Self-Monitoring and Auditory Verbal Hallucinations in Schizophrenia

The positive symptoms of schizophrenia present one striking and debilitating aspect of the disorder and include delusions of control (where subjects take actions they perform to be done by others), thought insertion (where subjects have thoughts that they attribute as belonging to others), and auditory verbal hallucinations (AVH), where subjects hear the voices of others. A central question is how to explain why these symptoms occur. A central model is one that appeals to self- or source-monitoring. The general interest of this model is that it makes use of predictions, something of recent general interest and relevant to discussing the visual stream hypothesis and consciousness (a few posts from now).

Irwin Feinberg first suggested the application of self-modeling to schizophrenia, but Christopher Frith (and coworkers) is perhaps the first to come to mind when philosophers think of these models in schizophrenia. Judith Ford has done a lot of recent interesting empirical work to test the model. The basic idea is attributed to Helmholtz who raised it to explain spatial constancy in vision (why the world appears stable despite eye movement), and modern formulations of the model draw on the notions of  efference copy or, as I prefer, corollary discharge. Put poetically, the proposed mechanism involves a signal (I’ll call it the corollary discharge) that allows a system to distinguish self from other. Given how I characterized the positive symptoms above, you can see a potential resonance in talk of an “other”. Yet while self-monitoring, as a family of models, is plausibly the received view in explanations of the positive symptoms, I do not think it can explain AVH.

Here’s a simple recounting of application of self-monitoring in the case of visual spatial constancy (my take on this phenomenon was first given here). When the eye moves, the projection of the world on the retina (the retinal image) also moves. When a projection of an object moves on the retina (while our eyes are still) we experience the motion of the object. So, why don’t we experience motion when we move our eyes. The Helmholtzian idea is that when our eyes move, our brains know that our eyes move, and with that information about the self, can tag moving retinal images as generated by movement of the self rather than the other (the external world). The signal of self-movement is carried in the corollary discharge, a copy or correlated signal generated by the motor system. Obviously, there’s more to the explanation than that, and indeed it gets quite complicated. Bruce Bridgeman, by the way, has done a lot of interesting work on this, and perhaps the best empirical evidence for a corollary discharge circuit in cortical control of eye movement comes from work by a former CNBC professor, Marc Sommer (see review here). The point is that spatial constancy is determined by a kind of self-monitoring based on corollary discharge, and the use of corollary discharge is quite prevalent in forward models of movement control. But how do we extend these ideas to AVH, and auditory phenomenon?

We can think of AVH as the result of failure of self-monitoring. First, a common view is that the substrate of AVH is inner speech, where the subject generates inner speech, but given failure of self-monitoring, loses track of that speech as self-generated. The application of self-monitoring is easiest here if we think of inner speech itself as involving a motoric component and so we can again call upon a corollary discharge signal. But say one loses track of the speech as self-generated by failure to process the corollary discharge signal. One then experiences that speech, but it is not tagged as one’s own speech (cf. the changes in retinal image not tagged as self-changed). So, the speech is experienced as “other” and this leads to AVH. This gets at the basic outline of the proposed explanation. I should emphasize that we have good empirical evidence that such self-monitoring mechanisms seem to be altered in patients with schizophrenia.

I think theorists of schizophrenia reach for this model in part because at a metaphorical level, it makes sense. This talk of the “other” and of “self-tagging” slots easily in our attempts to understand AVH. But it doesn’t make much sense once you start thinking about it in more purely mechanistic terms. Let me just raise two points:

  • It is not clear how to understand the idea of a “self tag” such that it explains the sense of “otherness” in positive symptoms. Mechanistically, it is the corollary discharge signal, but that signal is used in a variety of domains where no “sense of self” is involved. This is where the poetic aspects of the model don’t have good correlates with plausible mechanisms.
  • Self-monitoring models have lost track of the phenomenology: It seems that the substrate of AVH is not inner speech, or at least not in  central cases. The reason is that AVH is auditory whereas inner speech often is not (many patients will attest to this difference when you ask them). Further, there are cases where patients respond to their AVH with inner speech which is problematic for the model since the self-monitoring has to go on and off in time with the inner dialogue, namely off with AVH, on with inner speech. This doesn’t seem very plausible.

Against this, I want to draw on the concept of automaticity in my last post to suggest another model that is much simpler. Notice that we can see the self-monitoring model as postulating AVH as the result of a failure in a control mechanism. Alternatively, AVH is just the spontaneous (automatic) activation of auditory representations of another person’s speech. As the surgeon Wilder Penfield showed when he directly stimulated auditory cortex in his patients, many then experienced auditory hallucinations. The Spontaneous Activity model holds that AVH is the result of overactivity in auditory cortex. Two things to note about this model:

  • We know that such hallucinations can arise from within spontaneous activity of the auditory system.
  • It is simpler in its explanation of “otherness” since the voice that is auditorily experienced is precisely another person’s voice. It’s built into the representation, and we do not need a convoluted explanation of how “otherness” arises due to failure of corollary discharge mechanisms.

These reflections don’t settle what I take to be an empirical question: what causes AVH? Self-monitoring models could be right, though I think that they are surprisingly undeveloped as explanations of AVH. Further, both models face a challenge in explaining why AVH has the properties that it does: voices are often heard as located within the head, are of specific voices, and almost always negative in semantic content.

Still, I think the Spontaneous Activity model should be the default view. What the Spontaneous Activity account gives you is a simple explanation of two phenomenal aspects of AVH: the sense of otherness and the involuntariness. You get otherness by virtue of the auditory representations literally representing the voice as of another and you get involuntariness precisely because AVH is automatic and, as I defined in the last post, against one’s intentions.

What of the other positive symptoms? I actually think something like the self-monitoring account will work for delusions of control since (a) I noted there is evidence of defective self-monitoring and (b) this model is best elaborated for the motor domain. I’m less sure about thought insertion, but that’s a difficult symptom to precisely pin down given different forms of thought. I’m more inclined to the spontaneous activity account for explaining why AVH arises, but it is in the end an empirical question.

A more detailed discussion of AVH is given my recent paper, and I am currently working with Raymond Cho (University of Pittsburgh Psychiatry) on thinking about how to test the model and perhaps actually testing it!


  1. Hi Wayne,

    I’m glad someone is working on these important and fascinating issues. Some questions for you:

    1. What do you think of the “Hearing Voices Network”? The organization tries to educate people about the fact that many (but not all) voice-hearers are not only non-psychotic but find great comfort and assurance in voice-hearing, which is in direct opposition to the “clinical” model of voice-hearing that only sees it as a problem to be fixed rather than a potential expression of neural diversity.

    2. What do you think of the “Third Man Factor”? ( The most famous example of the phenomena is the artic explorer Ernest Shackleton’s report. John Geiger has collected a large amount of these adventure and extreme survival stories in his gripping book The Third Man Factor as well as a website where thousands of people have submitted personal stories that have been gathered into a searchable database.

    If we assume that “Third Men” are auditory hallucinations and not guardian angels, it is impresses me that (1) these hallucinations are triggered by extreme stress and (2) seem positively correlated with increased survival in situations of life and death. This suggests the striking hypothesis that auditory hallucinations are not purely dysfunctional as per Frith’s clinical model but in fact have a potential adaptive significance, and were possibly selected for in ancestral populations. Though the first emergence of AVH could have been a side-effect of language learning, the advantages of self-regulation could have provided selection pressures for “command hallucinations”. This is similar to Dan Dennett’s story in Consciousness Explained about the hypothetical evolution of verbal self-stimulation as a means of adaptive self-control (which I am almost positive is a modified version of Julian Jaynes’ hypothesis that such verbal self-stimulation was hallucinatory)

    Several hypotheses present themselves for how AVH could be adaptive:

    (1) The hallucinations serve as a mechanism of persistent motivation in times of extreme stress e.g. many “third men” are entirely encouraging, saying things like “keep going!” in a warm, friendly voice as the person is trying to gather enough energy to keep doing and not just lie down helplessly to die (as in common in polar or mountaineering situations.

    (2) the hallucination offers “new” information not previously available to reflective consciousness that has been culled from what Timothy Wilson calls the adaptive unconscious e.g. Some “third men” give specific and detailed directions related to surviving these extreme situations.

    Looking forward to your response. Thanks!

  2. Wayne Wu

    Hi Gary:

    It’s worth pointing out that a significant portion of the non-clinical population experiences auditory hallucinations. Such hallucinations need not be negative in content, though as I understand it, the preponderance of AVH in schizophrenia is or becomes negative. I’ll have a look at the network, thanks for alerting me to it!

    I’ve certainly experienced the “third man”, in a moment of vivid stress when I was younger. At the time, I thought it was God speaking to me in an encouraging and authoritative way! (I was raised in a very strict religious household.) But I wouldn’t be surprised if many of us have had similar experiences. These days, I have more often the cell-phone buzzing in my pocket illusion.

    There are, I suspect, many reasons why they auditory system might be activated to give rise to auditory experiences that philosophers would define as hallucinations: recalling things in an auditory way, thinking in inner speech where this might be auditory in structure, etc. These can have positive influences on our ability to adapt to situations.

    What continues to puzzle me about AVH in schizophrenia are some of its fairly consistent phenomenal properties: second or third-person voice, typical internal localization (though plenty of external localization) and negative content.

  3. Dear Wayne

    Thank you for your fascinating and new ideas on this ever so fascintating topic of verbal hallucinations.

    1. Can you tell me your thoughts on how your model deals with the topic of insight – whether on is able to recognize that one has a mental illness and whether one is able to relabel unusual mental events as hallucinatory or delusional – into psychosis?

    To put it differently: how come that a significant amount of people suffering from verbal hallucinations will underlline the for them ‘essential non-pathological realness’ of this phenomenon, even if you would be able to respectfully explain them all scientific/common sensical/medical knowledge we have about them till now. I think most people would accept the, for instance, Muller-Lyer illusion for what it is – an illusion – even if experience tells them differently.
    Often not so for hallucinations occuring during psychosis. This to the point that sometimes someone suffering from verbal hallucinations will – I think sincerely – not believe that you actually don’t hear them.

    I (the lack of) insight/anosognosia is an issue every model of psychotic phenomena should address at some point. It is as much part of the phenomenology of verbal hallucinations like any other; it should be accounted for.

    2. What are your thoughts on the following:

    The number of concepts patients have to express their mental states is verylimited; and because they can only have recourse to the restricted number of concepts uniquely adapted to the needs of normal life, they easily end up using the same ideas to fill the void consuming them. Thus they are condemned to express their problems as well as they can using concepts fitted for problems of a different nature. This does not stop them from believing they are right, but also forces us to be on our guard and not literally transpose their sayings to our own mental lives. (Minkowski, 1933/1995, p.211)

    Maybe the reason why the properties of verbal hallucinations are so diverse (and sometimes bizarre) is that people suffering from them are trying to communicate experiences so different from the ‘normal’ situation that they actually do not have the words for it. Voices having no acoustical properties, phenomena like auditory insertion (which I think is more common than your Mind&Language paper footnote 16 suggests), the sometimes bizarre explanations for them…

    I agree that every investigation into psychotic phenomena has to start somewhere. Every investigation has to start from the phenomenon it wants to explain and take that (very/respectfully) seriously.
    But maybe taking the reported properties (loudness, location,…) literally is a mistake. Maybe we simply do not understand what someone experiencing verbal hallucinations is trying to tell us.

    How do you think/feel about this possibility?

    Minkowski, E. (1933/1995).Le temps vécu.. Paris: Quadrige / Presses Universitairesde France.

    interesting (older) reads on this topic:

    – Mental Automatisms (Hermes Whispers Press 2002): a collection of translated papers of Gatian de Clerambault; de Clerambault inspired Minkowski, who in turn inspired Merleau-Ponty on this topic (certainly read his discussion of hallucinations in Phenomenology of Perception, at the end of section ‘The thing and the natural world’

    • Wayne Wu

      Hi Jan Pieter (or just Jan?)

      Thanks for these comments! I’m grateful for the citations you’ve given. We’ve (humanity) been studying schizophrenia for a very long time, and I know there is a wealth of wisdom to be gained from this additional literature. I’ll be sure to look at them.

      Let me start with your second point first, which I think is spot on. Patients do have a more limited “phenomenal” vocabulary than we theorists. So, I agree that we need to proceed with caution and that taxonomies of the symptoms needs to proceed with care, especially when what they express seems to us to be rather strange. Unfortunately, the situation in clinical environments can’t be first concerned with theory as once in a clinical setting, people need to be cared for. But to the extent that patients are interested, it would be helpful to have a more open dialogue with them, to get them interested in theory.

      Actually the M&L paper you cite (thanks for looking at it) tries to make the case that in a way, we haven’t been paying enough attention to their reported phenomenology.

      That said, I think when they use everyday concepts like loudness and location, I’m inclined to take them at face value since they are not high-browed theoretical concepts. Of course, there might be indicators that the patients are using these terms in a different way (…it’s as if it was in my head, but it can’t be there, it sounds faint, to faint to be audible but I can hear it…). And then, we need to proceed with more care.

      Do you know of cases of auditory insertion? That would be interesting. You might be right, the idea of having someone else’s experience might seem really weird, so that someone might not know how to express what turns out to be auditory insertion. But they do attest to thought insertion!

      On the first point, I think there are other aspects of the experience of AVH that are not covered by my model. There must be a doxastic/belief component to AVH as well in that subjects beliefs are also in some way off kilter. You’re absolutely right that we must also look at this.

      Part of the sense of reality you point to might be nicely explained if my model is correct. Why shouldn’t a patient be surprised that you can’t hear a voice that they “hear”. After all, when auditory normally acts that way in the patient when there is a sound, we both will hear the same sound. So that part seems to me intelligible.

      Why some patients are resistant to recognizing AVH as a hallucination must have the doxastic component tied to how they deal with reality, of which I think there is good evidence. That being said, some patients do recognize that AVHs are Hs (hallucinations) and this, I suppose, can be helpful in their coping. So, maybe the place to start is to understand the differences between these two patient populations.

      Does that seem right to you? I wonder if you have further insights on this?

      • Dear Wayne

        Thank you for considering my comments and your kind reply.

        – On the ‘phenomenal vocabulary’/’theoretical concepts’:

        It is not so much that I think that people experiencing AVH’s lack the theoretical or more ‘fine-grained’ concepts but that our ‘everyday’ concepts are simply not up to describing what the experience is like. My point is (that it is a possiblity) that even someone who ‘mastered’ all possible theoretical concepts would not be capable of ‘correctly’ (the term is awkward here) describing his experience of AVH’s. Some philosophers would argue that thought insertion is a contradictio in terminis/irrational/un-understandable. (see the exchange between Annalisa Coliva and John Campbell in Philosophy, Psychiatry, & Psychology).

        Further, I think that it is a possibility that psychotic phenomena like hearing voices, thought insertion, hearing your own thoughts out loud (Gedankenlautwerden), thought broadcast, external agents speaking through your own vocal apparatus, voices with or without acoustical properties, etc.. are all ‘attempts’ to describe the same experiential phenomenon.

        Still I think we have to start somewhere if we want to explain and understand these phenomena. And that is exactly by taking at face value what people who experience them report about them. I cannot agree with you more on that point.

        – on auditory insertion:

        “To explain these remarkable [visual] hallucinations I had made a special theory during my illness, “the theory of psychic induction”. One cannot doubt that I am hallucinating; I thought by myself, the centres of my organs of perception must be highly excited. But some of my hallucinations differed in such a way from the others that, “they did not fit my character at all”, so I had to conclude that the other patients had infected me; a heavily hallucinating person can induce in the brain of a nearby person,especially when he has a certain disposition, things that he himself sees or hears, like the current of an induction coil induces another current in a nearby wire. Later, with the help of this theory of “psychic induction” by other patients, I tried to explain one of the most baroque delusions, mainly those that had the character of compulsive ideas. (Kandinksy 1881)

        I am not sure if this would fit your description of auditory insertion, but please correct me if I misunderstood your point. I think that Kandinsky is actually describing a case of auditory insertion:
        “a heavily hallucinating person can induce in the brain of a nearby person,especially when he has a certain disposition, things that he himself sees or HEARS.” I read that as Kandinsky ‘receiving’ auditory experiences of someone else.

        – on “the differences between these two patient populations”
        I think it would be difficult to separate the groups you mention as ‘insight’ in psychotic phenomena often changes over time within the same person, there can be ‘insight’ in one aspect but not in another aspect of psychotic phenomena, ‘partial insight’ etc..
        In a way I doubt whether the phenomenal and doxastic components can really be separated (but I fully agree that it is easier to express this doubt than to work it out)

        Kind regards

        Jan Pieter

        Kandinsky, V. (1881). Zur Lehre von den Hallucinationen. Archiv fur Psychiatrie und Nervenkrankheiten, 11, 453-464.

        Lerner, V., Witztum, E., & Dening, T. (2003). Victor Kandinsky, MD: psychiatrist, researcher and patient. History of Psychiatry, 14(1), 103-111.

        • Wayne Wu

          Hi Jan Pieter:

          Thanks for your reply. I think you suggested that the various positive symptoms might be in fact really the “same experiential phenomenon”. Perhaps, though I suppose I have my doubts. But one should not rule out this possibility (would you think different neural activations, for different sorts of experiential phenomena be relevant in assessing the claim?).

          Thanks for the case that you cite. That does look like a case of potential auditory insertion, although Kandinsky seems to have arrived at it by a highly theoretical influence whereas, I took the idea of insertion to be somewhat more in the face experiential. But I’m going to have to read the article, it’s a great reference.

  4. Hi Wayne

    – would you think different neural activations, for different sorts of experiential phenomena be relevant in assessing the claim?

    Certainly. But I must admit that I am not well read in the vast amount of empirical studies available. An ideal study would seem a set-up in wich subjects experiencing different psychotic phenomena (or reporting in different ways on the same phenomenon) are scanned while experiencing them.
    Is this your point? Do you have certain studies in mind?

    I do think there will be a difficulty in interpreting these data. If there are clearly different neural activations corresponding to the different psychotic phenomena would that mean that these phenomena are different on the experiential level, or would the different activations only ‘mirror’ the different interpretations/translations people make of the same experiential phenomenon. Feels a bit like a philosophical minefield here.

    – on Kandinsky’s auditory insertion:

    I agree that Kandinsky examines his experiences and tries to explain them by looking for theories. But that does not rule out that
    1) he arrives at exactly this ‘theoretical’ conclusion because it is the best fit for what he experiences
    2) that in ‘ordinary’ cases of thought insertion there also is a non-experiential component of the person looking for explanations that influences his report of the phenomenon

    Kind regards
    Many thanks for your reply

  5. This is not intended as a criticism, but it’s worth keeping in mind that the relationship between brain activity and reported experience can be quite indirect. Even if hallucinations are caused by altered auditory cortex activity, it could just as well be underactivity rather than overactivity. Underactivity could cause an increase in gain in the downstream targets of the auditory cortex, resulting in amplification of noise. That’s apparently what happens in phantom limb illusions — also there is recent evidence that the hallucinations caused by psychedelic drugs are associated with reduced cortical activity.

Comments are closed.

Back to Top