Conspiracy theorists' are on the 'psychosis continuum' —Yale psychiatrist
If you can't figure it out, ask a computer
As part of their ongoing struggle to be perceived as genuine scientists, psychiatrists are now trying not only to diagnose people with “mental illness” but to actually predict who might be susceptible to “diseases of the brain.” Much attention is currently being given to utilizing artificial intelligence (AI) models by training them to distinguish between people who are “ill” and those who appear ill but are actually not.
“Sonia,” who was interviewed for this article, describes what strikes her as odd in this approach.
If you’re a psychiatrist and you need an AI model to figure out if someone’s crazy or not, then perhaps you should be looking for a different job. My husband’s been psychotic a few times, and I could see it in his eyes right away.
An exact science?
Perhaps Sonia is right, but there’s nothing in the Diagnostic and Statistical Manual (DSM, the psychiatrist’s guide to disease) about “seeing it in the eyes.” One imagines that psychiatrists would be the objects of ridicule if they tried to claim that they diagnosed patients based on such a criterion.
Under the heading, “Psychotic Disorders,” the latest edition of the DSM has a grand total of nine sub-categories. At the top of the list is “schizophrenia.”
The characteristic symptoms of schizophrenia are described as:
- Delusions
- Hallucinations
- Disorganized speech (e.g. frequent derailment or incoherence)
- Grossly disorganized or catatonic behavior
- Negative symptoms (i.e. diminished emotional expression or involition)
A person needs to have two or more of these symptoms for a “significant” portion of the time for at least a month (or less, if “successfully treated”). The exception is when the delusions are defined as “bizarre” or hallucinations consist of a “running commentary on the person’s behavior or thoughts, or two or more voices conversing together.”
None of these symptoms can be defined in any established scientific way. Despite decades of talking about neurotransmitters, measuring and observing hormones and brain activity, and experimenting with drugs that play with all of these (as well as so-called genetic markers for schizophrenia), there’s no physical test that can diagnose schizophrenia.
Are you a quasi-nutter?
More problematically, as a recent paper published in JAMA Psychiatry claimed, plenty of people sometimes experience symptoms one-to-five (and even several of them) without being psychotic.
Hallucination-like experiences and delusion-like beliefs are common in individuals without psychoses.
How so? The paper (written by Dr. Philip Corlett, an associate professor at the Yale University School of Medicine’s Department of Psychiatry) begins by quoting an early psychiatric text (dating back to 1901) which refers to delusions and the like which are linked to religious experiences. Today, that category has been broadened with the addition of “people who believe in conspiracy theories” who are on the “psychosis continuum,” though not (perhaps) technically psychotic themselves.
If madness is a continuum, then psychiatrists can’t be blamed for failing to correctly determine exactly where on that continuum a particular patient lies, so goes the thinking.
Most people on the spectrum?
Continuums are useful things, and they often come with numbers attached; that is, on a scale of 1 to 100, those between 80 and 100 (for example) can be said to be schizophrenic, those between 60 and 80 are potentially so, and so forth. It turns out that psychiatrists actually have a form of diagnosis based on a numbered continuum — it’s called the “21-Item Peters et al. Delusions Inventory (PDI).”
The PDI is made up of a list of questions on how people view all kinds of issues such as being spied upon, thinking that people can read your mind, feeling that you have a special mission in life, and so forth. The final score a person receives depends not only on what they believe but how distressed they are by it.
The JAMA paper describes how the PDI can be used to decide whether or not someone has a “psychotic illness”:
People without a psychotic illness endorse a mean (SD) of 7 (4) unusual beliefs compared to 12 (6) in patients with schizophrenia…
An SD (standard deviation) of 4, and certainly 6, is huge, and means that there was a huge range of scores in both categories. Indeed, when we look at the raw data in the original study (which was based on 35 questions, later whittled down to 21), we find that “healthy” participants scored between 0 and 31 on the 35-point scale; “deluded” participants scored between 2 and 35 points.
And the JAMA paper acknowledges this, admitting in delicious understatement that,
… clearly, these ranges overlap…
Conspiracy theorists as lab rats
So, since symptoms of schizophrenia can’t be scientifically determined, brains can’t be analyzed to decide whether someone is psychotic or not, and the 21-Item Inventory produces rather unsatisfactory results, what is a psychiatrist to do if he wants to diagnose accurately?
One answer is AI — let the computers figure it out.
But AI models have to be trained, and this is where the conspiracy theorist becomes useful. Just as an AI model being taught how to detect a cat against a variety of backgrounds needs to be tested against objects that are very similar to cats but nonetheless not cats, so too the AI psychosis-detection model needs to have “foils” who are very similar to psychotics but nonetheless not psychotic.
We would like to suggest that people who believe in conspiracy theories or new religious movements and people who hear voices but do not have clinical diagnoses might serve as adversarial examples to a network trained on the phenomenal features and cognitive and neural mechanisms of psychosis that we have accrued as a field so far.
Psychotics in-the-making
At first glance, this could appear encouraging for those who believe in what psychiatrists define as conspiracies. They’re not psychotic according to this approach. They’re just… something as yet undefined. The JAMA paper suggests that distinguishing between genuine psychotics and conspiracy theorists could reveal what divides them:
In this way, we can hone in on the important features that portend illness, disability, and distress, rather than features that are distributed more widely, even in individuals whose experiences closely resemble those of people with schizophrenia.
Intriguingly, the author of the paper does not distinguish between people who are ill and those who are not. The distinction is between people who might become ill (“features that portend illness…”) and people who apparently won’t.
If you like your conspiracy theory, you can keep it (just don't appear distressed by it)
Who are these mysterious people who could become psychotic?
According to Yale’s Dr. Corlett, these could be people (conspiracy theorists or others) who are upset by their delusions and find that they make it harder for them to behave normally:
These might include people in clinical high-risk studies who do not quite meet the threshold for psychosis risk state but who nevertheless evince some degree of distress and impairment.
He would like an AI model to be able to predict whether these people will develop full-blown psychosis or not:
Any differential prediction of conversion might impact the ongoing debate regarding the risk state and whether it merely represents an earlier phase of the illness course.
Who will get trampled under the inexorable march of psychiatrists' feet?
So, while Dr. Corlett admits that “such individuals might be classified as having a psychotic illness when in fact they do not,” he is not content to leave things there. If he (or other psychiatrists, or anyone else) decides that the field of psychiatry needs more scientific backbone and that the continuum theory of psychosis is to be confirmed, he may seek to use conspiracy theorists to “prove” this theory.
After all, he states,
… individuals with schizotypal dispositions who experience attenuated psychotic symptoms may also provide useful foils as the field of psychiatry joins the inexorable march toward precision approaches driven by machine learning.
If psychiatry is to become a precise science, it will be very helpful if psychiatrists can identify people who are likely to become psychotic and predict, based on a pseudo-scientific AI model, that they will become so.
In which case, Sonia’s recommendation, if you happen to be a “conspiracy theorist,” is:
Stay far away from psychiatry.
