Commentary on Wan <i>et al</i> .: Optimizing the neurofeedback protocol in addiction treatment

The inconsistent efficacy of EEG-NF is driven by a mechanistic mismatch compounded by methodological heterogeneity. Future research must shift from general efficacy to mechanism-driven protocol design and unified standards (CRED-nf) to identify effective parameters and specific patient subgroups, enabling reliable clinical application. Neurofeedback (NF) is a form of biofeedback grounded in clinical and behavioral neuroscience that enables individuals to manipulate neural processes by learning to self-regulate brain physiological activity in real-time [1]. Although neurofeedback of brain oscillations based on electroencephalography (EEG-NF) has become the predominant NF modality because of its low-cost and accessibility, its therapeutic utility in drug addiction remains inconsistent [2]. This inconsistency stems from multiple factors, including lack of mechanistic specificity inherent to conventional EEG-NF paradigms to fully engage the dynamic, prefrontal cortex-mediated imbalance that sustains drug-seeking behavior [3]. Another limitation is the small number of well-powered studies available to conclusively establish efficacy and to define optimal NF protocols [4]. The recent systematic review and meta-analysis by Wan and colleagues [5] provides quantitative evidence supporting efficacy of EEG-NF in both substance and behavioral addiction. Importantly, it identifies feedback modality and the number of sessions as primary factors influencing outcomes. The reported findings highlight the need to move beyond demonstrating overall efficacy and instead focus on detailed protocol development and a rigorous mechanistic understanding of which parameters drive therapeutic effects. Across the trials included in this analysis, outcome measures focused primarily on craving and symptom severity linked to relapse vulnerability. Craving is a core clinical manifestation driven by sensitized mesolimbic dopaminergic signaling and heightened attentional bias, consistent with the incentive sensitization framework [3]. Chronic drug exposure exaggerates cue-driven motivational salience, culminating in hyper-reactive ventral striatal responses [6]. Although most EEG-NF protocols aim to reduce dysregulated processes by training oscillatory patterns linked to arousal or inhibitory control [7], the observed symptom reductions suggest these modulations can influence craving-related neural states. However, the mixed findings point to a critical mismatch: broad target selection does not adequately capture the specific, dysregulated neural circuits that shape individual relapse profiles. The meta-analysis identified ‘feedback modality’ as a key parameter mediating inconsistency. This parameter is often treated as a neutral technical choice, even though the initially observed benefit of using auditory feedback disappears on outlier exclusion. Therefore, a precise mechanistic interpretation necessitates further research and should be considered provisional. Visual feedback, which dominates the literature, risks engaging fronto-striatal reward circuits in unintended ways. For individuals with heightened cue-reactivity, visual stimuli trigger non-specific attentional capture and striatal responsivity [3]. This incidental circuit engagement can compete with the self-regulation task, potentially weakening therapeutic impact [8]. Conversely, non-visual modalities (auditory or haptic) may provide a cleaner neurophysiological channel by minimizing reward-circuit biases. Future protocol design must align sensory delivery channels with known neurocognitive vulnerabilities in addiction phenotypes. The heterogeneity revealed in this meta-analysis further highlights the urgency of methodological standardization in neurofeedback research. Despite growing interest, the field still lacks consistent reporting of protocol architecture, session dynamics, control conditions and neurophysiological targets. The Consensus on the Reporting and Experimental Design of Clinical and Cognitive-Behavioural Neurofeedback Studies (CRED-nf) provides a robust framework for addressing this gap [9]. Unlike modality-specific guidelines, CRED-nf focuses on principles unique to neurofeedback: signal quality, real-time processing, reward contingencies, training structure and mechanistic justification of the targeted neural markers. Adopting this framework, alongside Consolidated Standards of Reporting Trails (CONSORT) standards, is vital for improving reproducibility and systematically determining which protocol components (e.g. frequency bands, sensory modalities or dose parameters) drive therapeutic change. This methodological rigor is key to advancing EEG-NF from experimental trials to an evidence-based, reliable clinical application in addiction. Taken together, this review positions EEG-NF as a promising neuromodulation strategy for addiction. However, future impact of EEG-NF depends critically on a paradigm shift toward mechanism-driven protocol design and transparent methodological rigor. Interventions must increasingly incorporate a clear neuroscientific rationale, particularly concerning how protocol choices, such as feedback modality, interact with addiction circuitry. Emerging multi-modal imaging approaches and personalized state monitoring create opportunities for more adaptive and precision-oriented paradigms in treatment [10]. Moreover, targeted clinical trials are necessary to identify which subgroups (e.g. high cue-reactivity or impulsive/compulsive phenotypes) benefit most from distinct NF strategies. EEG-NF will only achieve reliable clinical utility through mechanistic rigor and methodological unity. Di Zhao: Conceptualization (equal); writing—original draft (equal). Björn Apfel: Conceptualization (equal); writing—review and editing (equal). Surjo R. Soekadar: Conceptualization (equal); writing—review and editing (equal). D.Z. is supported by National Natural Science Foundation of China grant (82422029, 82271530), the Fundamental Research Funds for the Central Universities (YG2025ZD07), the Science and Technology Commission of Shanghai Municipality (24Y22800200), and China Scholarship Council (202506230324). B.A. and S.R.S. are supported by the German Research Foundation (DFG: Project-ID 402170461–TRR 265), the European Research Council (ERC) under project BNCI2 (101088715) and the Einstein Foundation Berlin, Germany. None. This is a commentary with no data.