views from the borders of mental illness...,

guardian | When the psychologist Peter Chadwick explained that he was trying to research psychosis he was given short shrift by one of his patients. "You're trying to climb rain, Peter, or sweep sun off the pavement." The desire to build a science of disabling mental states can sometimes seem like wishful thinking, especially to those who have experienced the turmoil of an unquiet mind.

It is therefore no accident that critics of psychiatry have always had a particular dislike for the use of diagnosis. There are those on the outer fringes who still argue that classifying anything as a "mental illness" is fundamentally flawed, but most of the debate centres on the possibility of distinguishing different forms of psychological disability. One of the key issues is whether different diagnoses such as schizophrenia, bipolar or depression represent distinct disorders that have specific causes or whether these are just convenient and perhaps improvised ways of dividing up human distress for the purposes of treatment.

This is a hot and newly contentious topic. The fifth edition of the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders (DSM-5), the book that lists psychiatric diagnoses, is due out in May. The latest revision has emerged after a decade of unfriendly debates over what should be included and where the boundaries should lie.

The most medical approach sees each diagnosis as a separate disease with specific causes. For example, the National Institute of Mental Health, a US government research agency, describes schizophrenia as "a chronic, severe, and disabling brain disorder", something akin to a distinct condition linked to specific genetic risks and brain changes. But diagnoses are not usually derived from scientific discovery but are based on descriptions of experiences and behaviours, which are then tested for their coherence in scientific studies. For those who see mental illness as something best understood at the level of the brain and genetics, the discovery of specific biological differences associated with a particular diagnosis is considered to be good evidence for its validity.

An alternative approach is to see the definition of schizophrenia as a makeshift way of classifying mental distress that clinicians happen to agree on. From this point of view, rather than schizophrenia being a scientific discovery, it's a tradition – varying in its usefulness depending on your point of view. This difference of opinion turns out to be remarkably politicised: the medical model traditionally favours diagnosis, medication and biomedical science, while the social model is linked to the championing of individual experience, psychotherapy and social interventions.

But a growing body of evidence suggests that this divide is both unhelpful and misleading because some of the best evidence that diagnoses do not represent distinct disorders comes not from social criticism but from medical genetics. Observers may note that this is a deliciously uncomfortable situation for both parties. The hardline biological psychiatrists have had diagnoses undermined by exactly the techniques they use to support them and the social constructionists may have to accept that the best evidence for their "humane" conclusions are biological studies which they reject as supposedly "alienating".