Human memory is very fallible, but lately cognitive scientists have found that our minds capture much more visual detail in a moment than once believed. A 2008 paper reported that people who saw thousands of images for three seconds each over five hours later identified ones they'd seen over similar alternatives with nearly 90% accuracy. They didn't just remember that they'd seen a cracked egg, they remembered that its egg white had formed a perfectly round puddle.
In other words, when we do retrieve a memorable image, a surprising amount of information comes with it, like a burr stuck to a sweater. That insight could have big implications for people who use visualizations in their everyday lives--graphic designers, for instance, or anyone on Tumblr. Above all, it suggests that memorability alone might enhance an infographic's effectiveness. But it also prompts a question: How does an image become memorable in the first place?
Doctoral student Michelle Borkin of Harvard's School of Engineering and Applied Sciences recently asked herself just that. To find an answer, she and several collaborators collected more than 2,000 informative images from a range of publications and websites for what became the "largest scale visualization study to date." The idea was to see which images were memorable in an intrinsic, automatic way that Borkin describes as "pre-attentive."
"You're not even consciously thinking about what this graphic is about," she tells Co.Design. "This is a low-level thing. You see it and things are going to stick or they're not."
The researchers first created a "visual taxonomy" of the images they gathered to determine which elements stuck most in people's minds. Then they chose 410 representative images and ran an online experiment where viewers looked at the series and had to indicate whenever they'd seen an image before. Repeat images always occurred between 91 and 109 images later, to control for the fact that timing can influence memorability.
The most memorable visualizations, by far, contained elements that fell under the category of "human recognizable objects." These were images with photographs, body parts, icons--things that people regularly encounter in their daily lives. "Human recognizable objects will instantly make it more memorable," says Borkin. All but one of the 12 most memorable images in the study had a recognizable component.
Beyond that, Borkin and company discovered a few secrets to infographic success. Color was key; visualizations with more than six colors were much more memorable than those with only a few colors or a black-and-white gradient. Visual density--what some of us might call "clutter"--wasn't a bad thing either. In fact, images with a lot going on were significantly more memorable than minimalist approaches. Roundness was another hallmark of memorability (after all, our brains do love curves).
On the contrary, basic bar graphs and charts were easy to forget. That surprised Borkin at first, who thought these visualizations might be memorable because people become familiar with them at an early age. Instead, they proved too similar to distinguish, eliciting a high false-alarm rate--meaning study participants thought they'd seen already them when they had not. As Borkin and her collaborators put it: "all the bar charts look alike."
"Every bar graph has the same format," she says. "These other visualizations which prove to be more memorable each has more variety in its layout and structure."
The findings--detailed in a paper presented at a conference last month--refute a common perception that a busy graphic is a bad one. Such "chart junk," as visual expert Edward Tufte has called it, can actually help an image stick in our minds. One of the most memorable graphics, Borkin says, depicted the phylogenic evolution of dinosaurs with images of the creatures that perhaps weren't entirely necessary. Yet it's hard to see such elements as excessive if they help people lock the graphic into mind.
Then again, the researchers emphasize that this study only scratches the surface of what makes a visualization effective. Borkin and the others didn't study how well people retained the information in the images, just that they retained the image itself. An image that's memorable without being comprehensible may not be worth much. Borkin has already moved on to a similar study of visual comprehension, and she suspects in this case that "chart junk" and extraneous design elements will have a negative impact.
"Icons, images, and human-recognizable objects will instantly make [a visualization] more memorable," she says. "But there's this very careful caveat--and this is me speaking as a viz design person: make sure they're helping your reader understand the main point of your data."