rt-fMRI Neurofeedback for Alcohol Use Disorder: Trial Reports Mixed Results

A randomized controlled trial of real-time fMRI neurofeedback targeting alcohol cue reactivity reports meaningful implementation challenges alongside its clinical insights, according to a preprint posted on medRxiv. The study adds a cautionary, method-focused data point to a neurofeedback literature that often leans on small, uncontrolled demonstrations.

The intervention under study is real-time functional MRI neurofeedback — not scalp EEG neurofeedback, and not any externally recorded EEG modality. Participants in the trial received concurrent feedback of their own blood-oxygen-level-dependent (BOLD) signal while viewing alcohol-related cues, with the goal of training voluntary down-regulation of cue-reactivity responses implicated in craving and relapse in alcohol use disorder.

The authors frame their report as much as a lessons-learned document as a clinical outcome paper. The title itself signals this dual purpose: "Challenges and Insights from a Randomized Controlled Trial."

Why a randomized controlled trial matters here

Most neurofeedback findings in addiction research come from single-arm studies or small pilot samples, which cannot separate specific feedback effects from placebo, expectancy, or nonspecific scanner exposure. By randomizing patients with alcohol use disorder, this trial addresses that evidentiary gap directly.

The randomized design allows the investigators to compare feedback-trained regulation of cue-reactive BOLD activity against control conditions, and to examine whether learned regulation translates into clinical measures such as craving. The preprint emphasizes that running such trials in this population is hard — recruitment, retention, and protocol adherence in patients with alcohol use disorder all present obstacles the authors document in detail.

What the modality is — and is not

Real-time fMRI neurofeedback differs fundamentally from the EEG-based neurofeedback familiar to most clinical neurophysiology readers. The signal is hemodynamic, acquired in the scanner, with seconds-long temporal resolution but high spatial localization to deep and cortical structures involved in reward and salience processing.

The trial targeted cue reactivity — the neural response elicited by alcohol-associated stimuli — rather than resting-state oscillations or event-related potentials. That target choice aligns with the dominant mechanistic account of relapse risk in alcohol use disorder, in which heightened cue-elicited reactivity drives craving and subsequent drinking.

Practical reading for neurofeedback practitioners

For labs and clinics running neurofeedback protocols in any modality, the trial's methodological disclosures are the actionable content:

• Randomized controlled designs are feasible in alcohol use disorder populations, though attrition and compliance demand explicit planning.

• Real-time fMRI feedback protocols carry costs and logistics that limit scalability relative to EEG-based approaches, a constraint the field must weigh when translating findings.

• Null or mixed results, when they occur, carry information: the authors use their data to interrogate what works and what does not in cue-reactivity training.

Because the paper appears on medRxiv, it has not yet completed peer review; readers should treat the findings and the authors' framing of the challenges as provisional until journal publication.

Investigational status

Real-time fMRI neurofeedback for alcohol use disorder remains investigational. No device or protocol of this class holds FDA clearance for substance use disorder treatment, and nothing in this trial changes that regulatory picture. The work is human research, conducted under trial conditions, and distinct from any cleared therapeutic neurofeedback application.

What comes next

The preprint signals that the field's next step is not more demonstrations that regulation can be learned, but rigorous tests of whether learned regulation changes clinical outcomes — and clearer accounting of why trials in this population are difficult to run. Peer review and full publication of the trial's outcome data will determine how much weight this study can carry in that debate.