What Part of the Brain Causes Hallucinations?

No single brain region flips a switch that produces hallucinations. Instead, hallucinations arise from disrupted communication across networks of brain areas, and the specific regions involved shift depending on whether someone hears voices, sees things, or smells phantom odors. A landmark lesion-mapping study of 89 patients found that brain damage in many different locations could trigger hallucinations, but all those locations fell within a single functionally connected network rather than clustering in one spot. That finding captures the field’s current understanding well: hallucinations are a network problem, not a location problem, though certain hubs in those networks come up again and again.

Where Auditory Hallucinations Originate

Hearing voices is the most studied type of hallucination, largely because it is so common in schizophrenia. The brain regions that light up during voice-hearing episodes overlap heavily with the regions you use for normal language. Two areas stand out: the left superior temporal gyrus, which includes Wernicke’s area (involved in understanding speech), and the left inferior frontal gyrus, which includes Broca’s area (involved in producing speech). Their mirror-image counterparts on the right side of the brain are also frequently activated. Imaging studies show that in people with auditory-verbal hallucinations, the connection running from Wernicke’s area to Broca’s area is significantly weakened, along with reduced input from the right-hemisphere equivalents of both regions.¹ In other words, the speech-production machinery in the frontal lobe appears to be running without proper guidance from the regions that normally process incoming sound, as though the brain’s language system is talking to itself with the volume turned down on the outside world.

Direct electrical stimulation during brain surgery has helped confirm these locations. When surgeons stimulate the superior temporal gyrus, patients report simple noise-like hallucinations such as buzzing. But stimulating the medial prefrontal cortex can produce something far more elaborate: one patient heard a young woman’s voice speaking a full sentence, a hallucination that matched the content of the patient’s habitual seizures.² Broader mapping work shows that auditory hallucinations tend to be triggered from more medial (inner) parts of the temporal lobe, while simpler auditory illusions come from more lateral (outer) parts, though there is significant overlap.³

Why the Brain Mistakes Its Own Thoughts for Outside Voices

One of the more compelling explanations for voice-hearing involves a mechanism called corollary discharge. Every time you speak or even silently rehearse words in your head, your motor cortex sends a copy of that command to your auditory cortex, essentially a heads-up signal that says, “You’re about to hear something, and it’s coming from you.” This lets the auditory cortex dampen its response to self-generated sounds so you can tell them apart from sounds originating in the environment. Recent recordings from neurosurgical patients pinpointed the source of this signal in the ventral speech motor cortex, and showed that its strength predicted how much the auditory cortex quieted down during speech.⁴

If that copy signal fails or arrives garbled, inner speech could arrive at the auditory cortex without its “from me” tag, making it feel like an external voice. This idea has been proposed as a mechanism behind auditory hallucinations in schizophrenia, where the corollary discharge from speech motor commands may not adequately prepare the auditory cortex for self-generated sound.⁵ It is an elegant explanation, though it likely accounts for only some types of auditory hallucination. Voices that carry elaborate narratives or speak in unfamiliar accents are harder to explain as misattributed inner speech alone.

The Brain Areas Behind Visual Hallucinations

Visual hallucinations engage a different set of regions, anchored in the visual processing pathway that runs from the primary visual cortex at the back of the brain through the temporal and parietal lobes. In Parkinson’s disease, where visual hallucinations are common, researchers have found abnormal activity in the inferior temporal gyrus, particularly the fusiform region, which is the area the brain uses to recognize faces and complex objects.⁶ It makes intuitive sense that hallucinations of people and animals, which are among the most commonly reported visual hallucinations in Parkinson’s, would involve the brain’s face- and object-recognition hardware.

A related finding concerns the default mode network, a set of brain regions most active when you are daydreaming or mind-wandering rather than focused on external tasks. In Parkinson’s patients who experience visual hallucinations, the default mode network shows abnormally high internal connectivity, especially in the right middle frontal gyrus and the posterior cingulate cortex.⁷ The implication is that the brain’s internal imagery system may be running too hot, generating vivid mental pictures that leak into conscious perception. Systematic reviews confirm that altered connectivity within default mode, attentional, and sensory networks is a recurring theme in neurodegenerative patients who hallucinate, though the exact pattern differs by condition.⁸

Interestingly, whether or not a person realizes their hallucination is not real also has a neural signature. In people with Lewy body disease, those who were unaware their visual hallucinations were hallucinations showed weaker connectivity between visual areas and regions involved in evaluating whether perceptions match reality.⁹ The hallucination itself may arise in visual processing areas, but the failure to recognize it as false involves a separate breakdown in self-monitoring.

When Lost Vision Fills Itself In

One of the most striking examples of visual hallucinations occurs in people with significant sight loss. Charles Bonnet syndrome produces vivid, complex visual hallucinations, sometimes of faces, animals, or entire scenes, in people whose eyes can no longer deliver a full visual signal to the brain. The leading explanation is the deafferentation hypothesis: when the visual cortex stops receiving adequate input from the eyes, it becomes hyperexcitable, essentially filling the void with self-generated imagery. A study testing this hypothesis found that people with Charles Bonnet syndrome had measurably increased visual cortex excitability, and the degree of that excitability correlated with how severe their hallucinations were.¹⁰

This pattern aligns with a broader principle that runs through hallucination research: lesions in a sensory pathway tend to produce hallucinations in that same sense. A systematic review of focal brain lesions causing hallucinations in a single modality found that the damage was essentially always located somewhere along the brain’s pathway for that specific sense.¹¹ After the lesion, nearby tissue in the same pathway appeared to overactivate in a compensatory way. The brain seems to abhor a sensory vacuum, and the visual cortex in Charles Bonnet syndrome is a particularly clear case of that principle at work.¹²

The Chemical Side of Hallucinations

Brain regions do not act in isolation; they communicate through neurotransmitter systems, and several of these chemical messengers play direct roles in hallucination. The most thoroughly studied is dopamine, which is central to how the brain assigns importance to stimuli. In people who are prone to psychotic experiences, the striatum (a deep brain structure involved in reward and motivation) shows elevated dopamine activity, and this excess dopamine is linked to a tendency to tag irrelevant stimuli as significant.¹³ That misplaced sense of importance may be part of why hallucinated voices feel commanding and real rather than like random background noise. Antipsychotic medications work primarily by blocking dopamine receptors, which underscores the chemical’s role.

Serotonin provides a different window. Psychedelic drugs like psilocybin produce visual hallucinations primarily by activating a specific serotonin receptor called 5-HT2A. Research in healthy volunteers showed that psilocybin sharply reduced alpha brainwaves in the visual cortex, which normally act as a kind of rhythmic gatekeeper on visual processing, and also suppressed an early brain response to visual stimuli. Both effects were completely blocked when participants were pretreated with a drug that selectively blocks 5-HT2A receptors.¹⁴ The result is a state where the visual cortex becomes flooded with spontaneous activity that overwhelms incoming signals from the eyes, producing the swirling, morphing imagery people associate with psychedelic experiences.

Glutamate, the brain’s main excitatory neurotransmitter, adds another layer. Drugs like ketamine and PCP block a glutamate receptor called NMDA, and both can cause vivid hallucinations. In animal studies, blocking NMDA receptors caused two normally independent rhythmic oscillations in the visual cortex to lock together, forcing layers of the cortex that process top-down predictions and bottom-up sensory information to synchronize in an abnormal way.¹⁵ The practical effect is that the brain’s internal expectations start overwriting what the senses actually report.

Acetylcholine also plays a role, particularly in Parkinson’s disease. Postmortem studies have found that Parkinson’s patients who experienced visual hallucinations had about 40% fewer cholinergic neurons in the pedunculopontine nucleus, a brainstem structure involved in arousal and attention, compared with healthy controls.¹⁶ Loss of these neurons may undermine the brain’s ability to maintain a stable distinction between real perceptions and internally generated imagery, especially during transitions between wakefulness and sleep.

The Thalamus as Gatekeeper

Sitting deep in the center of the brain, the thalamus relays almost all sensory information to the cortex. One theoretical framework proposes that hallucinations emerge when the thalamus loses its ability to let sensory signals constrain cortical activity. Normally, incoming sights and sounds shape the rhythmic patterns that the thalamus and cortex generate together. But when sensory input can no longer adequately steer that self-organizing activity, attention alone may determine what gets perceived, and the result can be a hallucination.¹⁷ This framework helps explain why hallucinations tend to occur in states of reduced sensory input: sensory deprivation, darkness, drowsiness, and the kind of visual-system decline seen in Charles Bonnet syndrome.

Predictions Gone Wrong

An increasingly influential way of thinking about hallucinations is the predictive processing model. In this view, the brain is constantly generating predictions about what it expects to perceive, then comparing those predictions against what the senses actually deliver. Perception is the brain’s best guess, updated by sensory evidence. Hallucinations, under this framework, represent a state where the brain’s predictions become so strong that they override the sensory evidence, or where sensory evidence is too weak to correct them. Researchers have proposed that hallucinations can be understood as “top-down effects on perception,” driven by inappropriately strong perceptual expectations.¹⁸

This framing does a good job unifying otherwise disconnected observations. It explains why reducing sensory input (as in blindness or sensory deprivation) increases hallucination risk: with less bottom-up evidence, predictions dominate. It explains why dopamine excess might trigger hallucinations: dopamine helps encode the precision of predictions, so too much could make the brain overly confident in its internal model. And it explains why the same brain regions, frontal areas involved in generating expectations and sensory cortices that receive those expectations, keep appearing in hallucination research across different conditions.

A Brain Fold That Predicts Risk

Beyond connectivity and chemistry, the physical shape of the brain itself may predispose some people to hallucinations. A fold in the medial prefrontal cortex called the paracingulate sulcus, one of the last brain folds to develop before birth, shows a consistent relationship with hallucination risk. In people with schizophrenia, each centimeter of reduction in the length of this fold on the left side was associated with a roughly 20% increase in the odds of experiencing hallucinations, regardless of whether those hallucinations were auditory, visual, or another type.¹⁹

A follow-up study added nuance: schizophrenia patients with auditory hallucinations had paracingulate sulci roughly 30 millimeters shorter than healthy controls, but healthy people who also reported hearing voices did not show the same reduction.²⁰ This suggests the structural difference is tied specifically to clinical hallucinations rather than the broader human capacity for voice-like experiences. The paracingulate region is involved in reality monitoring, the ability to distinguish internally generated thoughts from external events, so less cortical tissue there could mean a less reliable filter between imagination and perception.

Phantom Smells and Bodily Sensations

Hallucinations are not limited to hearing and seeing. Olfactory hallucinations, the perception of smells that are not present, involve their own neural territory. Studies of patients with phantosmia (phantom smell perception) have linked the phenomenon to decreased activity of the inhibitory neurotransmitter GABA in specific brain regions, and treatment that boosted GABA activity suppressed the phantom smells. The hallucinations appear to follow the same general principle seen in other modalities: reduced inhibition within a sensory pathway allows internally generated activity to break through into conscious perception.

Somatic hallucinations, such as the sensation of insects crawling on the skin or internal organs being manipulated, similarly localize along the body’s sensory pathways. The systematic review of lesion-caused hallucinations found that when a brain injury produced hallucinations in one and only one sense, the damage was consistently somewhere along that sense’s processing route, with nearby tissue seemingly overcompensating after the lesion.¹¹ The brain’s sensory systems, it seems, are each independently capable of producing hallucinations when their normal function is disrupted, whether by a tumor, a stroke, a chemical imbalance, or neurodegeneration.

How a Lesion Network Connects It All

The study that mapped 89 brain lesions causing hallucinations found damage scattered across many different regions: the thalamus, the brainstem, parts of the temporal and occipital lobes. On the surface, the locations looked random. But when the researchers checked how each damaged spot was functionally connected to the rest of the brain, all the lesion sites fell within a single shared network.²¹ This network-level convergence is perhaps the most important finding for understanding hallucinations as a whole. It means that damage at any node in the network can produce hallucinations, much like cutting any wire in a circuit can shut it down. The relevant question is rarely “which exact region is broken” and almost always “which network connection has been lost or distorted.”

Targeting Brain Circuits to Treat Hallucinations

The fact that specific brain regions consistently appear in hallucination research has opened the door to targeted treatments using brain stimulation. Repetitive transcranial magnetic stimulation (rTMS), which uses magnetic pulses delivered through the scalp to either ramp up or quiet down activity in a specific brain area, has shown promise for auditory hallucinations that do not respond to medication. In a study of schizophrenia patients with persistent auditory hallucinations, low-frequency rTMS applied twice daily to the left temporoparietal junction produced a roughly 39% reduction in hallucination severity, with a large effect size.²² A sham-controlled trial confirmed that active stimulation to this region reduced positive symptoms, including hallucinations, while placebo stimulation did not.²³

The same approach has been tried for rarer hallucination types. In a case of persistent musical hallucinations, a woman who constantly heard music that was not playing experienced substantial relief after low-frequency rTMS was applied to her left temporoparietal junction, with weekly maintenance sessions sustaining the improvement.²⁴ These interventions work by calming down a hyperactive hub in the network. They are not cures, and they do not work for everyone, but they represent a concrete payoff from decades of mapping which brain regions are involved.

Hallucinations During Sleep Transitions

Many people who never hallucinate during the day experience vivid perceptions while falling asleep or waking up. These hypnagogic and hypnopompic hallucinations are extremely common and generally harmless. They can include seeing faces, hearing someone call your name, or feeling a presence in the room. Researchers have explored whether these experiences share neural mechanisms with clinical hallucinations, and there are genuine similarities: both involve internally generated sensory experiences that feel real, and both may involve some of the same temporal and visual brain regions. However, the evidence is not strong enough to conclude that most clinical hallucinations are simply REM sleep processes intruding into waking life. The overlap is real but partial, and the causes appear to be distinct in important ways. If you occasionally hear your name called as you drift off, that is a normal feature of a drowsy brain relaxing its sensory gatekeeping, not a sign of a hallucination disorder.