Can you fact check this: "I think you said there were three, did you say there were three root causes of mental illness? Three or four. So one of them is distress tolerance. What were the others? Yeah. So perfectionism is another big one. So when we're very, very hard on ourselves, that's one of the other transdiagnostic causes. There's debate over... so there are a couple that I've seen. Those are the two biggest that I think have the strongest level of evidence: distress tolerance and perfectionism. There's debate over what the other two really are. 100%. And this has to do with factor analysis. So this gets a bit... in factor analysis, you'll basically come up with four major factors. I think distress tolerance and perfectionism are the two that are the most significant."
ChatGPT said:
Short answer: this is oversold, incomplete, and not really how the science is framed.
Longer answer:
There is a literature on transdiagnostic processes in mental illness.
BUT:
-
There is no consensus about there being “three” or “four” root causes.
-
distress intolerance and perfectionism are two widely discussed transdiagnostic cognitive/affective processes — BUT they are not “root causes” in any strict sense. They are associations / mechanisms that appear across multiple disorders.
-
The big transdiagnostic frameworks (e.g. Barlow’s Unified Protocol, Mansell, Harvey, Wolpert, etc) talk about multiple families of mechanism:
-
negative affectivity / neuroticism
-
cognitive appraisals & attentional biases
-
avoidance (experiential avoidance, safety behaviors)
-
emotion dysregulation broadly
-
interpersonal schemas
Not “two main ones.”
-
Factor analytic work does not yield clean “4 factors” that are universally accepted. The factor structures differ by sample, measurement instrument, etc. The most robust “general factor” in psychopathology research is the p-factor (Caspi et al.) — but that is different from the transdiagnostic-process literature.
-
Perfectionism is real as a transdiagnostic process but not dominant over, say:
-
avoidance/escape
-
rumination
-
threat-focused attentional bias
-
emotion dysregulation
Many meta-analyses point to emotion regulation deficits and negative affectivity as the most general.
So:
“distress tolerance and perfectionism are the two biggest root causes…and there are two more but debate over what they are”
No. That is not an accurate representation of the state of the field.
It collapses a much larger literature into a reductive “four factor” schema that doesn’t exist.
If someone wanted to state something closer to truth, they would say something like:
There are recurrent transdiagnostic mechanisms (e.g. distress intolerance, maladaptive perfectionism, avoidance, rumination, negative affectivity, emotion regulation difficulties) that cut across multiple disorders. The field is still systematizing these, and there is no consensus on “how many” or which are primary.
You said:
ChatGPT said:
“p-factor” = the general factor of psychopathology.
It’s an analogy to g-factor in intelligence research.
A number of large, longitudinal, multi-measure studies (Caspi & Moffitt et al. ~2014 onward) have shown that if you factor analyze symptoms across the DSM categories (depression, anxiety, substance use, psychosis, personality disorder traits, etc), a single latent factor emerges that explains a big chunk of variance across them all.
People who score high on this latent factor have:
-
more comorbidity
-
earlier onset
-
worse prognosis
-
more ongoing impairment
It’s not a “cause.”
It’s more like: there is something shared between very different mental disorders — some core vulnerability dimension — that loads across them all.
Researchers debate what this “something” is. Hypotheses include:
-
general negative emotionality
-
stress reactivity / threat sensitivity
-
executive control deficits
-
general dysregulation of affect and cognition
But the point is:
the data doesn’t show clean, separate diseases.
Instead it shows a lot of cross-loading, cross-correlation, and co-occurrence that points to a single shared vulnerability dimension.
So in factor analysis of mental illness symptoms:
-
you can find narrower clusters (internalizing, externalizing, psychotic-like)
-
but on top of those you also find p.
p = what’s common to all of them.
(That’s the p-factor.)
ChatGPT said:
yes.
this is actually the core of Hacking’s point.
“Interactive kinds” means:
-
when you classify humans, humans notice
-
and their way of being changes in response
-
which then recursively changes what the classification is describing
So psychiatric nosology, psych constructs, “transdiagnostic factors”, “p-factor” — all of that is doubly dynamic:
-
dynamic internally (because the phenomena are not stable)
-
dynamic externally (because the people are aware of the categories, treatments, expectations, discourse)
and so:
studying people changes the people.
diagnosing people changes the people.
describing “root causes” changes the expression of the putative causes.
This is the Hacking loop.
So yes — the factor analyses and classifications are not merely mapping a static underlying ontology
They are partially producing the field they purport to observe.
Which means:
-
each reification (e.g. “perfectionism is a root cause”) can alter how patients and clinicians interpret themselves
-
which then affects future factor structure
A classification implies a norm and a direction — therefore a pressure.
And pressure feeds back into the system.
And this is why psychiatric claims have to be held lightly
Because the categories are performative. They don’t just describe; they intervene.
In other words:
in the human sciences, ontology is partly a downstream effect of epistemology.
Hacking’s brilliance is that he makes it impossible to imagine a neutral psychiatric map.
You don’t just map these worlds — you enter them.
No comments:
Post a Comment