Royal Institute Lecture

The Dancer's Body

I've been a ballet dancer for thirty-three years: over three quarters of my life. I started to dance – according to my mother – before I could walk, and took my first ballet class – like hundreds of thousands of little girls all over the country – when I was seven years old. Unlike most of those little girls, however, I stayed the course, became a professional dancer and, for twelve years, a principal with The Royal Ballet. A ballet dancer.

The average man in the street looks on ballet dancers with a mixture of incredulity, horror, fascination, squeamishness and awe. And it is true that we are the physical equivalent of a polymath, if such a thing exists. A pentathlete – at least. It's not enough to be able to jump like a high jumper. No, we have to spin like a figure skater, achieve the flexibility of a gymnast, the balance of a high wire walker, the strength of a weight lifter, the coordination of a master juggler, the endurance of a middle distance runner and the speed of a sprinter. On top of that, we have to be able to learn, master and remember impossibly complex sequences of movement, all the time displaying the dramatic skills of an Olivier. All this in a single body. A dancer's body.

The dancer's body is without doubt an extraordinary machine: a particular blend of good genetics and good fortune, adapted through years of training to meet the intense physical demands of classical ballet technique.

An extraordinary body, yes. But without the brain, it wouldn't be dancing. And it's the brain I'm going to talk about tonight: mine... and yours.

It doesn't matter whether you're dancing the charleston or the third act of Swan Lake. Moving muscles – however striking, however impressive – are just the visible consequence of an invisible ballet of firing neurons, electronic impulses and chemical reactions which is more complex, faster and, frankly, more beautiful than anything a choreographer could possibly invent.

Every human movement - from simple breathing to soft shoe shuffling - is controlled in the brain: the most complex and the least understood component of the human body - home to a million million nerve cells - or neurons: more neurons than there are stars in the milky way.

It's the almost infinite potential for connections between those neurons that allows the brain to produce the blend of precision, coordination, balance, speed, accuracy, learning, memory and spatial awareness which we call dancing.

Honing all those neural processes to the levels of refinement required in classical ballet demands an early start. The neurons in the brain increase in number for the first seven or so years of a child's life – and then commence a slow but steady decline towards old age. We can lose brain cells at the astonishing rate of between 10 and 100,000 every day.

Losing brain cells isn't necessarily bad news. Even before a baby is born, its developing brain is involved in an ongoing sculpting process known as apoptosis, or programmed cell death: strengthening and rationalising connections which are useful and pruning out those which aren't. It's possible that this process of programmed cell death, while essential to create a brain suited to the life we live, also strips us of the neural connections which allow the sort of skills we often refer to as gifts: so-called photographic memory, for instance, or the phenomenal recall of idiots savants. Apoptosis can, however, run amok, destroying too many connections. This excessive apoptosis is believed to be one of the causes of impaired intelligence in Down's Syndrome.

The brain is at its most plastic in infancy, all the time tailoring to the demands a child makes of it. In fact, it is so adaptable that if you were to remove one half of an infant's brain, the other half would rewire itself to take on the tasks of both – even those tasks which are exclusively the domain of the missing section. It becomes increasingly rigid as we grow older. Developing the highly refined motor – or movement - skills of a ballet dancer - or a concert pianist – demands that you take advantage of that early plasticity, developing and strengthening the necessary neural connections before the nerve cell count in the brain starts to go down.

My first ballet classes at age seven were at the Janice Sutton School of Dance in Skegness. Lots of skipping and hopping, and a bit of improvisation where we all pretended to be trees. But it wasn't all fun and games - in between the improvisation sessions we began the process of setting up neural connections through endless repetition of movement patterns which are still ingrained thirty years later: good toes, naughty toes.

By looking at my toes and attaching names to the position - good toes, naughty toes - I was enhancing an inherent feedback process without which we would be incapable of controlling our movements: A kind of sixth sense which we all have, but very few of us are even aware exists. Proprioception.

Proprioception is, literally, our sense of self: information about the state of our bodies sent to the brain from 'sentry points' at different sites around the body: joints, tendons, muscles, skin and eyes. It's a kind of movement monitoring process.

Our nervous system provides an information superhighway for commands to go out to the body from the brain and it's a fantastically complex process. To make any single movement, I must first of all access a stored representation of the movement and gather together its various components from the left hemisphere of the parietal cortex.

Once assembled, the movement signal passes to the basal ganglia, which act like a traffic cop, allowing or inhibiting action. The basal ganglia provide, if you like, the drive for action: damage to the basal ganglia, in Parkinson's Disease, causes patients to freeze when they try to start moving. But the basal ganglia are also responsive to the motivational areas of the brain. So if I really really really want to do an arabesque, the drive from the basal ganglia is dramatically increased.

The movement signal is then sent on to the motor cortex, where all the voluntary movement pathways converge.

The motor cortex contains a somatotopic map: a representation of the muscles stretched out along the cortex like a back to front image of our own body. The map is distorted, with huge areas representing the bits that need fine, detailed control – like the hands and the face - and smaller areas representing those bits – like the legs - that don't.

The movement signal goes to the section of the map representing the part of the body that needs to act – so if I want to lift my leg, the signal goes to the leg region of the somatotopic map.

It's the motor cortex that fires the starting pistol, sending out the action signal to the body. From the motor cortex, the signal travels along the outgoing branch of the nerve cell - the axon - and crosses to the relevant side of the body at the top of the spinal cord. The signal is pulsed along at an extraordinary rate: 5 times faster than the fastest Olympic sprinter. The hand, for instance, can be moving as little as 20 milliseconds after the motor cortex gives it the green light.

All this just to move. But that initial command only initiates movement. The control and execution of movement – the skill which is dancing – is dependent on a second set of brain circuitry coming into play, and this is the process known as proprioception.

Whenever the body carries out the motor cortex's movement commands, the sensors in the muscles and the joints send a progress report back to the cerebellum, a beautiful, flower-like structure just above the brain stem, about the size of a child's fist, and home to half the total number of neurons in the brain. Based on this progress report, the cerebellum can order new commands to be sent out from the motor cortex, making sure the body achieves exactly what the brain was intending.

This conversation between our body and our brain – proprioception – continues all the time, and yet we're completely unaware it's going on. You can see it in action, though: look closely at my wobbling ankle as I struggle to stay on balance. That's proprioception at work.

The brain is constantly bombarded with masses of proprioceptive feedback. It copes with the sheer volume of information by dealing with it hierarchically. If all of it went to the highest, conscious part of our brains, we'd be overwhelmed. So signals we've come to expect – like the sensation of our skin stretching when we walk, or the pressure of the soles of the feet on the ground – are monitored lower down in the brain stem. You're never aware of it until you alter the regular pattern - perhaps when a plaster slightly tightens the skin, or you wear unfamiliar shoes. Dealing with familiar feedback at an unconscious level leaves the conscious brain free to take on new skills – or, in a dancer, to deal with the more important business of interpretation and artistry.

Given the importance of proprioception to dancers in achieving their high level of physical control, it's surprising that we compromise our proprioceptive abilities all the time - by working in studios with mirrors. And despite the teacher constantly begging us not to get obsessed with our own reflection, we all do. Visual feedback will always dominate over proprioceptive feedback, partly because the information we get from our eyes is more precise and a lot more interesting.

It may be more interesting, but it's much slower. Visual information reaches the brain approximately 70ms after feedback from the body, and there's an added delay while the brain works out what to do with it. Because visual feedback comes via a different system – it's in a different code, if you like – it has to be analysed and translated into a movement command before it can be put into use.

I was never quite sure why my teachers went on and on about not looking in the mirror. I thought it was a question of vanity. If they'd told me it was because a dancer relying on her reflection to make corrections will always be less fluent than a dancer relying on proprioception, I might have listened...

Like anyone trying to learn a new physical skill, inexperienced dancers inevitably wobble as they try to master ballet technique. Their brains might be sending accurate motor commands to the muscles, but the message back to the brain - the progress report, if you like - is not getting through fast enough to make the appropriate corrections in time. Operating on out of date feedback is a bit like trying to play the stock market with nothing more than a week old copy of the Financial Times for guidance.

As the young dancer matures, the feedback loop speeds up and the wobbles become less obvious. Myelination, a process which continues until late adolescence, speeds up nerve transmission by surrounding the nerve branches like insulation on an electric cable. Until the process of myelination is complete, the movements of a pre-adolescent dancer will inevitably be less smoothly controlled than those of an adult.

But young dancers also improve – and here, once again, science proves what we know - through experience.

The brain is a fast learner, and it learns by its mistakes. All those early, fumbling attempts are not wasted - they're the visible effects of the brain working out the most efficient and direct way to send the appropriate message to the muscles. Once a movement has been repeated often enough, the cerebellum is able to predict, based on past experience, the likely consequences of the movement command.

The cerebellum knows, for instance, because it's happened so many times before, that if you lift your right leg, you're likely to over balance and fall to the left. So it short cuts the normal feedback route and uses a second, even faster process for executing precise and skilled movement – a fast, internal loop from motor cortex to cerebellum.

At exactly the same time as the motor cortex sends its GO signal to the muscle, it sends a copy of the command to the cerebellum – a facsimile, if you like.

Based on what it knows from experience, the cerebellum can take an executive decision and cut the muscle feedback out of the loop. It can check the copy command, predict the problems involved and send new, improved instructions to the motor cortex within 25ms.

If the cerebellum were to wait for the muscle feedback before it sent the correction, there would be a delay of another 100ms. That might not sound a lot, but it's the difference between a dancer who wobbles and one who doesn't.

This fast correcting process is totally unconscious, but it's dependent on a slower correcting mechanism which we consciously control ourselves. When the fast process fails and the movement doesn't take place as planned, we register the failure - generally because we fall over. So we try again, this time with a different strategy. This slower, conscious correction process not only improves performance, it is also the 'teacher' of the fast process: the cerebellum makes its predictions based on past experience. It's through conscious practice that we stock our unconscious with the experience that allows it to operate.

Developing the controlled physical skills required in dancing involves fine tuning all our feedback systems – and short cutting them wherever possible by using the faster process of control.

So in fact, what children are really learning, when they learn to dance, or even just to walk, is to process feedback - to make better and faster predictions about the relation between motor commands and the movement the muscles will make in response to them. The ability to deal with feedback from the body – to use it to increase speed, accuracy and consistency – clearly improves with practice, and it's almost certainly the ability to process this information at speed that gives dancers such exceptional physical control.

Having learnt the basic ABC of dance technique – the alphabet, if you like, from which ballet choreography is written – a professional dancer's life is spent learning new ways of combining those movements in an ever growing repertoire of ballets, from 19th century classics to modern day creations.

One of my strengths as a dancer was my ability to learn new sequences at speed. I had various tricks for doing this. To encourage new movements to stick, I would consciously associate unusual patterns with something familiar. So if a choreographer asked me to do an unfamiliar collection of movements, I would remember the sequence by connecting it with a movement I knew, registering it as something like brush hair, check watch. Every time we got to that bit of the dance, I would trigger the movement pattern by consciously saying the code name in my head. Brush hair, check watch. Not an exact movement match, but close enough to give me a hook on which I could hang the learning.

What I didn't realise was that in using this short hand - consciously associating new movements with familiar ones and attaching an easily recognised label – I was actually mimicking the way the brain learns. I was consciously reinforcing an unconscious process.

The brain has other tricks for learning new tasks. It takes elements of existing skills and recombines them in a new way. A dancer learning to play football will, in the first instance, kick the ball very much like she does battements frappes. Really. Take it from me. Eventually, with enough practice, the brain makes a new motor command, with a new set of triggers, which is filed under 'football' and can be called upon at will.

But learning new movement patterns takes time. Lee, Swinnen and Verschueren's 1995 experiment showed that even after 60 practice trials, the brain will still go for the movement patterns it knows. It was only after 180 practice trials that the brain consistently reproduced the new movement pattern.

Which is why for every three hours you see on the stage, there are around sixty spent in the studio. A ballet performance, I have always maintained, is like a swan. The bit you see, gliding along on top of the water, conceals a massive amount of hard work hidden beneath the surface.

Much of the learning that goes on in the cerebellum is unconscious - that is, it goes on without us even thinking about it. The process of consolidation – laying down the movement memory, making it more permanent - continues for around five hours after rehearsal has finished. So while a dancer is in the shower, getting changed, going home from rehearsal, cooking dinner and watching TV, the learning is made permanent. It's a bit like making a jelly. You can stir it as much as you like, but at some point, you have to put it to one side and leave it to set.

The bad news is that if you try to learn a second movement skill in those five hours - a different type of dance, for instance - you compromise the process of laying down the first. The brain's representation of how to execute the movement is initially fragile, only becoming fixed over time. This is another place where scientific knowledge and dance practice part company. A typical dancer's day can involve learning four or five unrelated ballets in a series of consecutive rehearsals. It's no wonder we sometimes wake up wondering what the hell we learnt the day before.

When I was at my busiest, trying not only to take on new repertoire but also to change and improve the way I danced things I already knew, I would often lie in bed at night and go through the steps and the corrections in my head before I went to sleep. This was the early 1980s - I had no idea what a fast twitch muscle fibre was, let alone a neural pathway. But I seemed to know instinctively that a really effective way of getting corrections into my body when I had no time to physically rehearse was to go through them in my head: to imagine myself doing the movement perfectly, with all the corrections in place.

Nowadays, this all sounds very familiar. We've all heard about athletes who use imagery to improve their performance or conquer pre-match anxiety. And we've all read newspaper reports about how you can think yourself strong without going to the gym.

On the face of it, it sounds miraculous. But the science is quite straightforward. Functional Magnetic Resonance Imaging has, for the last dozen or so years, allowed the brain to be scanned in action. Like the muscles, the brain's demand for oxygen increases when it is active. And as oxygen is delivered via the blood stream, you can build an accurate picture of which areas are working by measuring blood flow changes within the brain.

Scans of my brain taken last year by Donna Lloyd at the Centre for the Functional Magnetic Resonance Imaging of the Brain in Oxford, proved that the pre-motor cortex – the area involved in planning the execution of movement – was highly active when I simply imagined myself dancing a role I had recently rehearsed. The limited space in the scanner meant there was no chance at all of me actually moving, yet my brain was clearly planning the procedures which would set my body dancing.

As far as the brain is concerned, imagined movement is the same as actual movement – just without the muscle contraction. By rehearsing the thought processes involved in movement, you improve the quality of instructions you send to your muscles. The quality of instruction has – obviously – a direct influence on the quality of the movement you produce. And as efficient planning of movements is one component of muscular strength, it makes sense that visualising a movement – rehearsing it in your head – will promote visible improvements in strength. Whether this means you can think yourself fit – or thin – is another matter.

For dancers and athletes, who continually tread that fine line between the stress necessary to produce adaptation and the stress that produces injury, this is a really useful tool. It allows you to strengthen the neural pathways involved in movement without stressing the muscles that produce it. It means you can rehearse effectively at any time and anywhere – even in bed.

I started my research into the dancer's brain convinced that learning something new had to be different from remembering something we already knew. I got myself into heated arguments with eminent neuroscientists because I couldn't understand why they couldn't tell me where I kept my memories. Surely, I argued, memories must be stored somewhere - like the videos you keep at the back of the cupboard and only bring out when there's absolutely nothing else on the box. And then, as I began to understand more about the brain and how it works, I found myself patiently explaining to a colleague that of course learning and memory are the same thing. It was obvious. At that point, I knew something in my brain had changed. I'd gone native.

Just as there is no single centre for movement in the brain, there is no single centre for memory - the best guess seems to be that memories are distributed across various zones - in the same way that bytes of information are stored on a computer's hard disk. And learning and memory are inextricably linked: Without learning, the brain has no memories and without remembering, the brain doesn't learn. Every time the brain sends an accurate command, hundreds – sometimes thousands - of nerve cells fire at the same time, strengthening the junction between the nerve cells – the synapse - and creating a mini network which holds a representation of the whole movement. You and I might call this mini network, this representation of a procedure which we can repeat at will, a memory.

A single nerve cell can form links with thousands of other nerve cells, so the total number of movement memories we can hold in our heads is enormous. The parietal cortex of the left hemisphere - the 'recipe book' for movements - seems to specialise in storing movement memories. But it doesn't store them as individual muscle instructions. It stores them as representations of the act, and we can access that representation in different sizes and different time frames. This is why a signature etched in the finest pencil can still be recognised when it's writ large with a brush and a pot of paint. And it's why I can choose to dance an assemble on the spot, or travelling half way across the room.

Scattered across my brain there are memories of some of the 120 ballets I've danced over my twenty years as a professional dancer. Some I can remember completely, but others are no more than sketches – the odd jump or turn. But they're all there, somewhere, and usually, a snippet of the music or a single step can often be enough to bring it all flooding back.

Roles I've danced more times than I can count - like the second act solo from Swan Lake for instance – don't need any kind of trigger to set them free. They seem to be stuck in my unconscious and I can't shake them off.

In fact I know it so well that while I'm dancing the solo, my mind can be anywhere. In the TV series of The Dancer's Body, I delivered a section of the script whilst dancing Swan Lake. The script was highly appropriate: it explained Paul Fitts and Michael Posner's acknowledged classic learning stages model, from 1967, which proposed that learning a motor – or movement – skill involves three stages: the cognitive stage, where you have to think about every move, engaging in cognitive activity as you concentrate on sending the correct instructions to your body. At this stage, mistakes are frequent and while you generally know you're doing something wrong, you're not sure how to put it right.

The second stage is the associative stage, where you start to associate certain cues with the movement. Performance standards become a bit more consistent and you start to be able to detect and identify some of your own errors.

After serious and sustained practice – which can take many years – some people – not all – move into the third stage, the autonomous stage. Now, the skill has become almost automatic. You don't have to think about what you're doing and you can often carry out another task at the same time – like talking to a camera while you're dancing, or holding a conversation while you're driving.

In fact, anyone who can drive a manual car can identify with Fitts and Posner's model. Think about when you first learnt to drive – remember how impossible those hill starts were. Foot brake, clutch, engage gear, hand brake off as you let the clutch out. Stall. Try again. Then, after a bit of practice, you could do a hill start as long as you really, really thought about it. Finally, after hours of practice, you could do a hill start at the same time as arguing with your girlfriend about whether it's better to go via the South Circular or the A2.

That's the body on auto pilot.

We all have a massive repertoire of every day skills we can perform on auto pilot. Everybody's body knows more about some pretty complicated movement sequences that you might realise. Think about the hundreds of instructions your brain had to send to your body just to get you here tonight...or even the multiple muscle contractions involved in sitting down, or standing up from a chair.

An extraordinarily complex sequence of actions, yet we never even stop to think about it. That's because the brain lets the body get on with all those movements it knows well. It delegates them to the cerebellum, leaving the conscious centres of the brain free to think about something else. Like an exam invigilator, it stands silently back, only intervening when it senses trouble. If you step on a wobbly paving slab, for instance, or the heel breaks off your shoe, you don't fall over. Your conscious brain instantly snaps to attention and takes control.

The brain, like the body, has an amazing capacity to adapt to the demands we make of it. Day by day, perhaps even hour by hour, the brain redesigns the somatotopic map to ensure that it's fit for purpose, apportioning more of it to dealing with tasks you perform on a regular basis. As far as I'm aware, no one has examined the somatotopic map of a dancer, but brain scans of string players have found that the area of the map that corresponds to the fingers of the left hand - the fingers that do all the fingering - is enlarged to cope with the repeated demand for precision and speed.

And the brain will go on improving for years. Crossman's famous 1959 experiment, with workers in a cigar factory, revealed that even after seven years employment, even after the employees had made over 10 million cigars, their cigar rolling technique was still becoming both faster and more efficient.

An extensive study of experts in various disciplines by Ericsson, Krampe and Tesch-Romer in 1993 concluded that expertise is the result of intense practise for a minimum of ten years. Ericsson and his colleagues described the specific type of practice a person needs to achieve expertise in any area as deliberate practice: optimal instruction as well as several hours of intense, workmanlike practice for hours each day. Sounds like ballet training to me...

Dancers' heads might be full of foot steps, but don't make the mistake of assuming their brains reside entirely in the tips of their toes.

The dancer's brain is a fit for purpose computer, skilled at organizing fragments of movement knowledge and strengthening the connections between them. Specially adapted to remember more information, solve problems more efficiently and make accurate decisions faster.

The dancer's brain is like the dancer's body - a highly developed collection of cells fine tuned over years of training to produce the amazing feats of human movement that we call dancing. And it's all done for one reason - to stimulate the brains of the people who come to watch.

So for the last ten minutes or so, I'm going to turn the spotlight on you, the audience.

Over recent years, I've become increasingly fascinated by the power of dance to engage and communicate with people who know little or nothing about the art form or its technique. Normal people – you lot – who are moved, excited, challenged, entertained and, frankly, turned on by dance. Without using words, dance appears to be capable of communicating ideas and emotions, and your brains appear to be able to read not only movement, but meaning within that movement.

I've frequently listened to surprisingly articulate explanations of the meaning behind ballets, meaning that I, the dancer, never imagined could be gleaned from the dance I'd just performed.

Scientists are generally too busy researching cures and eradicating disease to spend much time on the arts. But a number of eminent scientists have shown themselves to be more than willing to engage with the debate and discuss ideas about how our brains interpret the meaning in movement. And some of them are game enough to do it the most unusual circumstances. I've played tennis with Sarah-Jayne Blakemore and danced tango with Semir Zeki. Had I had the good fortune to meet Ramachandran, I can only speculate on what we might have got up to in the name of art....

In any investigation into how the brain reads dance, the most obvious place to start is with body language. It's often said that 70% of everything we understand in a conversation is based not on the words we hear, but on the gestures, inflexion, postures, winks, nods and facial contortions which accompany them. Body language. A language older than words: without doubt, man has been communicating to man for far longer than we have had codified language. So there must be some part of our brain, some primitive hangover, which is able to read movement without the additional contribution of verbal language. It seems entirely reasonable that this could be one of the ways in which an audience reads meaning into dance.

Marc Salem, a Body Linguist who attracted a lot of attention last year with his one man show, Mind Games, demonstrates what we probably all know – that if someone's voice and body are telling different stories, you tend to believe the body. Why else does a television director focus in on sweating palms and fidgeting fingers when DI Frost or Inspector Morse interrogates a murder suspect pleading his innocence? I saw Salem perform in London and I hope he wouldn't mind me saying that half of his act is pure showbiz. It was certainly astonishing, but my cynical brain alerted me early on to the fact that there must be some sort of trickery involved when a guy can reveal the multi digit code on a bank note folded in a stranger's pocket. The other half, however, is a very clever demonstration of how the human body endlessly gives away much more than it intends through what Marc calls micro movements. At one point, Marc invited several members of the audience to draw a picture, and then respond with a 'no' when he asked each in turn whether or not they were the artist. After two or three fast paced turns around the group, he correctly identified the person forced to lie. Not science, but hard not to marvel, especially when he explains the gestures, tics and glances through which people give themselves away. And very hard not to be impressed when he did the same trick on me.

According to Salem, body language is harder to control than words because it is often unconscious; sometimes, even, a biological response to emotion which cannot be concealed without years of practice and massive conscious effort. A small example: most of us widen our eyes before we tell lies. Try hiding that.

Ramachandran says the same thing, but with a scientist's circumspection: 'certain moves and postures' he says, 'evoke emotions in the brain.'

With a dancer's instinct, I know this to be true. Dance can tell us about feelings far more eloquently than words. Emotions have a physical consequence: in the crudest terms, sadness leads to hunched shoulders, elation to an uplifted chest and gleefully thrown back head. It's not hard to write that in dance terms. And if we are all habitual users of body language, it stands to reason that we are all habitual readers of it, too. Could we therefore stop ourselves reading the body language of dancers on the stage?

Whether consciously or unconsciously, choreographers certainly use elements of body language when they create dance. Kenneth MacMillan's choreography for Princess Stephanie and Prince Rudolf in his ballet, Mayerling, provides clear examples. He is brutal, depraved, unloving. She is a new bride far from home: terrified, but resigned to what her royal upbringing has taught her: a wife's role is to make her husband happy. So for every vicious assault he makes on her – wrenching back her head, torturing her and then tossing her aside, quite literally, like a rag doll – she dutifully goes back for more, clinging around his neck at one point like the encumbrance he clearly sees her to be. Centuries of abusive relationships, and the frightening paradox of women who return time and time again to men who mistreat them, is all danced out in their pas de deux. And as the curtain slowly falls, we just have time to see Rudolf advance menacingly towards the exhausted Stephanie, lying crumpled on the bed. No one is left in any doubt as to what will happen next.

I've never danced this duet, and never had it explained to me. All of this I read into the choreography as clearly as if it were printed on their costumes.

What dance can't tell us, not without copious programme notes or cumbersome ballet mime, is the catalogue of events explaining why the character feels the way they do. So while we can assume certain things about Rudolf and his attitude to women – distant mother, starved of affection – we cannot possibly know that the four menacing Officers who keep appearing from behind curtains are Hungarian separatists intent on converting Rudolf to their cause. As the choreographer Georges Balanchine so neatly put it, explaining why the overwhelming majority of his ballets are plotless: how do you explain mother in law in ballet mime?

Semir Zeki equates choreographers with neurologists, insisting that they are both engaged in the same research. As we tangoed around the Rivoli Ballroom in Brockley, South East London, Zeki put forward the idea that what choreographers and neurologists have in common is that they are both investigating how and why the brain responds to certain external stimuli.

Choreographers know, from a mixture of instinct and experience, how audiences will respond if they group their dancers in certain patterns or ask them to move in particular ways. Wayne McGregor is a contemporary choreographer who describes his own choreography as dysfunctional. He delights in creating movement which doesn't take the obvious course – movement which might twist without warning, or appear to go against the physically possible. And he knows that the audience will feel slightly uncomfortable but immensely stimulated by this type of movement. You don't tend to snooze down into your seat during a performance of Wayne McGregor's work. It forces you to sit up and pay attention.

It was fascinating to have almost the same conversation, but with brain zones replacing dance terms, choreographic context replaced by neurological context, with Sarah-Jayne Blakemore. Over a game of tennis, Sarah-Jayne explained to me how the brain is able to predict movement pathways based on the stance, velocity and angle of the human body. With the kind of speed that puts your average PC to shame, it calculates the predicted consequence of those postures, enabling it to get ahead of the game. That's how tennis players know where the ball will land when their opponent squares up to serve – they couldn't possibly see the ball in flight at that speed. And with skilful playacting, that's how footballers trick the goalkeeper into diving the wrong way during a penalty shoot out. Our brains cannot help but read intention into action – and when the prediction is not fulfilled, the brain is fascinated. It perks up and pays attention. So when Wayne McGregor's dancers give every indication of jumping to the left, yet somehow jump to the right, we are alert, interested....and vaguely unsettled.

By creating a discrepancy between what the brain predicts will happen and what actually happens, choreographers stimulate the audience, grab their attention, set them on the edge of their seat.

Conversely, the brain experiences a massive sense of fulfilment when the prediction is accurate – the 'ah' feeling. This is one of the reasons why symmetry is so appealing... why we feel relaxed in classical surroundings. I suspect it's also why we feel comfortable when the tune returns to the tonic. Things which are asymmetrical, arrhythmic or discordant make us edgy, restless and unsettled. Order makes us feel comfortable, at ease, that all is well in the world. In Sleeping Beauty, for instance, the dancers are frequently arranged in equal ranks on either side of the stage, mirroring each other's movement and barely concealing the underlying message that all is in order in Russia's Imperial Court. Classical architecture, music and art generally adhere to the same principles of balance and harmony. Contemporary culture, on the other hand, is frequently asymmetrical, discordant, arrhythmic. I suspect there's an interesting piece of research to be done on how current trends in music and culture influence the state of society by leaving us feeling vaguely unsettled without quite realising why.

But according to Ramachandran, there's more than just comfort in symmetry. Our eyes continually feed our brains with images. It has learnt to filter out those things which aren't important and focus on the information which could have serious consequences. And it's particularly interested in information with biological significance. As most mammals move (and are) symmetrical, symmetry instantly alerts our brain to the presence of another human being or an animal - the potential of predator, mate or prey. Something which could fulfil our most basic needs or cause us harm.

Huge areas of our brain are dedicated to detecting biological motion. And it seems that we can detect biological motion even when it doesn't appear to come from a biological source. A fascinating experiment attached lights to the joints of two people and then asked them to move in the dark in specific ways. The only information available to the spectator is dots of light, but the brain is acutely sensitive to human movement. Using less than a ten thousandth of the information you'd get from a normal television picture, it's perfectly obvious that these are human bodies and they are walking, bowing, kneeling or dancing.

Given that our fellow man represents both our only hope for survival and our greatest threat, it's not surprising that our brains have evolved to recognise biological movement and predict the physical consequences of certain postures, movements and preparations. It seems that choreographers play with this by creating extraordinary, almost inhuman movement – movement which shouldn't be possible: legs extending upwards past an ear, bodies bending in directions which appear to conflict with reality. Movement which exaggerates postures, facial expressions, and gestures, provoking a heightened response from the neurons. Ramachandran cites the example of classical Indian dance, with its exaggerated femininity, extravagant hand gestures and extreme use of the eyes.

Some of the inspiration for Ramachandran's research into the effect of exaggeration in art came from an unlikely source: zoologist Niko Tinbergen's study of the feeding habits of baby seagulls. The mother seagull has a distinctive long yellow beak with a red spot. When the chick pecks at this spot, she regurgitates food and the hungry chicks are sated. The chick quickly learns that the red spot equals food. Not surprisingly, Tinbergen discovered that the chicks were equally attracted to a long piece of yellow cardboard painted with a red spot. But his real discovery was that even though it looked nothing like a beak, a thin stick painted with three red stripes got the chicks more excited than the real thing. It was as if the feeding neuron in the chicks' brain was responsive to red, so as far as they were concerned, the more red, the better. Ramachandran's punch line was that if the seagulls had art galleries, they would put the stick with its red stripes on a wall, pay millions of dollars for it and call it a Picasso. And over drinks and canapés they would sqwauk to the other birds, 'we don't know what it is about it. We just love it.' Similarly, we aren't sure why we respond to the highly stylised, grossly exaggerated postures of classical ballet, but we do.

I've taken part in numerous discussions over the years about whether we appreciate dance in our heads or in our bodies. Certainly, people often describe getting involved in the movement they watch, and dancers find it impossible to see other dancers in action without swaying from side to side and feeling their every move. Away from the stage, most of us can't watch a boxer being punched or a character in a movie beaten up without wincing in sympathy. It's as if seeing a physical action stimulates a physical response.

Again, science is beginning to prove what we instinctively know. In the late 1980s, a group of neurophysiologists in Parma, Italy, discovered a class of neurons in the premotor cortex of monkeys which fired not only when the monkey carried out an action itself – in this case, picking up a peanut – but also when the monkey observed another monkey making the same action. The neurons were not just involved in direct imitation - they fired when the monkey was completely still. They appeared to be involved in observing and understanding the actions of others, by relating them to one's own potential actions. It seems that mirror neurons provide the brain with a system not just for copying actions, but for empathising with them.

Brain imaging studies in man have confirmed the existence of a similar premotor area, which responds in a similar way. Science is only just beginning to uncover the implications of mirror neurons, but it seems logical to me that they must play a part in what's often referred to as a kinaesthetic appreciation of dance.

An experiment carried out just this year on 30 males – 10 ballet dancers, 10 capoeira dancers and 10 control subjects - aimed to discover whether there is a difference in the way the brain watches unfamiliar movement and movement in which it is an expert. Dr Patrick Haggard, of the Institute of Cognitive Neuroscience, has recently been appointed the first Associate Scientist at the Royal Opera House and used dancers of The Royal Ballet both to develop the experiment and as subjects to be scanned.

Haggard and his colleagues were particularly interested in the premotor cortex, the area where mirror neurons involved in action observation were found in the monkeys. The experiment involved showing all three groups video clips of a ballet dancer and a capoeira dancer performing similar movements. The results showed that the ballet dancers had more activation when they watched ballet than capoeira while for the capoeira dancers, the reverse was true. The brain clearly identified with the moves it knew and the premotor cortex responded by planning the action it would take were it asked to generate the movement itself. Haggard thinks that this is the first experiment to demonstrate that the human mirror neuron system responds not just to observation of physical body movement but to observation of specific learned skills: in Haggard's words, we interpret not just the movements of others relative to our own, but also the higher level action plans of others relative to the action plans that we have learned ourselves.

Once again, the data suggests a neuroscientific basis for what professional dancers already know: that watching the performance of others can illuminate and even improve one's own performance.

In addition, though, the scans revealed something unexpected: greater brain activity in a separate set of mid line areas which are involved in familiarity and emotional engagement – the areas which say 'I can do that' or 'I like that'. Clearly the experts were engaging on an emotional level with the technique they knew.

It's fascinating stuff and I'm looking forward to working with Patrick on more and more investigations into the dancer's brain. But I still want to know why a non-dancing individual can get so involved with dance. If you can't execute a series of fouettes or entrechat six, you're unlikely to have a mirror neuron response when you see them performed. But this raises a new question: much of dance is based on biological movement, a variation on walking, kneeling, bending, running, jumping. Movements we all make all the time. So might we have some kind of low grade mirror neuron activity when we watch dance whether or not we can actually do the steps? A kick is a kick is a kick, after all.

Even without the relevant mirror neurons, there are plenty of people who get turned on, excited and moved by dance, extracting meaning from its particular physical language. So as part of The Dancer's Body – the series for BBC2 which screened last year – I thought I'd conduct a little experiment of my own. We gathered together a varied audience for a performance of a dance specially choreographed by David Bintley, director of Birmingham Royal Ballet. It was made up of people with all sorts of backgrounds and all sorts of attitudes to dance. People who love it and people who hate it. People with a range of physical and artistic skills: sportsmen and women, painters, scientists, designers and musicians. Performed to Eric Satie's Gymnopedies, the dance was a duet based loosely on Greek wrestling. Myt partner and I were, depending on your interpretation, engaged in a ballet of wills, a physical struggle or a passionate love affair. We danced the three minute duet twice, without giving the audience any verbal or written clues at all, and then interviewed many of them individually to find out what they had read into Bintley's choreography and our performance. The responses were fascinating.

I was surprised and genuinely delighted that everyone had found something in the dance which I believed was there, and had, in most cases, added a little twist of their own. The choreography frequently called on the dancers to come close together and then pull apart and most people interpreted this as indicative of their relationship. 'It's like they want to be together but they can't quite work out a way'. Michelle Griffiths, the hurdler, called it a 'love battle' and Mark Hamilton, a relationship counsellor, said that if those two had come to see him professionally he'd conclude that they desperately needed to have sex. Mina Anwar, the actress who danced until she was in her late teens, described the physical sensations she experienced in her muscles as she watched and said she felt as if she'd performed the whole thing herself.

But it was a ten year old girl who most astonished me. The final position in the dance had happened quite by chance in rehearsal: Facing away from one another, me perched in the nook created by my partner's bent knee. I had raised my hand to my head, completing the picture, only because there was a note in the music left to fill. As rehearsals progressed, that ending never changed. The young girl said it was like two pieces of a jigsaw slotting together, and what it meant was that our relationship could only work if we faced in opposite directions. Lifting my hand to my head symbolised a sort of tired resignation to our fate. This image, of two halves making a whole but finding their own particular way to accommodate each other, described exactly the relationship I thought Bintley had tried to create through the dance.

Studying the workings of the brain has, inevitably, had an impact on my own. Still plastic, after all these years. For all its extraordinary physical skills, I've come to see the dancer's body as nothing more than a medium – a conduit for the communication of information between two brains: mine, the dancer's, and yours, the spectator's.

Now that it's no longer what my body does for a living, I've finally realised that dancing is all in the head.