1998 WINNER

How to compose your thoughts

By Andrew McGregor

YOUNG Mozart was fed up. He wanted to see the score of Allegri's Miserere but no one would show it to him. This was, however, no big deal for a genius. After listening to just two church performances he simply wrote out the entire score from memory. Over 200 years later, although such skills are rarely displayed, the potential for the rest of us to exhibit them may be on the verge of being realised.

All of us are musical to some extent; we know what music we like; we often associate memories with certain pieces of music and we all seem to be able to recognise music we have heard before. We are, however, often unable to pinpoint exactly what we like about a composition, understand why we always cry at certain music or sing back a tune we immediately recognise. So we fail to capitalise on our innate musical thoughts. If only it were possible to express ourselves without the need for years of musical training. But how?

This is the question a group at the University of Glasgow is trying to answer. Led by Dr Eduardo Reck Miranda, a researcher at the Sony Computer Science Laboratory in Paris, its ultimate aim is to bypass the need for the technical skills required to make music by playing instruments directly from a person's brainwaves.

Ten years ago this idea would have been consigned to the realm of science fiction. Now, advances in artificial intelligence, computer science, psychology and musicology are making such an ambitious goal realisable. There are, however, major challenges.

The first main task is to determine how the brain represents the music it hears or imagines - it is, however, very reluctant to divulge these secrets. There are several methods of obtaining information about what the brain is doing at any particular time. The most promising is by taking electroencephalograms or EEGs.

These are obtained using electrodes placed on the scalp to detect the electrical activity. It is anticipated that by analysing the EEG signals, the researchers will be able to tie specific musical ideas to particular types of activity in the brain.

In practice this is incredibly difficult, since the brain is never working exclusively on processing music. Although taking the average of many EEG recordings can help, the complexity of the brainwaves means that finding recurring patterns, let alone linking specific patterns with musical thoughts, is like looking for needles in haystacks. Luckily for the Glasgow group, they are not the only ones looking for needles.

Progress came when two pieces of apparently unrelated research came together. Firstly, psychologists produced experimental evidence that suggested the brain processed music and language in a similar way. Secondly, Heinrich Schenker, a German musicologist, proposed the idea that all good musical compositions have essentially the same structure.

The fact that this structure has very obvious parallels with grammar in linguistics has allowed researchers of both disciplines to profit from each other's findings. This breakthrough has enabled them to come much closer to answering important questions concerning what it is about certain sounds that identifies them as music.

Using this understanding of the structure of music makes finding the equivalent patterns in the brain much easier. It gives clues on how to relate brainwaves to music, using currently available pattern matching technology, a type of computer called a neural network.

Once this is accomplished the potential for progress is almost unlimited. Computers controlled by a person's brain activity could be used to play instruments and write scores, musically expressing emotions otherwise inexpressible by most people.

Roughly speaking, a computer, linked to the brain and to a synthesiser, is programmed to learn how to associate classes of sound patterns with the patterns of brain activity produced by a subject thinking of, or hearing music. Once the computer is trained to make such associations, it can playback the sound patterns associated with specific brain activities. This opens up a world of possibilities presently only available to those with years of musical training.

But what would the musical greats of yesteryear have to say about it all? If Dr Miranda and his team are successful, will the Mozarts and Beethovens need to make way for a new era of music since their listeners will have become composers in their own right?

It is impossible to speculate about the quality of the music each one of us could be capable of generating. Will the schizophrenic mind excel at creating twopart harmonies? Will the ordered mind of a brilliant scientist challenge the classical composers? Only this research can answer such questions.

Regardless, however, of the likely quality of the music which you and I could produce surely we all deserve the right to realise our full musical potential . . .it is only a thought.