Are mental illnesses real? (Part Three)

(Part One, Part Two)

This is the third part in a series on the philosophy of mental illness. The series is looking at the long-standing debate about the legitimacy of mental illness. It covers some of the classic contributions to the literature. For example, in part one, we considered Thomas Szasz’s infamous arguments for the “myth” of mental illness. And in part two, we considered Robert Kendell’s attempted defence of the legitimacy of mental illness, as well as Bill Fulford’s account of the logical geography of illness concepts.

In this, the final, part we will do two further things. First, we will step back from the particular arguments for and against the legitimacy of mental illness, and focus on Neil Pickering’s meta-philosophical diagnosis of the problems inherent in the debate. Then, having sharpened our appreciation for the meta-philosophical issues, we will consider what is probably the most recent and widely-discussed attempt to define “illness” in such a way that it (properly) includes mental illnesses: Jerome Wakefield’s Harmful Dysfunction analysis.

As mentioned in previous entries, this series is based heavily on chapter one of Tim Thornton’s book Essential Philosophy of Psychiatry, though, as ever, I will feel free to critically expand upon what it says.

1. The Likeness Argument and its Flaws…
Disagreement about the legitimacy of mental illness persists despite decades of debate. This is not too surprising: philosophical disagreements have a remarkable knack for persistence. Nevertheless, every now and then, philosophers who are fed up with the endless argumentative back-and-forth associated with these disagreements, like to take a step back and offer a diagnosis. That’s exactly what Neil Pickering did with the mental illness debate in his 2006 book The Metaphor of Mental Illness.

According to Pickering, the major problem with the current debate is that it relies on something he calls the Likeness Argument. The Likeness Argument tries to determine the legitimacy of mental illness by comparing it to a paradigm case. In other words, it takes a condition or illness that everyone agrees is an illness, and tries to argue that mental illnesses are sufficiently like (or not like) that paradigm case. This is usually done by abstracting the essential properties of that paradigm case and applying them to mental illnesses. In the case of Szasz and Kendell the paradigm illness is either some specific physical illness (e.g. hypertension) or the general class of physical illnesses.

The following is a more formalised version of the Likeness Argument template:

• (1) A paradigm case of illness properly-so-called has properties P₁….P_n

• (2) Mental illnesses share a sufficient number of these properties.

• (3) Therefore, probably, mental illnesses are illnesses properly-so-called.

The Likeness Argument is, in effect, an argument from analogy and so suffers from the associated logical shortcomings. It is not formally valid: the conclusion does not really follow the premises. The similarity between the two cases gives us, at best, a probabilistic, defeasible reason for endorsing the conclusion.

Typically, disputants in the mental illness debate will argue about the premises of the likeness argument. Thus, they’ll offer different accounts of the essential properties of the paradigm case, and different accounts of the similarities between the paradigm case and the case of mental illness. Pickering’s goal is to show how hidden assumptions undermine these arguments.

In particular, his goal is to highlight two hidden assumptions that make the Likeness Argument work. They are:

Hidden Assumption 1: That there are necessary and sufficient conditions that determine category membership for things like illnesses.

Hidden Assumption 2: That a putative mental illness such as schizophrenia is describable in terms of its properties without that description presupposing which category it belongs to.

The first hidden assumption plagues many philosophical debates. Consequently, it is difficult to say whether or not it is truly problematic, without engaging in a much wider debate. Philosophical analysis of concepts and phenomena often proceeds on the assumption that there are objective conditions that determine why X is one thing and not another. To some extent, this is just good old-fashioned commonsense: surely it is true that the cup upon my table is distinct from the saucer? And surely this distinction is determined by the properties they both exemplify? The concern is that the game of philosophical analysis, whereby ever-finer distinctions between concepts and categories are drawn, ends up with arbitrary and stipulative conditions for category membership. If it does, then the claim that mental illnesses, or indeed illnesses more generally, have some objective essence that determines whether or not they belong to the category of illnesses properly-so-called might be undermined. But this is a big debate, not one that can be settled here.

The second assumption is rather more interesting, and it is the one that Pickering thinks is particularly problematic in the mental illness debate. As he sees it, people like Szasz and Kendell work from the assumption that category membership is determined from the bottom up. In other words, that first you identify the properties associated with a particular condition (e.g. hypertension or schizophrenia) and then you work out which category it belongs to (illness/not an illness). Pickering argues that the relationship between property description and category membership is more holistic than that. Oftentimes, assumptions about category membership determine how we describe a phenomenon. The process is more top-down than we may realise.

He illustrates this by using the example of addiction. Let’s say there are two categories to which this phenomenon could belong: (i) it could be a blameworthy moral defect; or (ii) it could be a blameless mental illness. Pickering’s point is that hidden assumptions about which category it belongs to will affect how we describe it. So, if we think it is a moral defect we will describe an addict’s behaviour in terms of “choice”, “autonomy”, “vice”, “weakness of the will” and so on. On the other hand, if we think it is an illness, we will describe their behaviour in terms of “chemical dependency”, “helplessness”, “addiction” and so forth. I have tried to illustrate this in the diagram below.

What Pickering is pointing out here is the theory-ladenness of observation, something which has long been recognised in the philosophy of science. The point is that data doesn’t simply present itself to us in an objective, category-neutral fashion. We need to make theoretical assumptions before we can even make sense of the data and distinguish it from the background noise.

I have no doubt that observations are theory-laden. But is this a major problem? Does it undermine any attempt to argue rationally about the status of mental illness? I’m not so sure. It’s true that we have to start from somewhere — i.e. with some set of theoretical assumptions — but that doesn’t mean we are forever wedded to those assumptions. I could start out believing that addiction was a moral defect, but gradually adjust my view — through a process of reflective equilibrium — to the belief that it is a mental illness. I’m not sure I would be irrational in so doing: my readjustment could be driven by sound reasoning and argumentation. Furthermore, as Thornton points out, we can accept the epistemic-dependence between theory and description, without thereby needing to accept their ontological-dependence. In other words, we can accept that we would not be able to know the properties of a particular condition without also knowing its overall classification; but that doesn’t mean that we must accept that our judgments about those properties are constituted by the overall classification.


2. Jerome Wakefield’s Harmful Dysfunction Analysis
With that meta-philosophical point out of the way, we can proceed to consider one final attempt to define illnesses in a way that mental illnesses are (properly) included within their scope. The attempt comes from Jerome Wakefield, and it is probably the most widely-debated and discussed analysis of illness in recent times. (Terminological note: Wakefield uses the term “disorder” instead of illness, as he thinks it is the broader term. I’m going to stick with “illness” since I have used it throughout this series. This should not lead to confusion. For the purposes of this discussion, the terms “illness”, “disease” and “disorder” can all be taken to refer to the same kind of thing: a condition that is a legitimate subject of medical scrutiny and treatment).

Wakefield thinks that the major challenge for psychiatry is to show why the so-called illnesses (or disorders) are different from other mental traits. After all, the challenge raised by the likes of Szasz is that all mental illnesses are really just socially deviant or disvalued forms of thought and behaviour, and therefore shouldn’t be subject to excessive control or treatment from the psychiatric profession. This is probably the main weapon with which the anti-psychiatrists attack the legitimacy of psychiatry. Wakefield agrees that there is a problem here. As he sees it, the legitimacy of psychiatry depends on having a definition of mental illness that distinguishes true mental illnesses (like the various forms of psychosis and depression) from other socially undesirable traits (e.g. illiteracy, aggression, infidelity, lack of skill etc.).

He thinks this can be done by first accepting that all definitions of illness include a value judgment that is linked to social norms, and then by adding to that a value-free element that distinguishes illness from other forms of norm-violation. This is exactly what the Harmful Dysfunction analysis tries to do. It says that any illness properly-so-called will include the following two elements:

Harm Element: The condition will be harmful.

Dysfunction Element: The condition will involve the divergence of a biological/mental mechanism from its natural function.

The harm element incorporates the value judgment, with harm being measured in relation to social standards (e.g. harm = setback to interests). So things like illiteracy, infidelity, aggression, addiction, psychosis, depression, anxiety and so forth will all match this criterion. What distinguishes the latter from the former, however, is the dysfunction element.

There’s quite a bit of philosophy/science underlying the dysfunction element. According to Wakefield, dysfunction is determined in relation to the selected-for function(s) of a particular biological or mental mechanism. In this manner, his account of illness is explicitly evolutionary in nature. Natural function is equivalent to selected-for function. To give an example, the heart is a biological mechanism that performs certain functions. Its selected-for function is the function that explains why it has continued to exist over evolutionary time. Wakefield suggests that the selected-for function of a heart is its ability to pump blood around the body. It was this function that increased the inclusive fitness of the organisms that had a heart. (To be clear: a single organ or mechanism can have many selected-for functions). Heart disease then is any harmful condition that prevents the heart from fulfilling its selected-for function.

Wakefield argues that the harmful dysfunction analysis can apply just as well to mental illnesses. The brain/mind is made up of a variety of mental mechanisms: perceptual, cognitive and emotional. These mechanisms perform many different functions, some of which were selected-for over evolutionary time. A mental illness is simply any harmful condition that prevents a mental mechanism from fulfilling its selected-for function(s). For example, schizophrenia is a mental illness because it is harmful and because it prevents our visual, auditory and cognitive mechanisms from performing their selected-for functions.

This gives us Wakefield’s paradigm argument for the legitimacy of mental illness:

• (4) A condition X is an illness properly-so-called if it is (a) harmful and (b) involves the divergence of a biological/mental mechanism from its natural (read: selected-for) function(s).

• (5) At least some mental illnesses fulfill both of these criteria.

• (6) Therefore, at least some mental illnesses are illnesses properly-so-called.

Wakefield’s analysis is certainly interesting, and to be fair to the guy, there’s a lot more detail to it than I have been able to cover in this discussion. Nevertheless, there are some obvious problems that are worth mentioning here.

The first is that Wakefield’s attempt to have a value-free element in his definition of disease is doubtful. Some would argue that the concept of a dysfunction, even if it is related to Darwinian natural selection, involves a value judgment of some kind. The idea is that function depends on the purpose or telos of the organism, and that the purpose or telos of an organism is a value-laden thing. I’m not too enamoured with this critique. I think evolutionary purposes might be a good deal more objective than other kinds of purpose; and I also think the whole desire to have a value-free element in the definition of illness is misguided. I don’t think the legitimacy of psychiatry depends on this: what matters is whether it is guided by the right value judgments.

A bigger problem with Wakefield’s definition is that, in many cases, it can be difficult to identify the selected-for functions of a particular mechanism. Take language as an example. Has the brain evolved, in part, to allow us to learn and speak a language? Was that one of the selected-for functions of the brain? Some people argue yes (e.g. Pinker), others argue that language acquisition is a by-product of other selected-for functions. The same is true for many other mental mechanisms. Some are viewed as clear adaptations, others as by-products, with lots of arguments about these classifications. The evidence is complex and nuanced, and yet if Wakefield’s definition were to be followed it would make a big difference whether something was classified as an adaptation or a by-product. If it’s the former, then an inability to perform that function might lead to a diagnosis of illness; it it’s the latter, it would not.

Tied to this is the fact that some people — indeed, some psychiatrists — argue that certain classic mental illnesses like depression are in fact evolutionary adaptations. If they are right (and they may well not be), then depression shouldn’t be classified as a mental illness. Of course, this isn’t necessarily a problem: it could be that depression doesn’t deserve to classified as a mental illness. Furthermore, Wakefield is clear that certain psychiatric diagnoses are insufficiently attentive to the question of natural function at the moment. This is a basis on which the diagnostic criteria can be revised. Still, if one his goals is to defend the legitimacy of mental illness, then it is possible that his definition will incentivise psychiatrists to ignore or call a halt to any attempt to explain mental illness in terms of evolutionary adaptation.

Finally, there is the concern that the definition is under-inclusive. It’s possible that functions for which we were biologically selected are harmful in modern environments (e.g. having a sweet tooth). Does this mean that their harmful manifestations should not be classified as illnesses or disorders?

In the end, I feel like the classification of illness is largely (and perhaps appropriately) driven by pragmatic, treatment-oriented factors: if it can be treated by medical means, then it is a illness; if not, it's not. Of course, I recognise that even this definition assumes that the concept of “medical treatment” is more stable than it really is.