Schizophrenia Hallucinations: What They Are & Why They Happen

Hallucinations in schizophrenia are vivid sensory experiences that occur without any corresponding external stimulus. They feel real to the person having them, and roughly four in five people with schizophrenia-spectrum disorders will experience at least one type of hallucination during their lifetime. While hearing voices is by far the most recognized form, hallucinations can involve any sense, and the neuroscience behind them points to something more intricate than simply “hearing things that aren’t there.”

How Common Are They, and Which Senses Are Involved

Auditory hallucinations dominate the picture. Across large clinical samples, the lifetime rate of auditory hallucinations in people with schizophrenia-spectrum disorders runs between about 64 and 80 percent, making them far more prevalent than any other type. Visual hallucinations come in a distant second at roughly 23 to 31 percent lifetime prevalence, followed by tactile hallucinations (around 9 to 19 percent) and olfactory hallucinations (about 6 to 10 percent).¹ When researchers looked at a given month rather than a lifetime, the numbers dropped sharply: only about a quarter of patients reported auditory hallucinations in the past month, and single digits for every other modality.

Most people who hallucinate do so in just one sensory channel, and that channel is almost always hearing. About a third experience hallucinations in two modalities, most often auditory and visual together. One finding that gives a sense of how tightly the auditory experience anchors the others: while only 30 to 37 percent of people who hear voices also see things, the reverse relationship is much stronger. Between 83 and 97 percent of people who have visual hallucinations also hear voices.¹ In one study of schizophrenia-spectrum patients, 53 percent reported hallucinations spanning multiple sensory modalities over their lifetime.²

Having hallucinations in more senses at once is not just a curiosity. Pilot data suggest that people who experience both visual and auditory hallucinations tend to show worse cognitive performance across the board compared with those who hallucinate in just one modality.³ Multi-modal hallucinations seem to track with a more severe illness course, which is why clinicians pay attention to the range of senses involved and not just whether voices are present.

What Auditory Hallucinations Actually Sound Like

People sometimes imagine that “hearing voices” means hearing a single sinister whisper, but the reality is more varied and more linguistically specific than that. Research analyzing the actual speech content that patients report hearing has found a consistent linguistic fingerprint. The voices overwhelmingly address the person in the second or third person (“you” or “he/she”) rather than the first person. They tend to speak in short, isolated statements with little grammatical linking between them, almost like receiving individual text messages rather than hearing a flowing conversation. Despite the fragmented structure, the statements are grammatically correct, and the content is almost always personal rather than abstract.⁴

Voice content is also not random. Voices frequently comment on what the person is doing, issue commands, criticize, or occasionally praise. Some patients hear familiar voices (a relative, a friend), while others hear strangers. The emotional tone can range from neutral to deeply threatening. And contrary to the common assumption that voices repeat the same thing like a broken record, the research shows they are generally not stereotyped. Patients report variation in what the voices say, which partly explains why the experience can be so consuming: you cannot tune out a voice the way you tune out a predictable noise.

The Brain Mechanisms Behind Hallucinations

Several overlapping brain processes contribute to hallucinations, and researchers have been fitting together pieces of the puzzle for decades. No single mechanism explains everything, but a few stand out as well-supported.

Corollary Discharge Failure

When you talk aloud or even think in words, your brain sends an internal copy of the motor command (called a corollary discharge) from frontal speech-production areas to the auditory cortex in the temporal lobe. This signal essentially tells the auditory cortex, “This is coming from us, not from the outside world,” and the cortex dampens its response accordingly. In people with schizophrenia, particularly those who hallucinate, this signaling appears to break down. Electrophysiological studies show that control subjects dampen auditory cortex activity during speaking and inner speech, but patients with hallucinations do not. The frontal-temporal coordination that normally flags internal speech as self-generated is weaker or absent, which leads to inner thoughts being experienced as external voices.⁵

Predictive Coding Gone Wrong

A broader framework that has gained traction views the brain as a prediction machine. Rather than passively receiving sensory input, the brain constantly generates expectations about what it is about to perceive and compares those expectations against incoming signals. When there is a mismatch, a “prediction error” signal updates the brain’s model of reality. In psychosis, the balance between these prior beliefs and incoming data goes awry. Prior expectations can become weighted too strongly or in the wrong context, causing the brain to perceive things that are not there.⁶

Neuroimaging work has put concrete numbers behind this idea. Patients with auditory hallucinations show weaker prediction-error signals in the auditory cortex compared to healthy controls. And the weaker those error signals are, the more spontaneous activity the auditory cortex produces during silence. In other words, when the brain’s error-checking is poor, the auditory system fills the quiet with its own generated content. The correlation between weak prediction-error signaling and greater auditory cortex activation during blank trials was remarkably strong in one study, suggesting this is not a minor secondary feature but a core part of the hallucinatory experience.⁷

Structural Brain Differences

The brain regions most involved in hearing and language processing are physically different in people who hallucinate. MRI studies consistently find that the superior temporal gyrus, a strip of cortex crucial for auditory processing, is smaller in patients with schizophrenia, especially on the left side. Critically, this shrinkage is not just a generalized loss of brain volume; the reduction is concentrated in auditory areas and correlates selectively with the severity of hallucinations, not with other symptoms of schizophrenia.⁸ Further volumetric work found that the left anterior portion of this gyrus was particularly affected, and that its size negatively correlated with hallucination severity.⁹ The structural changes align with the functional problems: a smaller, differently wired auditory cortex is more likely to produce aberrant signals and less able to properly filter internal from external sounds.

Dopamine, Glutamate, and the Chemistry of Salience

On a neurochemical level, the leading explanation for why irrelevant internal signals get treated as significant involves dopamine. The “aberrant salience” hypothesis proposes that dysregulated dopamine signaling in the striatum causes the brain to tag unimportant stimuli as meaningful or threatening. Instead of background noise staying in the background, it gets flagged as a signal that demands attention, which can drive both hallucinations and delusional thinking.¹⁰ This is also why most antipsychotic medications work by blocking dopamine D2 receptors: reducing the inappropriate salience signal can quiet the voices.

Dopamine is not the whole story, though. Drugs that block a certain type of glutamate receptor (the NMDA receptor) can produce hallucination-like experiences in healthy people and worsen hallucinations in those who already have schizophrenia. These observations have fueled a parallel “glutamate hypothesis” that implicates the brain’s main excitatory neurotransmitter system in the disorder.¹¹ The dopamine and glutamate systems interact extensively, so both hypotheses likely describe different parts of the same underlying dysfunction rather than competing explanations.

The Source Monitoring Problem

There is also a cognitive-level explanation that works alongside the neural one: people who hallucinate have difficulty with “source monitoring,” the everyday skill of knowing where a thought or perception came from. Every time you recall a memory, you unconsciously tag it: did I actually see that, or did I just imagine it? Did someone say that to me, or did I think it to myself? In experimental settings, patients with schizophrenia who hallucinate are specifically impaired at recognizing their own self-generated thoughts and tend to misattribute them to an external source. They do not have trouble distinguishing between two external sources (for example, telling apart two different speakers); the deficit is selective to the internal-versus-external boundary.¹² This fits neatly with the corollary discharge findings described earlier. If the brain does not properly label inner speech as self-generated, the cognitive system that monitors source will classify it as external, and the person will genuinely hear a voice that is not theirs.¹³

How Culture Shapes the Content of Voices

Although the mechanisms behind hallucinations are universal, what the voices say varies strikingly by culture. An influential study comparing voice-hearing experiences across the United States, South India, and West Africa found that while the basic phenomenon was similar everywhere, the character of the voices differed. Americans were more likely to describe their voices as intrusive, unwelcome thoughts. South Indian participants more often reported the voices as providing useful guidance. And in the West African sample, voices were more frequently described as morally good and having real causal power in the world.¹⁴

A separate comparison between Saudi Arabian and British patients found that the physical characteristics of the voices (loudness, clarity, frequency) did not vary much between the two groups, but their content did. Saudi patients heard voices with heavily religious and superstitious themes, while British patients more often heard running commentary and direct instructions.¹⁵ These patterns suggest that the brain generates the experience of hearing a voice through the biological mechanisms outlined above, but the local cultural context furnishes the script. This has real clinical implications: a therapist trained in one cultural setting might misinterpret the meaning of a patient’s hallucinations in another.

Hallucinations as Early Warning Signs

Mild perceptual disturbances often show up before full-blown psychosis, during what clinicians call the “clinical high-risk” period. This might include hearing faint whispers, feeling that sounds are slightly distorted, seeing brief shadows, or sensing that one’s surroundings look subtly unreal. The tricky part is that these attenuated experiences, taken as a single combined score, do not reliably predict who will actually convert to psychosis and who will not.

However, researchers found something useful when they split the scoring into auditory and visual components rather than lumping them together. Auditory perceptual abnormalities and visual perceptual abnormalities independently predicted conversion to psychosis, but in opposite directions: greater auditory disturbances predicted higher conversion risk, while greater visual disturbances were associated with lower risk. This was masked when both were combined into one overall score.¹⁶ If replicated, the finding could sharpen how clinicians identify people most likely to develop schizophrenia and intervene earlier.

Treatment and What Happens When Medication Falls Short

Antipsychotic medications remain the first-line treatment for hallucinations in schizophrenia, and they work by reducing dopamine activity at D2 receptors. For many people they quiet the voices substantially, though rarely eliminating them entirely. The bigger problem is that an estimated 20 to 30 percent of patients do not respond adequately to standard antipsychotics. When first-line drugs fail, clozapine is the established second option, but it requires regular blood monitoring because of rare but serious side effects, and even clozapine does not work for everyone.

For treatment-resistant hallucinations, the evidence base is thinner and more uncertain. A review of available strategies found that simply increasing antipsychotic doses did not improve any symptom domain. More promising approaches included augmenting antipsychotics with agents that act on the glutamate system’s glycine site, which showed moderate benefits for positive symptoms, and non-invasive brain stimulation (such as transcranial magnetic stimulation aimed at the temporal cortex), which also moderately reduced positive symptoms including hallucinations. Adding structured psychotherapy to medication had a similar moderate benefit.¹⁷ The quality of evidence for all these alternatives was rated low to very low, meaning more research is needed before any of them becomes a standard recommendation.

One of the more creative recent developments is AVATAR therapy, in which a patient creates a digital avatar that represents their persecutory voice, controlling its appearance and pitch until it closely matches the hallucinatory experience. A therapist then speaks through the avatar, gradually shifting the dialogue so the patient learns to assert control over the voice. A meta-analysis of available trials found that AVATAR therapy reduced hallucination severity by a clinically meaningful margin compared to control conditions, and the benefits held at three-month follow-up. It also improved depression, anxiety, and quality of life with small-to-medium effect sizes.¹⁸ AVATAR therapy is still considered experimental and is not widely available, but the results so far are encouraging for people whose voices have not responded to medication alone.

Voice Hearing Outside of Schizophrenia

It is worth knowing that hearing voices is not exclusive to schizophrenia. Population surveys consistently find that a small percentage of otherwise healthy people hear voices at some point in their lives without ever developing psychosis. These “non-clinical voice hearers” differ from people with schizophrenia in important neurobiological ways. Neuroimaging research has found that decreased language lateralization and dopamine dysfunction, both consistently observed in schizophrenia, are absent in healthy voice hearers.¹⁹ In other words, the experience of hearing a voice can arise from more than one brain pathway. The pathological version in schizophrenia involves a constellation of structural, chemical, and connectivity differences that healthy voice hearers do not share.

The boundary between schizophrenia hallucinations and other kinds of hallucinatory experience is also relevant in older adults, where conditions like Parkinson’s disease and Lewy body dementia produce hallucinations with a very different sensory profile. In those conditions, visual hallucinations predominate (often detailed images of people or animals), while in schizophrenia auditory hallucinations carry the heaviest impact. When comparing the two, auditory hallucinations showed the highest “repercussion” score, meaning the most distress and functional disruption, in both conditions, but the overall pattern of which senses are involved reliably distinguishes them.²

Language, Lateralization, and an Evolutionary Puzzle

One of the more provocative ideas in the field connects schizophrenia hallucinations not to a flaw in the brain but to the evolutionary cost of human language itself. The hypothesis goes like this: the genetic changes that gave Homo sapiens the capacity for language involved a shift in how the two brain hemispheres specialize, with the left hemisphere taking on more language function and the right more spatial and non-verbal processing. This “cerebral torque,” reflected in structural asymmetry from right frontal to left occipital regions, is a defining feature of human brain organization. Normal variation in the degree of this lateralization tracks with differences in verbal and non-verbal ability. But at the extreme end of that variation, the same architecture that enables language may also produce the misattributed inner speech and disordered connectivity that underlie hallucinations.²⁰ If this idea is correct, the persistence of schizophrenia-related genes in the human population, despite the disorder’s clear disadvantages, makes more sense: they are maintained because they are bound up with the genetic package that makes complex language possible. The hypothesis remains debated and is difficult to test directly, but it reframes hallucinations as something other than pure malfunction, positioning them instead as a byproduct of the very capacity that defines us.