Revitalizing EEG Neurofeedback

A prominent group of scientists in Tulsa, Oklahoma that investigates EEG and rtfMRI neurofeedback (see earlier BioSource post related to worry and rumination) recently published an editorial in the American Journal of Psychiatry (Aupperle et al., 2025) in which they describe how EEG neurofeedback (EEG NFB) may experience a revitalization based on its integration of individualized precision training methods that address the hypothesized mechanisms of particular disorders. Their editorial references an article by Gou et al. (2025) in the same issue, which we summarized below.

What is the Science?

Addiction is increasingly understood not simply as a failure of will but as a disorder of brain systems that govern attention, salience, and control. Among people with methamphetamine use disorder, one of the most disabling impairments lies in response inhibition, the ability to stop or withhold an action that has been triggered automatically.

When a person with this disorder encounters drug-related cues, such as images of pipes, crystals, or familiar drug-using contexts, those cues become powerful signals that hijack attention and bias behavior toward use.

Neuroimaging and electrophysiological studies have repeatedly shown that these cues activate frontal, striatal, limbic, and insular circuits that overlap with the very systems responsible for inhibiting behavior. As a result, the brain becomes simultaneously pulled toward the drug and less able to stop itself.

The impaired response inhibition and salience attribution model of addiction proposes that drug cues acquire abnormally high motivational salience, crowding out cognitive resources that would otherwise support self-control. In practical terms, this means that when a cue appears, the brain of a person with methamphetamine use disorder becomes both more reactive and less regulated at the same time.

Event-related potential studies support this account by showing that the N2 component, a neural marker associated with inhibitory control, is reduced in people with methamphetamine dependence when they perform tasks that require them to suppress responses.

Traditional treatments, including cognitive behavioral therapy and standard rehabilitation programs, can improve coping and reduce use, but they do not directly target these cue-driven neural patterns.

The central scientific question addressed by Gou and colleagues (2025) is whether it is possible to teach people with methamphetamine use disorder to recognize and actively suppress their own cue-reactive brain states in real time, thereby restoring the neural conditions necessary for effective response inhibition.

What Did They Study?

The authors studied whether a form of cognition-guided electroencephalography neurofeedback (EEG NFB) could improve response inhibition in men with methamphetamine use disorder by training them to deactivate personalized brain patterns associated with methamphetamine cue reactivity.

Ninety-nine men undergoing residential rehabilitation in China participated in two related experiments. All met DSM-5 criteria for at least moderate methamphetamine use disorder and were abstinent during the study.

In the first sample, 66 participants were randomly assigned to either a real EEG NFB group that received feedback based on their own brain activity or a yoked control group that received feedback based on another participant's brain activity. In the second sample, 33 additional participants were assigned to either a second real EEG NFB group or a standard rehabilitation group that received no neurofeedback. This second sample served as an independent validation of the main findings.

Across both samples, the key outcome was performance on a go/no-go task using methamphetamine-related and neutral images. This task measures how well participants can withhold a response when they see drug cues and is a well-validated index of response inhibition in addiction research.

How Did They Do It?

The study combined machine learning, EEG signal processing, and closed-loop neurofeedback in an unusually sophisticated manner for clinical addiction research. Each participant first completed a methamphetamine cue reactivity task while an EEG was recorded from sixty-four scalp electrodes. During this task, they viewed methamphetamine-related images and neutral images.

From these data, the researchers extracted features in both the time and time-frequency domains that distinguished brain responses to drug cues from those to neutral cues.

Using these features, they trained a personalized support vector machine classifier for each participant. This classifier learned to recognize, with significantly better than chance accuracy, when the participant's brain was in a methamphetamine cue-reactive state versus a neutral state. These individualized classifiers formed the core of the neurofeedback system.

During 10 neurofeedback visits, each participant completed multiple cycles in which their ongoing EEG was fed into the classifier in real time. The classifier produced a probabilistic score indicating how strongly the current brain activity matched the participant's methamphetamine cue-reactivity pattern. That score was displayed to the participant as a line graph and was also used to update images on the screen.

When the score was high, indicating strong cue reactivity, the system showed images associated with higher craving. When the score was low, suggesting a more neutral brain state, the images became less provocative. Participants were instructed to discover mental strategies, such as relaxation or focused attention, that would lower the line on the screen, even though they were not told explicitly that this meant suppressing drug cue reactivity.

The yoked control group saw the same types of images and displays, but their feedback was based on another person's brain rather than their own. This ensured that any differences could be attributed to true self-regulation rather than simple exposure to cues or expectancy effects. The standard rehabilitation group in the second sample received only the facility's usual program of exercise, vocational training, and supportive therapy.

Before and after the 10 sessions, all participants completed questionnaires, a craving assessment, and the methamphetamine-specific go-no-go task. In the first sample, EEG was also recorded during the task to assess neural markers of inhibition.

What Did They Find?

The results showed a clear and coherent pattern across behavioral, neural, and learning measures.

In the first sample, participants receiving real neurofeedback progressively learned to deactivate their methamphetamine cue-reactive brain patterns across sessions.

The classifier's probabilistic scores decreased significantly over time, whereas no such trend was observed in the yoke control group. This learning also transferred to periods without visual feedback, indicating that participants could maintain a more neutral brain state even when the screen was blank.

This neural learning translated into meaningful behavioral change. On the go-no-go task, the real neurofeedback group showed a large and statistically robust increase in d-prime in the drug no-go condition, indicating improved ability to distinguish when to respond and when to inhibit responses in the presence of methamphetamine cues.

The yoked control group did not show a comparable improvement. Moreover, the degree to which a participant reduced their cue-reactivity score across training was significantly correlated with the degree to which their response inhibition improved, demonstrating a direct link between neural self-regulation and cognitive control.

Craving also decreased more in the real neurofeedback group than in the yoke group. Participants who learned to quiet their cue-reactive brain patterns experienced a greater reduction in subjective urge when exposed to a drug-use video.

In the second sample, the findings were replicated. Participants who received real neurofeedback showed significant improvements in response inhibition, whereas those who received standard rehabilitation alone did not.

A machine learning model trained on data from the first sample was able to predict who would benefit most from neurofeedback in the second sample, based on early training performance and baseline characteristics, indicating that treatment response is both measurable and, to some extent, foreseeable.

What Were the Strengths and Limitations?

A major strength of this study lies in its rigorous experimental control. The use of a yoked neurofeedback group ruled out the possibility that improvements were due merely to seeing drug images, receiving attention from researchers, or believing one was in a treatment condition.

The inclusion of a second independent sample with a standard rehabilitation comparison further strengthened causal inference and external validity.

Another strength is the intervention's personalized nature. Instead of targeting a generic EEG frequency band, the researchers identified individualized patterns of cue reactivity across the whole scalp and trained participants to suppress those specific signatures. This approach aligns closely with the neurobiology of addiction and avoids the oversimplifications that have limited some earlier neurofeedback studies.

The study also had limitations. All participants were men in a controlled residential setting, which means the findings cannot yet be generalized to women or to outpatient populations.

The follow-up period was short, so it remains unknown how long the improvements in inhibition and craving persist after training ends or whether they translate into reduced relapse in daily life.

Finally, although EEG provides excellent temporal resolution, it does not localize brain activity with the precision of imaging methods such as fMRI.

What Was the Impact on Neurofeedback?

This study provides some of the strongest evidence to date that neurofeedback can be used not just to modulate generic brain rhythms but to target and retrain clinically meaningful cognitive processes in addiction, as well as personalized EEG signatures of addictive cue reactivity.

By showing that methamphetamine cue-reactive brain patterns can be identified, decoded, and voluntarily suppressed, the authors bridge the gap between laboratory neuroscience and therapeutic intervention.

The implications extend beyond methamphetamine use disorder. Cue-induced failures of inhibition drive many forms of addiction and compulsive behavior. A cognition-guided neurofeedback approach that teaches people to recognize and dampen those brain states could, in principle, be adapted to other substances and even to non-substance compulsions.

Perhaps most importantly, the study reframes recovery as a form of neural skill learning that enables the reallocation of cognitive resources needed to maintain sobriety. Participants were not merely avoiding drugs or practicing coping strategies; they were learning to reshape the real-time dynamics of their own brains in the presence of temptation. That conceptual shift opens a new avenue for precision, brain-based interventions in the treatment of addictive behaviors.

More generally, Gou et al.'s (2025) research demonstrates how EEG NFB can be successfully used to target individualized EEG patterns related to the cognitive mechanisms underlying psychiatric disorders.

Aupperle et al. (2025) suggest that this, together with the greater accessibility and cost-effectiveness of EEG NFB compared to rtfMRI NFB, can revitalize interest in EEG NFB among basic scientists and clinical researchers.

Takeaways

1. Methamphetamine cues directly undermine response inhibition by activating brain networks that compete with self-control.

   
   

2. Personalized EEG classifiers can reliably identify each individual's cue-reactive brain state.

   
   

3. Cognition-guided neurofeedback enables patients to learn how to deactivate those states in real time.

   
   

4. Reductions in cue-reactive brain activity predict improvements in behavioral inhibition and craving.

   
   

5. This approach offers a promising, mechanistically grounded addition to standard addiction treatment.

Glossary

Barratt Impulsiveness Scale–11: a widely used questionnaire that measures different aspects of impulsivity.

classifier: a machine learning algorithm that assigns data to one of two or more categories.

cue reactivity: a pattern of neural, physiological, and behavioral responses triggered by drug-related stimuli.

d-prime: a signal detection index that quantifies how well a person discriminates between target and non-target stimuli.

electroencephalography: a noninvasive method for recording electrical activity of the brain from the scalp.

event-related potential (ERP): a time-locked brain response to a specific sensory or cognitive event.

go-no-go task: a behavioral paradigm used to measure response inhibition.

methamphetamine use disorder: a psychiatric condition characterized by compulsive methamphetamine use and associated cognitive and behavioral impairments.

neurofeedback: a technique that provides real-time information about brain activity so that individuals can learn to regulate it.

response inhibition: the cognitive ability to suppress an automatic or prepotent action.

support vector machine: a supervised learning algorithm used to classify data by finding the optimal separating boundary between categories.

References

Aupperle, R., White, E. J., & Misaki, M. (2025). Revitalizing EEG neurofeedback using precision-based approaches. The American Journal of Psychiatry, 182(9), 816–818. https://doi.org/10.1176/appi.ajp.20250657

Gou, H., Bu, J., Cheng, Y., Liu, C., Gan, H., Liu, M., Zhao, Q., Chen, X., Ren, J., Hong, W., Wang, R., Cao, Y., Yu, C., Chen, X., & Zhang, X. (2025). Improved response inhibition through cognition-guided EEG neurofeedback in men with methamphetamine use disorder. The American Journal of Psychiatry, 182(9), 861–877. https://doi.org/10.1176/appi.ajp.20240475

About the Author

Dr. John Raymond Davis is an adjunct lecturer in the Department of Psychiatry and Behavioural Neurosciences at McMaster University's Faculty of Health Sciences. His scholarly contributions include research on EEG changes in major depression and case studies on neurological conditions. ​

Support Our Friends