Neurofeedback for ADHD: Why a Credible Sham Changes the Evidence

Evidence scope: Prepared for an October 30, 2026 editorial reference date using the dated primary sources linked below. This is not an exhaustive search through that date; later evidence, corrections, and regulatory decisions require a fresh check before publication. No claim of current approval or availability is implied.

Disclosure: This article is not diagnosis, treatment advice, or a substitute for individualized care. Do not change treatment based on this article.

Neurofeedback offers an appealing account of ADHD treatment: measure brain activity, provide real-time feedback, and help a person learn a different pattern. The account is testable, which is a strength. But improvement during training does not by itself establish that feedback based on the person's brain signals caused the improvement. Attention from staff, repeated practice, expectations, and time can all contribute. A credible control condition is therefore central to evaluating the specific claim.

A double-blind placebo-controlled randomized trial with 13-month follow-up tested a neurofeedback protocol for ADHD against a control designed to resemble training without delivering the same brain-signal contingency. Both groups improved, but the study did not demonstrate a specific advantage of active neurofeedback on its primary outcome. That distinction is crucial: an observed improvement and a treatment-specific effect are different results, even when both can sound encouraging in a promotional summary.

What the control condition is trying to separate

A training program can involve regular appointments, a supportive setting, an engaging task, and the expectation of improvement. If those features are present only in the active group, the comparison cannot isolate the role of the brain-signal feedback. A carefully matched sham aims to hold more of the experience constant. It asks whether the specific component advertised as neurofeedback adds benefit beyond the surrounding program, not whether any structured contact is better than nothing.

This is a demanding question, but not an unfair one. If a service attributes its value to training a particular physiological process, evidence should test that process rather than rely only on before-and-after scores. The primary trial's result therefore matters even if families in both groups noticed improvement. It helps identify what is and is not established about the mechanism and the added value of the tested protocol.

Blinding deserves more than a label

Readers should ask who was blinded, whether participants and raters could guess assignment, and which outcomes were prespecified. ADHD research may include ratings from parents, teachers, clinicians, or participants, and these perspectives can differ. Expectations can affect some measurements more than others. A strong report explains these details rather than collapsing all ratings into a single statement that the intervention worked. Agreement across different outcome sources can be informative, but disagreement also deserves transparent reporting.

The site's research-reading guide provides a broader framework for evaluating these design choices. A study with a credible sham is not automatically the final word, but its comparison should not be ignored in favor of less controlled evidence simply because the latter sounds more positive. Editorial judgment should track the question each design can answer, with particular attention to whether the intervention's claimed active ingredient was actually isolated.

A protocol-specific result is not a verdict on every possibility

Neurofeedback includes different signals, algorithms, session structures, and target populations. A trial of one protocol does not establish that every possible approach is ineffective. The reverse is equally important: variation across protocols does not justify treating every untested version as promising by default. Each claim needs evidence appropriate to its own design. A provider should be able to identify the protocol and the primary research supporting the outcome being advertised.

Mechanistic success and clinical success also need to be distinguished. Even if a participant changes a measured brain signal, researchers must show that this translates into meaningful benefit. A laboratory target can move without improving daily functioning. Conversely, improvement in a symptom rating does not prove that the proposed physiological mechanism caused it. Future studies should specify how these links will be tested rather than infer them from the existence of a sophisticated measurement system.

Burden, cost, and opportunity cost matter

Repeated training can require substantial time, travel, and family coordination. These burdens should be measured alongside outcomes. A program may feel attractive because it is described as non-drug, but that label does not establish effectiveness, suitability, or the absence of harms and costs. The relevant comparison includes what other assessment, educational support, or clinical care might be delayed or displaced while a family invests in the program.

No article can determine an individual child's diagnosis or select treatment. The site's treatment-change questions can help families ask about evidence, goals, and follow-up. Useful questions include which outcome is expected to improve, how progress will be measured independently, what would count as insufficient benefit, and when the plan would be reconsidered. These are safeguards against indefinite participation based only on the promise that more sessions may eventually produce a result.

What stronger future evidence would look like

The next credible studies would use transparent protocols, appropriate controls, blinded assessments where feasible, and meaningful longer-term outcomes. They should report functioning at school and home as well as symptom ratings, account for concurrent treatment, and make discontinuation visible. Replication across independent teams would help establish whether any specific protocol delivers a reliable added benefit. This is a more useful research agenda than asking whether neurofeedback as a broad brand category is good or bad.

The trial's central lesson reaches beyond ADHD: technology can make a treatment look precise without proving its incremental value. Our precision psychiatry pillar examines the same distinction in treatment-matching research. Families and clinicians deserve an explanation of what the active comparison demonstrated, not just reassurance that participants improved. Future brain-based interventions will earn credibility by surviving that question rather than avoiding it.

Sources

• Double-Blind Placebo-Controlled Randomized Clinical Trial of Neurofeedback for ADHD With 13-Month Follow-up

For general medicine-safety background, see NIMH mental health medications; this is background, not independent confirmation of this article’s specific study.