Having discovered that Pierre Boulez didn't think very highly of a performance I'd given of Stravinsky's Petrushka, I thought I should find out why. This was a long time ago, but my abiding memory is that most of his suggestions differed from what was in the score. Once I'd plucked up courage to mention this, he explained that he approached music as a composer not as a conductor. He felt the fixed nature of printed texts can run contrary to a constantly evolving composer's mind and what Stravinsky wrote on one day wasn't necessarily what he wanted to hear on another. Coming from many conductors this could sound arrogant but if someone like Boulez says it, the idea of changing what's written cannot just be dismissed as dilettantish. Mahler asked conductors to alter whatever sounded wrong. Perhaps it takes a composer's mindset to do it well.

Composers engage in music creatively, and when they conduct, that genuine creativity remains. This is clear when you listen to recordings they've made of their own music and hear where they contradict what they've actually written. This leaves performers with a dilemma as to which is the more reliable source. Rachmaninov's recordings vary significantly from what he composed. Should we aim to perform what he wrote or what he plays? My view is that the score should remain the authority. I've made enough CDs to appreciate that final versions don't always represent exactly what was envisioned, and in the early days of recording there must have been several variables that prevented composers feeling they'd created definitive performances. Nevertheless Rachmaninov was a wonderful conductor and a great pianist too. His performance legacy cannot be dismissed. Those who argue that his practical experience as a performer should override his achievements as a composer have a valid point.

Elgar's recordings are more victims of circumstance. Fabulous documents though they are - one can almost hear the twirling of Edwardian moustaches - not many would uphold his choice of tempi as being preferable to those he wrote. Other composer-conductors lacked the purely technical ability to achieve their goals. When a timorous record producer remarked to Stravinsky that his second take was faster than the first, the composer said he liked the second version better. This retort was probably more of a smokescreen for his conducting failings, but one cannot really be sure. How do we know if something we hear is the result of a composer's weakness as a performer or a confession that their subsequent ideas are preferable? The choice is ours, which I suspect is exactly how most composers would like it.

Performers try to embody a sense of ownership of the music they play. That's easy when composer and performer are one. But the principle infuses composers' thoughts on other music too - an identification that liberates their own creativity. Shostakovich said it's 'better to love a score than respect it'. Respect keeps its distance. Love gets involved. There's no denying which is more powerful.

www.markwigglesworth.com