Schizophrenia affects roughly one in every hundred people worldwide, and while antipsychotic medications control many of its features, a substantial fraction of patients continue to experience persistent auditory hallucinations, hearing voices that no one else can hear, as well as entrenched negative symptoms such as social withdrawal, emotional flatness, and loss of drive. These residual symptoms are among the most disabling aspects of the illness and are notoriously resistant to pharmacological escalation. This therapeutic gap has fueled decades of interest in brain stimulation techniques that might modulate the dysfunctional neural circuits thought to underlie them, offering hope for patients who have exhausted conventional options.

The technique tested in STIM’Zo belongs to a family of methods known as low-intensity transcranial electrical stimulation. Unlike transcranial magnetic stimulation, which uses rapidly changing magnetic fields to induce currents strong enough to trigger neuronal firing, transcranial electrical stimulation delivers currents so weak, typically on the order of one to two milliamps, that they do not directly cause neurons to discharge. Instead, these faint currents subtly shift the resting electrical potential of cortical neurons, nudging their excitability up or down depending on the stimulation protocol. Transcranial direct current stimulation applies a steady current, whereas tRNS delivers current that fluctuates randomly across a range of frequencies, a design intended to prevent the brain from habituating to a constant signal and to exploit the tendency of neural networks to resonate with stochastic input.

Earlier evidence for tRNS and its close cousin tDCS in schizophrenia rested on a fragile foundation of case reports and small pilot studies, some of which reported striking reductions in hallucination severity. Encouraged by these signals, the STIM’Zo investigators, led by Jerome Brunelin, Marine Mondino, and Emmanuel Poulet and their collaborators, designed a definitive test. The trial was registered as NCT02744989 and received support from the French Ministry of Health, the NeuroDis foundation, and the Centre Scientifique de Monaco, with no funder involvement in the design, analysis, or reporting of the study.

The trial’s methodology reflects the highest standard of clinical rigor. Triple-blind means that patients, the clinicians assessing outcomes, and the statisticians analyzing the data all remained unaware of who received active stimulation and who received sham. The sham condition was engineered to mimic the sensations of real stimulation, typically a brief tingling or itching at the electrode sites at the start of each session, without delivering therapeutic current. Patients were randomized across multiple centers, and the primary outcome was defined in advance as clinical response at Day 5, operationalized as a reduction of at least 25 percent on the total score of the Positive and Negative Syndrome Scale, the most widely used standardized instrument for measuring schizophrenia symptom severity. Secondary outcomes tracked symptoms at one, three, and six months using additional instruments including the Auditory Hallucinations Rating Scale, the Calgary Depression Scale for Schizophrenia, the Clinical Global Impression scale, and a schizophrenia-specific quality of life questionnaire.

The primary analysis included 132 patients, 64 in the active stimulation group and 68 in the sham group, with balanced sex distributions across arms. The result was unambiguous. At Day 5, only 6 of 64 patients in the active group, or 9.4 percent, met the response criterion, compared with 9 of 68 patients, or 13.2 percent, in the sham group. The difference between groups was not statistically significant, with an effect estimate of negative 0.38 and a 95 percent confidence interval spanning negative 1.64 to 0.89, and a p value of 0.56. In plain terms, the sham procedure performed slightly, though not significantly, better than the real intervention. None of the secondary outcomes, measured over the following six months, showed significant differences either, meaning the stimulation neither reduced hallucinations, nor improved negative symptoms, nor lifted mood or quality of life beyond what the placebo ritual itself achieved.

Safety, at least, was not in question. Serious adverse events unrelated to treatment occurred in five patients in the active group and seven in the sham group, a difference that was not statistically significant. The technique was well tolerated, and the trial confirmed that low-intensity electrical stimulation of this kind carries no meaningful physical risk when properly administered. But safety without efficacy is a hollow consolation for patients and clinicians, and the authors’ conclusion was blunt: tRNS did not show superiority over sham in reducing persistent symptoms of schizophrenia.

Why did a technique that showed promise in early studies fail so decisively here? The most instructive answer lies in the trial design itself. Small pilot studies and case reports are exquisitely vulnerable to placebo effects, publication bias, and regression to the mean, and psychiatric symptoms are especially susceptible to the powerful psychological ritual of receiving what feels like an active treatment. The sham group’s 13.2 percent response rate illustrates how much apparent improvement can arise from expectation, attention, and the natural fluctuation of symptoms alone. When these factors are rigorously controlled, as they were in STIM’Zo, the true effect of the intervention can shrink dramatically or vanish. The trial also raises deeper questions about mechanism: whether currents this weak, applied over the scalp and attenuated by skull and skin, can reach and meaningfully alter the deep and distributed circuits implicated in hallucinations remains uncertain. Some researchers argue that stimulation parameters, electrode positioning over regions such as the temporoparietal junction and dorsolateral prefrontal cortex, or the duration of treatment may need optimization, but the STIM’Zo data provide no support for the technique in its current form.

The implications extend well beyond tRNS. The result tempers enthusiasm for a wave of consumer and clinical enthusiasm around electrical brain stimulation for psychiatric conditions, and it underscores the necessity of large, triple-blind, sham-controlled trials before any brain stimulation protocol is adopted into routine care. It also offers a model of how the field should proceed: preregistered endpoints, multicenter recruitment, transparent funding, and honest reporting of negative results. For the patients who participated, and for the millions living with persistent voices and blunted lives, the trial did not deliver a new therapy, but it delivered something arguably just as valuable, a clear and trustworthy answer that redirects research toward approaches with a genuine chance of working. In science, knowing what does not work, and knowing it with confidence, is the first step toward finding what does.

Subject of Research: A triple-blind randomized trial testing transcranial random noise stimulation as an add-on treatment for persistent symptoms in schizophrenia.

Article Title: Low intensity transcranial electrical stimulation for schizophrenia: a triple-blind, randomized, sham-controlled multicenter trial (STIM’Zo)

Article References: Brunelin, J., Mondino, M., Plaze, M., Attal, J., Benoit, M., El-Hage, W., Galvao, F., Haesebaert, J., Jardri, R., Llorca, P. M., Magaud, L., Nathou, C., Szekely, D., Fakra, E., & Poulet, E. (2026). Low intensity transcranial electrical stimulation for schizophrenia: a triple-blind, randomized, sham-controlled multicenter trial (STIM’Zo). BMC Medicine. https://doi.org/10.1186/s12916-026-05227-7

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05227-7

Keywords: schizophrenia, tRNS, brain stimulation, auditory hallucinations, negative symptoms, placebo-controlled trial, BMC Medicine, psychiatry, tDCS, clinical trial, PANSS, noninvasive stimulation