My baby could not look more like a subject in a laboratory experiment. Wearing a soft white skullcap attached by long wires to an EEG machine measuring his brain activity, he is also surrounded by computer equipment and fussing researchers at University College London. "Hopefully you'll be contributing to high-powered science!" one coos at him.

Before I'm written off as a bad mother, I should explain: this is the London Babylab, part of University College London's cognitive development research group, which studies how infants perceive the world around them. The tests aren't uncomfortable, and are supposed to be fun. They're also a rare chance for me to peer inside my baby's mind.

Scientists have him, a healthy 15-week-old, look at shapes and cartoon characters while they track his gaze and brain responses. Cradled in my lap, he watches the screen, and thinks.

I have spent hours wondering what he's thinking. The problem is that getting inside the head of a baby is like deciphering the thoughts of a kitten.

"Lots of people don't like working with babies because it's super difficult. With adults, you can just ask them questions. With animals, you can make them do things. Not with babies," says UCL researcher Zita Patai.

But with creative, highly targeted experiments (the key, as any parent knows, is to turn everything into a game) scientists are starting to understand the baby brain. At the same time, this growing body of research is adding weight to a popular theory that our little bundles of joy are far more intelligent than we have assumed.

The immediate motivation behind this particular research at UCL is to compare the brains of healthy babies with those who may have suffered a lack of oxygen as newborns because of illness.

"We want to see if we can spot any early behaviours that suggest a baby will have problems in later life," says Patai. "The idea is that you show two types of stimuli, in this case different sounds and images on a screen - one that appears frequently, the other infrequently - and then look at how they respond. There should be a significant difference between the two responses, even in young babies," she adds.

The expectation is that brain-damaged babies may not be as sophisticated in their reactions.

Such studies are also changing the way we think about child development. Laura Schulz, an associate professor of cognitive sciences at the Massachusetts Institute of Technology, describes it as an "infant revolution" in science.

She credits it partly to the changing status of women in research, which has helped mothers and children be taken more seriously as subjects.

The popularity of the field has also snowballed as surprising results roll in.

"Babies know much more about the world than we previously believed. They have a lot of prior knowledge, right from birth," says Schulz. "They're very sophisticated learners."

Adds Michelle De Haan, a developmental cognitive neuroscientist at UCL: "Until recently, babies weren't thought to be active choosers of information. We now know that's not true."

Babies don't randomly engage with the world around them. They have preferences, betrayed by how long they stare at one thing over another. Looking-time studies, which track what holds an infant's gaze, have allowed scientists to get a stronger handle on what babies really know.

It was once believed, for instance, that young babies had a limited understanding of physical properties such as gravity and the fact that objects are solid. But looking-time studies have shown that they stare longer at a toy car that seems to be moving through a solid wall than at actions that don't betray the laws of physics, implying that they find it odd when these universal rules are broken. This isn't limited to human babies.

"There's evidence now that even newborn chicks can do that, and lots of other animals," says Schulz.

Another discovery is that babies appear to understand rational action. In 1997, Hungarian researchers showed babies who were less than a year old an animation of a circle jumping over a rectangular block. When the block was taken away but the circle still performed a jump to get behind where the block had been, babies were more surprised than when the circle moved in a straight line to reach the other side. It was evidence, the researchers concluded, that babies expected the circle to behave with some degree of common sense.

More recently, a paper published this year by psychologists at the University of California, Berkeley, claimed that children as young as 6 months may be able to reason using probabilities. When they were shown a box filled with coloured balls, almost all of them pink and the rest yellow, babies watched longer when someone began picking more yellow balls out of the box than pink ones. The experiment suggested that the babies knew to expect more pink balls and were surprised when that wasn't what happened.

Schulz thinks that much of the scientific enthusiasm around babies stems from the fact that they have so much to teach us about intelligence.

"You can't get computers to reason about others' intentions, for example. If you want to understand how human intelligence and learning works, there's exactly one organism that solves all those problems: babies," she says. "Much like if you want to really understand flying, you should look at birds."

Cutting-edge technology is pushing the boundaries of baby research even further. "Early infant work relied on looking-time studies but our technology has advanced so we can complement them with brain imaging and eye tracking," says Victoria Southgate, a cognitive neuroscientist at the Birkbeck Babylab in London.

Using neuro-imaging, for example, researchers from Birkbeck and Padua have found that day-old babies can tell the difference between social interactions and non-social actions, such as an arm throwing a ball.

Southgate's research focuses on the motor cortex - a region in the brain responsible for planning and carrying out movement - which has been found to activate when adults look at people doing physical things such as reaching for an object.

Her results point to the possibility that a baby's does, too, even though they aren't physically capable of doing most of these things themselves.

This all promises to answer the question of just how developed we humans are when we are born. Do newborns come pre-programmed, or are they blank slates? "We know that the adult brain is divided into separate functions, with different regions for different things. But there is still a controversy about how this happens," says De Haan. Scientists remain split between those who think that our brains develop this over time, through experience, and those who think our brains arrive into the world this way.

The fact that babies show at least some of the same differences between brain areas as adults, as Southgate's work so far seems to prove, is lending more credence to the idea that localised brain activity is something we are born with.

But in a field so new, there are minefields. Tempting though it is to want to "see" what babies' brains look like on the inside, brain imaging doesn't necessarily offer many more insights. Knowing what our grey matter looks like doesn't answer the important question of how the brain solves tasks.

More fundamentally, though, there is the possibility that baby researchers in general are reading too much into their results.

"It's a contentious issue," says Gert Westermann, a professor of psychology at Lancaster University, who studies children up to 18 months old. "Researchers are ascribing ever more spectacular qualities to infants. Everybody likes to hear that infants can do great things. But when you run these kinds of studies, there are different ways to interpret them."

For instance, he says, babies may stare at an object for a long time not because they have a deep understanding of physics or maths but because it just happens to be more interesting than what they've seen before.

Westermann doesn't agree with the idea that babies are born with a sophisticated, innate knowledge of the world. A simpler explanation, he says, is that newborns learn at such lightning speed that even at the point at which they're studied, they've acquired a good understanding of the things around them.

Baby studies, however, continue to feed the parent-pleasing possibility that we are all spawning baby Einsteins. While it is true that babies are smarter than it was thought 50 years ago, Southgate, like Westermann, cautions against taking this too far.

One study by psychologists at the University of St Andrews in 2005, for example, compared how young children and chimpanzees copied people's actions. It found that children imitate more than they need to, even when it should be clear to them that there's no point in it.

"Chimps copy only what is useful," says Southgate. "Make of that what you will."

Baby science: a growth industry

In the early 20th century, the Swiss psychologist Jean Piaget was among the first to write about the baby mind, forming his ideas by observing his own children. "There's a lot of Piaget's influence in the field to this day," says Laura Schulz, a cognitive scientist at the Massachusetts Institute of Technology.

Piaget thought that babies were born with hardly any understanding of the world, learning everything through their senses, by looking, sucking and feeling. He claimed that children were as old as eight months before they realised that a hidden object hadn't vanished out of existence.

In the 1960s, baby science came into its own. American psychologist Robert Fantz found that babies stare longer at something new or unexpected, such as faces or patterns. The idea of "preferential looking" turned out to be an experimental goldmine. Researchers used it to prove that a baby could recognise his or her mother's face at birth.

In the 1980s, the Harvard University psychologist Elizabeth Spelke deployed this approach to suggest that Piaget was even farther off the mark than people thought.

Babies, she says, are born with a core understanding of the physical and social world. She has claimed that newborns do understand that objects still exist when they can't see them any more. She has also run experiments to demonstrate that toddlers have a natural sense of geometry.

Today, there are baby labs throughout the world, including at the University of Oxford and University College London.

- Observer