In the third and final part of our art and science debate, Mark Hanson claims that both disciplines depend on observation and interpretation; WE TRY TO PASS ON OUR SYNTHESIS OF FEELINGS AND EVENTS
Click to follow
"Never Trust facts unless they are supported by theory" remarked a colleague at a recent scientific conference in the USA. I had heard that adage before, as a flippant reversal of the commonly accepted order of things but, this time, it was very much in accord with my own thoughts. We had met, about 150 of us, to discuss progress in research to prevent Sudden Infant Death Syndrome. Our meeting seemed to consist of a series of papers which reported anecdotal observations or, worse, ideas not backed by careful experiment, seemingly plucked from thin air. In view of the importance of the problem we'd met to discuss, it seemed depressingly trivial. But as I left I knew that it had been a worthwhile conference. Why?

Scientists involved in research for more than a few years know, whether or not they admit it, that the simple view of scientific method expounded by Karl Popper - one sets up a hypothesis and then designs experiments to test it - applies considerably less than half of the time. The deductive approach (moving from the general theory to particular examples) is frequently overshadowed by the need to work by induction (from the particular to the general). I ponder this as I sit in the meeting's darkened auditorium, reassuring myself that jet lag does wonders for lateral thinking, even if a few logical connections may slip by unnoticed.

Travelling can be a good time to concentrate on some work, away from the accusatory ring of the telephone and the gleeful bleep of incoming e- mail. I'm reviewing papers submitted for publication in a major journal and one in particular is causing me trouble. I read it carefully, concentrating first on the methods used and the duality of the observations reported. Then I turn to the inferences, the conclusions derived. I'm not convinced and, as often happens when one scientist is confronted with the work of another, I've formed my own ideas on how the data was arrived at - maybe partly artefactual, perhaps instruments not accurately calibrated, some dodgy statistics, etc. My interpretation of the results is different: I don't support the paper's publication in this form.

We all try to make sense of the impressions that make up our daily lives, to fit them into some kind of structure, some system which we can understand and maybe even predict and control. Our experience stretches from the intangible - a meeting does not go well and we wonder what lay behind this, what signs we had failed to read - to events which flood the consciousness with the brutality of "fact" - a close friend dies and we are only gradually able to come to terms with the loss.

Then, at one remove from this, in what Wordsworth termed "emotion recollected in tranquillity", we try to pass on our synthesis of feelings and events, the things which we can explain and those that we cannot, to others. We interpret. And at this point the artist and the scientist work in the same way. The scientist tries to account for his observations and those of his colleagues, the better to order and communicate them. He assembles them into a structure and may accord this the status of a "theory". Sometimes the framework is logical, but often it is associative or consists of things collected under a useful heading, for example "various conditions under which new-born infants show prolonged pauses in breathing", the significance of which he may be able to establish later.

Meanwhile the artist is also endeavouring to discover a vehicle for his observations. His goal is "to make you feel ... to make you see" as Conrad put it. He may work meticulously through a series of experiments in technique, and it is unlikely that his success in synthesising his experience and his feelings will be any less hard-won than the scientist's. Both of them will also, if they are any good, be attuned to the possibility of accident, chance, a serendipitous observation, opening up a highly productive avenue. Here is the exercise of human intelligence, experience and judgement operating in similar ways, to create the scientific theory or the work of art. We admire Cezanne, working so slowly and tirelessly until the painting, the interpretation, was perfect. We admire that ability to synthesise so much experience, so much "fact", into a communicable statement, an expression of what the world is like, just as we admire the greatest scientific minds for their ability to synthesise, to describe nature. I was struck by reading recently that, when asked how he was able to make his discoveries, Einstein replied "I used my imagination".

The problem with interpretations is that, by definition, they are potentially endless. Where do you stop? Can a consensus ever be achieved because, if not, what's the point of trying to communicate anything? Thankfully, a sort of consensus can be agreed, for a while at least, on purely functional grounds. In art we privilege one "reading" of a work over another as it is more telling, it reveals and communicates more, or it makes us think. Similarly in science, an interpretation is retained if it is useful: it brings the observations together nicely, and it indicates a way forward. Matisse drew attention to the problem that a precise image of a place or a person does not necessarily capture the feeling of being there, of knowing them. One hundred and fifty years of photography confirms that view. It cautions us that scientific precision doesn't necessarily yield the truth. We have to go beyond the observations themselves, assembling them into systems; we can call some of these theories; those that become widely accepted we call facts. Perhaps we have discovered a kind of truth. The theory may be useful, and it may survive for a while. That is enough to be going on with.

There were several other people from the conference returning on the same flight as me. I listened to my colleagues during the delay of a couple of hours in the departure lounge, piecing together their impressions of the conference, putting the new data they'd seen into a framework of ideas, different for each of them. By the time we boarded each had a serviceable sketch, personal and biased maybe, but ready to be worked into a clear picture of what we'd learned by the time we got back to our laboratories.

Mark Hanson teaches at University College London Medical School. He is serving on the Wellcome Trust Sci-Art Committee.