No sign yet of that smoking gun

Hardly a week goes by without our reading of a particular psychological disorder that has been traced to a rogue brain chemical, or brain area, or beyond that to a genetic variant. These two books, in their different ways, reflect the fact that psychiatric research today sits at the intersection of neuroscience and genetics, and a very long way from the psychodynamic terrain it once occupied. Nancy Andreasen is one of the ushers of this ongoing paradigm shift. Twenty years ago, one of her books, The Broken Brain , became the general reader's herald that the scientific consensus about psychological disorders was shifting. Here she returns to survey the ground, choosing this time the new genetics and brain scanning as the tools that are making light the dark recesses of the mind.

Andreasen's book is a solid enough overview of some aspects of modern biological psychiatry. Her summaries of the techniques of contemporary human genetics, and of brain scanning and the cerebral architecture it reveals, are extremely clear and useful. By the time she comes to survey the main psychological disorders, though, she has run out of steam slightly. Thus, while these chapters constitute useful introductions to schizophrenia, affective disorders, anxiety and dementia for those who have no prior knowledge, there is little in the way of synthesis of what we now understand to be the nature of these disorders or of the puzzles that remain.

In the case of schizophrenia, for example, far from the twin searchlights of genetics and brain-scanning locating a nice clean smoking gun, they have revealed much more complexity and variability than we could possibly have imagined. Despite some welcome words about the need to avoid obstructive dichotomies (nature vs nurture, brain vs mind), Andreasen is not prepared to engage with the possibility that neurobiological research might make us question the whole conceptual framework - a set of neat, discontinuous disease categories - that her book employs. Anxiety and mood disorders, for example, are treated as separate chapters in the book without comment, while genetic and neurobiological evidence leads us increasingly to think of them as related or even different aspects of the same thing. Many of the brain abnormalities (and some of the genetic loci) associated with schizophrenia are also associated with bipolar mood disorders, a crossover difficult to accommodate within a discrete framework.

Perhaps most interestingly, contemporary neurobiological research has revealed the presence of abundant variation within (and overlap between) psychiatric populations and "normal" controls. Because of this variation and overlap, for all our powerful scanning and molecular techniques, we are still a long way from finding the smoking gun, particularly in the case of schizophrenia. The gun in this case would be some biological or genetic feature that was shared by only and all individuals who suffer from schizophrenia. This feature would be to schizophrenia what the cholera bacillus is to cholera and the HIV virus to Aids, the pathognomon.

R. Walter Heinrichs's book takes us on a journey through the scientific literature on schizophrenia in search of the pathognomon. The fascinating thing about schizophrenia is not, as is sometimes alleged, that no physical basis has been found to the illness. On the contrary, physical differences between patient and control brains have been found in terms of overall size, size of various sub-parts, gray-matter thickness, neuronal density, neuronal orientation, size of cerebral ventricles, dopamine activity levels, density of D2 and D4 dopamine receptors, serotonin activity, glutamate... The list is almost endless. The problem is knowing which of these manifold differences is significant, especially since many of them have not proved consistent.

Heinrichs's methodology is so simple as to be beautiful, and produces wonderful clarity in a confusing field. He meta-analyses the literature on each of the pathognomic contenders. This is no mean feat since 2,000 research papers are published on schizophrenia each year. He concentrates not just on the statistical significance of differences, but on effect size, that is the magnitude of difference between patients and controls. For each abnormality, he comes up with a mean-effect size from the literature - a kind of balance sheet of the research world's findings.

The results are most illuminating. First, despite a lot of variability and non-replication, the balance of evidence shows that the brains (and behaviours) of schizophrenia sufferers do differ significantly from those of the rest of the population in multiple ways. Second, there is no single trait that is much more pathognomic than any other - no smoking gun. Third, the degree of difference is in no case very great. On many measures, schizophrenic brains differ from the rest of the population by about one standard deviation. This means that they are measurably different, but there is a range of variation in both patients and controls, and there is huge overlap between those who receive the diagnosis of schizophrenia and those who do not. Thus we have a huge amount of physical information that is all relatively poor at identifying the individuals with the disorder.

These are important results to have. The physical basis of schizophrenia is an area dogged by researchers narrowly pursuing their own favoured abnormality, be it dopamine receptors, glutamate or neuronal migration, without reference to other paradigms that are out there, and without an attempt to achieve an overall synthesis. The frequency of non-replication makes it difficult to know which results to rely on, and Heinrichs's method solves that problem. The work is pursued with great clarity and thoroughness. The only omission I noted was the failure to cover differences of asymmetry and lateralisation in the schizophrenic brain.

Heinrichs struggles slightly to accommodate his findings within his chosen conceptual framework. One explanation is that the abnormalities we have discovered so far are not very illuminating because they are not the key ones, but weakly related ones or side-effects, and further research will find the real pathognomic culprit. The second explanation, which he favours, is that there is causal variability within the schizophrenia population, and so any single brain trait may characterise only some sufferers. But I would have liked to see him go further. Taken individually, the symptoms of schizophrenia are not uncommon within the "normal" population. There are grey areas at the boundaries of psychosis, and there is precious little reason to think that the behaviours labelled schizophrenic represent an abrupt discontinuity with the behaviours not so labelled. Thus it is perhaps not a huge surprise to find that there is no abrupt discontinuity in the brains either. On the other hand, and disproving the extreme constructionist arguments of the anti-psychiatry movement, the differences are real. It is just that they are about continua, rather than the all-or-nothing pathognomon that the dichotomous nature of medical labels can make us assume should exist.

In short, the new techniques of genetics, brain-scanning and the like mean that we have no shortage of information about the basis of psychological disorders. Heinrichs in particular has given us a magisterial assessment of some of that evidence. The goal now must be to synthesise it into a more satisfactory conceptual framework.

Daniel Nettle is lecturer in biological psychology, Open University.