“`html
The globe is saturated with data representations, ranging from graphs that accompany news reports about the economy to diagrams monitoring weekly temperatures to scatterplots illustrating correlations between baseball metrics.
Fundamentally, data representations communicate knowledge, and individuals digest that knowledge in various ways. One individual might skim the axes, while another could concentrate on an anomalous data point or scrutinize the height of each colored bar.
But how do you interpret that knowledge if it’s beyond your sight?
Rendering a data representation accessible to blind and visually impaired readers often entails composing a detailed description that encapsulates several essential aspects in a concise paragraph.
“However, that implies blind and low-vision readers lack the opportunity to analyze the data independently. What if they have a different inquiry regarding the information? Suddenly, a straightforward caption fails to provide that. The fundamental concept behind our team’s efforts in accessibility has been to preserve autonomy for blind and low-vision individuals,” states Arvind Satyanarayan, a recently tenured associate professor in the MIT Department of Electrical Engineering and Computer Science (EECS) and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL).
Satyanarayan’s team has investigated methods to make data visuals usable for screen readers, which vocalize content displayed on computer screens. His team developed a hierarchical system that enables screen reader users to navigate different details within a visualization using their keyboards, transitioning from broad insights to specific data points.
Within the sphere of human-computer interaction (HCI) research, Satyanarayan’s Visualization Group also creates programming languages and authoring tools for visuals, examines the sociocultural aspects of visualization design, and employs visuals to explore machine-learning models.
For Satyanarayan, HCI is about fostering human autonomy, whether it’s empowering a blind user to discern data patterns or ensuring designers maintain control over AI-fueled visualization systems.
“We truly adopt a human-centered philosophy toward data visualization,” he remarks.
A knack for technology
Satyanarayan stumbled upon the realm of data visualization almost unintentionally.
As a youth growing up in India, Bahrain, and Abu Dhabi, his early fascination with science originated from his passion for experimentation.
Satyanarayan remembers his father bringing home a laptop, which he filled with basic games. The internet advanced alongside him, and as a teenager, he became deeply involved in the famous blogging platform Movable Type.
A natural educator even in his teenage years, Satyanarayan provided tutorials on how to utilize the platform and organized a contest for individuals to customize their blogs. Throughout this journey, he self-taught himself the capabilities to develop plugins and add-ons.
He relished creating visually appealing and user-centric blogs, laying the groundwork for his exploration in human-computer interaction.
Upon arriving at the University of California at San Diego for his undergraduate studies, he was sufficiently intrigued by the HCI field to enroll in an introductory course.
“I’ve always been a history enthusiast, and this introductory course genuinely fascinated me because it focused on the historical evolution of user interfaces, tracing their origins and development,” he explains.
Almost as an afterthought, he conversed with the professor, Jim Hollan — a forerunner in the domain. Even though he hadn’t previously considered diving into research, Satyanarayan spent the summer in Hollan’s lab, investigating human interaction with wall-sized displays.
As he geared up to pursue graduate studies (Satyanarayan split his doctoral work between Stanford University and the University of Washington), he was indecisive about whether to concentrate on programming languages or HCI. When the moment to decide arrived, the human-centered focus of HCI and the interdisciplinary nature of data visualization drew him in.
“Data visualization is incredibly technical, but it also integrates aspects of cognitive science, perceptual psychology, and visual arts and aesthetics, with significant implications for civic and social duty,” he notes.
He recognized the role visualization plays in civic and social accountability during his inaugural project with his PhD advisor, Jeffrey Heer. Satyanarayan and his collaborators designed a data visualization interface for journalists in newsrooms lacking the resources to hire data teams. That drag-and-drop tool empowered journalists to create the visualizations and narratives they wanted to communicate.
This project planted many ideas that later formed his thesis, wherein he examined new programming languages for visualization and constructed interactive graphical systems utilizing them.
After obtaining his PhD, Satyanarayan searched for a faculty position and endured an intense interview season traveling across the country, attending 15 interviews in just two months.
MIT was his final location.
“I remember feeling drained and operating on autopilot, thinking that this wasn’t going well. However, the first day of my interview at MIT was filled with some of the most enriching discussions I had. Individuals were so enthusiastic and curious about comprehending my research and its connections to theirs,” he recounts.
Mapping a collaborative journey
The collaborative essence of MIT remained crucial as he established his research group; one of the group’s pioneering graduate students was pursuing a PhD in MIT’s program in History, Anthropology, and Science, Technology, and Society. They continue to collaborate closely with faculty examining anthropology, humanities topics, and clinical machine learning.
With interdisciplinary collaborators, the Visualization Group has investigated the sociotechnical consequences of data representations. For example, charts are often shared, disseminated, and discussed on social media, where they lose their contextual background.
“What occurs as a result is they can become conduits for misinformation or misunderstanding. However, this is not due to poor initial design. We invested significant time analyzing those particulars,” Satyanarayan states.
His group is also examining tactile graphics, typical in museums to assist blind and low-vision individuals interact with displays. Often, producing a tactile graphic boils down to 3D-printing a chart.
“Yet, a chart was intended to be observed by our eyes, and our eyes function very differently than our fingers. We are now delving deeper into what it entails to create tactile-first visualizations,” he adds.
Co-design is a fundamental principle behind all his group’s accessibility initiatives. In numerous projects, they collaborate closely with Daniel Hajas, a researcher at University College London who has been blind since age 16.
“This has been tremendously vital for us to ensure that as sighted individuals, we are developing tools and platforms that genuinely benefit blind and low-vision users,” he states.
His group is also investigating the sociocultural implications of data visualization. For instance, during the height of the Covid-19 pandemic, data visuals were frequently transformed into memes and social artifacts used to endorse or contest information from experts.
“In truth, neither data nor visuals are neutral. We’ve been contemplating the data you use for visualization, and the design decisions behind specific visuals, and what those communicate beyond insights about the information,” he mentions.
Illustrating a real-world difference
Interdisciplinarity is also a recurring theme in Satyanarayan’s interactive data visualization course, which he co-facilitates with faculty members Sarah Williams and Catherine D’Ignazio from the Department of Urban Studies and Planning; and Crystal Lee from Comparative Media Studies/Writing, with shared appointments in the School of Arts, Humanities, and Social Sciences and the MIT Schwarzman College of Computing.
In the popular course, students not only acquire the technical expertise to create data visuals, but they also develop final projects focused on social significance. Over the past two years, students have concentrated on the housing affordability issue in the Boston area, collaborating with the Massachusetts Area Planning Council. The students appreciate the chance to make a tangible impact with their work, Satyanarayan notes.
And he relishes the course just as much as they do.
“I adore teaching. I genuinely enjoy engaging with the students. Our students are so intellectually inquisitive and dedicated. It reassures me that the future is in capable hands,” he shares.
One of Satyanarayan’s personal hobbies is jogging along the Charles River Esplanade in Boston, which he does almost daily. He also takes pleasure in cooking, particularly with unfamiliar ingredients.
Satyanarayan and his wife, who met while pursuing graduate studies at Stanford (her PhD is in microbiology), also delight in cultivating their plot in the Fenway Victory Gardens, brimming with lilies, lavender, lilacs, peonies, and roses.
Their latest addition is a miniature poodle puppy named Fen, acquired when Satyanarayan received tenure earlier this year.
Looking ahead at the trajectory of his research, Satyanarayan is eager to further explore how generative AI may effectively aid individuals in creating visualizations, and its ramifications for human creativity.
“In the realm of generative AI, this issue of agency pertains to all of us,” he states. “How do we ensure that, amid these AI-driven systems, we’ve retained the aspects of the work we find most engaging?”
“`