Most of us share our thoughts, ideas, and beliefs online, and the sheer volume of words now floating in the public sphere
presents an unprecedented opportunity for psychologists who study language. What we say and write can reveal more about us than our words alone convey: Within the vast trove of communiqués in cyberspace lie patterns that can provide rich insights into how our minds work. But psychologists who study language and social cognition have struggled to make sense of such huge amounts of data using traditional methodologies. In the field of computer science, recent advances in machine learning have begun to produce tools that could be used to mine these datasets. An SFI working group, which met online in April, brought together psychologists and computer scientists to explore how the two fields can collaborate.
“You could have all the content in the world, but if you don’t have methods [to analyze it], you’re stuck,” said working group organizer and External Professor Mahzarin Banaji, a psychologist at Harvard. “But we do now.”
The collaboration will focus on “the coming together of these large sets of data that contain human expressions of language and methods that are being designed by computer scientists — algorithms — that plow through these massive amounts of words to be able to tell us something about these hidden patterns,” she added. At the same time, psychologists will share their own methods for analyzing text, which could help computer scientists refine their algorithms. “We’re trying to jump-start a collaborative, interdisciplinary approach, to be able to speak to each other.”
In addition to Banaji, the 11-member working group included researchers with backgrounds in computer science, psychology, brain science, and social dynamics. Among them: Aylin Caliskan, a computer scientist at George Washington University who studies machine learning; SFI Professor Mirta Galesic; and psychologist Jamie Pennebaker of the University of Texas at Austin.
The idea for the working group came to Banaji, who is known for her research on implicit bias, when she read a paper about machine learning. The study got her thinking about what psychologists might be able to learn from computer scientists, especially when it comes to unearthing the layers of meaning embedded in the torrent of words flowing across the Internet.
“Although I don’t study language, I’ve always been interested in how language reflects our beliefs and values and on the other hand how we use language to shift the way we think,” Banaji said.
Originally scheduled to meet in Santa Fe, the working group, “Language as a Window into Human Minds: Explorations with Computer-Resident Language and Naturalistic Conversation” was held online on April 23.