What happens when seeing isn’t believing?

When scientists need to communicate research findings to professional and lay audiences, they typically employ some form of data visualization, such as charts, posters, and other media. Their goal is to illustrate and describe the findings in a way that is both clear and compelling. This often requires a balance between giving a user sufficient information to understand a phenomenon but not so much that it becomes overwhelming. This reflects the concept of abstraction, a core idea in the field of computer science. According to Dr. Arvind Satyanarayan, associate professor of Computer Science at MIT, “As readers have limited time, attention, and expertise, visualization authors need to decide which aspects of a complex dataset to focus on, and how to further simplify it for easy consumption. Audiences are rarely aware that these decisions have been made on their behalf.”

In the current climate, the spread of misinformation and the trend of mistrusting scientists has made scientific communication tricky. Given that the researcher is making decisions about what information to display, why should the audience trust the information as it’s presented? To Dr. Graham Jones, linguistic anthropologist at MIT, this brought to mind a subject he has studied extensively: magic. “Like scientists assembling data visualizations, magicians make decisions about what to reveal and what to conceal. A key difference, though, is that audiences approach a magic show with the foreknowledge that the magician will deceive them.” 

Dr. Jones says, “Misinformation and disinformation are often based on abuses of trust, like a con game. We are constantly surrounded by a din of information vying for our attention. How do we make decisions about what information is trustworthy?” Together with Dr. Satyanarayan, he hopes to explore how insights from visual communication strategies used in performance magic might be applicable to creating data visualizations that open channels for even skeptical audiences to engage productively with scientific findings.

Under no illusions

The challenges facing the team—which includes postdoc Dr. Michelle Morgenstern—are many. First, the project represents an unusual mix of linguistic anthropology and human-computer interaction; lacking sufficient precedents, the project is unlikely to attract traditional sources of funding. Computer science funders, in particular, would likely be unsupportive of the central role of magic in the research plan. In fact, the team intends to create and consult closely with an advisory panel comprising professional magicians who will also be asked to take part in public events for the MIT community.

Pick a cue, any cue

The team intends to design a series of metacommunicative cues that will be displayed on, within, or around the ultimate visualization, and examine their effect on readers’ awareness of what kinds of information might be concealed from view. Following an iterative design prototyping process, the team will draw on the expertise of magicians, who will offer feedback on the efficacy of the cues and suggest refinements or alternatives. When the approved cues are ready, the team will conduct a series of user studies to determine how effectively the cues help to build trust and increase insight into the information being presented.

Ultimately, the team hopes to create new guidelines for how to design charts in anticipation of adversarial readings and audiences, as well as develop a new generation of visualization tools that account for oppositional stances. Additionally, says Dr. Jones, “One of our bigger ambitions is to chart a new way forward for collaborations between MIT’s School of Humanities, Arts, and Social Sciences (SHASS) and the College of Computing. As computer scientists and anthropologists, we hope to serve as a model of collaboration to inspire new relationships and lines of inquiry into these kinds of topics.”