NSF grant to fund project on behavioral psychology

The results of a project that combines computing technology with massive data collection and analysis could help scientists better understand the connections between the brain and speech as well as other aspects of communication and behavior.

Initiated by the Psychology Department, the project is a test bed for a new approach to research in the social sciences and is being funded with a two-year, $2 million grant from the National Science Foundation.

“Much of the research in social sciences now consists of single researchers working in labs, isolated from others. This creates silos of knowledge,” said Bennett Bertenthal, Professor in Psychology and principal investigator for the project.

“What we intend to do is to gather data in a wide variety of ways, collecting information on speech and gesture and facial expression, for instance—and make that data available via the Web to researchers around the world for analysis,” he said. Because scholars from a wide variety of institutions and a large number of disciplines will be able to access the same data, new insights will develop as scholars investigate the interaction of various human responses that were previously examined in isolation.

The $2 million grant is one of two the NSF made as the first awards in its Next Generation Cybertools program, an initiative designed to extend the boundaries of social and behavioral research and lead to fundamental advances in cyber-infrastructure.

The initiative will serve two purposes. First, it will help social and behavioral scientists push their research through the use of vast new webs of computers, networks and data resources that are becoming increasingly important to science as a whole, and to the activities of NSF in particular. Second, the scientists’ efforts will guide the development of future computational tools, which will advance cyber-infrastructure itself.

“Over the past 15 years, the social and behavioral sciences have arguably been most transformed by new cyber-infrastructure,” said David Lightfoot, head of the NSF’s Directorate for Social, Behavioral and Economic Sciences. “Many areas have undergone dramatic changes in the kind of research that has become possible. The goal of this initiative is to explore further innovations that will continue to drive new research possibilities in scale and detail.”

The Chicago project, which also includes researchers at Argonne National Laboratory and the University of Illinois at Chicago, will develop tools for collecting and analyzing human behavioral data on an unprecedented scale and level of sophistication, Bertenthal said. In a “SuperLab,” a multidisciplinary team of researchers will be able to track human behavior in both individual and group settings, while collecting exquisitely detailed data on the participants in real time, he said.These data, in turn, will help the Chicago group address research questions that far exceed the capacity of any laboratory today, Bertenthal said. Those questions include: How is social behavior correlated with the participants’ neural activity, for example? How is it connected with their movements, postures, gestures, facial expressions and speech — or for that matter, their state of development, environmental context and cultural norms?

Central to the Chicago group’s effort will be the creation of a distributed data warehouse known as the SID Grid (Social Informatics Data Grid), which will encourage data sharing and accelerate the development of standards for collecting and coding physiological and behavioral data.

The SID Grid will be deployed as part of the larger TeraGrid, a suite of grid- computing resources available to the scientific community at large. The Chicago group will develop, test, and refine their data collection and analysis tools through research in three areas of inquiry: multimodal communication, neurobiology of social behavior, and cognitive and social neuroscience

“We see many other uses for the tools we will create,” said Bertenthal. The efforts of the researchers will show how information on human behavior can be automatically extracted, and even interpreted, from media such as audio and video recordings.

“Such research may open the way toward mining the truly vast amounts of data on human behavior that are recorded every day,” he said.