It would be nice to know what to believe. In many areas of opinion, though psychology is a particularly good example, it is easy for an idea to be attractive, sometimes almost irresistibly so, without there being much reason for thinking that it is true. If epistemology existed – a science of evidence – it would sort these things out for us, and tell us when the evidence really backs up our inclinations. But, as seems with hindsight inevitable, epistemologies are as hard to evaluate as anything else. And so instead of asking, ‘Should I believe it?’ one often asks apparently easier questions: ‘What kind of reasons are these?’ ‘How does the evidence for this theory compare with the evidence for this other theory?’ Or simply: ‘Is this science?’
In psychiatry it is always hard to see how much to count as science, and how much one should want to. And this seems inescapable, for one doesn’t want therapy to be guided by guesswork or dogma, while wanting to acknowledge the particular humanness of the situations it confronts. Grünbaum and Barham are concerned with the limits of science from completely opposed points of view and yet drive at peculiarly similar conclusions. Grünbaum considers the evidence for psychoanalysis. He is concerned mostly with psychoanalysis as presented in Freud’s writings, and only with the less extravagant theories, ignoring the ‘metapsychology’. He concludes that the evidence Freud cites cannot support his claims, and he hints at a more interesting conclusion: that there are features of Freudian theory which make it inevitable that scientific evidence for it will be hard to find.
The heart of Grünbaum’s case is in the second and third chapters, in which he goes carefully through an argument, the ‘tally argument’, and its variations, which is crucial for Freud’s claim to have solid evidence of a causal link between repression and neurosis, and shows how it fails. Arguments like the tally argument are only worth such detailed examination, however, if it really matters to psychoanalysis whether or not they work. They would be peripheral to psychoanalysis if any of a number of what Grünbaum calls ‘scientiphobic reconstructions’ of Freud were right. According to these, what Freud really accomplished was to establish a way of thinking which has its own logic, its own way of giving reasons for conclusions, quite distinct from that of natural science. To dispose of this kind of objection, Grünbaum devotes a careful, fierce and often tiresome 94-page preface to attacking writers such as Habermas, Ricoeur and George Klein, who read Freud as having begun a rigorous but not scientific approach to the mind. ‘Scientific’ has to bear a fairly complicated burden here. The emphasis in some of these writers is on the absence of anything like a relation of objective causality in their reconstructions of Freud, and in others it is on the difference between the ways claims are established in psychoanalysis and in science. Certainly in Habermas we find both. Grünbaum establishes pretty conclusively that no such reconstruction can do justice to Freud’s intentions: he wanted to establish a science, both in terms of the causal force of its predictions and diagnoses and in the character of its evidence. Freud’s picture of how theory and evidence work in science is definitely not a restricted positivistic one, as is evident from a well-known passage from ‘Instincts and their Vicissitudes’. But it is a picture which is to apply to his enterprise and to that of the physical sciences; there are to be no unbridgeable divisions between them. And it is definite enough for the standard criteria of evidence to apply: he would be bothered by what Grünbaum says.
What is not so clear is whether, even without Freud’s blessing, these reinterpretations of his heritage work. Grünbaum succeeds in showing that Habermas and Ricoeur have rather naive views of science and also succeeds, as far as I am concerned, in showing how hard it is to extract from their writings an account of what the rules of an alternative ‘hermeneutic’, non-scientific psychoanalysis would be – in particular, what it would count as a ground for belief. This is not quite the same as showing that no such project can work, that the only hope for psychoanalysis is to give causal hypotheses supported by evidence of a kind that would be adequate in natural science. But it does shift the burden of proof somewhat.
For Grünbaum such questions are peripheral to his central point, the analysis of the ‘tally argument’. It goes like this. Psychoanalytic patients sometimes escape their neurotic symptoms after a course of analysis, and analysts explain their recovery in terms of various claims about the connections between such symptoms and various kinds of repressed material. One might expect some direct evidence for the truth of these claims. And to know that it was the analysis that produced the cure one would expect to see detailed statistics about the rates of recovery both of neurotics of various kinds who had had no treatment and of neurotics who had undergone other forms of treatment. But direct evidence of either kind is not actually required, Freud sometimes argues, because of the principle that recovery from neurosis can only occur when the patient has accepted interpretations of his or her motives which ‘tally with reality’ – that is, with the truth about the patient’s childhood experiences or fantasies. Why does he think this principle is true? There is a little direct evidence, of an anecdotal kind, of correlations between what turns up in analysis and what can be established to have occurred in childhood. But some of the anecdotes go the wrong way, and in any case Freud does not seem to have tried very hard for this sort of confirmation. Instead, he has a variety of complicated arguments for the principle derived from his case-studies. It is hard to see how such considerations could possibly do what is required, and it is not surprising that Grünbaum concludes that the assumption behind them is, not very thickly disguised, the very principle itself.
Much the same argumentative pattern, according to Grünbaum, is found in Freud’s defence of the theory of dreams, the analysis of slips of the tongue – and in most of the not too abstract bits of psychoanalytic theory. In the end a large part of the argument is circular: what is left unsupported is the assertion of a causal connection between unconscious memory or desire and present symptom, and what creates an illusion of evidential support is a claim, itself unsupported, that insight brings relief. Now one reason why this is interesting is that Freud’s central aim always seemed to be scientific curiosity: his medical duty to particular patients was real enough, but its main interest to him was as a source of data about the human organism. It would be ironical, then, if his claims of therapeutic effectiveness were the weak point in his argument. Grünbaum suggests, moreover, that it is of the nature of psychoanalysis that this will happen, that, as he puts it, therapeutic assets are likely to be epistemic liabilities; for – I am here supplying reasons from remarks scattered through the book – whatever therapeutic force psychoanalysis possesses is surely related to the influence of the analyst over the patient and the emotional bond between them. This is evident from the outside, and is made intrinsic to the psychoanalytic theory of therapy in the concept of the transference. But this influence and this closeness are the very things that make it hard to use clinical evidence to support the theory. The interaction between analyst and patient is so complex that one cannot come to any safe conclusions about the ways in which the course of the treatment is related to changes in the patient’s personality or behaviour. There are too many possibilities.
Could there be non-clinical evidence of psychoanalysis? Grünbaum gives a number of nice, almost playful examples of how there could be unproblematic evidence for psychoanalytic hypotheses. A decline in the stigma attached to homosexuality could be correlated with a decline in the incidence of paranoia; the course of a satisfactory analysis could be correlated with a decline in the number of the patient’s dreams (measured by rapid eye movements, say). The trouble is that none of these things seem to have been observed. There is another difficulty, though, and it is as much a difficulty with Grünbaum as it is a difficulty with Freud. Any such evidence will confirm one particular prediction of psychoanalysis rather than the central bits of the theory. And no doubt other predictions will be disconfirmed and yet others left uncertain. How much support does this give to the central claims from which these predictions derive? Grünbaum cannot say, because although his book is about the evidence for a particular theory, he avoids stating any general theory of evidence.
This is the main disappointment of the book, taken as philosophy rather than as a criticism of Freud. His discussion is in various ways indebted to that of Clark Glymour, but he explicitly fails to endorse Glymour’s ‘bootstrap’ theory of evidence. And his attitude to Popper is equally obscure. He disagrees very fundamentally with Popper, in that he thinks, against everything Popper has written on the subject, that psychoanalysis has the form of a scientific theory, though unfortunately the evidence for it is lacking. But he leaves it completely unclear how much of the Popperian machinery he accepts while differing on the particular application. It is rather frustrating: he tells you that the evidence is bad without telling you enough about what would count as better.
This is an important gap in his case. For if, nowadays, your picture of science is centred on the physical sciences, you are unlikely to take psychoanalysis very seriously. As I see it, this is just a social fact. So if you do take it seriously, if you are in a position to be troubled by Grünbaum’s conclusions, then your assumptions about evidence will very likely be somewhat looser. But since Grünbaum has not stated his assumptions about evidence you have the option of blaming the trouble on those unstated assumptions rather than on psychoanalysis itself. That would be a mistake, I suspect, but it would be attractive. The effect would be to push psychoanalysis further away from science, to give even less emphasis to the causal connections and even more to the illuminating stories and the impression of insight – what one might call narrative understanding.
Barham certainly does not underestimate narrative understanding. It is at the heart of the picture of schizophrenia which he presents, influenced in part by Alasdair Mac-Intyre and Richard Rorty, who themselves are transmitting a message from the hermeneutic writers Grünbaum attacks in his long introduction. His aim is to insist, without denying inherited dispositions, biochemical imbalances, and occasionally inescapable destinies, that individual schizophrenics and their histories are as different from one another, and in the same ways, as anyone else’s, that the differences between schizophrenics and non-schizophrenics are a part of the natural variability of the human race, and that fundamentally the same strategies of understanding will work with them. He presents this by means of theorising and storytelling in roughly equal parts. The storytelling works best. He tells a story of schizophrenia, from its emergence out of the situation of the mad in the 19th century (a gap here – his historical data are English, while the original thinkers among the medical people he discusses are Continental), and he tells the stories of a number of individual contemporary schizophrenics, largely in their own words. The stories are meant to show how a change in the ways of society can leave some people unable to function in it, and to give a picture of what characterises these people. The picture is extremely vague; I think it comes down to the idea that schizophrenics do not operate with standard narrative conventions. He cannot say what is different about the way they tell and understand stories (including the deep and influential stories we tell of where our lives are going), or why this should be so. He can only let an impression of it come out in the examples, particularly those of a man called Joseph whose speech seems to consist entirely of brilliantly impossible metaphorical twistings of whatever is said to him.
This style of argument would not do as scientific evidence. But that is not what Barham is trying to provide. ‘The kind of progress which matters most,’ he writes, ‘is that which is available to the social sciences conceived of as a form of moral science, as distinct from the kind of progress that is available to the natural sciences.’ In this he has a peculiar agreement with Grünbaum. Grünbaum suspects that human relations (between analyst and patient) make it impossible to gather scientific evidence; and Barham is sure that what scientific evidence leads to doesn’t help human relations.
I will end by stating my own disagreement with both of them on just this point. Neither book, despite its insights and the force of its argument, convinces me that science and narrative understanding are inevitable enemies. For neither can show that there cannot be a powerful scientific social psychology, a theory throwing light on sociality while tracing our causal influences on one another. If we understood social psychology, then we would understand why, or at any rate describe what happens when, Barham’s schizophrenics cannot adjust to the narrative conventions of industrial life, and we would understand the possible interactions between analyst and patient, and be able to draw conclusions from their results. The right social psychology would not face us with a choice between thinking of people as agents and thinking of them as objects. Perhaps that is a sign that no such theory is possible. But I am not yet convinced.