Murdering to dissect. For all the advances of science, we are no closer to understanding the essential mystery of the self. But perhaps the strange world of autism offers clues. By Edward Skidelsky

The Cradle of Thought: Exploring the Origins of Thinking

Peter Hobson <em>Macmillan, 296pp, £20 </

Of all the sciences, psychology promises most but delivers least. Its aspiration is the oldest and most basic of all; it is the aspiration, as the Delphic Oracle put it, to "know thyself". Yet how much progress has modern psychology made towards this goal? Yes, we know more about the processing of spatial information in the hippocampus than we did 50 years ago, and much else besides. But in knowing all this, do we really know any more about ourselves? Do we not still turn for self-knowledge to the old authorities, to novelists, philosophers and theologians? Or if we turn to psychologists, it is to those whose status within the discipline is, for that very reason, suspect. It seems that psychology, in becoming an exact science, has lost sight of its central, defining goal. It can no longer fulfil the Oracle's mandate.

Peter Hobson is well aware of these paradoxes. He has unusually broad experience for an academic psychologist; he is a practising psychotherapist and is well read in philosophy and poetry. His aim is to restore psychology to the status of a humane discipline; he wants to study the whole person and not just disjointed faculties. He quotes Wordsworth's famous lines about our "meddling intellect" that "misshapes the beauteous forms of things" and "murders to dissect". All this is very much in the old tradition of German romantic psychology, which arose in opposition to the associationist psychology of Locke, Hobbes and Hume. It insisted that our experiences cannot be broken down into collections of sense data, but always constitute, even at the most primitive level, a total "Gestalt", or form. This applies in particular to our experiences of other people. We do not have to infer that other people have thoughts and feelings; we just see that they do. A baby's first experiences are not of meaningless sounds and patches of colour, but of soothing voices and friendly faces. The human world is not built up out of non-human elements; it is there right from the beginning.

But attractive as it is, this theory has always found it hard to win scientific credibility. Science, as Max Planck put it, tries to measure all things measurable and to render all unmeasurable things measurable. But how can one render a happy face measurable? There seems to be no straightforward correlation between changes on the physical level and changes on the psychic level. A slight alteration of the eyebrows, and the happy face has become a sarcastic face. Or has it? It all depends on your interpretation. Many psychologists have concluded that it is best to ignore emotions and other inner states altogether, and to concentrate purely on measurable behaviour. This conclusion is methodologically unimpeachable, yet it deprives psychology of its distinctive subject matter - the psyche. Everything that is human about the human world seems to slip through the net of scientific method. Hobson is aware of the dilemma, although he is far from resolving it.

Hobson pursues what it is to be human by investigating a group of people who are in many ways distinctly unhuman - autistics. Autistics are not always retarded; they may be exceptionally intelligent. Their specific problem is understanding other human beings. It is only with great difficulty that they learn to recognise human beings as human beings, distinct from mere things. "I really didn't know there were people until I was seven years old," said a young autistic adult. "I then suddenly realised there were people. But not like you do. I still have to remind myself that there are people." Autistics, in other words, have to "work out" that there are other people. This is what Descartes and many other philosophers following him thought that we all do. But the very strangeness of autistics demonstrates that this is not how most of us relate to others. It illuminates, by contrast, how our knowledge of other minds is direct, not inferential. Hobson quotes Wittgenstein: "My attitude towards him is an attitude towards a soul. I am not of the opinion that he has a soul." Autistics, you might say, are of the opinion that other people have souls.

To demonstrate this, Hobson devised an experiment that could almost have been inspired by a remark of Wittgenstein. Wittgenstein noticed that it is very hard to detect the expression of a face drawn upside down, even if it is accurate in all its physical details. Expression, in other words, is not simply a summation of physical details; it is a total "Gestalt". Hobson asked two groups of children, one autistic and one non-autistic, to sort upside-down faces according to their expression. The autistic group performed much better. This is because, Hobson concludes, "the 'emotions' were no longer recognisable as emotions when the faces were presented upside down. Effectively, the task was reduced to one of pattern or feature recognition." The exercise having thus been rendered meaningless from an emotional point of view, the autistics had the upper hand. The experiment is a beautiful demonstration of how philosophy and experimental psychology can corroborate one another.

However, Hobson's ambitions go beyond the rather easy demonstration that autistics have difficulty understanding other people. He wants to show that this difficulty hampers their understanding of symbolism. Autistic children, he notes, do not engage in symbolic play like normal children. They will not spontaneously pick up a matchbox and pretend it is a car; they instead spend their time in meaningless, mechanical rituals, such as spinning a wheel. Hobson offers an interesting explanation for this. To use symbolism is to treat one thing as another thing. Our ability to do this is based on our capacity to step outside our minds, to see things from another person's point of view. From about the age of 12 months, normal infants can "shift perspectives" in this way. They can identify their mother's attitude and incorporate it into their own attitude. This is precisely what children with autism cannot do. Hobson argues that this basic human capacity for empathy prepares the ground for language and all other specialised forms of symbolism. "One can use symbols only if one has the kind of emotional life that connects one with the world and others." Emotion is, as the title of the book suggests, "the cradle of thought".

Hobson's target is the prevailing view that language is gov- erned by a specific "module" or "programme" in the brain. First suggested by Noam Chomsky, this theory has had a powerful influence. It is associated with a picture of the mind as divided into discrete faculties, each controlling a separate aspect of thought or behaviour. These various faculties can then, in theory, be modelled using computers. All this is anathema to Hobson's romantic holism; it is, in Wordsworth's phrase, murdering to dissect. Hobson's underlying ambition is to demonstrate that language is not governed by a specific module in the brain, but grows out of a general symbolic competence rooted in our ability to form relationships with people. It is the product of nurture, not nature.

All this has a nice, New Age feel to it. One wants it to be true. But it is contradicted by one simple fact that Hobson himself mentions in passing without, astonishingly, noticing that it refutes his theory. Many autistic children actually do learn to speak. Many others don't, but that doesn't matter. Were Hobson's theory true, it would be impossible for any autistic child to speak, because no autistic child has the kind of emotional life that is, in Hobson's view, a precondition for the use of symbolism. That autism does not preclude speech is in fact a powerful point in favour of the very theory Hobson is trying to refute. If language can remain relatively unaffected by a more general impairment in emotional and symbolic competence, this strongly suggests that it is governed by a discrete module. Some autistic children can even, as is well known, perform remarkable feats of computation. All this suggests that emotional and intellectual ability form a less cosy unity than Hobson wishes us to believe.

Edward Skidelsky is a lead reviewer for the NS