Taking Bad Arguments Seriously

Ian Hacking on psychopathology and social contruction

The idea of social construction is wonderfully liberating. It reminds us, for example, that motherhood and its meanings are not the fixed and inevitable consequence of child-bearing and rearing, but the product of historical events, social forces and ideology. Mothers who know but fear standard canons of emotion and behaviour may see that the ways they are supposed to feel and act are not ordained by human nature. And if they don’t obey either the old rules of family, or whatever is the official psycho-paediatric rule of the day, they need not feel quite as guilty as they are supposed to feel.

On the other hand, the words ‘social construction’ work like cancerous cells: once seeded, they replicate out of hand. Consider Alan Sokal’s hoax. In May 1996, Sokal, a physicist at New York University, published a learned pastiche in a special issue of Social Text, a more or less literary journal. Believing that some fashionable intellectuals have abandoned rigour for glory and empty jargon, he brilliantly parodied some of their idiotic remarks about physics, often using recent quotations footnoted with all the veneer of academic rigour. Derrida was perhaps the most famous of his targets. Sokal fooled the editors of Social Text, who printed the piece thinking it was the real McCoy – wow, a physicist understands Derrida et al.

In an almost simultaneous issue of Lingua Franca, a sort of People magazine for professors and their ilk, Sokal owned up to the mischief. His confession used the phrase ‘social construction’ just twice in a five-page essay. Stanley Fish, who teaches at Duke University, which publishes Social Text, and a man widely regarded as the very dean of advanced (literary) ‘theory’, replied on the oped page of the New York Times. There, he used the phrase, or its cognates, 16 times in a few paragraphs. If a cancer cell did that to a human body, death would be immediate. Sokal has since been an immense draw on American campuses, pretty well guaranteed to attract 800 people, many of them angry (for or against him). A few days ago, one of the search engines on the Internet found 84,272 distinct items in which ‘Sokal Affair’ was a key phrase. For comparison, it found 7767 for Ludwig Wittgenstein and 11,334 for Quantum Mechanics, the science whose abuse furnished Sokal’s illustrations. I find the whole event rather embarrassing. Do I want to live in a scholarly community where so many bad arguments are taken seriously?

Many people were unhappy about using the idea of social construction in discussing the natural sciences. Sokal fanned the fire, to the point where people doubted if it even made sense for motherhood. Given the inflamed passions going the rounds, you might think that the first thing we want is a definition of social construction. On the contrary, we need first to confront the point of social constructionist talk. Its primary use is for consciousness-raising. I have an immense list of social construction booktitles, going all the way from Authorship to Zulu Identity. In almost every case the aim is not to refute an idea, but, to use Karl Mannheim’s words, to ‘unmask’ it, and often to show its ‘extra-theoretical function’, with the hope of ‘disintegrating’ the idea. Or at least to undermine its authority, showing that an idea or a present state of affairs need not have been the way it is. The project is to de-inevitabilise something that we take for granted. Nowhere has this been more successful than with feminist work on gender, and now sex. Some constructionists are as rebellious as radical feminists, but many seem to want only to understand the world. They adopt what Richard Rorty calls an ‘ironic’ stance. We thought that such and such was inevitable, in the present state of things. We now see it is the product of a highly contingent social history; but we are stuck with it, or so many constructionists seem to say.

I have never made use of the social construction approach. My topics in the past have tended to be too complex, and to demand too much nuance, as well as conceptual analysis, for the simplistic phrase ‘social construction’ to be serviceable. And I hope that will continue to be the case. I used to believe that the best way to contribute to the debate was to remain silent. To talk about social construction would be to entrench the use of a confusing idea. But philosophers of my stripe should analyse, not exclude. Even in the narrow domains called the history and the philosophy of the sciences, many historians and many philosophers won’t talk to each other, or they talk past each other, because one side is so contentedly constructionist while the other is so dismissive of construction. In larger arenas, public scientists shout at sociologists, who shout back.

There is a felt tension between the notion that something is real and the notion that it is constructed. It results partly from the ways in which the ‘real’ and the ‘constructed’ interact, and comes out most brazenly when the usual (if currently unpopular) distinction between mind and body breaks down. Take schizophrenia. Most research scientists believe that it is a biochemical or neurological or genetic disorder (or perhaps all three). A minority of critics take the illness or its manifestations to be socially constructed. It is not claimed that people diagnosed as schizophrenic are not in a terrible state, but that the way the symptoms are organised, and the way in which they develop and express themselves, are the product of psychiatry and society. I do not want here to take sides on these issues, but to elaborate on the rival attitudes involved.

Whenever someone talks about ‘construction’ it is important to be clear about just what is supposed to be constructed. For an extreme example take the book by Andrew Pickering called Constructing Quarks. Pickering does not mean that quarks, as ‘objects’, are socially constructed, but that the idea of the quark was constructed, along with the relevant theories, scientific institutions and apparatus for detecting them. Well, isn’t that obvious? Pickering also maintains, however, something that is, to say the least, controversial: not that quark theories are false, but that physics could have been successfully developed in other and non-equivalent ways. Good physics does not have to be quarky. He wants to de-inevitabilise the idea of the quark, with all its material infrastructure. He is thereby also challenging the authority of high energy physics. I do not agree with him, but I take him seriously. For one thing, he shows how important it is to say exactly what is supposed to be socially constructed. When we talk about ‘the social construction of X’, does X refer to objects (quarks) or to the idea and all that it involves?

Idea and object are often confused; I have confused them myself. In my book, Rewriting the Soul, I referred to a paper by a paediatrician entitled ‘The Social Construction of Child Abuse’. In order to forestall tedious discussion about whether child abuse is a socially constructed or real phenomenon, I wrote: ‘it is a real evil, and it was so before the concept was constructed. It was nevertheless constructed. Neither reality nor construction should be in question.’ What ‘it’ is a real evil? The object is: the actual behaviour or practice of child abuse. What ‘it’ is said to be socially constructed? The concept. My switch from object to idea conceals the most difficult problem of all. Concepts and practices interact, and such interaction is often the whole point of talk of social construction. My original project of studying child abuse – I once wrote an essay called ‘The Making and Moulding of Child Abuse’ – was largely motivated by an attempt to understand this type of interaction; but I conflated the two fundamentally different categories of object and concept.

If we were to say that child abuse was socially constructed, what would we mean? Not, trivially, that child abuse, the ‘object’ or activity, takes place in a given historical and social setting. The point, rather, would be that the idea of child abuse, our ways of conceptualising what some people do or have done to them, came into being at a given moment, and evolved thereafter. The idea was brought to the fore, in 1961, by a group of Colorado doctors, who insisted that a lot of babies were being battered: their bones were being broken, and hospitals were treating these fractures as accidents. When, subsequently, a journal was established, a professional organisation founded, and a campaign mounted to introduce legislation, the full name of the phenomenon was ‘child abuse and neglect’. By 1975, the old-fashioned practice of cruelty to children had become intertwined with that of incest, thanks to the forceful intervention of feminists: child abuse now came to mean first and foremost sexual abuse, and neglect seemed largely to drop out of popular consciousness.

I could not now improve on ‘making and moulding’ as a metaphor for what happened. It suggests plasticity, not building or assembling from parts (the picture conveyed by ‘construction’). What it doesn’t answer is the question: ‘What happened to what?’ To the idea of child abuse, or to the practice now called by that name? The answer was to both, but not in the same way. Primarily, I was referring to the idea of child abuse, including its moral connotations and ramifications. Battering babies in 1961 was wrong and cruel, and intelligible only when one reflected on the conditions of squalor, poverty and oppression in which much of it took place. After 1975, when ‘child abuse’ came increasingly to mean sexual abuse, it was entangled with the two ancient themes of incest and innocence. The old doctrine of childish innocence (‘suffer the little children to come unto me’) plus the late 19th-cenutry idea that children develop naturally at a certain pace, mixed in with incest, and with feminist hatred of patriarchy, made a potent brew. ‘Child abuse’ came to name the worst private evil of the Eighties, with genocide as the worst public evil. Despite the resultant excesses, this was one of the most valuable pieces of consciousness-raising of the century.

Child abuse reminds us of a curious fact about the present state of our civilisation. We are supposed to be overwhelmed by moral relativism; it is said that there are no longer any stable values. Try speaking out in favour of child abuse, not under the guise of man-boy love (already much spat on in most quarters), but going the whole hog. Or take literacy. Can we imagine anyone speaking out seriously in favour of illiteracy for the poor? Child abuse and illiteracy are nowadays absolutely bad.

One of the reasons social construction theses are so hard to nail down is that in the phrase, ‘the social construction of X’, the X may implicitly refer to entities of different types that interact. Classifications may interact, for example, with those classified. People who are classified as being of a kind – abusers or abused – can become aware that they are so classified, and those who apply new classifications to others then relate to those others in new ways. We make tacit or even explicit choices, adapt our ways of living so as to fit into, or else escape from, the classification applied. Then there may be a secondary effect. What was previously believed about people of a particular kind may no longer be true because people of that kind have changed by virtue of how they have come to think of themselves. For some time, I have been calling this phenomenon ‘the looping effect of human kinds’. Looping effects are everywhere. Think what the category of ‘genius’ did to those Romantics who saw themselves as geniuses, and what their behaviour did in turn to the category of genius. Think about the transformations effected by the classifying of people as fat, overweight or anorexic. If someone talks about the social construction of genius or anorexia, they are likely to be talking about the idea, about the individuals classifiable under the idea, the interaction between the idea and other people and the manifold of social practices and institutions that these interactions involve.

I coined the phrase ‘human kind’ to match the philosophers’ term, ‘natural kind’, which has been in use for well over a century. It is derived from John Stuart Mill and William Whewell, who both wrote of ‘kinds’, although the Cambridge logician John Venn may have been the first to use it in a work of philosophy. The term was intended to get rid of a lot of philosophical baggage, like the word ‘essence’; but that is never effective for very long. ‘Natural kind’ is now encumbered by a multitude of porters carrying all the old valises to the old destinations, but with new name tags.

We need some new labels. In talking about ‘human kinds’, my interest is in kinds of people, their behaviour and their experiences involving action, awareness and self-awareness. The awareness may be personal, or may be shared and developed within a group. I am concerned with classifications that can change the ways in which individuals experience themselves – and may even lead them to alter their behaviour. The name I now give to such kinds and their behaviour is ‘interactive kinds’, an ugly phrase that has the merit of incorporating the notion of agency, while also suggesting the way in which the classification and the individual so classified may interact, and the ‘looping effect’, that may result.

The notion of an ‘interactive kind’ is fuzzy bur not useless, because plenty of classifications differ fundamentally from the interactive kinds. Quarks are not aware. A few of them may be affected by what people do to them in accelerators, so our knowledge can act on them. But it acts on them without their becoming aware of what we know. What name to give to classifications of this sort? ‘Indifferent’ will do. Quarks are indifferent: they do not behave as they do because they are aware of how they are classified. Perhaps the fundamental difference between the natural and social sciences is that the natural sciences investigate indifferent kinds, while the social sciences are on the whole concerned with interactive kinds.

Why do the social sciences not prosper to the same extent as the natural ones? Because they proceed as if they were trying to find ‘indifferent kinds’, while in fact they are most often examining interactive kinds, which, to put it crudely, do not stay still under the microscope. There have been over ninety major statistical studies of the causes and consequences of child abuse (with hundreds and sometimes thousands of individual human beings involved in each study), with little collective upshot. Weak methodology, mutter the critics. No, I say, a wrong idea of what the studies are doing.

Unlike the human sciences, the natural sciences use indifferent kinds. ‘Indifferent’ need not be passive. Plutonium is singularly non-passive: it kills. We made it, yet it kills us. But plutonium is not aware of being called plutonium, or experiencing existence in plutonium institutions like reactors, bombs and storage tanks. It is indifferent.

Here a dilemma arises. There has been a constant thrust, for two hundred years, to biologise human characteristics. Nowhere has this been more resolute than with the psychopathologies. We have many reasons to expect that some of these – some types of schizophrenia, for example – will come to have a known biological pathology – biochemical, neurological, genetic or some combination of these. On the other hand, schizophrenia has been an interactive kind, so that the behaviour of schizophrenics, although to some extent cross-cultural, is also strongly interactive with conceptions of schizophrenia. How can it be both an interactive and a biological or ‘indifferent’ kind? This is not an abstract question, for it is one way in which to pose some of the fundamental conflicts between psychiatry and anti-psychiatry.

I could have chosen many other examples: mental retardation, childhood autism, even attention deficit hyperactivity disorder. There are competing theses about the aetiology of such conditions. One school tends, speaking very loosely, to the constructionist, the other to the biological. In the constructionist camp, these disorders are seen as interactive kinds, in the biological camp, as indifferent ones. This is the sharpest instance of the fundamental tension I mentioned at the start between ‘real’ and ‘constructed’ explanations.

As a philosopher, I want to create room for both the constructionist and the biological schools, without favouring one or the other, so that each is able to work without interfering too much with the positive aspects of the other school’s research. First, then, the constructionist attitude to schizophrenia. I take, for my example, Mary Boyle’s 1990 book, Schizophrenia: A Scientific Delusion? Critics like Boyle argue that the criteria for diagnosis of schizophrenia have varied enormously from decade to decade, yet psychiatry still asserts it to be a definite illness or group of illnesses. Boyle is a rebellious constructionist, who wants to unmask the ideologies that give rise to the diagnosis and disintegrate it.

The classification of schizophrenia and its implications in terms of treatment are matters of which patients, for all their periodic deficits of logic and a sense of reality, are intensely aware. Thanks to the continually developing arsenal of psychotropic drugs that can bring some semblance of ordinary life to a large and increasing proportion of those patients diagnosed with severe schizophrenia, they are also more aware than they used to be of who and what they are.

Classification profoundly affects the sensibilities of schizophrenics. One reason why the symptom profile of schizophrenia has changed over the years is that it is a moving target. Phenomena such as auditory hallucinations, for example, often lead to its diagnosis, but even the ways in which people so diagnosed describe these hallucinations have changed, as has their content. Moreover, the role of hallucinations in diagnosis is itself mobile. The founding fathers, Emil Kraepelin and Eugen Bleuler, emphasised ‘flat affect’, and held that many other mental illnesses are accompanied by hallucinations. Just before World War Two, Kurt Schneider produced a list of 12 ‘First Rank Symptoms’, with auditory hallucinations top of the list. When First Rank Symptoms ruled the diagnostic roost, a lot more people became schizophrenic than would ever have been so classified in the wards of the Burghölzli hospital during Bleuler’s reign.

There may be a remarkable looping effect here. For Bleuler, auditory hallucinations were not important; there were other aspects of the patient’s life to be taken into account and many of those he did not diagnose as schizophrenic also had hallucinations. Hallucinations having thus become unproblematic, schizophrenics took them more casually. Schneider, finding them to be universal, made them almost a sine qua non of the diagnosis. As psychiatrists became more cautious about using the diagnosis of schizophrenia, ‘flat affect’ came back, and in the most recent diagnostic manuals, hallucinations are no longer key. In some jurisdictions, there are fewer than half as many diagnosed schizophrenics today as there were twenty years ago. So what’s up? I am suggesting that part of what’s up is the interaction between the diagnoses, the illness and the rest of the world, occurring not only at the material and neurological level, but also at the level of ideas and their interaction.

Let me now turn to the biological side. There is a deep-seated conviction that schizophrenics suffer from one or more fundamental neurological or biochemical problems which will one day be identified. It is not claimed that every person now diagnosed will be found to have the same problem. Some researchers conjecture that there are at least two distinct disorders, one of which declares itself in late adolescence and is genetic, while a second type of schizophrenia may not be inherited. For simplicity’s take, suppose there is just one such pathology called Z, and that in the future we discover what Z is. It will be quite in order to report a great discovery: ‘Schizophrenia is Z.’

Pathology Z will be an indifferent kind, a neuro-geno-biochemical state which is not aware of what we have found out about it. (Alas, nothing is simple, some kind of biofeedback may occur; we have evidence that behavioural modification of depressive states changes serotonin levels.) At the same time, our new knowledge may, with luck, enable us to intervene and either prevent or ameliorate the pathology.

How can schizophrenia be both an interactive and an indifferent kind? When I wrote that child abuse is both socially constructed and real, I could avoid any contradiction by means of a trifling distinction: the idea of child abuse and its material consequences can be socially constructed, while child abuse is real. With schizophrenia it is not so easy. We need something more sophisticated than a crude distinction between ‘object’ and ‘idea’.

At this juncture, philosophers may like to think of schizophrenia and a postulated pathology Z in terms of the theories of reference advocated a quarter of a century ago by Hilary Putnam and Saul Kripke. Kripke and Putnam would call ‘schizophrenia’ a natural-kind term, if indeed this is a kind of illness that exists in nature. Putnam presents the meaning of a term as a sequence of items, much like a dictionary entry: part of speech, category, down through stereotype, but ending with an item no dictionary could ever include – the extension of the term being defined. That is, the class of things to which the term applies. This is called the reference of the term. In my example, the final entry in the definition of ‘schizophrenia’ would be the pathology Z, or perhaps all instances of the pathology Z.

We can perfectly well keep Putnam’s machinery, but suppose that into the definition of the word ‘schizophrenia’ (and a great many other words) we put an enriched stereotype: the current idea of schizophrenia, accompanied by descriptions of prototypical schizophrenics. (So-called definitions of mental disorders commonly proceed by giving such clinical examples.) In the case of schizophrenia, the stereotype should also include theories, hypotheses, therapies, attitudes, the lot.

Now for the bottom line. Someone (not me) writes a scientific paper entitled ‘The Social Construction of Schizophrenia’. The author could perfectly well maintain that there probably is a definite unknown neuropathology Z that is the cause of prototypical and most other examples of what we now call schizophrenia. But he could also argue that the idea of schizophrenia, expressed in the stereotype, is a social construct that interacts not only with therapists and psychiatrists in their treatments, but also with schizophrenics, who find the current mode of being schizophrenic compelling.

We might go further and imagine different authors, at ten-year intervals, writing papers with much the same title. What is constructed, the extended stereotype, might differ from decade to decade, even though the underlying pathology would, by hypothesis, remain constant. This is because the ways in which schizophrenics are classified and helped could be expected to change from decade to decade, causing the individuals themselves to acquire different profiles, and hence altering the prototype of what a schizophrenic is. The identification of Z would profoundly affect the schizophrenic’s self-conception. (This would be especially striking if several organically different underlying pathologies were identified, so that the patient would be schiz-1, schiz-2 or whatever, each type involving different therapies and prognoses). There would thus be ample room for looping effects and for one or more definite references for the kind-term ‘schizophrenia’.

My own position here is uncomfortable to the point of being dissociated. I do not favour the language of social construction. I also have endless philosophical difficulties with the Kripke-Putnam semantics. I am discussing both these ideas (so seldom joined) in connection with psychopathologies because many deeply committed critics of psychiatric establishments find social construction talk liberating. It enables them to launch a critique of practices about which they are deeply sceptical. I respect their concerns, and have, I hope, represented them fairly, if cautiously. On the other hand, I also respect the biological programme of research into the most troubling of psychiatric disorders. That creates a dilemma. I have suggested a semantic way for a philosopher to make peace with the dilemma. Some would even say that it is a tidy resolution of the dilemma. I am ambivalent about the use of semantics in connection with disease. Semantical theories are not literally correct descriptions of natural language, they are artificial ways of construing natural languages for this or that purpose. The old Kripke-Putnam semantics is wonderfully suitable for many purposes, and in the present case, putting a theory of reference alongside social construction talk shows how to reduce a felt dilemma. And in doing so enables us to move on to more significant issues, to what I would call the dynamics of classification.

This is where the interest is. How would the discovery of Z affect the way schizophrenics and their families conceive of themselves, or affect their behaviour? Which patients, formerly classified as schizophrenic, would now be excluded on biological grounds, and what would that do to them? The bulk of constructionist writing has examined the dynamics of this or that classification. Studies of Authorship, Brotherhood, the Child Television-Viewer and Dangerousness (to start an alphabetical list) have, in their various ways, been concerned with just that: the construction of the idea of X, of the object X, of the experience of being X, etc, and how these interact. Is there anything to say in general about such dynamics, over and above particular examples? How does the ‘making and moulding’ of an interactive kind, be it child abuse or schizophrenia, help to form people? How do people form themselves, as they act in ways that conform to, or avoid, such powerful classifications? Semantics may intrigue the logician, but the dynamics of classification is where the action is.