EEG Study Links “Hearing Voices” to Misperceived Inner Speech

For people with schizophrenia, hearing voices can feel vividly real – but why this happens has remained a mystery.

A recent study by the University of New South Wales (UNSW) showed that the brain of someone experiencing auditory hallucinations reacts to inner speech as if it were external, offering the first direct biological clue.

How inner speech relates to schizophrenia

Auditory verbal hallucinations – hearing voices that aren’t there – are one of the most common and distressing symptoms of schizophrenia. Around 60–80% of people with the condition experience them at some point in their lives.

“Inner speech is the voice in your head that silently narrates your thoughts – what you’re doing, planning or noticing,” said lead author Dr. Thomas Whitford, a professor in the school of psychology at UNSW. “Most people experience inner speech regularly, often without realising it, though there are some who don’t experience it at all.”

For decades, researchers have suspected that hearing voices might happen when the brain misidentifies this inner speech as coming from someone else. Normally, when you talk or even think words silently, your brain predicts what that speech will sound like and reduces its own auditory response. This internal feedback loop – known as corollary discharge – helps the brain tell the difference between self-generated and external sounds.

“This idea’s been around for 50 years, but it’s been very difficult to test because inner speech is inherently private,” said Whitford.

The challenge is: you can’t record it directly. “How do you measure it? One way is by using an EEG [electroencephalogram], which records the brain’s electrical activity,” explained Whitford.

Previous studies showed that people with schizophrenia have weaker suppression of brain activity when they speak aloud; however, until now, it hasn’t been clear if the same problem happens during silent inner speech. The new study set out to answer that question directly.

EEG study reveals brain activity patterns

The team recruited 3 groups: 55 people with schizophrenia who had recently heard voices, 44 with schizophrenia but no recent hallucinations and 43 healthy participants. All participants wore EEG caps to record their brain’s electrical activity.

Participants were asked to imagine saying either “bah” or “bih” in their heads while hearing one of those syllables played through headphones. Sometimes the imagined and heard sounds matched, sometimes they didn’t and sometimes participants just listened passively. This setup allowed the team to see how the brain’s auditory cortex reacted when inner speech lined up with external sound.

In healthy participants, hearing a sound that matched their imagined one led to reduced activity in the auditory cortex – showing the expected “suppression” effect when the brain predicts its own sound. In people who were currently hearing voices, the pattern flipped: their brains became more active when inner and external speech matched.

“Even though we can’t hear inner speech, the brain still reacts to it – and in healthy people, using inner speech produces the same kind of reduction in brain activity as when they speak out loud,” said Whitford. “But in people who hear voices, that reduction of activity doesn’t happen. In fact, their brains react even more strongly to inner speech, as if it’s coming from someone else. That might help explain why the voices feel so real.”

“This reversal of the normal suppression effect suggests that the brain’s prediction mechanism may be disrupted in people currently experiencing auditory hallucinations, which may cause their own inner voice to be misinterpreted as external speech,” he added.

Those with schizophrenia who weren’t currently hallucinating showed a middle-ground pattern – some suppression, but weaker than healthy participants.

Implications for diagnosing schizophrenia

This study links the experience of “hearing voices” to measurable brain activity, offering biological support for a long-standing theory of schizophrenia. The reversal of suppression – or “inner speaking-induced enhancement” – might eventually serve as a biomarker to assess psychosis risk or track treatment progress.

Advancing Neurodegenerative Disease Research With iPSCs

This eBook explores the latest advances in iPSC-derived models, highlighting how they bridge preclinical and clinical research to accelerate discovery in neurodegenerative disease research.
View eBook

Advertisement

At present, schizophrenia is diagnosed purely through symptoms. There are no lab tests or scans that can detect it. If EEG markers like this prove reliable, they could one day help identify people at risk before psychotic symptoms appear.

“This sort of measure has great potential to be a biomarker for the development of psychosis,” Whitford said.

However, the study has caveats. The study shows correlation, not cause. Most participants were on medication, which may influence brain signals, and the work will need replication and long-term follow-up to confirm if these EEG patterns predict psychosis.

“Our research shows that when we speak – even just in our heads – the part of the brain that processes sounds from the outside world becomes less active. This is because the brain predicts the sound of our own voice. But in people who hear voices, this prediction seems to go wrong, and the brain reacts as if the voice is coming from someone else,” said Whitford.

“It was always a plausible theory – that people were hearing their own thoughts spoken out loud – but this new approach has provided the strongest and most direct test of this theory to date,” he added.

“Ultimately, I think that understanding the biological causes of the symptoms of schizophrenia is a necessary first step if we hope to develop new and effective treatments,” Whitford concluded.

Reference: Whitford TJ, Chung LK, Griffiths O, et al. Corollary discharge dysfunction to inner speech and its relationship to auditory verbal hallucinations in patients with schizophrenia spectrum disorders. Schizophr Bull. 2025:sbaf167. doi: 10.1093/schbul/sbaf167

This article is a rework of a press release issued by the University of New South Wales. Material has been edited for length and content.