The imminent publication of DSM-5 (due to appear in May) is a call to action for those concerned about the role of diagnosis in psychiatry. A number of campaigns are afoot, including the Committee to Boycott the DSM-5 (see https://staging-madinamerica.kinsta.cloud/2013/02/dsm-5-boycott-launched/) established by Jack Carney, although some are concerned that the committee advocates the use of the ICD in parts of the world where remuneration depends on the use of diagnostic codings. There is also an International DSM-5 Response Committee coordinated by a group of clinical psychologists, and which is in the early stages of coordinating a public campaign to coincide with the launch of DSM-5.
Although its position on the ICD is not yet clear, it’s hard to imagine that the criticisms raised about DSM-5 won’t also apply to the ICD. One of the principal concerns is that both lack any sort of scientific basis. The changes to the DSM also raise the deeply worrying prospect of widespread medicalisation of human variation and yet more neocolonialism by pharma. In this article I want to go back to basics, and ask what is the problem with scientific research in psychiatry? I will do this by examining psychiatric diagnosis in two ways. First, I will examine the failure of research to establish empirical evidence to support the scientific validity of diagnostic categories like ‘schizophrenia’; second, to consider through philosophical arguments why this failure has arisen.
Validity and the Natural Sciences
The DSM claims to be a scientific system of classification. The validity of any system of scientific classification is the extent to which it can be shown to reflect the real world. A good example is Mendeleev’s classification of the chemical elements in the Periodic Table. He noticed that if he arranged the elements according to their atomic numbers, a pattern, or periodicity, was observed so that every eighth member of the series shared roughly similar physical properties. For example, the metal lithium is a light, highly reactive element with the atomic number 3. It shares these properties with the metals sodium (atomic number 11) and potassium (19). Ordering all the known elements at the time into groups based on atomic number, he noticed that there were gaps. His predictions about the properties of these elements were subsequently found to be highly accurate when the missing elements were discovered later on. Furthermore, we now understand why Mendeleev’s system of classification was so precise. It reflected fundamental natural laws concerning atomic structure and the physical properties of matter. Bill Fulford and colleagues (2006) point out that such a rigorous definition of validity is beyond the ability of medical science, which is forced to rely on watered-down concepts of validity, such as construct validity and predictive validity. In empirical terms, both are highly problematic as far as psychiatric diagnosis is concerned.
This concerns the extent to which a diagnosis predicts a specific course and outcome for a particular diagnosis. Included in this is a specific response to treatment or other therapeutic interventions. This is a matter of vital importance in psychiatry, where the diagnosis of schizophrenia, if you accept Kraepelin’s view of the condition, is associated with poor outcome and ‘deterioration’ in social and psychological function. The late Robert Kendell (formerly Professor of Psychiatry in Edinburgh University, and an international authority on the problems of classification and diagnosis in psychiatry) drew attention to the importance of predictive validity in psychiatry nearly forty years ago. Diagnostic concepts stand or fall by their clinical value in terms of their accuracy in predicting prognosis and therapeutic response, not because of what validity might tell us about the relationships between the symptoms of a disease and the disease process (Kendell, 1975). This is yet another example of how validity has been watered down in psychiatry compared with other branches of medicine. Is there any evidence that the diagnosis of ‘schizophrenia’ predicts poor outcome?
In his original study, Kraepelin (1913) reported that only 13% of his patients suffering from dementia praecox recovered. The results of four long-term outcome studies of over 1,200 people diagnosed with schizophrenia followed up for between 21 to 32 years challenge this, with recovery rates between 49% to 68% (Bleuler, 1978; Huber, 1979; Ciompi, 1980; Harding, 1987). In broad terms 50% or more of people with the diagnosis make a significant recovery. Other work confirms that people who are given a diagnosis of schizophrenia have a wide range of outcomes, throwing doubt on the view that the condition has a poor prognosis. The work of Strauss and Carpenter (1974a, 1974b, 1977) in the USA shows that social factors such as work status and social contacts, not illness or diagnostic features, are the best predictors of outcome, along with family environment (Leff and colleagues, 1983).
Cultural factors are important too. Richard Warner (1985) uses results from the international pilot study of schizophrenia to show that the outcome of the condition in non-Western cultures is much better than in the Western world. Recent work confirms this. Kua et al (2003) found that two thirds of their patients in Singapore had a good or fair outcome at twenty years. In Madras, Thara et al (2004) found that only five out of their sixty-one subjects followed up over twenty years had been continuously ill. More than three quarters were in employment. Taken overall the results of these studies indicate that the predictive validity of the diagnosis of schizophrenia is very poor. There is simply no evidence that the diagnosis of ‘schizophrenia’ predicts uniformly poor outcome.
Setting aside the limitations of predictive validity, the crux of claims for the scientific basis of psychiatric diagnosis is construct validity. There are different definitions of construct validity, but in broad terms it concerns the extent to which a diagnosis is related to an underlying theory about the cause of a disease and (what follows is key) the extent to which there is empirical evidence to support this relationship. Construct validity is at the heart of scientific medicine, because the validity of a diagnosis stands or falls by the extent to which the symptoms of the diagnosis can be tied to an underlying causal pathological mechanism. In psychiatry this has proved impossible to achieve, and we will see that there are philosophical reasons for this. Since 1970 there have been four major articles in the psychiatric literature dealing with the issue of construct validity; Robins and Guze (1970), Kendler (1980), Andreasen (1995) and Kendell and Jablensky (2003). These papers are high on aspiration but low on realisation. Robins and Guze’s (1970) highly influential paper on the validity of psychiatric diagnoses refers to laboratory studies, including chemical, physiological, radiological and anatomical findings, as well as psychological tests, and family studies of the inheritance of psychiatric disorders. They assert that since psychiatric illnesses like schizophrenia run in families, this indicates that they have a biological basis. Despite this, fifty years of research has revealed no evidence of a genetic basis for the condition (see for example Jay Joseph’s excellent blog at https://staging-madinamerica.kinsta.cloud/2013/02/five-decades-of-gene-finding-failures-in-psychiatry/ ) Forty years on from Robins and Guze, what empirical evidence is there that disordered brain function is causally related a psychiatric diagnosis like ‘schizophrenia’?
Early studies failed to establish the construct validity of the diagnosis of schizophrenia (see Kendler’s 1980 overview). Bentall and Jackson (1988) point out that the problem with earlier work was that it failed to take into account the effects of years of toxic and dangerous physical treatment and institutionalisation that patients with the diagnosis experienced. There is growing evidence that physical treatments such as neuroleptic drugs and ECT bring about physical changes in brain structure and function that lead to chronicity. Furthermore, many of the studies were correlational in nature and it is not possible to establish a causal link between the physical ‘abnormality’ under investigation and the diagnosis. Much the same criticism applies to the studies of structural brain changes in people with schizophrenia.
Andreasen’s (1995) editorial in the American Journal of Psychiatry conceded that the long hoped-for laboratory tests anticipated by Robins and Guze (1970) had not materialised; ‘…we still lack definitive diagnostic tests equivalent to the measurement of blood sugar for diabetes or the ECG for myocardial infarction’ (Andreasen, 1995:161). Nancy Andreasen was writing at the mid-point of the ‘decade of the brain’, and was immersed in a range of neuroscientific studies using the latest brain imaging technologies to map the ‘broken brain’. Her editorial refers to ‘unstable genes’, the maturation of the frontal cortex in autism, and evidence from brain potential studies of an auditory sensory memory deficit in schizophrenia, all of which pointed to what she called a ‘second programme’ for the validation of psychiatric diagnoses. This involved careful clinical descriptions (neo-Kraepelinism) tied to external validators including studies based in molecular genetics, neurochemistry, neuroanatomy, neurophysiology and neuro-imaging. These technologies are, she argued, capable of ‘…prob[ing] beneath…surface features and seek to identify actual neural or genetic mechanisms’ (162).
Twenty years since Andreasen’s panegyric where does the validation of the diagnosis of schizophrenia stand? Anckarsäter (2010) used the Robins and Guze criteria for validity to assess meta-analyses and review papers for neurobiological markers and treatment effects in major psychiatric disorders. Apart from conditions like Huntington’s Chorea, which has an established basis in molecular genetics (and which is arguably a neurological condition) no laboratory marker has been found to support the construct validity of any diagnostic category in psychiatry, neither is there any evidence to support the view that these categories have specific outcomes and responses to treatment (predictive validity). He summarises the situation as follows:
Despite the obvious lack of empirical support for today’s diagnostic models, it is not without a sense of heresy one has to conclude that most, if not all, of the mental disorders known today, i.e. the categories that have structured both the psychiatric praxis and the research into their prevalences, patterns of distributions, “comorbidities”, and aetiologies, simply do not exist as such.
Although the literature abounds with studies claiming to find differences in the brains of those with a diagnosis of schizophrenia and those not so diagnosed, the results of replication studies either fail to confirm initial findings, or are inconclusive. Even the most recent NICE guidelines on the treatment of schizophrenia (the UK clinical practice guidelines) acknowledge the lack of evidence for a biological basis for schizophrenia:
The possible causes of schizophrenia are not well understood. Research has attempted to determine the causal role of biological, psychological and social factors. The evidence does not point to any single cause. Increasingly, it is thought that schizophrenia and related psychoses result instead from a complex interaction of multiple factors.
(National Collaborating Centre for Mental Health, 2010:22)
Biological research has failed to reveal a point of discontinuity between people with the diagnosis of schizophrenia, and those without. Psychiatric diagnoses like schizophrenia simply do not carve nature at the joint. Kendell and Jablensky (2003) acknowledge that since Robins and Guze’s (1970) paper, the validation of the diagnosis of schizophrenia remains unresolved, either in terms of its symptom profile, or its genetic (and thus biological) basis. They note that an ‘… air of disenchantment…’ is apparent ‘…in the light of the failures of the revolutionary new nosology [classification] provided by DSM-III and its successors to lead to major insights into the aetiology of any of the main syndromes.’ (Kendell & Jablensky, 2003:7) They conclude that psychiatry is two hundred years behind other branches of medicine because it can only define most of its conditions in terms of syndromes (i.e. groups of symptoms that tend to occur together).
…most contemporary psychiatric disorders, even those such as schizophrenia that have a pedigree stretching back to the nineteenth century, cannot yet be described as valid disease categories.
(Kendell and Jablensky, 2003:10)
This is because unlike somatic medicine, psychiatric diagnoses are almost exclusively descriptive and based in symptoms rather than aetiology. A negative interpretation of this is that psychiatry is ‘scientifically primitive’. Until the seventeenth century diagnosis in somatic medicine was almost entirely descriptive, consisting of syndromes as Kendell (1975) pointed out. In psychiatry twenty-first century technology is being wasted in attempts to solve a nineteenth century problem based on seventeenth century assumptions about the brain-mind relationship. Despite this, those involved in neuroscience research in psychiatry still maintain, as did Robins and Guze over forty years ago, and Andreasen twenty years ago, that it is only a matter of time before science will lift the veil that hides the biological basis of madness. The key question is whether any form of technology will ever be capable of establishing the validity of any system of classification in psychiatry.
The problem of body-mind dualism
The idea that there could be such a thing as a casual biological fault or mechanism underlying a psychiatric diagnosis is the keystone of scientific validity in psychiatric diagnoses, but the idea that this fault could be revealed by technology is based in a flawed and reductionist set of assumptions about the relationship between brain and mind. Let us take as an example the latest brain imaging technologies. Until quite recently it was not possible to study the part of the body most closely tied to mental states, the brain. Scientific studies of psychiatric diagnoses relied on indirect observations of brain function, through skull X-rays or examination of the cerebrospinal fluid that bathes the brain and spinal cord.
This changed with the introduction of new imaging technologies, first computerised tomographic (CT) scans then nuclear magnetic resonance (NMR) scans, making it possible to examine the brain in much greater detail. It is even possible through regional cerebral blood flow (rCBF) and fMRI (functional magnetic resonance) imaging to examine brain activity more or less in real time. For example, studies have investigated if there are differences in the patterns of brain activity when people hear voices. A recent study by van Lutterveld and colleagues (2013) appears to show that activity in several areas of the brain is associated specifically with the experience of hearing voices, and this activity is not related to non-specific brain activity associated with the methodology specific to the research study (i.e. stimulus detection and motor response).
This raises a fundamental problem. How can we know that there is a link between a specific pattern of brain activity and a mental state like hearing the voice of James T. Kirk? This in essence is the problem of body-mind dualism. Such research assumes that the observed brain events cause the mental states. How can we be certain that that is the case?
On the face of it this is a very attractive proposition. The new generation of brain imaging technologies, what my colleague the Dutch psychiatrist Dirk Corstens calls the ‘pinball’ view of the brain, assumes that brain events cause mental events. This is a neat and apposite metaphor worth thinking through. When we play a pinball machine we know that the purpose of the game is to manipulate the ball to trigger the switches and relays that increase the score. We do this through physical actions that include using the paddles and tilting the table. Coincidentally, some of these switches and relays operate lights in different parts of the display and trigger off electronic sound effects. The lights are incidental to the purpose of the game. The only purpose they serve is to enhance our enjoyment. We do not need to know how these processes occur. We do not need to know anything about the machine’s wiring diagram if we are to play the game successfully.
But we do not hold the belief that these incidental properties, the lights, cause my physical manipulation of the paddles as I play the game. In truth most people would consider me mad to hold such a belief. The ultimate cause of all these physical aspect of the game, paddle, table tilting, lights and sounds, is my intention to play the game and beat the other player. Note that this brings us back to the issue of intentional causality that I raised in my last blog about narrative psychiatry.
The tacit assumption in all brain imaging studies is that brain activity causes the experience of hearing voices. This is equivalent to saying that the flashing lights in the pinball machine cause the game. Brain imaging studies, indeed all scientific investigations of the experiences of madness assume that there is only one form of causality, and this operates only in one direction. Brain events are a priori, and mental events are secondary, in other words only non-intentional causality matters.
This is an example of the epiphenomalist position made famous by Thomas Huxley in 1874 who said that mind (in relation to brain) is of no more significance than the whistle of a steam engine. It disregards the possibility that intentional mental states can lead to similar or identical patterns of brain activity. This is what McGuire and colleagues (1995) found in a rCBF study that examined patterns of brain activation in three tasks; a control situation in which subjects were asked to read silently from a list of words; an ‘inner speech’ task in which they had to recite mentally a sentence ending in a given word; and an auditory verbal imagery task, similar to the second task, but in which they were told to imagine the sentences being read by an unknown neutral voice. They found that mental state events, in this case intentional mental state events representing inner speech, produce brain events. This makes it difficult to sustain the simplistic notion of causal determinism, that brain events only lead to mental events.
There is an absurdity at the heart of the assumption that brain events cause the experience of hearing voices that needs further explication. Let us say, just for the sake of argument, that when specific areas of my brain light up, or are activated, I hear the voice of James T. Kirk telling me that I am the next Klingon Emperor. These events may be seen to take place in my brain, but the voice I hear is a content of consciousness in my mind. This raises the question of how brain and mind interact. How is it possible for physical processes in my brain to give raise to unique and specific contents of my consciousness? This question arises as one consequence of the distinction between body and mind introduced by the French philosopher Renee Descartes in the seventeenth century. Why he found it necessary to make this distinction isn’t relevant here (if you want, you can find more details in Bracken and Thomas, 2005), but Descartes located his solution to the problem of brain-mind interaction in the pineal gland. He believed this was the site in the brain where it all came together, sensation, imagination, memory, bodily movement and action.
The problems this raises are set out by the philosopher Daniel Dennett (1991) who likens the role of the pineal gland in Descartes’ theory to a theatre. In this theatre “I”, the subject, sit, watch and listen as the brain activity that represents my experience of the voice of James T. Kirk is presented before me. This, however, still fails to address the question, how do I become aware of and experience this specific brain activity as the voice of James T. Kirk, and not a different hallucinatory voice saying something else. The ‘theatre’ metaphor also fails to account for how, in the same subject, activity in the same areas of the brain give rise to voices with quite different content. One way around this is to postulate that my mind must have an inner ‘centrum’ where this brain activity is presented yet again before another observer.
This is plainly ridiculous because we run into an endless regress, an infinite array of Babushka dolls nested inside each other, spiralling away to infinity. Of course at one level brain events bear a relationship to the contents of consciousness, but this arguably is at a level beyond neuroscience, and in any case it is not of necessity a causal relationship. Neuroimaging studies fail to account for two things, the precise way in which brain and mind interact, and the specificity of individual experience. Free will and autonomy (not that these are unproblematic) are swept aside in the pinball game that clinical neuroscience has become.
The latest generation of brain imaging studies deal with this problem by denial. They fail to acknowledge the complexity of the problematic relationship between brain and experience, preferring instead naive materialism and reductionism. I have not encountered a single study that has engaged in a serious discussion of this problem (if anyone can direct me to one that does so convincingly, I’d be very grateful). They fail to engage with the specificity problem by limiting any concern with the content of voices to those aspects that are only necessary to make a diagnosis of schizophrenia.
It is possible to extend this argument more generally to any scientific attempt to set out a causal basis for the experiences of madness through diagnostic categories, and thus the validity of these categories. This is why fifty years of study and investigation, and huge sums of money spent across the Western world on neuroscientific research institutes, on careers and equipment, has failed to establish the validity of a single psychiatric diagnosis. How much more money must be wasted on the doomed project of neuroscience in the service of psychiatric diagnosis?
Anckarsäter, H. (2010) Beyond categorical diagnostics in psychiatry: Scientific and medicolegal implications. International Journal of Law and Psychiatry, 33, 59–65.
Andreasen, N. (1995) The Validation of Psychiatric Diagnosis: New Models and Approaches. American Journal of Psychiatry, 152, 161 – 162.
Bentall, R. & Jackson, H. (1988) Abandoning the concept of ‘schizophrenia’: Some implications of validity arguments for psychological research into psychotic phenomena. British Journal of Clinical Psychology. 27, 303 – 324.
Bracken, P. & Thomas, P. (2005) Postpsychiatry: Mental Health in a Postmodern World. Oxford, Oxford University Press (see especially Chapter Four, pp. 105 – 138)
Bleuler, M. (1978). The schizophrenic disorders: Long-term patient and family studies. New Haven, Yale University Press.
Ciompi, L (1980) The natural history of schizophrenia in the long term. British Journal of Psychiatry, 136, 413-420.
Dennett, D. (1991) Consciousness Explained. London, Allen Lane.
Harding, C. M., G. W. Brooks, et al. (1987). The Vermont longitudinal study of persons with severe mental illness: I. Methodology, study sample, and overall status 32 years later. American Journal of Psychiatry 144(6): 718-726.
Huber, G., G. Gross, et al. (1975). A long-term follow-up study of schizophrenia: Psychiatric course of illness and prognosis. Acta Psychiatrica Scandinavica 52: 49-57.
Kendell, R. (1975) The Role of Diagnosis in Psychiatry. Blackwell Scientific Publications, Oxford.
Kendell, R. & Jablensky, A. (2003) Distinguishing Between the Validity and Utility of Psychiatric Diagnoses. American Journal of Psychiatry; 160:4–12
Kendler, K. (1980) The Nosological Validity of Paranoia (Simple Delusional Disorder) Archives of General Psychiatry, 37, 699 – 706.
Kraepelin, E. (1913) Psychiatrie, ein Lehrbuch fur Studierende und Artzt (Psychiatry, a textbook for students and practitioners) (8th edition), Vol. 3. Leipzig, Barth.
Kua, J., K. E. Wong, et al. (2003). A 20-year follow-up study on schizophrenia in Singapore. Acta Psychiatrica Scandinavica 108(2): 118-125.
Leff, J., Kuipers, L., Berkowitz, R., Vaughn, C. & Strugeon, C. (1983) Life events, relatives’ expressed emotion and maintenance neuroleptics in schizophrenic relapse. Psychological Medicine, 13, 799 – 806.
McGuire, P., Silbersweig, D.A., Wright, I., Murray, R.M., David, A.S., Frackowiak, R.S.J. & Frith, C.D. (1995) Abnormal monitoring of inner speech: a physiological basis for auditory hallucinations. Lancet, 346, 596 – 600.
National Collaborating Centre for Mental Health (2010) Schizophrenia: The NICE Guidelines on Core Interventions in the Treatment and Management of Schizophrenia in Adults in Primary and Secondary Care. Updated Edition. London, The British Psychological Society and The Royal College of Psychiatrists. Accessed on 17th April 2010 at www.nice.org.uk/nicemedia/pdf/CG82FullGuideline.pdf
Robins, E. & Guze, S. (1970) Establishment of Diagnostic Validity in Psychiatric Illness: Its Application to Schizophrenia. American Journal of Psychiatry, 126, 983 – 987.
Strauss, J. & Carpenter, W. (1974a) The prediction of outcome in schizophrenia: II. Relationships between predictor and outcome variables. Archives of General Psychiatry, 31, 37 – 42.
Strauss, J. & Carpenter, W. (1974b) Characteristic symptoms and outcome in schizophrenia. Archives of General Psychiatry, 30, 429 – 434.
Strauss, J. & Carpenter, W. (1977) Prediciton of outcome in schizophrenia. III. Five year outcome and its predictors. Archives of General Psychiatry, 34, 159 – 163.
Thara, R. (2004). Twenty-Year Course of Schizophrenia: The Madras Longitudinal Study. Canadian Journal of Psychiatry 49(8): 564-569.
van Lutterveld, R., et al., (2013) The influence of stimulus detection on activation patterns during auditory hallucinations, Schizophrenia Research, http://dx.doi.org/10.1016/j.schres.2013.01.004
Warner, R. (1985) Recovery from Schizophenia: Psychiatry and Political Economy. London, Routledge and Kegan Paul.