 1;37(2):119-127. doi: 10.1097/YCT.0000000000000723.
• Kuhnel A, Teckentrup V, Neuser MP, Huys QJM, Burrasch C, Walter M, Kroemer NB. Stimulation of the vagus nerve reduces learning in a go/no-go reinforcement learning task. Eur Neuropsychopharmacol. 2020 Jun;35:17-29. doi: 10.1016/j.euroneuro.2020.03.023. Epub 2020 May 11.
• Kamel LY, Xiong W, Gott BM, Kumar A, Conway CR. Vagus nerve stimulation: An update on a novel treatment for treatment-resistant depression. J Neurol Sci. 2022 Mar 15;434:120171. doi: 10.1016/j.jns.2022.120171. Epub 2022 Jan 29.
• Hewitt SRC, Norbury A, Huys QJM, Hauser TU. Day-to-day fluctuations in motivation drive effort-based decision-making. Proc Natl Acad Sci U S A. 2025 Mar 25;122(12):e2417964122. doi: 10.1073/pnas.2417964122. Epub 2025 Mar 17.
• Ferstl M, Kuhnel A, Klaus J, Lin WM, Kroemer NB. Non-invasive vagus nerve stimulation conditions increased invigoration and wanting in depression. Compr Psychiatry. 2024 Jul;132:152488. doi: 10.1016/j.comppsych.2024.152488. Epub 2024 Apr 16.
• Conroy SK, Holtzheimer PE. Neuromodulation Strategies for the Treatment of Depression. Am J Psychiatry. 2021 Dec;178(12):1082-1088. doi: 10.1176/appi.ajp.2021.21101034.
• Biernacki K, Myers CE, Cole S, Cavanagh JF, Baker TE. Prefrontal transcranial magnetic stimulation boosts response vigour during reinforcement learning in healthy adults. Eur J Neurosci. 2023 Feb;57(4):680-691. doi: 10.1111/ejn.15905. Epub 2022 Dec 30.
• Amlung M, Marsden E, Holshausen K, Morris V, Patel H, Vedelago L, Naish KR, Reed DD, McCabe RE. Delay Discounting as a Transdiagnostic Process in Psychiatric Disorders: A Meta-analysis. JAMA Psychiatry. 2019 Nov 1;76(11):1176-1186. doi: 10.1001/jamapsychiatry.2019.2102.

Study record dates

Study Major Dates

• Study Start (Estimated): September 1, 2026
• Primary Completion (Estimated): August 1, 2028
• Study Completion (Estimated): August 1, 2028

Study Registration Dates

• First Submitted: September 10, 2026
• First Submitted That Met QC Criteria: September 10, 2026
• First Posted (Actual): September 16, 2026

Study Record Updates

• Last Update Posted (Actual): September 16, 2026
• Last Update Submitted That Met QC Criteria: September 10, 2026
• Last Verified: September 1, 2026

More Information

Terms related to this study

• Keywords: Depression; repetitive transcranial magnetic stimulation; Brain Stimulation; ECT; VNS; Reinforcement Learning; tVNS
• Additional Relevant MeSH Terms: Mental Disorders; Behavioral Symptoms; Mood Disorders; Depressive Disorder; Behavior; Depression; Depressive Disorder, Major
• Other Study ID Numbers: BON009

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: After the publication of the key results of the study, anonymized data will be made publicly available (e.g., at osf.org).
• IPD Sharing Time Frame: Anonymized data will be shared upon publication of the results
• IPD Sharing Access Criteria: Until the data is publicly available, researchers may contact the lead PI to gain access.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No