The debate about classification in psychiatry can tend to get stuck on the question of whether mental health diagnoses represent actually existing entities or not. One one side of this debate are constructivist anti-nosologists, who claim that they are not. They invoke the argument that such concepts are actually social constructions wielded by whoever happens to have the power. The opposing view is positivist realism, which sees diagnoses as real entities in the world with real biological underpinnings. Between these two extreme positions we can locate an enormous range of views, but the fundamental divide comes down to whether diagnosis does or does not usefully pick out something that really exists. Twenty years ago, philosopher of science Ian Hacking drove right through this dialectic impasse to create a position that took elements of both. His position allows us to pick a course between the possibility that diagnosis are in some way real sorts of things and the obvious truth that they are not a neutral, harmless sort of thing.
Hacking recognised that when scientists classify "natural kinds" like quarks, or trees the classified entities in question don't know or care that they have been classified. Researchers may or may not have "carved nature at the joints" but, if they are picking out a reliable pattern, there is a natural kind that can be of value to scientific inquiry Kraeplinian psychiatry seeks to divide people up into natural kinds, but Hacking points out that when people are classified it is into "Human Kinds" ("I choose the label 'human kinds' for it's inhumane ring" he said), which do not operate in the same way. We may legitimately be able to pick out patterns in groups of people, but an interesting phenomenon arises when we group them together and label them. People can be (and obviously usually are) aware of the fact that they have been classified, and this changes their behaviour. For this reason, the classification itself almost always has an effect on the individuals that have been classified, thereby changing the classification itself. Hacking called this phenomenon "The Looping Effect of Human Kinds" (the relevant book chapter by him can be found here and it is fantastically entertaining and beautifully written so I strongly recommend it).
Universal Psychological Processes?
The Looping Effect of Human Kinds idea provides a useful way of thinking about the variation of psychological categories across cultures. Many people flag up the way that mental health diagnoses manifest themselves differently in different cultural contexts. This is taken as evidence that there are no categories out there in the world. Hacking confronts us with a different possibility. There may yet be mental illness constructs "out there" (in fact he explicitly says he thinks there are) but the nature of Human Kinds is such that this knowledge will have different effects in different cultural/historical contexts. This must be what gives rise to at least a portion of cultural variation in mental health problems. In London, being "Schizophrenic" has different connotations than in Ethiopia, and any differences that are caused by being identified as such across the two contexts (or by only being identified as such in one culture and not in the other) will be contingent on the knowledge and cultural associations available to those operating in them. At the same time, it leaves open the possibility that entities like "Depression" or "Psychosis" are underpinned by at least some psychological processes that are common to people across cultures.
Diagnosis Changes Who You Are:
For me the most important issue raised is the social desirability of diagnosis. Hacking actually regards Human Kinds as "unavoidable", but in clinical psychology this is contested, and the examples he raises ("Juvenile Delinquent" and "Multiple Personality Disorder") serve as potent reminders that being told that you are a particular sort of thing has tangible effects on how you behave. This will be familiar stuff to anyone who regards psychiatric diagnosis as merely "labelling" of a disenfranchised group of people by a more powerful group. I have always felt ambivalent about this notion because it seems that partly what people are objecting to here is the insulting metonymic substitution of one entity (a "Schizophrenic") for another (a "Person with Schizophrenia). All medical diagnoses can be regarded as a form of labelling, so their appropriate use may to come down to a question of sensitivity. If diagnoses have any clinical value at all then stigma is an issue that needs to be addressed separately (as it has been in physical conditions).
The power of Hacking's idea lies in the way it captures that "labelling" isn't just about sensitivity or politeness, but concerns the way in which people think of themselves. If you are told you have Schizophrenia, you come to view yourself in a particular way in terms of your genetic constitution, your future prospects for life without medication, and your capacity to remain stable. This information obviously changes the way you behave and relate to others. It changes you, and in some sense it changes what is meant by "Schizophrenia".
Many (perhaps most) psychiatrists and psychologists accept that diagnoses reflected a useful pragmatic way of classifying people according to the problems they have experienced and the sorts of help that are likely to be useful. However, the Looping Effect suggests that their are downsides to this, and that these may outweigh the benefits. If the focusing illusion suggests that diagnosis limits the possibilities that clinicians will offer to service users, the Looping Effect suggests that diagnosis in some way limits the range of behaviours people have open to them. The ramifications of this are that psychologists don't need to win the argument about whether diagnoses exist before they argue that they shouldn't be used. The question of whether diagnoses are valid constructs is a complex and multifaceted debate that takes place on shifting philosophical sands. The question of whether they do more harm than good might be rather simpler.