CHILDREN / The development of mind-reading: Fibbing means a child knows others have minds - and a crucial part of the brain has switched on, says Karen Gold

ANY ADULT who plays with small children expects to slurp an imaginary cup of tea, bandage teddy's invisible wounds or speak into a banana. Sharing children's make-believe can be infuriating or charming, but it never seems very important. Yet psychologists now think that the moment when a two-year-old first embarks on shared pretence is a moment of crucial mental change: the first step from a child's understanding of other people to an adult's.

Imagine a child who pours a pretend cup of tea from a toy teapot, and then - with a conspiratorial smile and grand gesture - tips the non-existent contents of the cup into your lap. A toddler cannot do this. Adults who try to engage under-twos in shared pretence are met with a blank stare. But the two-year-old, having spilt the invisible tea, will expect you to pretend to be wet and to mop it up. He or she is becoming aware, though without yet realising it, that you have a mind.

This is the first indication of what philosophers call 'theory of mind', the conscious knowledge of other people's thoughts that develops between the ages of three and four. It is second in importance only to language in the creation of human society. Almost everything we do and say is affected by what we think other people think, believe, intend or desire. Yet until recently, no one thought to ask whether human beings have a theory of mind from birth. If not, then at what age do we acquire it? And are there people in whom it never develops?

This issue was raised when animal researchers discovered that chimpanzees have a primitive theory of mind. Primate-watchers have seen baboons deceive their assailants by staring suddenly at the horizon, as if trouble were about to come over it, before making a nifty getaway. A chimpanzee has been caught second-guessing a rival by faking retreat from a feeding station, only to return just in time to snatch the bananas away.

Both episodes imply one animal calculates what is in another's mind. Almost identical scenarios could have happened in a playgroup. Yet when the chimp evidence was collected in 1978, most psychologists believed that children had no theory of mind until at least the age of six. They began to devise ways of testing young children's beliefs about other people's minds, without asking complicated 'What did you think he thought . . . ?' questions.

They came up with something called a 'test of false belief'. If a child talks about what someone else is thinking, that does not prove he or she has a theory of mind. The child may simply believe that everybody else's mind is the same as his or her own. But if children understand that the contents of someone else's mind are different from their own, then they have a theory of mind. It works like this:

The child watches as a doll called Sally puts a marble in a drawer. Sally then goes out of the dolls' room. In comes naughty doll Anne; she removes Sally's marble from the drawer and puts it on a shelf. Sally does not see this. Anne leaves and Sally returns. The question is, where does Sally look for her marble? If you are an adult, it is obvious. Sally left her marble in the drawer; she doesn't know Anne moved it, so she looks where she saw it last - in the drawer.

What researchers discovered was that most four-year-olds understand this. They tell you Sally looks in the drawer; they have a theory of mind. But three-year-olds are different. Again and again they will insist that Sally looks on the shelf. They know the marble is on the shelf; so they think Sally must know it too.

As psychologists repeated the experiment throughout the 1980s with the same results, it became clear that they had discovered a turning point in children's development. Four- year-olds have an overt theory of mind; three- year-olds do not. Suddenly all kinds of other differences across this age gap made sense.

For example, any parent knows the impossibility of persuading a three-year-old that he or she is mistaken about anything. In 1988 American researchers found out why. They handed each of a group of small children a Smarties tube containing not Smarties but a pencil. Before they gave each child the tube, they asked what the child thought was inside it. Understandably, each one said 'Smarties'.

Each child then took the tube, opened it, and found the pencil. When the four-year-olds were asked what they had imagined was in the tube before they opened it, they said 'Smarties'. But the three-year-olds all said 'A pencil'. To make sure the younger children were not just hiding embarrassment at being wrong, researchers played the trick using dolls, then asked the three-year-olds what the dolls had thought. The result was the same. Three-year-olds can't grasp their own false belief any better than someone else's. They have no theory of mind.

Theory of mind is not obviously practised or learnt gradually, like talking. Only in the last two years have psychologists realised that pretend play at the age of two foreshadows theory of mind. But when theory of mind becomes explicit, the change is sudden and striking. Three-year- olds rarely tell lies, though they may say what they wish was true, because they can't conceive of putting an idea into someone else's mind. In various experiments in which children would gain chocolate if they tricked a robber puppet, three-year-olds were incapable of the required deception even though they got crosser and crosser as the robber took all the chocolate.

Four-year-olds not only deceive, but anticipate others' deception. They see the point of competitive games, understanding their opponent's desire to win as well as their own. As other people's motives become clear, they become both more wary and more socially competent. They appreciate jokes and stories based on one character's knowledge and another's ignorance.

Four-year-olds doing none of these things would seem unusual. As researchers began to understand more in the 1980s about when and how children's theory of mind develops, they began to ask themselves what a child with no theory of mind would be like. Such children would find people's words and actions incomprehensible. They would be social misfits, probably withdrawn and unresponsive. They would not play pretending games. The picture tallied precisely with autism.

Psychologists working in London for the Medical Research Council tried out the Sally- Anne doll test on a group of autistic children. To check that the results did not simply reflect intelligence, they also gave the test to Down's syndrome children whose IQs were lower than the autistics. The outcome was stark. The children with Down's syndrome passed. Like four-year-olds, they could comprehend Sally's false belief. The autistic children failed. Though aged between seven and their mid-teens, they performed like far younger children. Their theory of mind was missing.

Researchers are now beginning to describe autistic children as 'mind-blind'. The yes-or-no nature of children's mind-reading suggests there is a specific theory of mind 'switch' in the brain, programmed like language development to flick on at a particular age. In autistic children, the theory of mind part of the brain appears never to switch on.

But could it be encouraged? Dr Josef Perner of Sussex University has discovered that children with two close siblings, whether older or younger, succeed in Sally-Anne tests almost a year earlier than only children. Playing and arguing together appear to be the main factors, he says. 'It's not the siblings themselves that have the effect, but the opportunities for parental interference - the number of times when there is conflict and the parent is drawn into explaining what went wrong, or why someone did what they did.'

If parents and nursery teachers know that, they should be able to promote theory of mind even among only children. Certainly, once children have explicit theories of mind they are already encouraged to train them: to think about other people's points of view; to comprehend irony; to write stories from different perspectives; to read Romeo and Juliet.

Whether autistic children could be helped by training would depend on whether their brains house potential for their theory of mind to develop. Last year, for the first time, psychologists at St Andrews University taught autistic children to pass false belief tests. It may lead nowhere, says St Andrews reader in psychology Dr Andrew Whiten, but it adds a piece to the jigsaw: 'We know now that at a certain stage the child sees into the power of attributing mental states. By adulthood, mind-reading is so natural that we never think about what it's like to be mind-blind. If we understand that, then even if we can't change autistics we can at least understand them better.'