Hallucinations: seeing what you expect

Peter MolemanArticles, Psychiatry6 Comments

Verbazingwekkend en paradoxaal | Peter Moleman

Translated by Rumia Bose

The film “A beautiful mind” ends with the Nobel prize being awarded to John Forbes Nash Jr. (Russell Crowe), the brilliant scientist who, for a large part of his life, suffered from psychoses and hallucinations. As Nash is leaving the venue in Stockholm he sees the Russian agents Charles, Marcee and Parcher. The agents however existed only in his psychotic world as hallucinations. When asked how he could ever believe in their existence he answered: “Because the ideas I had about supernatural beings came to me the same way that my mathematical ideas did.”1(https://www.goodreads.com/author/quotes/14312.John_Nash). How can this be? How can you see things, whole scenes and events which are not there, and experience them as real?

Is the brain an analysing machine?

Till recently we thought that it worked as follows. You are continuously bombarded with information coming via your eyes, ears, nose, all your sensory organs. From all this information you have to choose what is relevant, what you will react to. All perceptions are – consciously or unconsciously – analysed: what looks good, what was helpful in the past, what do I want to do, what is dangerous, and more. You make choices and react on the basis of this analysis. But this process turns out to be impossible. What should you look at? There is so very much to look at. And even if you knew which way you should be looking, which objects in your field of vision should you be focusing on? Or is it what you hear that is more important? This means that there needs to be a preconceived plan which directs which way you look, what you focus on, what you perceive. What would such a plan look like? It would need to be based on what you like, what you have earlier found to be useful, what you want to do, what you usually consider to be dangerous and much more. Therefore all the above-named aspects of analysing your perceptions need to be included in a master plan beforehand.

No, the brain is a prediction machine!

But what I refer to as a plan is in fact too limited. Within your brain there is a coherent model of the whole world outside you. That model is based on all your previous knowledge, perceptions, needs, feelings, emotions. Based on that model, your brains form a prediction about how the world will look in the following moment. In your brain this prediction is checked against what you perceive. This is very efficient, because if these coincide, then your brains do not have to do anything different. If the perception does not coincide with the prediction then one of two things happens: either your brains adjust the model and the prediction with it, or you change the world in such a way that the perception does coincide with the prediction. Two examples can perhaps help understand this.

Je kijkt met je ogen, maar ziet met je hersenen. Peter Moleman
Fig. 1 What can be seen here?

It usually takes some effort to identify anything in this illustration2See also You look with your eyes, but see with your brain.. It is a Dalmatian. Once you know this and have been able to visualise it, on a following occasion you see a dog at once, because you have adjusted your expectation.

Anil Seth: Your brain hallucinates your conscious reality

Anil Seth demonstrates another example in a TED talk. Without the preceding information on the basis of which you can predict what you are about to hear, it is difficult to isolate words from the noise. Some people do not manage this even after they are told what can be heard. It is not always easy to find this sort of example, because predictions and perceptions are almost always being instantaneously and imperceptibly coordinated.

Hallucinating through sensory deprivation

The correction of predictions by perceptions can go wrong and then you may hallucinate. The simplest example is when you cannot sense anything. If you put someone in a space, wrapped in cotton, in the dark, in absolute silence and free of odour, then it does not take long before the person hallucinates. Your brain always makes predictions, and these begin to lead a life of their own because they can no longer be corrected by perceptions.

Schizophrenia and hallucinations

People with schizophrenia, such as Nash, often suffer from hallucinations. In their case there is nothing wrong with their sensory organs, but if the perception does not coincide with the prediction then their brains attach more significance to the prediction than to the perception. They do not adapt the model of reality in their brains (sufficiently) on the basis of their perceptions. That starts with seeing a hazy figure or hearing murmurs that are not there. This precursory hallucination then drives the next prediction. And in this way in the course of time entire new plots develop which are indistinguishable for the affected person from real life. The model of reality begins to lead a life of its own which has little or nothing in common with what is actually taking place.

Psychosis proneness and brain development

Psychosis proneness is related to brain development after conception and in childhood3See The basis for most of the severe mental disorders is clear. This usually does not manifest in childhood with hallucinations and other psychotic phenomena. It is however possible to determine psychotic predisposition with subtle tests. These reveal that a psychosis-prone person has a predilection for experiencing normal perceptions as unusual or surprising. This hampers the construction of a stable model of the world with accompanying predictions, because each time something unusual is perceived, the prediction has to be adapted. Later, mostly during or shortly after puberty, this can lead to the perceptions having less and less corrective effect on predictions. As described above, this can lead to hallucinations and other psychotic phenomena. The usual advice to such people is to avoid stress, but I think that they should avoid (too) surprising sensory perceptions to avoid developing a psychosis. Often this avoidance does not work sufficiently and antipsychotics are required. Antipsychotics work on a brain system that follows the discrepancies between predictions and perceptions and activates mechanisms for correcting these4See Where are the new medicines for psychiatry?. This also explains that behavioural adjustment and antipsychotics complement each other. The more a person can avoid surprising sensory perceptions the less antipsychotics he will need.

Do you hallucinate?

Your brains therefore make predictions over what is happening in the world around you, and these are corrected by perceptions of the world entering through your sensory organs. In this way the model in your brain corresponds to reality. When you hallucinate, this correction is missing and what is playing in your head does not match the reality. But because that what is happening in your head does not appear any different from normal, a hallucination is not distinguishable from reality. This means that you can never tell if what is playing in your head is being corrected by your senses. Even if you are not psychosis-prone you never know for sure if you are hallucinating.


Poletti M, Tortorella A, Raballo A (2019): Impaired Corollary Discharge in Psychosis and At-Risk States: Integrating Neurodevelopmental, Phenomenological, and Clinical Perspectives. Biol Psychiatry Cogn Neurosci Neuroimaging DOI: 10.1016/j.bpsc.2019.05.008

Corlett PR, Horga G, Fletcher PC, Alderson-Day B, Schmack K, Powers AR (2019): Hallucinations and Strong Priors. Trends in Cognitive Sciences 23:114–127.

Cassidy CM, Balsam PD, Weinstein JJ, Rosengard RJ, Slifstein M, Daw ND, et al. (2018): A Perceptual Inference Mechanism for Hallucinations Linked to Striatal Dopamine. Curr Biol 28:503-514.e4.

Sterzer P, Adams RA, Fletcher P, Frith C, Lawrie SM, Muckli L, et al. (2018): The Predictive Coding Account of Psychosis. Biol Psychiatry 84:634–643.

Adams RA, Stephan KE, Brown HR, Frith CD, Friston KJ (2013): The computational anatomy of psychosis. Front Psychiatry 4:47.

Hoffman DD (2011): The Construction of Visual Reality; in Blom JD, Sommer IEC (eds): Hallucinations, p. 7-15. Springer Verlag. ISBN 978-1-4614-0959-5

Vaughanbell A (2009): Hallucinations in sensory deprivation after 15 minutes [Internet]Mind Hacks [cited 2019 Jan 28];Available from: https://mindhacks.com/2009/10/…

Corlett PR, Frith CD, Fletcher PC (2009): From drugs to deprivation: a Bayesian framework for understanding models of psychosis. Psychopharmacology (Berl) 206:515–30.

Fletcher PC, Frith CD (2009): Perceiving is believing: a Bayesian approach to explaining the positive symptoms of schizophrenia. Nat Rev Neurosci 10:48–58.

Merckelbach H, van de Ven V (2001): Another White Christmas: fantasy proneness and reports of ‘hallucinatory experiences’ in undergraduate students. Journal of Behavior Therapy and Experimental Psychiatry 32:137–144.

6 Comments on “Hallucinations: seeing what you expect”

  1. Interesting article Peter as well as the comments of Paul. Thank you.
    I have a friend who suffered from psychosis due to (childhood) trauma. Unfortunately he ended up in solitary confinement. What he needed was human contact, genuine personal attention. He was confused, not a danger to others.
    While in many countries solitary confinement is against the law, in The Netherlands it is still being used rather often. Not really something to be proud of….
    My friend is now doing very well by the way.
    He is now also conducting workshops on psychosis. With special virtual reality glasses people can experience what it is like to have a psychosis.

  2. Thanks Peter
    I enjoyed reading this. I’m interested in your advice regarding the avoidance of surprising sensory experiences if one is prone to say hallucinations. I don’t feel clear on how one might do this in practice but I wonder if some of the potential value of psychotherapeutic approaches would lie in finding alternative (prediction error-minimising) higher explanations for mismatch/surprise. I find in the clinic that talking to people about the mechanisms for visual illusions (e.g. Kanizsa Triangle) gives them a stronger sense of how there is not necessarily only one interpretation for their perceptual reality but that there may be alternative – and more fruitful or less distressing – ways to construct experiences. So if avoiding the sensations is difficult, another approach is to look at re-evaluating the priors
    Just a thought

    1. Paul, nice to have you here commenting. What you suggest as a psychotherapeutic approach is kind of a meta-instruction. I don’t know, but this may work for people who are not 100% sure their hallucination is reality. Also, a certain degree of intelligence and reflective ability may have to be present. So, having different options may be helpful.
      I see avoidance of surprising sensory experiences as a more precise way of avoiding stress. When and where does a patient get overwhelmed, confused by sensory input, noise, visual input etc? Also, which personal, social interactions are confusing, unsettling (so, more than only direct sensory experiences)? So being overwhelmed, confused, unsettled may be more of an indication of unresolved mismatch/surprise than focusing on stress in general. This has to be personalised of course. Probably nothing new to you and most psychiatrists. But an important point is that you may reduce the dose of the antipsychotic in parallel if this succeeds. Every bit less of an antipsychotic is of benefit to the patient. I think the mechanism of “surprise reduction” with antipsychotics or effective behavioural interventions is similar..
      For people not yet psychotic, but at high risk, to administer preventive antipsychotics is researched. I am wary of that possibility. I would welcome research into the effect of behavioural interventions as discussed here, as an alternative. Or did I miss developments in that direction?

      1. I think you’ve got this backwards. Avoiding novel stimulus would just shift the threshold for what is novel, and you can see in anxiety disorders a pattern where increased avoidance leads to ever-expanding ranges of triggers to be avoided. (Quickly the thing that most of the anxiety is about is… experiencing the anxiety.)

        Maybe people who are aware that their brain is potentially mis-editing their interpretation is better addressed by controlled exposure until the experience gets boring.

        This certainly explains my personal experience with amphetamine generated psychosis: when certain patterns show up, check perception for plausibility. (Real people don’t find me interesting enough to talk about with no interruption. So if my hearing keeps reporting that input, odds are it’s not really happening.

        This is obviously highly anecdotal, and may not model schizophrenia appropriately. I’m not schizophrenic, but I’ve abused drugs heavily at various times in my life and I’m no stranger to the chemical induction of states where sensory input becomes excessively salient + novel. The first thing I want to point out is that absent the additional component of heightened threat, dropping these expectation-constructed filters is central to some of the most pleasurable and transformative experiences: psychedelics and MDMA.

        Add threat and you get paranoia. Usually this is meth or cocaine driven, with a substantial (maybe even primary) factor being sleep deprivation. (I feel like sleep is an underappreciated factor in psychological health.)

        Years later I still catch myself thinking that the people in the next room are talking about me. It’s not particularly distressing, but it certainly means that I need to “go the f*ck to sleep”.

        Please excuse the light editing, I hope it makes sense, the framing of the experience as excessive novelty makes (ironically the kind that sets off some reflexive caution for being perhaps too perfect.)

        1. Thank you for your comment, which adds valuable real world experience to my post. I do not think I got it backwards, but it is complicated. Your input might help clarify some points.
          First your remark about sleep may be spot on, but I would not know how that affects the mechanism I explained.
          I do not think avoiding novel stimuli would shift the threshold. Avoidance in anxiety disorders may expand, but I think that is a matter of increasing generalization of anxiety provoking stimuli.
          Anxiety during psychotic experiences (be it amphetamine induced or otherwise), I think, signifies a mismatch between prediction and sensory perception that can not or is not easily corrected. Anxiety could also signify, I think, a mismatch between hallucination (based on uncorrected prediction) and other internal processes, thoughts etc., which is different from a mismatch with actual perception. When you say: “add threat and you get paranoia”, may be based on unresolved mismatch.
          But hallucinations do not have to be unpleasant, threatening. Many (or some, I do not know) patients with schizophrenia live with their hallucinations and some would not want to live without (especially not when the hallucinations are suppressed with antipsychotics with their dulling effect). In that case there may be no anxiety, paranoia or other severe negative effect on their (daily) live. And perhaps they sometimes can look at it like you do when you say “check perception for plausibility”. Whether they experience “pleasurable and transformative experiences” like you do, I do not know.
          In the end, what I mean by “avoid ( too) surprising sensory perceptions” is probably the same you mean by “controlled exposure”. What I meant to emphasize is that this is more precise than avoiding “stress” in general. Every psychosis prone person may be best served if he learns exactly what is too surprising or what is beyond his controlled exposure. This will differ for every person.
          I am curious whether you agree and whether you have further comments.

Leave a Reply

Your email address will not be published. Required fields are marked *