A new exhibition at the British Library – its first ever science-based display – showcases some of the most striking visualisations of scientific data. One of the highlights is Florence Nightingale’s rose diagram, the graphic showing how dirty hospitals were killing more soldiers than the Crimean battles that had put them there.
Visualisation has always been vital to scientific progress. We are far better at spotting patterns or anomalies in pictures than in tables of numbers. The economists Arthur Briggs Farquhar and Henry Farquhar summed this up in 1891. “A heavy bank of figures is grievously wearisome to the eye,” they wrote, “and the popular mind is as incapable of drawing any useful lessons from it as of extracting sunbeams from cucumbers.”
When Crick and Watson were struggling to work out the structure of DNA, it wasn’t a table with a list of atomic co-ordinates that gave them the insight they needed; it was Rosalind Franklin’s X-ray crystallography. Her images provided a pictorial interpretation of the way DNA’s molecules are arranged in a double helix, and garnered Crick and Watson (though not Franklin) a Nobel prize.
In an era when data is cheap and plentiful, visual analysis plays an important role. There are pitfalls to pretty pictures, though. Data can be represented in various ways, and someone somewhere makes a choice. Sometimes, the chosen representation can obscure as much as it reveals.
Take a 1951 graphic showing the efficacy of three antibiotics on 16 kinds of bacteria. The way the designer chose to show the information emphasised the comparative effectiveness of the drugs. But the power of this representation masked a scientific insight. If it had been presented slightly differently, it would have been obvious that one of the bacteria had been classified wrongly. It took another 33 years for this oversight to be discovered, delaying effective treatments for some infections.
Today, we should be cautious of the brain scanning technique known as functional magnetic resonance imaging, or fMRI. This provides a way of comparing blood flow in various parts of the brain as a person thinks about specific things – perhaps a moral quandary, the face of a loved one, or a childhood memory. It seems that different types of thought cause blood flow to increase in some regions of the brain and to decrease in others.
Researchers hope that this technique will offer a way to read people’s minds; already in US courtrooms, it is being held up as a way of detecting lies. The trouble is, the reference fMRIs are usually an average over many individuals – and are gathered in artificial situations, such as students being paid to sit in scanners and tell lies. It remains unproven whether you can infer anything reliable from one person’s fMRI scan.
We need not even see the picture to be fooled. Research carried out in 2008 showed that people were more likely to believe a statement prefixed with “Brain scans indicate . . .”.
This was true even when the people observed were neuroscience students and the statements were scientifically flawed. What’s more, the pretty colours of fMRI scans make it easier for them to bypass the critical faculties of our picture-loving brains. Show a jury a picture of a scan, and they see it as scientific and convincing. In some fields, for all the help they are to many scientists, the road to hell is paved with visualisations.
“Beautiful Science” is at the British Library, London NW1, until 26 May