BioEd Online

"We must do this more often" was the constant refrain at a gathering of scientists, artists, film makers, designers, writers, editors and art historians at a meeting in Los Angeles. They were there to explore the use of images in science, for both understanding data and communicating it to others.

Image and Meaning 2, held from 23 to 25 June at the Getty art museum, was the successor to the first conference of this sort, staged in 2001 at the Massachusetts Institute of Technology (MIT) in Boston.

The conference series is the brainchild of Felice Frankel, a science photographer working at MIT. Frankel helps scientists present their work using imagery that is both informative and striking. Her photographs have graced many covers of Nature and Science.

Frankel convened the meeting because, she says, "we have a serious problem. There is an assumption that in science our graphics communicate. But they often don't." Frankel argues that many scientists don't see imagery as an integral part of the scientific process. This, she says, is doing the community a disservice.

Animated discussion

Many at the conference tackled the notion of how best to find patterns within their data. Creating graphs or maps becomes more challenging as the data get more complicated. Some have turned results into animations to wring meaning from them, whereas others have displayed their work using virtual reality 'rooms'.

There is now a wealth of computer tools for generating sophisticated imagery. But scientists at the meeting complained that many of these programs never get beyond the computer-graphics community that develops them.

The audience was impressed when David Salesin, a computer scientist at the University of Washington, presented the interactive visual tools that he is developing for Microsoft. Salesin showed software that can construct realistic-looking aerial photographs from maps after being trained with a few real photo/map combinations. He also had programs that could blend different faces, and automatically turn random objects into 'Escher tiles': these are shapes that can be rotated to fill a space without leaving any gaps (see some Escher tiles here).

Salesin's methods got many researchers thinking about new ways of looking at their findings. But they also raised a host of questions about proper limits to the manipulation of images. "I had not realised how easy it is to make changes in images seamlessly," says chemist George Whitesides of Harvard University, Cambridge, Massachusetts. He says the community should work out rules for what is acceptable in 'improving' images for publication.

Simple scribbles

On the opposite side of the coin, many argued that low-tech imagery can have advantages over high-tech in communicating research. A very polished image, says Frankel, can discourage a viewer from engaging with and thinking about a work. "If something is raw, it gives the viewer permission to participate," she says. Beautiful, computer-generated renderings can also mislead viewers, some argue, giving a false sense of certainty about the ideas they represent.

Cartoonist and self-described "lapsed mathematician" Larry Gonick argued that cartoons can be used to great effect in communicating science without having to turn to expensive or complicated techniques. "Cartooning's graphic style has certain features conducive to explanation," argues Gonick. "It invites the eye and draws the reader to the 'main character' of the illustration," he says. It also comes with a widely understood visual language, he argues, that can convey motion or narrative.

"Even the biggest-brained scientists at this conference have confessed an inability to comprehend some complex diagrams," says Gonick. Sometimes, he says, what one needs to understand a concept is a story.