Going Sane
by Adam Phillips
Hamish Hamilton £14.99, pp245

I once heard an Australian rabbi say that he wasn't looking for answers that made the questions go away. Adam Phillips's books share something of that spirit, a preference for enlarging the field of inquiry rather than fencing it in with solutions. The subject of his new book is sanity, which he sees as an impoverished notion compared to its traditional opposite of madness, badly in need of reformulating for a new time.

Phillips is a child psychologist and, in his capacity as general editor of new translations for Penguin, in some sense a curator of Freud's legacy. It's fascinating to see how something which was once a scientific project, claiming to unearth objective truths, has modulated into something closer to philosophy, with different ways of establishing its authority.

In the first section, he looks at the history of the word in literature. 'Madness' occurs often enough in Shakespeare, 'sanity' just once, when Polonius reluctantly admires the suggestiveness of some of Hamlet's replies ('A happiness that often madness hits on, which reason and sanity could not so prosperously be delivered of').

If this section is a little disappointing, it's not because Phillips isn't up to the job, just that many other writers could tackle it, while on home ground he is incomparable. There's a certain amount of overstatement here, too, not just in the description of Polonius as, 'for a moment at least', the first anti-psychiatrist, but in the claim that mad people 'are the stuff of iconography, the very people other than deities for whom forms of representation seem to have been invented'.

Phillips's version of authority is a blend of rigour and charm which has its own instability. Anyone writing about the mind is drawn to paradoxical formulations, since they echo the complexity and compression of mental processes, but in bulk they are exhausting to read. Sometimes, a love of epigrammatic conjuring can seem simply smug. For example: 'As if we aren't what we should be, but what we think we should be has all too little to do with who we actually are'.

Soon enough, he hits his stride. In the first two sections of part two, 'A Mad Start' and 'Sane Sex', respectively about infancy and adolescence, he is on top form, his insights informed by professional experience but extraordinarily free and lucid. If there is a better description of the extreme negotiations which go by the name of normal development, I haven't read it. These bracing essays should be enough alone to make whole shelvesful of parenting guides self-destruct.

Phillips's reading is very wide, but perhaps it isn't insanely presumptuous to suggest that he take a look at Bruce Bagemihl's book, Biological Exuberance, after which it's very hard to see animal sexuality, as he does, as a simple opposite of our own. Science has repressed or shied away from its own findings in this area for a century, clinging irrationally to the model of Noah's Ark, betraying its principles without Galileo's excuse of persecution.

In the fourth chapter of part two, 'Money Mad', Phillips has the courage to address the issue of greed: 'It is not merely that it is mad to be rich; but that it may also be a form of madness to have consented to the language of money.' The results of such a bold incursion are inevitably mixed. When he writes: 'To want money over and above the amount one actually needs to live is an essential part of modern people's passion for ignorance about themselves', he doesn't sound like someone with special perceptions or ideals. He sounds like someone with a good pension plan. Man may be the animal that knows it will die, but he is also the one who knows he must retire (or fall out of fashion if freelance). An appetite for money can represent not only desire but fear, fear of the future above all.

In part three, 'Sane Now', Phillips consciously changes gear with an almost incantatory coda. 'Sanity involves learning to enjoy conflict ... dissatisfaction is [for the sane] an inspiration rather than a refuge ... for the sane, the need to be recognised, like the need to be understood, is unnecessary; they are in no need of rescue ... the sane adult [does not] subscribe to the view that relationships are the kind of thing that one can be good or bad at, that one can succeed or fail at, any more than you can be good or bad at having red hair, or succeed and fail at being lucky.'

There's a conscious emotional openness here and a willingness to court embarrassment, which is touching, despite some odd moments. As an example (or 'emblem') of a sane person's wanting 'new forms of pleasure in which neither one's kindness nor one's excitement are overly compromised', he cites those gay men who experiment in coming without getting an erection. Which gay men are those, exactly? I seem to have lost their address, though I seem to remember a passage in Edmund White's States of Desire about such a sect, convinced by radical feminism that every erection was inherently an oppressive act.

But perhaps Phillips only throws this in, like a professor telling a dirty joke near the end of a lecture, to see who is paying attention. He must know, though, that in the third section of part two, 'Available Madness', he risks giving serious offence. To build up a picture of what sanity presupposes, he considers in turn three exemplary modern deficits - autism, schizophrenia and depression. He quotes child psychotherapist Frances Tustin's 'accurate account of what autistic children are like' and her theory that it represents 'a massive formation of avoidance reactions in order to deal with a traumatic awareness of bodily separateness from their mother'. The problem is that experts in autism are like physicists talking about what goes on in black holes. They can only describe 'what autistic children are like' from the standpoint of a world with which they don't engage.

As Phillips says, as if aware of some impropriety somewhere: 'It is always callous to use anyone as an example of anything', but he doesn't get the measure of it. It's shocking to read the phrase 'the insane child' in this context, even if autism does meet his definitions.

In these passages, the convention adopted in the rest of the book of alternating the genders of pronouns of unspecified persons is allowed to lapse. It's a fact that autism is pre-eminently a male condition, but there's something almost sinister about this tacit adjustment. To present autism as an irreversible choice arising from trauma, to ask: 'What could someone be hoping for when they exhibit this symptom, and why has their desire had to take this form?', comes uncomfortably close to the bad old days when mental health professionals had no great interest in multiplying the field of possibility and mothers were blamed as a matter of course for their sons' failures of development. Medicine and psychology have always competed for turf, but it doesn't seem helpful to claim autism so starkly as a mental act, using an argument that can no more be disproved than proved?