Real-Time fMRI Neurofeedback Modulation of Dopaminergic Midbrain Activity in Young Adults With Elevated Internet Gaming Disorder Risk: Randomized Controlled Trial

Introduction

Internet gaming disorder (IGD) had a global prevalence of 10.4% by 2022; it disproportionately affects younger individuals [1]. The effectiveness of traditional psychosocial and pharmacological treatments, however, remains inconclusive [2]. Real-time functional magnetic resonance imaging neurofeedback (rt-fMRI NF) training allows for self-regulation of brain activity patterns and shows promise for several psychiatric conditions [3]. This preregistered, randomized, single-blind study examined rt-fMRI NF as a potential intervention for IGD. Given the key role and potential modifiability of the ventral tegmental area (VTA), a dopaminergic midbrain region involved in reward processing functions [4,5], we validated the association between VTA response and gaming addiction severity in a small sample (N=9). In a larger study (N=20), we evaluated whether 2 neurofeedback training sessions for VTA downregulation reduced craving for internet gaming.

Methods

Overview

We recruited young participants with IGD risk through social media–based advertisements. In study 1, 9 participants performed a gaming video–based cue-reactivity task (Multimedia Appendix 1, Figure S1) to establish the association between VTA cue-reactivity and IGD symptom level [5]. In study 2, 20 different participants were screened and randomly assigned to 1 of 2 groups. The inclusion criteria and screening instruments are described in Multimedia Appendix 1, Figures S2 and S3. The experimental group received feedback from the VTA (Montreal Neurological Imaging [MNI] coordinates [1, –17, –13]; 246 voxels; Figure 1B), while the control group received sham feedback from the right middle temporal gyrus, specifically an 8-mm–radius sphere centered at MNI coordinates [53, 6, –18]. This region was chosen because it is unrelated to reward processing and is not functionally connected with the VTA. Multimedia Appendix 1 describes the training details. Briefly, during neurofeedback training, blood-oxygen-level–dependent signals in the feedback regions were presented to participants as line graphs in real time (Figure 1A). Both groups were instructed to downregulate the signals using mental strategies that are helpful against craving; they reported their effort and perceived success after each run (Multimedia Appendix 1, Table S1).

Participants completed a 1-item visual analog scale (VAS) for current craving levels at baseline, after the second imaging session, and a month after the last scan; VAS scores were the primary outcome measure. The neural-level VTA cue-reactivity and inhibitory control performance in an affective go/no-go task [6] (Multimedia Appendix 1, Figure S4) measured at baseline and after the second training session were the secondary outcome measures. Analysis of the outcome measures used JASP (version 0.18.3; JASP Team) and the nparLD toolbox [7].

‎

Ethical Considerations

Both studies were approved by the Panel on Research Ethics of the University of Macau (BSERE23-APP004-ICI-M1). The neurofeedback study was preregistered at ClinicalTrials.gov (NCT06063642). Participants provided written informed consent prior to participation and were compensated with MOP 200 (US $24.96) supermarket vouchers. All data were deidentified.

Results

In study 1, the cue-reactivity task successfully activated the VTA (P=.004 after correction for false-discovery rate at a whole-brain level; Multimedia Appendix 1, Figure S5A). Moreover, the VTA response was significantly correlated with addiction severity (Pearson r=.733; P=.02; Figure S5B).

Group comparison results for study 2 are shown in Table 1, and the baseline characteristics of the neurofeedback completers can be found in Multimedia Appendix 1, Table S2. Both groups showed significantly decreased subjective craving levels (P=.002; Multimedia Appendix 1, Figure S6) with no significant group differences (P=.10). Over the course of the neurofeedback sessions, the experimental group downregulated their VTA activity to a greater extent than the control group (P=.02). In contrast to sham feedback, VTA neurofeedback led to more pronounced decreases in VTA cue-reactivity (Figure 1C; P=.06), a trend that might suggest successful transfer of learned VTA regulation. No significant interaction or group effects were found for inhibitory control (accuracy and reaction time in the go/no-go task: P>.12 for all groups).

Discussion

We previously proposed rt-fMRI NF as a promising therapeutic strategy for individuals with internet addiction [3]. The efficacy of rt-fMRI NF, however, has not been empirically examined for any behavioral addictions. In this trial, we found that individuals with problematic internet gaming behaviors could learn to downregulate their VTA activity through genuine feedback, and that this effect translated into changes in VTA reactivity to gaming stimuli. Although we found no improvements at the behavioral level after NF, the neural activity findings underscore the potential of VTA NF training in restoring reward processing functions in individuals with IGD risk.

The study has several limitations. First, our participants were not diagnosed with IGD, meaning that their symptoms might not have been severe enough to benefit from the intervention. Furthermore, due to the small sample size, this study was potentially underpowered to detect group differences in behavioral changes. Larger-scale trials are needed to further examine the clinical efficacy of this intervention protocol.