The Politics of Psychiatric Classification
Scott Alexander discusses the incoherent wish for an apolitical psychiatric classification in a recent blogpost on Astral Codex Ten. His post is in response to assertions that it is better for classifications to eschew social and political influences because when they don’t do so, it results in errors like the DSM classifying homosexuality as a disorder. Alexander argues that “The people asking for apolitical taxonomies want an incoherent thing. They want something which doesn’t think about politics at all, and which simultaneously is more politically correct than any other taxonomy.”
I agree with Scott Alexander on this point, but I'll take this opportunity to clarify some aspects of the debate, especially since they usually aren't well explained in popular discussions.
The fundamental thing to appreciate is that we are looking at two different but interacting classification questions:
Is this condition in the class of mental disorders or not? Is this state psychopathological? Is this cluster of problems a psychiatric condition?
How do we distinguish between conditions within the class of mental disorders? How do we map the territory of psychopathology? Where do we set thresholds? How do we differentiate one problem from another?
The reason this is important is because the two questions require very different sorts of answers.
Here’s how I understand these issues:
There is no value-free answer to the question of what makes something a disorder.1 Disorder judgments – the way they are employed in medicine and psychiatry – cannot simply be answered by statistics or neurobiology or evolutionary history or something of the sort. These judgments are enmeshed with sociocultural and sociopolitical considerations. Disorder status is not a fact of nature that can be scientifically discovered. It doesn’t mean that scientific facts aren’t relevant. Scientific knowledge of how the brain and mind work is very important and helps us make decisions, but it isn't enough. It also doesn’t mean that disorder attribution is arbitrary or purely relative, nor does it mean that there aren’t good or bad answers. If we call something a disorder because of social prejudice, we have made an error. The nature of this error is normative. We have misunderstood and misapplied the norms for disorder judgments. These norms include judgments of typicality, rationality, adaptiveness, distress, disability, harm to others, and impediments to well-being in various conjunctions and combinations. Over time, we've learned that it's not enough to say that something is out of the ordinary, abnormal, irrational, or strange; it also has to be linked to the right kinds of harm. Disorder judgments are intended to identify states that threaten an individual’s adaptive functioning. Since this functioning is always in a particular social context, the judgments inevitably have a social and political dimension. When conditions involve a risk to others (e.g., pedophilia, pyromania, etc.), the needs of the individual have to be balanced against the needs of society. Disorder judgments, therefore, require a formal or informal process of negotiation among stakeholders. People with the condition are the most important stakeholders, but there are others as well, such as their families, clinicians, funders who will pay for healthcare and research, the legal system, etc. Just to make it clear: the process is, under ideal circumstances, constrained by the relevant considerations. It is not whimsical; it is not anything-goes relativism, where anything can be a disorder just because we want it to be.
Once we have identified a domain of disorders, or psychopathology, or psychiatric conditions, there are many different ways in which we can describe, categorize, and classify it further. And some of these ways may be as close to objective and apolitical as any method can be. Quantitative factor analytic methods can reveal the factors that underlie symptom co-variation. We can classify based on genetics, or functional neuroimaging, or neuropsychological testing, or a variety of other ways. Since psychiatry and clinical psychology are applied fields, the most important thing about a classification is how well it helps guide clinical practice. The classification is ultimately pragmatic. What method of categorization we prefer will depend on what we want to accomplish and how we can accomplish it. In situations where we are dealing with “natural kinds” — such as chemical elements in the periodic table — identification of natural kinds offers the best categorization. But when there are few or no natural kinds, as is the case in psychopathology, we have to deal with classification pluralism based on our pragmatic goals. Once we have identified certain goals, we have to undertake the process of discovery. If we want a classification based on patterns of symptom co-variation, we cannot simply make up the results. We have to do the scientific and statistical work and see what factors emerge. If we want a classification based on genetics or brain circuitry, again, we have to do the empirical work and follow the science.
Classifications like HiTOP and research frameworks like RDoC operate at this second level. They are working with implicit notions of psychopathology, or implicit notions of what constitutes conditions of medical/psychiatric interest, but they do not articulate what these notions are. In a narrow sense, they are objective and apolitical, to the extent that decisions are guided by empirical data or statistical findings. But in a broader sense, their objectivity is an illusion; they already assume an answer to what is included in the class of mental disorders. When HiTOP is contrasted with DSM, the contrast is usually intended to highlight DSM’s commitment to practical considerations. That is, considerations of insurance and reimbursement, or how thresholds are set to minimize inappropriate diagnosis, or how user-friendly the classification is. Classifications that are unconcerned with practical and pragmatic issues are also poorly equipped to serve those needs, and it is both an advantage and a disadvantage.
See also:
An answer that also adequately captures what we clinically and practically mean by “disorder.” Philosophers interested in natural functions in the abstract can work with weakly normative notions of dysfunction, but in my opinion, those notions don’t translate well into the clinical realm.
Thanks, Awais.
Both DSM and HiTOP have something in common - no mention of the brain. I don’t know how long these competing classifications (from both APAs) can keep the charade.
DSM offers some practical value as a communication tool, and it is readily accessible to the curious public. Also, the insurance companies and the CDC can collect data (thanks to the "S" in its name), and the FDA can give pharmaceutical houses a standardized language for advertising.
HiTOP can’t claim even that. The scheme is so convoluted and impractical that even seasoned psychiatrists can’t make heads or tails of it, not to mention primary care docs and OBGYNs who see most psychiatric patients.
DSM is ubiquitous but still makes a lousy taxonomy. For instance, PTSD diagnostic criteria list 20 symptoms and 4 situations, which in various combinations make 636,120 (!) possible diagnostic presentations. Lupus was called a disease with a thousand faces; how about more than 600,000? Imagine a medical disorder with the same conundrum. Or take, for instance, the DSM criteria for major depressive disorder (MDD) offering nine sets of symptoms and calling out five for the diagnosis. That makes more than 250 possible combinations, which means two people who do not have a single common symptom can still be diagnosed with major depression. How can you treat both of them with the same “FDA-approved” for major depression medications?
BTW, the third contender, RDoC, is as good as dead. The team bit off more than they could chew and got stuck in a tangle of social categories, disjointed symptoms, neurocircuits, neurotransmitters, receptors, and genes mixed with NIMH bureaucracy and politics. The statue of Laocoön and His Sons comes to mind.
All told, psychiatry painted itself into a proverbial corner. There is, however, a model that might work. It served physical medicine well for centuries. APA actively resists it despite its own statement that ”psychiatry is the branch of medicine.“ I am curious why they avoid mentioning the brain in its classification. A specific phobia, perhaps.
There is another point of confusion permeating both classifications. Despite the superficial appearance of continuity, etiology, pathology, diagnosis, and treatment are only loosely connected. We don’t treat etiology, not only in psychiatry but in the rest of medicine. The best we can do in psychiatry is to restore some missing functions and alleviate debilitating distress. The classifications should help us do that instead of muddling the picture with vague, contradictory symptoms and irrelevant categories, which continue to plague DSM.
The Comments format limits getting deeper into the medical model for psychiatric disorders. Another time perhaps.
"The classification is ultimately pragmatic. What method of categorization we prefer will depend on what we want to accomplish and how we can accomplish it."
The admission that classification is “pragmatic” is a direct endorsement of instrumentalism. But this raises a teleological question: Instrumental toward what end?
We can define the final cause as “the absence of dis-ease.” We can then define “problems” (mental disorders) as whatever obstructs that end. But because our understanding of evolution is contingent and shifting (because species are evolving, duh), these definitions are never fixed. They are temporary instruments for a moving target.
We are approaching DSM-6. Who here believes we will ever reach a “DSM-FINAL”? Who is prepared to defend an eternal taxonomy of suffering, for the rest of this universe? If we aren’t prepared to admit that our categories are transient —rising and falling across social contexts —then we are pretending our tools are "eternal truths". I'm saying psychiatry needs to embrace a dynamical universe where even “scientific laws” are provisional, not final.
The Fast-Slow-Defense model of psychopathology partially fulfill these criteria.
The real danger of instrumentalism isn’t that it’s pragmatic. It’s who the instrument serves: if it only serves the institution (insurers, legal systems) rather than the agent, there will be backlash. Because then, gents will inevitably invent their own causal languages of suffering to make sense of their lives. This is precisely why religious frameworks often “work” : they offer a language of meaning that serves the agent’s individual need for coherence and action, not the institution’s job security role.
According to Aristotle, who popularized the very concept of constitution, institutions emerge from the individual dynamics of persons interacting with each other. Therefore, individual persons constitute institutions a-priori. Therefore, institutions are not prime movers and should not be considered as such. The people trying to "rescue institutions" are failing to realize that "institutions" do not refer to individual physical people who can be held accountable. For every institutional scandal, no one goes to jail.