In the paper, I describe 2 barriers to progress that our field has not sufficiently grappled with.
Barrier (1): diagnostic literalism, i.e. mistaking mental health (MH) problems a person has with the diagnosis a person receives.
In short: MH problems ≠ diagnoses.
This becomes obvious when you look at the history of the DSM, which I briefly sketch in the paper. Diagnoses were meant as rough clinical proxies. They are not the kind of things that lend themselves well to e.g. biomarker discovery, or one-size-fits-all treatments.
Take DSM-5 depression symptoms. They go back to a paper by Cassidy in 1957, who established certain rules (depression has x symptoms; you need y symptoms to get a diagnosis).
When a journalist asked Cassidy in 1980 where these rules come from, he responded:
"It sounded about right".
And of course he did. Diagnoses like depression & schizophrenia are not the same kind of thing as helium or magnesium.
The former are pragmatic kinds, meant to be clinically useful. The latter are discovered in nature.
Science policy vs science.
This isn't a bad thing. High blood pressure predicts mortality, but the threshold remains somewhat arbitrary (different societies, different thresholds).
Other examples are from physics (e.g. there are multiple reasonable definitions what a planet is, cf the Pluto drama).
Conclusion part (1): we have devoted most of our resources to study diagnostic labels that summarize the complex MH problems states of people, rather than how biopsychosocial processes give rise to these problems.
This calls for studying MH systems *as systems*.
Barrier (2) I discuss in the paper is reductionism, i.e. the idea that you can figure out the properties of a system given the properties of its parts.
This works well in many cases; I briefly write about my own experience of someone giving me a bicycle that I broke (& repaired)
This works bc bikes are simple systems, & characteristics of bike parts remain unaltered when investigated in isolation.
But the last 5 decades have shown us that reductionism does not work well in complex systems (weather, stockmarket, ecosystems, internet, etc).
I then briefly discuss biological (or explanatory) reductionism as one example for a reductionist framework that has hindered progress in our field. The arguments will be familiar to most, so I won't discuss them in this thread in detail.
Conclusion (2): reductionist frameworks have led to insights into human biology, but told us little about MH problems—not bc biology is not crucially involved, but because diagnoses (i.e. labels) probably aren't the right targets to start with; cf great recent work by @NIMH_RDoC.
To sum up these 2 barriers, diagnoses & reductionism are useful epistemological tools for describing the world; once they become ontological convictions about how the world *is*—as they have in our field, at times—they start obfuscating scientific insights.
How to move forward? I briefly sketch the idea that MH states such as 'depression' or 'PTSD' emerge out of dynamic, biopsychosocial systems of components.
See figure (I copied the caption next to it) for a brief summary; this is explained in more detail in the paper.
I then showcase, with examples, that this perspective provides new lenses through which to study mental illness (e.g., attractor states, phase transitions), & new levers to treat them (e.g., early warning signals, novel treatment targets).
One of the examples I discuss is the idea of attractor states, and I use the analogy of a playing card held between two fingers to discuss what attractor states mean for system resilience.
Again, capation copied next to it for easier read on Twitter (looks better in paper ^^).
This is prob the most "shoulder of giants" paper I've ever written, & I'm happy the journal granted me 50 instead of the initial 15 allotted references.
But even 50 refs don't do the field justice. If you're looking for further reading, let me know the topic & I'll try to help.
Thx to folks who provided feedback and/or have contributed to my views on this via discussions & collaborations in the past years (incl @BorsboomDenny, @DianeOLeary, @RanyAbend, Ken Kendler, Don Robinaugh, & dozens of others I have had the opportunity to learn from).
End 🧵
(I apologize for the 🚨 in the first tweet; they look horrible and I vow by the old gods and the new to not do that ever again)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I often get the question which depression scale people ought to use for their work. Time for a short piece on the future of depression measurement. 🧵
1/ As most of you know, there are many symptom measures, for different purposes and populations. I see little value in creating new ones; pick what is validated for your particular question and population.
2/ Being careful here is important because different scales perform differently in eg sensitivity to treatment effects, or psychological vs pharmacological treatments, as we describe in this piece.
"Studying mental health problems as systems, not syndromes" online now. It's a short piece that I substantially rewrote in the last 6 months, and I think it has much improved. 1/19
The paper tackles to barries to progress in understanding, predicting, and treating mental health (abbreviated as "MH" from here on) problems, and suggests ways forward.
The two core roadblocks are diagnostlic literalism and reductionism. /2
First: *diagnostic literalism*, mistaking diagnoses for MH problems. I explain the differences between these 2 classes of things in detail using language from philosophy of science, & many examples other nosologies or taxonomies (e.g. chemical elements, biological species) /3
You may remember the recent NatureMedicine paper where authors claim to show—as the first author says below—that psilocybin "liberates the entrenched depressed brain". This led to considerable news coverage.
3 pretty remarkable things have happened since this was published 🧵
1/ First, authors have admitted they switched away from the registered primary outcome.
The justification reads like a clear concession of p-hacking to me: we did it because it worked better. Maybe I am missing something—curious how others see this.
2/ Second, in response to criticism of multiple testing & 1-sided tests, the authors appear to straight-up admit to doing something that, at least in my area of research, is considered by many a questionable research practice.
Got a new macbook & spent some time setting up a few useful apps I like. Some of them cost a little money, but the features were worth it for me.
A little 🧵
(COI: none, I am not involved in any of these apps or get anything for mentioning them here)
1/ Middleclick
Allows you to middle mouse click on touch pad and magic mouse by clicking with 3 fingers as a default. I use middle click a lot eg. to close browser tabs, so quite useful for me.
🔥 Folks—there's a new paper in town, and I'd love to tell you about it. In the paper I wrote with @JkayFlake & Don Robinaugh, we take a bird's eye view on depression measurement. History, present, future. In particular, we discuss shaky theoretical & methodological foundations🧵
2/ Paper in one tweet:
Limited progress in depression research is in part due to the lack of focus on measurement. We review the many problems with depression measurement, including validity and reliability. We discuss shaky methods & theory foundations, & provide paths forward
/3 Two tweets on how the paper came to be.
Don was my office mate in 2016, & encouraged me to write what he jokingly called my "manifesto" on depression measurement, given that a lot of my work was focused on that topic at the time. I started the first draft back then.