Saskatoon

Can a computer tell if you're lying? U of Sask. researcher says facial recognition software can breach privacy

A professor at the University of Saskatchewan says computer programs are getting better and better at analyzing emotions, which is opening up a huge variety of questions.

Companies can use webcams to gather information on everything from jury selection to selling products

Affective computing software can track a person's emotions through the camera in their cellphone. (The Associated Press/Jenny Kane)

A professor at the University of Saskatchewan says computer programs are getting better and better at analyzing emotions, which is opening up a huge variety of questions.

For the last several years, Will Buschert, a sessional lecturer in philosophy and political sciences at the University of Saskatchewan, has been researching "affective computing." That's where software studies facial expressions and predicts the emotional response lying behind them.

Not surprisingly, with the advent of social media and millions of webcams attached to a range of devices, the field is opening up debate, as well as questions about who is using the information — and for what.

"It's almost a who's who of major corporations across the world," Buschert told CBC Radio's Saskatoon Morning. "They use it for things like marketing research, public relations research — figuring out how to sell us things."

For years, researchers have been studying people's facial expressions in order to understand exactly what it is they are thinking. Through a series of microexpressions, researchers are able to tell, for example, whether a smile is genuine or forced by looking at which muscles in the face are being used.

However, advanced computer programs are now able to track these microexpressions with dizzying accuracy, tracking our responses to everything from advertising to videos.

Now, the technology has advanced to the point where the software only needs a web-connected camera, like the ones attached to smartphones, to be able to accurately predict people's emotions.

"It breaks down a final frontier of privacy," Buschert said. "Affective computing technologies can gauge the emotions that you're trying to conceal, the emotions that you're trying to keep private. Frankly, this has enormous implications."

Buschert is most interested in how the software exposes the barriers people put up when they want to disguise the truth.

"Up until now, there was a pretty firm distinction between the emotions you have and the emotions you express," he said. "That is being eroded by the technology."

Affective computing is also being used for therapy. Some of the software is being used to provide an early warning for depression, and was initially used by researchers to study people with developmental delays who couldn't communicate whether they liked certain forms of programming.