WELL over half a century ago, neurologists found that by placing electrodes on the scalp they could record the electrical signature of the brain at work. Most of the time they saw a random hotchpotch of signals-the combined activity of hundreds of thousands of brain cells. But they were astonished to find long stretches when this mess of activity became ordered into a pattern of elegant rhythmical waves. Ever since, scientists have wondered whether the secrets of our thoughts, perceptions and even consciousness itself might be hidden in the patterns of our brain waves.
The question of why we have brain waves-and what they tell us about how we think-is as hotly debated today as it was when the patterns were discovered. Researchers can see slow 鈥渁lpha鈥� waves in scalp recordings when the brain is relaxed, and 鈥渢heta鈥� and 鈥渄elta鈥� rhythms while we sleep. But the meaning, and even the existence, of faster 鈥済amma鈥� rhythms in the alert brain is highly controversial.
The problem is that you can鈥檛 see these faster rhythms directly. They are so well hidden in the noise of other brain activity that researchers have to uncover them by mathematically breaking up the scalp electrode trace-the electroencephalogram or EEG-into its component frequencies. And once you鈥檝e uncovered a rhythm, how do you know it is anything more than an artefact of the technique, or a meaningless by-product of neurons that are wired together into networks?
Advertisement
But many researchers are now coming round to the idea that these brain waves are for real, and far from meaningless. The latest suggestion is that the rhythms could be the key to detecting, linking and organising processes going on in different regions of the brain. Some believe that two of the rhythms-the theta rhythm, with between 4 and 8 waves a second, and the gamma rhythm, which oscillates up to ten times faster-might even interact, and in so doing help the brain to package information into coherent images, thoughts and memories.
Some of the first clues that brain waves might help to organise neural activity came from experiments with rats. By recording from large electrodes placed in the hippocampus-a brain area that is important for navigation, learning and memory-neurophysiologists detected a very prominent theta rhythm. 鈥淓arly on, there was a strong concern that this was an artefact of sniffing,鈥� says Howard Eichenbaum, a neuroscientist at Boston University. People thought that the rhythmic muscle activity was modulating the electrical signal. But theta waves are now known to be the result of genuine neuronal activity associated with the animal鈥檚 movement through the environment, he says. And they turn out to provide an elegant framework for organising the activity of hippocampal neurons.
Many of these neurons, called place cells, fire, or spike, when an animal is in a particular part of its environment. As a rat runs along, the firing sequence of place cells tracks the route. But each cell responds to quite a large area, so how does the rat get a continuous readout of its path, rather than a series of jumps? When researchers recorded the place cell spikes against the background of the theta rhythm they saw that when the rat first enters a place cell鈥檚 region, the cell fires late in the theta cycle. As the rat moves further into the area, the neuron spikes earlier and earlier in the cycle-a phenomenon described as 鈥渢heta-phase precession鈥�. So if you know on which part of the theta wave the spikes are surfing, you know which part of its place field a neuron is responding to.
Earlier this year, Michael Kahana and his colleagues at Brandeis University in Massachusetts found that this theta rhythm could also be important in people. Recording from electrodes implanted in the cortex of patients undergoing treatment for epilepsy, they found that theta waves appeared when people were navigating their way around a computer maze, and that the waves appeared more often if the maze was difficult. The brain wave patterns also reappeared when people tried to recall the route.
But it seems that the importance of brain waves is not limited to route finding. They might have a much more widespread organisational role. At the moment there is a big hole in many theories of perception, called the binding problem. The puzzle is how we bring together related sensory signals in the brain. Every time you see an object, for example, the brain processes the visual input by immediately splitting it into its component features (see Diagram). Some neurons in the brain react to particular colours, others to specific angles, some even to particular configurations of a face. But we perceive the object as a whole, which means that once the brain has analysed all of the individual features, it must somehow bring all these disparate bits of information back together again. This is the binding problem.
It would be fairly straightforward if you only ever saw one object at a time. But what happens when you see, say, an apple and a banana together? 鈥淩ed鈥� neurons, 鈥測ellow鈥� neurons, 鈥渞ound鈥� neurons and 鈥渆longated ellipsoid鈥� neurons would all be firing together, with no obvious way for the brain to link the right ones.
Wolf Singer and his colleagues at the Max Planck Institute for Brain Research in Frankfurt were among the first to study whether the features of an object could be bound together by synchronising the firing times of the neurons that encode them. Synchrony is the key, they believe, because groups of neurons that fire at precisely the same time reinforce each other and so could have a greater effect than neurons firing singly. This means that they would be able to activate other groups of neurons and pass on their message more easily. But how could you detect synchrony, especially when you鈥檙e looking at widely spaced areas of the brain?
Almost a decade ago, Singer鈥檚 group and others began to record neural activity in the visual cortex of cats, whose visual system has been well studied. They found that there was a background pattern of activity, and that this 鈥渇ield potential鈥� oscillated 40 times a second-the gamma rhythm.
Seen and heard
While he was monitoring this background activity, Singer measured what individual neurons in the visual cortex were doing. When the cat was looking at an image, the firing of specific neurons seemed to coincide with the peaks of the background gamma waves. It appeared that whole groups of neurons were firing at precisely the same time, and this meant that they could be 鈥渉eard鈥� over the din of other neurons firing randomly in the brain.
Singer鈥檚 team went on to look at groups of neurons monitoring neighbouring regions of the visual field that were all tuned to spot lines at a particular angle. They found that these neurons would synchronise their activity with the peaks of the gamma rhythm when a continuous line fell over them all. It seems that the neurons somehow know that they are looking at the same object. Widespread gamma oscillations could, according to Singer, give a time frame into which neuronal activity in widely separated areas is organised. In this way, gamma waves might be the key to perceptual binding.
Earlier this year, in Paris, Eugenio Rodriguez and his colleagues at the CNRS, the French national agency for scientific research, published intriguing evidence linking gamma oscillations in humans to perceptual binding ( Nature, vol 397, p 430). To try to work out what is going on in the human brain at the moment of perception, the researchers showed people high-contrast pictures known as Mooney faces. When they are the right way up, these black and white images are usually recognisable as faces, if not always instantly. When they are upside down, they look like a random jumble of shapes.
The researchers found that when people first looked at the pictures, there was a lot of gamma brain wave activity all over the visual cortex, but it wasn鈥檛 synchronised. As soon as a person recognised a face, however, the gamma waves in the different visual areas fell precisely into step.
To tell the researchers when they had recognised an image, the viewers had to press a computer key. The synchrony of gamma waves in the visual cortex dissolved just before the key press, and a second period of synchronised gamma activity began in the motor areas of the brain as the key-pressing movement started. So brain waves may not only organise sensory information, but could also package motor signals being sent out of the brain.
What鈥檚 more, it is not just in perception that brain waves could be useful for binding information together. It is also possible that we use gamma waves to connect information that we learn. Synchronous waves could link the different features of a face, perhaps even names and faces, or places and events. This idea is supported by the work of Wolfgang Miltner of the University of Jena in Germany and his colleagues. They showed a group of volunteers coloured lights. Every time one particular colour appeared, the volunteer was given a small electric shock on the middle finger. People rapidly learnt what the different colours meant, and after they had, the researchers found that the amount of gamma wave activity in the visual areas increased when the shock-associated colour appeared. At the same time, there was synchronised gamma activity in the part of the cortex that receives sensory signals from the fingers ( Nature, vol 397, p 434).
But even if gamma waves synchronise the representations of individual items, there is still a problem. How does the brain code for more than one item in a scene without muddling the features of each one up? By extrapolating the organisational power of theta waves to human cognition, John Lisman and his colleagues at Brandeis University have come up with a provocative theory for how several different items can be represented by gamma oscillations in the same parts of the brain without getting muddled together.
He points out that you can perceive things as separate events only if they occur more than about 25 milliseconds apart. If two clicks are played within this period you only hear one, but if they are further apart you hear both. And, Lisman says, 25 milliseconds just happens to be the time between successive peaks of a gamma wave. But how do you know whether a click falls in one gamma cycle or the next? Lisman believes that gamma and theta waves work together, with gamma waves packaging up features and theta waves acting as an internal clock that determines which gamma cycle, or set of features, is which.
He also carries this idea from perception to memories, suggesting that memories are represented by a group of neurons that fire in the same gamma cycle. 鈥淣ow, the really breathtaking idea,鈥� claims Lisman, 鈥渋s that from consideration of these oscillations, you can begin to appreciate why there would be capacity limits.鈥� If you are trying to remember a telephone number, say, your 鈥渨orking memory鈥� can keep track of only around seven items. Lisman believes this is because that鈥檚 how many gamma wave 鈥渟torage spaces鈥� fit inside the envelope of each theta wave (see Diagram). 鈥淩oughly speaking, you can fit five, six, seven gamma cycles in a theta cycle, and isn鈥檛 it interesting that this is roughly the limit of short-term memory?鈥� he says.
Lisman and his colleagues say that results from a task called the Sternberg test back their suggestion. Suppose you have to remember a list of three letters and then say whether the letter 鈥淏鈥� is in the list. It will take you a while to respond. If the list is increased to four letters, your reaction time will increase by about 30 milliseconds-roughly the period of a gamma wave. Lisman suggests that to identify whether a letter was on the list, you have to scan through each item, and that this requires a succession of gamma cycles.
However, Lisman is the first to agree that more research into the organisational powers of brain waves is necessary. As yet, his theory remains just that-a theory. But to see whether he is on the right lines, he would like to find out if the firing of hippocampal place cells at particular phases of the theta rhythm in rats is also coordinated by gamma waves.
Gyorgy Buzs谩ki of Rutgers University in New Jersey has shown that the size of gamma oscillations in the hippocampus can vary depending on the phase of the theta wave, but it is not yet possible to say whether there is any causal link between the two. Nevertheless, while Buzs谩ki and others might not yet be convinced that gamma and theta rhythms are the glue that binds together our perceptions, thoughts and memories, he does agree that brain waves are important. As he points out: 鈥淚t is such a ubiquitous pattern, that you can鈥檛 believe it鈥檚 not good for something basic.鈥�
