Food for thought: What Creates the Content of Hallucinations?

[whopper]

So you got a psychotic patient with auditory hallucinations to the degree where he can talk to them and they respond in a rational manner.

What's causing this? Is the frontal cortex devoting itself to create some of this content? Is it analogous to what's going on in a dream where the person interacts with someone? What's causing the new content there?

I got no idea.

---

Members don't see this ad.

 

[PsyDr]

So you got a psychotic patient with auditory hallucinations to the degree where he can talk to them and they respond in a rational manner.

What's causing this? Is the frontal cortex devoting itself to create some of this content? Is it analogous to what's going on in a dream where the person interacts with someone? What's causing the new content there?

I got no idea.

Luhrmann, T. M., Padmavati, R., Tharoor, H., & Osei, A. (2014). Differences in voice-hearing experiences of people with psychosis in the USA, India and Ghana: interview-based study. The British Journal of Psychiatry. doi: 10.1192/bjp.bp.113.139048

Laroi, F., Luhrmann, T. M., Bell, V., Christian, W. A., Jr., Deshpande, S., Fernyhough, C., . . . Woods, A. (2014). Culture and hallucinations: overview and future directions. Schizophr Bull, 40 Suppl 4, S213-220. doi: 10.1093/schbul/sbu012

 

[Fan_of_Meehl]

So you got a psychotic patient with auditory hallucinations to the degree where he can talk to them and they respond in a rational manner.

What's causing this? Is the frontal cortex devoting itself to create some of this content? Is it analogous to what's going on in a dream where the person interacts with someone? What's causing the new content there?

I got no idea.

Same phenomenon that creates dream content, I'd suppose...or daydreams/fantasies. Sometimes I think we don't fully appreciate the ocean of phenomena regarding consciousness or mental illness that remains mysterious to us and our prevailing scientific paradigm...it's one of the things that keeps the fields of psychology and psychiatry fresh and intriguing to me. I also think that we often forget that the metaphor of the brain and nervous system as being like a 'biological computer'--although quite useful--remains just that: a metaphorical description and not a completely exhaustive explanation of mentation. Stretching the metaphor, to the extent that the brain is a computational device, then it was 'programmed' by aeons of contingencies associated with successful/unsuccessful survival and reproduction of human and non-human ancestors over millions of years and what we refer to as concepts of 'consciousness' or 'ego' or 'self' are ultimately remote 'effects' of these cumulative selective pressures operating over inconceivable timescales. The 'source code' is irretrievably lost and non-localized. We may as well consult Jung or H.P Lovecraft as Medline.

 

[whopper]

It just goes to show you how limited our understanding is in knowledge.
There's so much I learned from a bachelor's in psychology that is very important in my day to day practice that was never taught in psych residency. (e.g. training psychometric testing, use of evidenced-based techniques in psychotherapy, knowing there's something outside of medication applications to treat patients, giving more structured discussions...)

I figure there must be something going on in the frontal cortex creating new content, but then that makes this aspect of the frontal cortex involuntary.

 

[Jules A]

Same phenomenon that creates dream content, I'd suppose...or daydreams/fantasies. Sometimes I think we don't fully appreciate the ocean of phenomena regarding consciousness or mental illness that remains mysterious to us and our prevailing scientific paradigm...it's one of the things that keeps the fields of psychology and psychiatry fresh and intriguing to me. I also think that we often forget that the metaphor of the brain and nervous system as being like a 'biological computer'--although quite useful--remains just that: a metaphorical description and not a completely exhaustive explanation of mentation. Stretching the metaphor, to the extent that the brain is a computational device, then it was 'programmed' by aeons of contingencies associated with successful/unsuccessful survival and reproduction of human and non-human ancestors over millions of years and what we refer to as concepts of 'consciousness' or 'ego' or 'self' are ultimately remote 'effects' of these cumulative selective pressures operating over inconceivable timescales. The 'source code' is irretrievably lost and non-localized. We may as well consult Jung or H.P Lovecraft as Medline.

I recently saw Elyn Saks speak. She described some of her experiences with psychosis as having a nightmare while you are awake.

 

[SmallBird]

So you got a psychotic patient with auditory hallucinations to the degree where he can talk to them and they respond in a rational manner.

What's causing this? Is the frontal cortex devoting itself to create some of this content? Is it analogous to what's going on in a dream where the person interacts with someone? What's causing the new content there?

I got no idea.

Predictive coding theories would suggest that the content would match what is required to minimize prediction error; the brain is an organ of inference and seeks to update its expectations of the world to minimize experiences of unexpected input. Psychotic patients tend to have very strong perceptual ‘expectations’ of the world which drive non-veridical percepts (hallucinations); they attribute excess salience to environmental information which should be ignored but instead leads to strongly held beliefs. Frequently one can reconcile the specific manifestation of a patients psychotic symptoms using this framework.

Google Scholar

 

[randomdoc1]

What I’ve always had an interesting experience with is when patients are stable from their psychosis, asking them about their delusions and hallucinations. Very interesting to hear what they have to say. Like dreams, they are often affected by events that recently occurred. They just about all started with a grain of truth somewhere that spiraled out of control.

 

[Fan_of_Meehl]

Predictive coding theories would suggest that the content would match what is required to minimize prediction error; the brain is an organ of inference and seeks to update its expectations of the world to minimize experiences of unexpected input. Psychotic patients tend to have very strong perceptual ‘expectations’ of the world which drive non-veridical percepts (hallucinations); they attribute excess salience to environmental information which should be ignored but instead leads to strongly held beliefs. Frequently one can reconcile the specific manifestation of a patients psychotic symptoms using this framework.

Google Scholar

Where do these 'expectations' come from?

 

[clausewitz2]

Predictive coding theories would suggest that the content would match what is required to minimize prediction error; the brain is an organ of inference and seeks to update its expectations of the world to minimize experiences of unexpected input. Psychotic patients tend to have very strong perceptual ‘expectations’ of the world which drive non-veridical percepts (hallucinations); they attribute excess salience to environmental information which should be ignored but instead leads to strongly held beliefs. Frequently one can reconcile the specific manifestation of a patients psychotic symptoms using this framework.

Google Scholar

I am as big a fan of Karl Friston and co as anyone, I have not yet seen a version of this account of psychosis that actually specifies the content of psychotic experiences.

The paper you linked also doesn't really have anything to say about that.

 

[clausewitz2]

Where do these 'expectations' come from?

This sort of model imagines the brain as constantly generating predictions about its environment and updating these predictions based on error signals conveying information about mismatch between prediction and experience. The whole network is constantly trying to minimize these signals. Often accounts of psychosis in this framework postulates erroneous error signal generation and/or inappropriate weighting of these signals when updating. You thus get elaborate predictions trying to minimize predictive error by any means necessary. They are often silent on why delusions often stabilize into a fixed belief system, but you can probably handwave something about it being a local maximum in n-dimensional space.

 

[Fan_of_Meehl]

This sort of model imagines the brain as constantly generating predictions about its environment and updating these predictions based on error signals conveying information about mismatch between prediction and experience. The whole network is constantly trying to minimize these signals. Often accounts of psychosis in this framework postulates erroneous error signal generation and/or inappropriate weighting of these signals when updating. You thus get elaborate predictions trying to minimize predictive error by any means necessary. They are often silent on why delusions often stabilize into a fixed belief system, but you can probably handwave something about it being a local maximum in n-dimensional space.

Thank you. This was what I was looking for. So, I suppose if the psychotic symptoms are generated (or exacerbated) by 'mismatches' between predictions and experience, minimizing those discrepancies would theoretically decrease the symptoms. I'm just musing on the topics of: 1) are these theories making any potentially falsifiable predictions (and, as a corollary, what are the extant 'competing' theories?) and 2) how could any of this be leveraged at the level of intervention? Are these mismatches postulated to be primarily occurring at a level of direct sensory experience below the threshold of awareness or are they postulated to be occurring at a 'higher' cognitive level (such as mismatches between expectations at an internal dialogue level (such as automatic thoughts, intermediate beliefs, or schemas from cognitive-behavioral therapy)? Or both, in some manner that is integrated by the patient's learning history and other networks of belief. For example, the psychotic patient has an implicit belief that there should not be any cars parked across the street from their apartment (and, if there is a cop car parked across the street from their apartment, it may activate paranoid/persecutory delusional beliefs). Is the idea that, in susceptible individuals, this level of 'mismatch' precipitates hallucinations (e.g., seeing a SWAT team in their bedroom)? I'm just trying to think about and understand the theoretical model. Sometimes I think we turned our back on the phenomenology of mental illness way too soon in our history as a field, considering the subject matter too 'unscientific.' However, it occurs to me that we also (in an attempt to be an 'empirical' field) operationally define almost all the meaningful constructs of clinical interest at the symptom report (which is ultimately, phenomenological) level.

 

[clausewitz2]

Thank you. This was what I was looking for. So, I suppose if the psychotic symptoms are generated (or exacerbated) by 'mismatches' between predictions and experience, minimizing those discrepancies would theoretically decrease the symptoms. I'm just musing on the topics of: 1) are these theories making any potentially falsifiable predictions (and, as a corollary, what are the extant 'competing' theories?)

One criticism of Bayesian modeling is, of course, that you can build a Bayesian model to capture just about any set of data and is thus suspect from a Popperian perspective. This paper has some more reasonably strong conceptual objections to the predictive coding model, but I wouldn't call it empirical evidence against:

https://www.sciencedirect.com/science/article/pii/S1053810017306219

and 2) how could any of this be leveraged at the level of intervention?

So far there are attempts to try and link receptor types or subpopulations to various putative functions in the model, the idea I suppose being that you could turn up/down the gain on various functions by selective pharmacological agents. I think this is somewhat foolish given how inadequate a post-synaptic-receptor-driven theory of mental functioning has been in accounting for the messiness of clinical reality but I am not sure how else to use the model for interventions at present.

Are these mismatches postulated to be primarily occurring at a level of direct sensory experience below the threshold of awareness or are they postulated to be occurring at a 'higher' cognitive level (such as mismatches between expectations at an internal dialogue level (such as automatic thoughts, intermediate beliefs, or schemas from cognitive-behavioral therapy)? Or both, in some manner that is integrated by the patient's learning history and other networks of belief. For example, the psychotic patient has an implicit belief that there should not be any cars parked across the street from their apartment (and, if there is a cop car parked across the street from their apartment, it may activate paranoid/persecutory delusional beliefs). Is the idea that, in susceptible individuals, this level of 'mismatch' precipitates hallucinations (e.g., seeing a SWAT team in their bedroom)?

The Williams paper I linked above takes this family of models to task for being vague about this; they tend to posit multiple hierarchical layers of processing with the hierarchy arrayed along some axis of complexity or abstraction, but this falls apart really quickly from a logical standpoint when you poke at it and try to make specific claims about how specific stimuli in specific contexts (environmental, learning history, or otherwise) ought to be processed. Here's another one on the predictive processing enterprise generally:

https://link.springer.com/content/pdf/10.1007/s11229-018-1768-x.pdf

I'm just trying to think about and understand the theoretical model. Sometimes I think we turned our back on the phenomenology of mental illness way too soon in our history as a field, considering the subject matter too 'unscientific.' However, it occurs to me that we also (in an attempt to be an 'empirical' field) operationally define almost all the meaningful constructs of clinical interest at the symptom report (which is ultimately, phenomenological) level.

Very much agreed about the premature abandonment of phenomenology. I think the more interesting Bayesian work that takes serious the specifics of psychotic phenomenology looks like this and yields potentially more useful results:

https://www.tandfonline.com/doi/full/10.1080/13546805.2010.548678?scroll=top&needAccess=true