(Credit: Ground Picture on Shutterstock)
Study confirms long-held beliefs about one of the most common and devastating symptoms of schizophrenia.
In A Nutshell
- Scientists found that when people with schizophrenia imagine speaking words silently, their brains do the opposite of what healthy brains do: they amplify the brain signals instead of quieting them.
- In 142 people tested, those currently hearing voices showed stronger brain responses when their imagined speech matched external sounds, while healthy people showed weaker responses, suggesting a brain signal turned the wrong direction.
- The more severe someone’s voice-hearing symptoms were, the more pronounced this backward brain response became, offering a measurable biological explanation for one of schizophrenia’s most distressing symptoms.
- This finding could eventually help doctors test young people at risk for psychosis and measure whether treatments are working at the brain level, not just based on patient reports.
Why do some people with schizophrenia hear voices that aren’t there? Scientists may have found a crucial piece of the puzzle by watching what happens in the brain when people talk to themselves silently.
In a study of 142 people, researchers discovered that when patients currently experiencing auditory-verbal hallucinations (AVH) imagined speaking syllables, their brains amplified the signals instead of quieting them like healthy brains do. This backward response might explain how “hearing” internal thoughts gets mistaken for external voices.
“The present study provides perhaps the strongest evidence to date that AVH commonly associated with schizophrenia spectrum disorders are related to an abnormality in the normatively suppressive mechanisms associated with the production of inner speech,” the researchers wrote in their paper.
For decades, scientists have theorized that hearing voices stems from a person’s inability to distinguish between their own inner voice and external sounds. The study, published in Schizophrenia Bulletin by researchers from the University of New South Wales, Chinese University of Hong Kong, and other institutions, offers brain-level evidence supporting the theory.
How Healthy Brains Quiet Internal Speech
Most people talk to themselves silently throughout the day. When healthy individuals produce inner speech, their brains send out what scientists call a “corollary discharge” signal. Picture a warning bell that tells the hearing system: “This sound is coming from me, so don’t pay too much attention to it.” This mechanism helps people tell the difference between sounds they generate themselves and sounds from the outside world.
Researchers track this suppression by measuring a brain wave called N1, an early signal that shows how strongly the brain responds to sounds.
Experts have known for years that people with schizophrenia show problems when they speak out loud. Their brains don’t suppress the sound of their own voices as much as they should. But testing what happens with silent, inner speech has been much trickier since inner speech is inherently invisible.
Thomas Whitford and his colleagues developed a clever workaround. Participants watched a line move across a computer screen at a steady pace. When the line reached a target point, they had to imagine speaking a specific syllable at that exact moment: either “ba” or “bi.” At the same instant, the researchers played an audible syllable through headphones. Sometimes the imagined and audible syllables matched. Other times they didn’t. In some trials, participants just listened without imagining anything.
Researchers measured brain waves from scalp electrodes to see how strongly the brain responded in each situation.

Three Groups, Three Different Patterns
The study included three groups: 43 healthy people, 55 patients who were currently hearing voices, and 44 patients who weren’t currently hallucinating.
Healthy participants showed a reduced brain response when their imagined syllable matched the audible one they heard. Their brains essentially treated predicted sounds as less important, quieting down the response to them. This only happened when the imagined and heard syllables aligned on both content and timing.
Patients currently hearing voices showed the reverse pattern. Their brain response increased when imagined and audible syllables matched. Instead of dampening the response to predicted sounds, their brains amplified it, making that sound more attention-grabbing and “real.”
Patients not currently experiencing hallucinations showed a different pattern altogether. They had reduced brain responses in the mismatch condition compared to both when they just listened passively and when syllables matched.
Perhaps not surprisingly, the more severe someone’s voice-hearing symptoms were, the more pronounced this backward brain response became. Patients with higher scores on assessments measuring how often and how intensely they heard voices showed stronger amplification when their imagined speech matched external sounds.
Why This Matters: When Thoughts Feel Like Someone Else Speaking
Auditory hallucinations affect roughly three-quarters of people diagnosed with schizophrenia at some point during their illness. People often describe hearing distinct voices speaking to them, about them, or commanding them to do things. The voices can be critical, threatening, or conversational.
The findings explain how internal mental speech transforms into an experience that feels externally generated. If your brain amplifies responses to predicted sounds instead of suppressing them, the boundary between “me thinking” and “someone else speaking” gets blurred.
Not every patient without current hallucinations showed the same pattern. Some had histories of hearing voices in the past but weren’t experiencing them during the study. When researchers compared those with and without past voice-hearing history, they found no significant differences. The abnormal pattern related specifically to current hallucinations.

Could This Lead to Better Treatment?
The research may eventually contribute to tools for identifying psychosis risk or measuring treatment response. Young people showing early warning signs could potentially be tested for inner speech problems before full symptoms emerge. Treatments could be evaluated based on whether they help normalize these brain patterns, not just based on what patients report.
Current treatments for auditory hallucinations include antipsychotic medications, cognitive behavioral therapy, and in some cases, transcranial magnetic stimulation. Having an objective brain measure of the mechanisms behind voice-hearing could help doctors track whether interventions are working at the biological level.
The results validate a longstanding theory about where voices come from. They confirm that something specific and measurable goes wrong with how the brain processes inner speech in people with schizophrenia. Auditory hallucinations aren’t mystical or inexplicable. They’re the result of identifiable brain circuits operating differently than they should.
The researchers note that the task used single syllables in a controlled setting, an important first step that will need to be extended to capture the full complexity of everyday inner speech.
Disclaimer: This article summarizes scientific research for general informational purposes only. It is not medical advice and should not replace consultation with qualified healthcare professionals. If you or someone you know experiences auditory hallucinations or other mental health concerns, please seek help from a licensed medical provider.
Paper Summary
Methodology
Researchers tested 142 adults split into three groups: people with schizophrenia spectrum disorders currently experiencing auditory hallucinations, people with these disorders not currently hallucinating, and healthy controls matched for age and gender. Participants watched a line move across a screen at a constant speed. When it reached a target point after 3.75 seconds, they had to silently imagine speaking either the syllable “ba” or “bi” at that precise moment. Simultaneously, researchers played an audible syllable (“ba” or “bi”) through headphones. Three conditions were tested: Match (imagined and audible syllables were the same), Mismatch (they were different), and Passive (participants didn’t imagine any syllable). Each person completed 144 trials across the three conditions. Researchers recorded brain activity using 64 electrodes placed on the scalp and measured the amplitude of a brain wave component called N1, which reflects how strongly the auditory cortex responds to sounds. Testing occurred at two sites: the University of New South Wales in Sydney, Australia, and the Chinese University of Hong Kong. Clinical symptom severity was assessed using standardized psychiatric rating scales.
Results
Healthy controls showed reduced N1-amplitude in the Match condition compared to both Passive and Mismatch conditions, a phenomenon the researchers call “inner speaking-induced suppression” or inner SIS. This means their brains responded less strongly to external sounds they had predicted through inner speech. Patients currently experiencing auditory hallucinations showed the opposite pattern, with enhanced N1-amplitude in the Match condition compared to both other conditions. Their brains responded more strongly to sounds matching their inner speech. Patients not currently hallucinating showed yet another pattern, with reduced N1-amplitude in the Mismatch condition compared to Match and Passive conditions. Across all patients, hallucination severity correlated significantly with the degree of N1-suppression in the Match condition, meaning more severe hallucinations associated with more pronounced N1-enhancement. Healthy controls also showed increased P2 and P3 brain wave components in the Match condition, but patient groups did not show these patterns. The effects remained significant when controlling for testing location, age, sex, and clinical variables like medication dosage.
Limitations
Many patients in the non-hallucinating group had experienced auditory hallucinations in the past, even though they weren’t currently hearing voices. Finding patients with schizophrenia spectrum disorders who have never hallucinated proved difficult given the symptom’s high prevalence. The study couldn’t differentiate between different subtypes of auditory hallucinations, such as voices commenting versus command hallucinations, due to sample size limitations. Inner speech varies considerably between individuals, with some people experiencing very clear verbal thoughts and others reporting more abstract internal language. The study couldn’t account for these individual differences. Data collection occurred at two sites with different EEG equipment, though statistical controls suggested this didn’t substantially affect results. The experimental task measured responses to single syllables, which may not fully capture the complexity of natural inner speecdh during daily life.
Funding and Disclosures
This work was supported by the National Health and Medical Research Council of Australia, the Australian Research Council, the Australian Government Research Training Program scholarship, the Daniel Beck Memorial Award for Schizophrenia
Publication Details
This research was published October 21, 2025 in Schizophrenia Bulletin by Thomas J. Whitford and colleagues from the University of New South Wales, the Chinese University of Hong Kong, and collaborating institutions. The paper, titled “Corollary Discharge Dysfunction to Inner Speech and its Relationship to Auditory Verbal Hallucinations in Patients with Schizophrenia Spectrum Disorders,” is available as open access (doi:10.1093/schbul/sbaf167).








Can a ‘normal’ person be trained to hear voices? Would make an interesting dialogue.
The puzzle is not that people hear voices, but why they always seem to believe them.