Scientists Put Human Interaction Under the Microscope
CMU-Q, Robotics Institute investigate what makes a successful negotiator
The video above offers an introduction to the Panoptic Studio.
By Angela Ford
Researchers at Carnegie Mellon University in Qatar and Pittsburgh have joined forces to explore the behavior of successful negotiators, very, very closely.
In the study, 10 students from CMU-Q engaged in a mock trading exercise in the Panoptic Studio at Carnegie Mellon's Robotics Institute in Pittsburgh, a two-story geodesic dome fitted with nearly 500 synchronized video cameras to capture the fine details of human interaction — facial expressions, eye contact, hand gestures and subtle body language.
Hanbyul Joo and John O'Brien inside the Panoptic Studio.
"Negotiation is the essence of business," said John O'Brien, associate dean at CMU-Q and associate professor of accounting. "It's in every transaction."
O'Brien brought the business-minded students to the Pittsburgh campus last year after they had won CMU-Q's Algorithmic Trading Hackathon, a competition in which students from different disciplines created algorithms for equity markets.
"They were already thinking about using technology for problems in business, and the dome seemed like an excellent tool," he said.
O'Brien approached Fuad Farooqi, assistant professor of finance, with the idea. Together they turned to Yaser Sheikh, associate professor of robotics, who specializes in dynamic motion reconstruction and human behavior analysis, and two robotics doctoral students, Hanbyul Joo and Tomas Simon.
"This was an opportunity to study negotiations at a higher level," Farooqi said.
The hundreds of cameras in the dome captured the negotiators' every little movement, generating a half terabyte of data from the cameras and audio feeds each minute.
"We were concerned that the dome would not be able to handle 10 negotiators, but it handled it just fine," O'Brien said.
O'Brien and Farooqi's team are analyzing the visuals, and they will feed the information back to Sheikh in Pittsburgh for machine learning purposes. Initially, some of the casual data suggests the more active traders and the students who initiated transactions performed better as did those who maintained eye contact.
"If we can understand what behaviors lead to success, we can train business students to be better negotiators," O'Brien said. "Down the road, these insights could lead to augmented reality applications."
Based on the pilot study, the team is planning a larger study on negotiation this summer.
Joo said the Panoptic Studio also is working on several campus collaborations including one with the Department of Psychology.