Watch young children at play for just a few minutes and you're reminded that they live in a far more wondrous, whimsical world than the rest of us. A pile of wooden blocks is a vast city, and some sticks the inhabitants. Indeed, often about the time babies begin to walk and talk, they also begin to pretend--giving a stuffed animal a sip from their cup or covering up a doll for sleep.

Is it all just flights of fancy? Or does it have implications for child development? That's what some researchers are trying to find out.

There is some indication that having a good imagination translates into more creativity as an adult. But even more intriguing--and the idea getting the most attention in labs these days--is a possible connection between pretend play and the ability to get along socially in the world. In particular, researchers are investigating whether pretend play facilitates the development of children's theory of mind--the ability to understand that others have thoughts and feeling all their own.

"It's been a big focus of research recently and some of the work is really fascinating," says developmental psychologist Alison Gopnik, PhD, of the University of California at Berkeley. "The only downside is, no one has an answer yet."

From studies of imaginary friends to those examining when and how well children distinguish reality from fantasy, researchers are developing theories about the role of imagination and pretend play in child development. For now, the questions are mainly academic, but some day the answers could lead to a better grasp of how imaginative play influences how well children get on in the world.

Pretending as actions or thoughts

The earliest theories about the connection between theory of mind and pretend play went something like this: Even though children don't develop a sophisticated theory of mind until somewhere between their third and fifth birthdays, it would seem that they need to understand something about others' thoughts if they're to grasp anything about pretending. For example, in order to play along when Mom pretends a banana is a phone, a child must have some idea that she is projecting the thought of a phone onto a banana.

From this general theory, researchers postulated that pretend play might be a tool that helps children realize that thoughts, not reality, guide people's actions, utterances and emotions.

The first task in supporting this theory is to show that young children do, in fact, understand that pretending involves thinking before they have a truly developed sense that others have thoughts. The results have been mixed.

Early on, University of Virginia psychologist Angeline Lillard, PhD, challenged the idea that children understand the mind's role in pretending. Based on several well-publicized studies, she concluded that before children develop theory of mind, they understand pretending more in terms of actions than in terms of mental states.

In an initial set of experiments she showed children a troll doll and told them that he had never seen a kangaroo and doesn't know they hop. She then showed them the doll hopping around and asked them whether he was pretending to be a kangaroo. Most of the 4-year-olds in the study said yes, that he was pretending to be a kangaroo despite the fact that he couldn't possibly have a mental representation of a kangaroo.

Based on that and several subsequent studies, Lillard postulates that young children think about pretending more in terms of actions than in terms of mental states. So, because the doll was actively hopping, he must be pretending. It doesn't matter what's in his head, because pretending does not crucially depend on mental involvement--its essence is in the action.

Challenging Lillard have been several studies, including one by University of Texas, Austin, psychologist Jacqueline Woolley, PhD, finding that children as young as age 3 can use information about someone's thoughts--as represented in a thought bubble with pictures to represent characters' thoughts--to infer what the person is pretending.

So, why does Woolley come to the opposite conclusion of Lillard? Because they use thought bubbles as an artificial cue to make the relationship obvious, says Lillard. In real life, there are no thought bubbles flying overhead.

Woolley and University of Oregon psychologist Marjorie Taylor, PhD--in what has become a friendly if sometimes spirited debate--argue equally forcefully that Lillard's experiment is flawed. "She tells them the doll is hopping around like a kangaroo--what else are they going to say?" says Taylor.

Woolley adds that, for sure, because pretending is often active, children understand it in terms of actions. In fact, in the real world, they probably often rely on the action component of pretending because it's most salient to them. But that doesn't mean the mental component is unavailable, she says.

"I'm convinced that they understand the mental aspect [of pretending] when we make it equally salient [by using thought bubbles]," says Woolley. "It's true that, in the real world, people's actions are usually what you have to go on, so that's what you use. But we're looking at capabilities, not what they normally do. These are equally important questions."

At the same time, adds Taylor, "Angel [Lillard] has made us back away from a really strong interpretation of explicit understanding of mental states," says Taylor. "But to talk about children not understanding anything about the mind's role in pretending is an overinterpretation of her findings."

In his new book, "The Work of the Imagination" (Blackwell, 2001), Harvard developmental psychologist Paul Harris, PhD, sums up the debate this way: "There are some indications that 3-year-olds are able to identify the thoughts that typically accompany an act of pretense....At the same time, it is not clear that their understanding of the role of thinking in pretense is sufficiently systematic and robust to help them understand the wider role of mental representation, especially in the context of belief-based action."

The role of role-play

Harris proposes a somewhat different way to think about the connection between pretend play and theory of mind that doesn't rely on whether children understand that pretending involves the mind. He posits that it's not all types of pretend play that facilitate the understanding of others' mental states, but rather specific types of pretending that involve role-playing.

So, for example, pretending to be a rocket ship would not facilitate theory of mind, but pretending to be an astronaut flying a rocket ship would. His argument is that both role-playing and theory of mind require children to make predictions about others' actions and thoughts.

"Children's role-play suggests that they can enact what people might do in a given situation by a process of simulation: They imagine themselves in that same situation and act vis-à-vis that imaginary situation," he writes in his book. "For example, when pretending to be Mother, the child pretends to be confronted by a crying baby, and ministers to it accordingly."

In the same way, when a child faces a real-life situation, he or she can use the same process of simulation to predict how another person may react. The difference is that when children are pretending, they act out a role and in real life they're simply predicting how another person might react.

This theory doesn't require children to understand others' mental states per se. Rather, it simply requires them to put themselves in someone else's shoes and imagine how they themselves will react.

So, while earlier theories predict that any type of pretend play will help children understand others' mental states by showing them a gap between someone's mental representation and reality, Harris argues that the benefit of pretend play is practice in the art of simulation. Therefore, he believes role-play, rather than pretend play in general, should be helpful because it enables a child to simulate what another person may be thinking.

"My idea is that pretend play doesn't teach children about the fact that there are mental states," says Harris. "But it does help them look at things from someone else's perspective. I would be happy if 3-year-olds didn't have the faintest idea that pretending involved mental states if they're busy not being themselves, and thinking about what others would do in certain situations. It's more the perspective-taking that's critical, not that they're thinking there's something in the head."

He has evidence to support his ideas. A few studies find that frequency of role-playing is a better predictor of performance on a common test of theory of mind than the overall frequency of pretending. In addition, Oregon researcher Marjorie Taylor has preliminary evidence that children with imaginary friends--an extreme form of perspective-taking--score higher on tests of theory of mind than children who score low on measures of fantasy play.

Into the lab

The main problem with these studies is that they're correlational and so unable to say anything about whether pretending causes children to develop a better sense of others' thoughts. Indeed, all of the theories that connect pretending with theory of mind suffer from the same conundrum: It is incredibly difficult to design a study that can tease apart whether early pretending facilitates theory of mind development or whether there's some other unknown factor that encourages pretend play and leads to better theory of mind.

"It's quite possible that role-play does help children develop a theory of mind," says Woolley. "But there might also be a third factor that leads some children to be good mind readers and that is also responsible for an interest in that kind of pretend play."

There's no study out there yet that can decipher this, but Harris' book has prompted many researchers into their labs to try.

Beth Azar is a writer in Portland, Ore.