Neurobiology of a hallucination

Hallucinations are often associated with psychiatric conditions such as schizophrenia or with LSD and related drugs. Hearing voices is a characteristic symptom which is reported by about 70% of schizophrenic patients, as well as by some 15% of patients with mood disorders such as depression; and those under the influence of LSD often experience extreme spatial distortions and surreal visions.

Most common are auditory and visual hallucinations, but the other senses can also produce mirages. Temporal lobe epilepsy or brain injury can lead to phantosmia, or olfactory hallucinations, during which one detects pleasing or foul smells (e.g. fresh flowers or rotting flesh) that are not actually present. More unusual is a tactile hallucination in which one experiences the sensation of insects crawling under the skin. This can be caused by shingles or cocaine abuse, and is referred to as formication (from the Latin word formica, meaning ‘ant’).

Hallucinations arise spontaneously and are transient in nature, and so the neural activity underlying them is difficult to study. Now though, Dominic Ffytche of the Institute of Psychiatry in London has devised a novel experimental technique with which he has studied the changes that occur in brain activity as an induced visual hallucination is taking place. His findings, which are published in the September issue of the journal Cortex, provide a new theoretical understanding of what happens in the brain during a hallucination.

Theories about how hallucinations occur can be broadly divided into two categories. Topological theories emphasize abnormal activity in specific regions of the brain. For example, visual or auditory hallucinations are associated with increased activity in the visual and auditory regions of the brain, respectively, and can be induced by electrical stimulation, during, for example, presurgical evaluation of epileptic patients.

Hodological theories, on the other hand, emphasize the changes in activity of pathways connecting different regions of the brain. For example, in schizophrenic patients who are prone to hallucinations, neuroimaging shows changes in the activity of pathways connecting the frontal and temporal lobes. Further evidence comes from patients who have undergone frontal lobotomy, which involves severing the white matter tracts which contain the connecting fibres; some lobotomized patients exhibit a reduced or abolished ability hallucinate. 

Hallucinations are not restricted to the mentally ill or those under the influence of psychotropic drugs. They are actually rather common, and are often experienced by healthy people under various circumstances. They can also be evoked in various ways. The Czech anatomist Jan Evangeliste Purkinje, one of the founding fathers of modern neuroscience, realized this at an early age: 

The lively mind of the child revels in the manifold stimuli of the external world…Who does not remember, if only dimly, such games from that beautiful time? One of them, which could keep us busy at a more serious age, is as follows: I stand in bright sunlight with closed eyes and face the sun. Then I move my outstretched, somewhat separated fingers up and down in front of my eyes, so that they are alternately illuminated and shaded. In addition to the uniform yellow-red that one expects with closed eyes, there appear beautiful regular figures that are initially difficult to define but slowly become clearer. When we continue to move the fingers, the figure becomes more compelx and fills the whole visual field.


Purkinje studied this and other visual phenomena for his doctoral thesis. In the laboratory, he evoked simple hallucinations in his subjects using a bright light and a rotating wheel; immediately afterwards, he would ask them to draw the patterns they experienced. The drawing above, which comes from Purkinje’s thesis, shows one of the visual phenomena, as drawn by one of his subjects.

Ffytche adapted Purkinje’s old method for his study. He induced these so-called Purkinje patterns in 6 subjects using custom-made goggles containing light-emitting diodes to produce stimuli consisting of repetitive flashes of light of a specific frequency. Functional magnetic resonance imaging (fMRI) and electroencephalogram (EEG) were used to observe the changes in brain activity that occur during the transition between the hallucinatory and non-hallucinatory states. As a control, the neural activity evoked by the flashes was compared to the activity that occurs in response to light stimuli of higher or lower frequencies that do not induce hallucinations.

The hallucinations induced in this way began several seconds after the light stimulation and contained not just geometrical patterns but also colour and movement. They were found to be associated with increased activity in a widely distributed network of brain regions, especially in the visual cortex in the occipital lobe, but also in regions of the frontal and temporal lobes. Activity in the white matter tracts (the connections) within the occipital lobes was found to first decrease and then increase, consistent with the subjects’ reports that the hallucinations began several seconds after the stimuli were presented to them.

In the EEG experiments, the activity recorded from two of the electrodes was found to become sychronous whilst the subjects were hallucinating.  Paradoxically, the neuroimaging also showed that the hallucinations were associated with a negative relationship between activity in the visual cortex and lateral geniculate nucleus (LGN), which is the first stop for visual information sent from the eye. When a hallicination was taking place, activity in the LGN was seen to decrease, whereas activity in the visual cortex increased at the same time.

Ffytche hypothesizes that the changes in connectivity could be due to changes in the firing mode of the thalamo-cortical connections (the connections between the LGN and visual cortex). He suggests that during a hallucination the LGN switches from “tonic” mode, in which the frequency of signals in the retina are faithfully translated to the cortex, to “burst” mode, in which the two become dissociated from one another. The latter is associated with an increase in the energy requirements of the visual cortex and a decrease in that of the LGN, which is consistent with the apparently paradoxical findings of the neuroimaging.

Overall, Fytche’s findings suggest that hallucination cannot be explained by a topological or hodological explanation alone, but instead by a combination of the two. And according to his hypothesis that the connectivity changes occur because of a switch by LGN neurons from tonic to burst mode, hallucinations (or Purkinje patterns at least) occur because of temporary “blindness”. During burst firing mode, certain aspects of the visual field do not reach the visual brain because of the uncoupling of activity in the LGN and cortex.      

Ffytche, D. (2008). The hodology of hallucinations. Cortex 44: 1067-1083. DOI: 10.1016/j.cortex.2008.04.005


  1. #1 David
    September 11, 2008

    Great post as always, Mo. There are examples of getting hallucinating schizophrenia patients into functional imaging machines, this one in particular, where the activation patterns seem broadly consistent with above. I seem to recall hearing this paper described in a seminar a while back, and one of the problems mentioned was the (loud) noise of fMRI masking auditory hallucinations. Though there are ways around this, using silent fMRI modes, other imaging techniques, and obviously, visual instead of auditory hallucinations.

  2. #2 Hesitant Iconoclast
    September 11, 2008

    Thank you very much for this informative post. Hallucinations are a subject of great interest for me and I eventually intend to engage in doctoral research on the subject. Although my interest is more to do with the mechanics of auditory hallucinations, it’s nice to read about the mechanics of visual ones too.

    I’m not sure if this is strictly true for those who experience visual hallucinations, but hallucinations in general are usually marked by an increase of larger temporal grey and white matter and frontal grey matter. More recent studies reviews found severity of hallucinations to be correlated with volume loss in the primary auditory cortex and left (inferior) supramarginal gyrus, as well as the right dorsolateral prefrontal cortex (Allen et al., 2007).

    Funnily enough, compared to the non-hallucinating brain, the hallucinating brain is characterised by reduced grey matter volumes in the temporal cortex, stronger activation in subcortical centres, reduced control by the dorsolateral prefrontal cortex, aberrant activation from emotional attention centres (rostral/ventral anterior cingulate), and attenuated activation of the dorsal anterior cingulate, supplementary motor area and cerebellum which are thought to be involved in monitoring processes.

    This is an interesting paper you’ve referenced, thanks again.

  3. #3 yengkhom
    September 12, 2008

    I am interested to and wanna peep into the symptoms and find out some link with some mysteries.

  4. #4 anon
    September 12, 2008

    I tend toward skepticism with regard to equating hallucinations produced by repetitive optical stimulation, those produced by psychedelic drugs in normal subjects, and those that occur in schizophrenics. The latter two categories are usually accompanied by other global changes in state of consciousness and may have subtly but significantly different mechanisms.

    Since this is nominally anonymous, I may as well report that I have self-experimented with psychedelic drugs, with repetitive visual stimulation, and with the hypnagogic state (that occurs between waking and sleeping). The types of imagery produced by each of these methods are subtly different, and the accompanying cognitive and emotional occurrences are substantially different.

    n-n-dimethyltryptamine (DMT) is powerfully hallucinogenic in doses moderate enough to be useful under lab conditions (by this I mean, one can obtain useful degrees of visual hallucination without radical changes in consciousness overall). The imagery is richly colored, and mildly geometric in a manner similar to stained glass illuminated from behind by sunlight. Brightness is less than for sunlight-illuminated stained glass, and colors are more diverse and include more subtle tones. Usually the imagery is “archetypal” in a manner that includes religious themes. There is a corresponding emotion of awe and wonder. From the latter point I would infer that the right temporal lobe is involved in the process.

    The overall subjective impression of the DMT imagery is that it originates from processes that occur “deeper” in the brain than the visual cortex, and is relayed to the visual centers of the brain for display. The foregoing impression was not an interpretation colored by the emotional state.

    At all times, I was aware that the experience was induced by the compound, and at no time did I mistake a hallucination for something actually present in the environment. This was also the case for the following two procedures.

    Note, do not try the above at home. Strictly speaking it is not safe to use these compounds except under proper medical supervision. Now that the US FDA has begun to authorize human subject research with psychedelics, there is no excuse for individuals taking these compounds other than in authorized projects.

    Note for researchers who would like to pursue authorization to use DMT in human subject studies: dosage has to be calibrated closely to individual body weight, and empirical testing should be undertaken to determine the appropriate ranges. Too low and you don’t get the phenomena you’re looking to investigate. Too high and the experience becomes “challenging” for the individual in a manner that would be useful in psychotherapy but not useful in research. At the correct dosage levels, the individual retains more than enough lucidity to be able to cooperate in research procedures, while also experiencing the intended visual imagery. DMT is in many ways an ideal research psychedelic because the peak effects last about 10 – 15 minutes, a significant alteration of consciousness is present for about another half hour, and a more moderate alteration is present for another hour or so. For safety, subjects should remain in the lab for 3- 4 hours after the last effect has worn off, and should not drive for the remainder of the day but should be driven home by a lab assistant or personal friend.

    Repetitive visual stimulation was obtained via a “sound & light device” commonly sold as a meditation aid, that used small incandescent bulbs behind a translucent white plastic cover mounted in a set of goggles. The flash rate could be manually varied; and interestingly enough, could be varied independently for each eye. In the latter setting, there were combinations of frequencies (left vs. right) that produced imagery that was in some ways similar to that produced by DMT, but in other ways dissimilar.

    The imagery in this case was more geometric and more brightly illuminated (the latter being no surprise since after all bright lights were flashing). However it did not tend toward any particular thematic content, and religious/archetypal themes were notably absent. Rather it appeared that the brain “organized” or “interpreted” the flashing light patterns in accord with whatever random cognitive content was occurring at the moment, for example memories, ideas, etc. The emotional state was one of calm relaxation, accompanied by curiosity. Perception of the physical body was decreased in a manner analogous to what occurs when engaged with a train of thought, as for example while writing.

    Note for researchers: The best results are obtained with sound/light devices using white lights (nowadays, white LEDs) and with manually variable flash rates, though it would also be highly interesting to have an output from the device that allows the flash rate to be logged and correlated with EEG output. In this case you can allow the subjects to control the flash rates in order to produce the most “interesting” results, and the flash rate / EEG comparisons would be publishable in and of themselves. One important precaution: you will need to pre-screen for photosensitive epilepsy. Aside from that, the procedure has no obvious safety issues.

    The hypnagogic state is for me most readily obtained by having to go to sleep while preoccupied with a task. There is not a feeling of sleepiness or tiredness, but rather a sense of being quite awake. When the imagery begins, it immediately is accompanied by (I want to say “produces” but that would be an interpretation) a strong sense of curiosity and attention, as if watching a stream looking for fish or frogs. Note, this imagery is not merely an outcome of lying down in a dark room, as the latter procedure in and of itself does not produce this result.

    The imagery appears to be more three-dimensional than that produced by DMT or by repetitive visual stimulation. Colors are richer and deeper, the “landscape” is darker but no less colorful. There is no thematic consistency here either, e.g. religious/archetypal themes are not particularly present, memories are not a substantial source of images, and so on. Rather, it appears that the images are a combination of the purely abstract elements, and random concrete elements as might occur by calling up photo albums online. As with DMT, it appears that the process was driven by events occurring at a deeper level of cognition, and the lack of the emotion of awe demonstrates that this inference is not an emotional artifact.

    The key difference between these procedures appears to be: The imagery produced by repetitive visual stimulation appeared to occur “from the outside going inward,” that is, from visual impressions first, to interpretive functions second. The imagery produced by DMT and by the hypnagogic state appeared to occur “from the inside going outward,” that is, from deeper cognitive processes first, to display of those processes as visual effects second.

    Getting the hypnagogic state into the lab reliably could be a very interesting task, as it is as sensitive to disruption as the sleep state. There have been attempts to induce this state through the use of audio stimulation (e.g. the Monroe Institute’s “hemi-sync” procedures) and these have been partially successful to an encouraging degree. However, as with DMT vs. flashing lights, I suspect the mechanisms differ between the hypnagogic state induced by attempting to sleep, and the analogous state induced by the Monroe Institute’s procedures (which were originally intended for use studying “out of body” experiences). (I’ll be eagerly looking out for any publications in this area because personally I find the hypnagogic state to be very interesting and to some degree “useful” as a creative source.) There appear to be no safety issues with respect to the hypnagogic state, as it appears to be a normal occurrence on the way to sleep for most people (and usually is not noticed due to the onset of sleep and not remembered the next day; in my case it is noticeable and memorable when I’m task-preoccupied and thus not sleepy).

    I’m going to guess that if you could get a human subject study comparing the effects of DMT to those of repetitive visual stimulation, you would get significantly different subjective descriptions, and also that you would see different mechanisms at work in the brain. I might be wrong. And someone might get an interesting paper to publish from doing this.

  5. #5 anon
    September 17, 2008

    anon: Thanks for that informative post. I am fairly well acquainted with DMT from non-academic, so-called “psychedelic” literature and online sources (discussion boards,, etc.) My interest in the chemical and the possibilities it holds for understanding human consciousness (and more) were spurred by the book “DMT: The Spirit Molecule” by Rick Strassman. In it, he postulates in an admittedly unfounded way, the possibility that DMT is the primary chemical responsible for all hallucinations as well as dream-type consciousness. An increase in its hippocampal production and distribution takes place when certain environmental triggers (flashing lights, physical exhaustion, REM sleep, etc.) are introduced/present.

    His study was conducted in the early 90s and his theories on DMT were uncorroborated by neural imaging or other neural probes. My question for you and any other readers is: have any such neurological studies been conducted since the publication of Strassman’s book and what is the scientific community’s overall consensus on the postulations made therein?

  6. #6 anotheranon
    September 19, 2008

    “Now that the US FDA has begun to authorize human subject research with psychedelics, there is no excuse for individuals taking these compounds other than in authorized projects.”

    I gotta say that this seems wrong. That’s like saying in 1640 that because the roman catholic church has allowed some select astrologists to look through telescopes, no other people should look through them.

    It’s really obvious that the tools Alexander Shulgin has developed, along with lsd dmt and psilocybin, have a huge role to play in studying how the brain produces consciousness. It’s really obvious that this research is stifled by the same irrationally dogmatic patterns of human behavior that stifled progress throughout history.

  7. #7 jerry
    October 12, 2008

    DMT discussions usually devolve into nonsense by people stuck in
    permanent DMT mode who start talking about the “sacrament”
    and mushrooms and The Last Supper and I’m not a druggy I’m
    a deep explorer of consciousness kind of guy etc. I’m glad the comments
    stopped where they did.

  8. #8 Bill Connelly
    October 12, 2008

    Hmm. I’m not convinced. I know the brain is ripe for chicken and egg conundrums, but I can’t see a simple visual stimuli whacking thalamocortical neurons into burst firing mode. A) Find me a study that shows any useable information is being passed by neurons in burst firing mode (i.e. the subject is awake and attentive). and B) How would the neurons get into this state? You need strong GABAergic drive presumably from the reticular thalamic nucleus to push neurons down into that ultrahyperpolarized state to get them to fire like that. What is driving the activity in the rTN? It has to come from the cortex.

    Maybe you could imagine a state where the cortex is specifically driving activity in the rTN to shut off the ‘nonsense’ coming from the eye, but essentially we are then talking about a mini- sense specific absence epilepsy event.

    No, it just seems a bit unlikely to me.

  9. #9 Barbara Peck
    January 20, 2009

    I am not a physician, therefore not an expert on the brain. But I have had daytime hallucinations while on the antibiotic Doxycycline during the treatment for neuro Lyme disease.
    The actual hallucinations were preceeded by ever increasing and vivid dreams, lucid dreams and eventually night terrors.

    I was fully aware when I was having a day time hallucination, and was fully able to hold a conversation with a friend, while a hallucination was occuring behind her.

    I had about 3 hallucinations before I went off the antibiotic. None of them took over 100% of my perception.

    One of them was when I was driving my car, and all of a sudden I felt like I had had 3 glasses of wine.. a few seconds later, when I looked at the stop sign to my right – it was not a stop sign, but a giant Tootsie Roll Lollypop standing on end- with a red wrapper atop. I knew immediately this was a hallucination, and looked away. WHen I looked back, everything was normal, and the stop sign was normal.

    Doxycycline crosses the blood brain barrier – so some neuroscientists must know where in the brain doxy may dock or accumulate.

    Maybe my experience with Lyme and Doxycycline will help understand what areas are stimulated, and understand hallucinations in more detail.

  10. #10 Ty Marbut
    February 11, 2009

    I have a question: I’m an undergraduate at Reed College and am working on a Purkinje Goggle device. I know that 5-30hz is the flash frequency, but I need to know the duty cycle as well to write my LED flasher program. Duty cycle is “on” time over period, usually expressed as a percentage. A duty cycle of 50% would be, for example, 20ms on and 20ms off, whereas a duty cycle of 10% would be, for example, 4ms on and 36ms off. So, what duty cycle works best? Are brief strobes best, or 50/50 on/off, or brief periods of off among relative long on times? Perhaps I should just experiment? If you have info, please respond!

    Thank you,
    Reed College, Portland, OR

  11. #11 Mo
    February 11, 2009

    @Ty: I have no idea what will work best, so you’ll have to do some experimenting.

  12. #12 Jennifer
    May 7, 2009

    Hi Barbara Peck, you are the first person I have found that has also had hallucinations with Doxycycline! When I took it, my apartment all of a sudden turned into a haunted apartment. It was really trippy and very vivid. I was scared. I was seeing things that I knew were not possible.
    I remember looking into my large ESPRIT purse and instead of the normal there was someones chopped off head in MY purse.
    I was too freaked out and decided to stop taking the Doxycycline. Anyway, I never told anyone about these hallucinations bcuz I didn’t want to sound crazy and abruptly discontinued Doxycycline and any further doctor appointments. I guess I probably still have MRSA, I dont know. . .this was back in 2005. Ive gone onto natural antibiotics, alkanizing the body, probiotics (yogurt with fruit=yum), vitamins, minerals and upping the oxygen.

  13. #13 Cindy
    May 12, 2011

    My daughter was 13yrs when she started taking Doxycycline 150mg she started to hallucinate and I had to hospitalize her after being on this medication for about a month, my question is if this medication is the cause of hallucinations!!!! why the would the docs perscribe this and way wouldnt the docs see what was causing this problem???? It was be helpful to know this was my daughters problems and that this would go away…….