"Emotion" is the last thing any scientist or design engineer wants to deal with, especially when it comes to developing computing systems.
"Oh, dear," Rosalind Picard, professor at the Massachusetts Institute of Technology (MIT) Media Laboratory, remembers muttering to herself, when it first became unavoidably clear to her that “emotion is vital to intelligent functions.” Picard was then working on machine learning systems.
"Scientists want to be rational. We develop machines that decide right or wrong in terms of 1’s and 0’s," said Picard. The first instinct about “emotion” among scientists and engineers is to roll their eyes at the very idea of feelings having any role in problem-solving, logical thinking or reason. But ignoring emotions is probably no longer an option for most scientists. Look closely, Picard said. Emotions play an important role in human intelligence, rational decision making, perception and learning.
"Emotions are necessary for intelligent day-to-day functions," said Picard in a recent interview with
EE Times. Behind what matters [to humans] emotionally, there is "a mass of information" to which computing systems today are totally blind, said Picard.
How can we incorporate emotions into models of intelligence, more specifically, in computers? And how can we make machines that pay more attention to our human’s affect? These questions led her to her seminal work on “Affective Computing.”
Picard (left), the pioneer of affective computing, is a keynote speaker at the Design East Conference, scheduled on Wednesday, Sept. 19th. During her keynote speech, she plans to demonstrate several applications developed by taking advantage of initial findings in the affective computing field. One example is a brand new iPhone app, called Cardio, released last Thursday (Aug. 16).
Developed by a team of Picard’s students at MIT, the app detects one’s heart rate through facial skin color changes, as the user looks at the iPhone camera.
Picard noted that the development of affective computing is an "evolving process." The research challenge still has a distance to go, as she acknowledged that “this isn’t the field where we can simply put a USB plug and read out emotions.”
Picard, however, made it clear that the mission for affective computing is not about making computers "cute or endearing." This is about “making machines function much more intelligently.”
Beyond reading a lot of neuroscience books and learning how the visual cortex works, Picard is involved in breaking down basic building blocks of human perceptions and emotion, identifying key emotional triggers. Picard’s silver lining is that she’s not alone. "We now have a large number of engineers, researchers and psychologists involved in the [affective computing] field."
Computer visionComputer vision is perhaps one of the first fields helped by affective computing. While humans can instinctively sense what matters to them most, balance reason with emotion (or not) and make decisions on where to concentrate their attention, a computer-vision system tends to survey all visual input objectively, with equal focus. Then, it needs to run a lot of data processing on every item and movement in its field of view before deciding what’s important, what’s not and what to do.
But what if the computer vision system knows it should pay its primary attention to human facial expressions or slight changes in skin color? Feeding computers certain algorithms that help decode human emotion and physiological clues could help computers extract the information that matters most -- more quickly. Success in this area could have major implications for computer vision applications ranging from toys and airport security to video surveillance systems and equipment used in hospital rooms, explained Jeff Bier, co-founder and president of Berkeley Design Technology, Inc.
Next: Fouding of startup Affectiva