Humour

Face to face with IT

T A Balasubramanian on the business of detecting the right emotions

Assigned the role of a friend and guide to Danny DeVito, the first biped walking humanoid, you, Papyrus Bytewala, CIO of Baffle Corporation, continue taking him around the many attractions and distractions at the Techno Over-exposition of Geeks and Gizmos for Lazy Enterprises (TOGGLE). DeVito’s learning curve, you observe, has been swinging up the scale as he absorbs real life experiences like a sponge.

As you enter the grand Virtual Reality pavilion you are greeted by a short masked man with a face that resembles a grinning version of King Kong.

“Hello, Papyrus. Recognize me? I’m your old pal, Gozo Gill.”

“Well, Gozo, with that face, how could I ever mistake you for anybody else?” you say.

“Ha, ha. This is just our promotional tool,” says Gill, removing the mask. “I’m the Chief Marketing Officer from Face Reality Solutions, and this is our stall here. We are known for adopting emerging technologies and applying them with great passion to give our customers unforgettable experiences. Our claim to fame is the new suite of interface products that we call FaceLook, which I will be only too happy to demonstrate to you.”

“Ah, I am sure you will, Gozo. This is my CTO, Danny DeVito, and he is eager to catch up with all the distractions you might want to offer.”

“Oh, sure, it’s my pleasure—Hello, Danny. King Kong and I are pleased to meet you,” Gill guffaws.

“Hey, Kong, you’re a handsome guy,” says DeVito to the mask. “So tell me, what is this all about?”

“If you recall the movie, ‘A Space Odyssey,’ you would have seen how HAL, the computer can—in addition to being capable of making independent judgments and showing intelligence—react to and imitate human emotions. That was science fiction. Now, many smart groups around the world have been racing to match one part of HAL’s capability—with what we call a ‘robust marker-free facial expression recognition tool.’ We are proud to be the first who have succeeded.”

“Wow. So what do you have here?”

“FaceLook. It can analyse your facial expression—your astonished look right now, for example—and accurately determine if you are displaying one of several emotions: happy, sad, angry, surprised, scared, disgusted, or neutral. All you need to use the product is a good quality Web camera, and bingo, your face is read. Emotions are represented as bar graphs and as a continuous signal. An additional dashboard summarizes how negative or positive the emotion is.”

“Ah, so what use can I make of that?”

“Good question, Danny. Spotting the right emotions is big business. Detecting liars is a tricky craft and one that most people—especially humans who are highly motivated to catch liars—are particularly bad at. There are all kinds of potential applications for this wonder product, from home protection to settings like police interrogations, security checks in airports and courtrooms. Everyone is trying to figure out who is telling the truth, and who’s not … we humans are just a wonderful society of happy deceivers pursued by unhappy detectives.”

“And you will be making some of those deceivers unhappy now?”

“That’s right, Danny. With FaceLook tools, you can have the power to collect, analyse and present observational data—mixing facial expression information with event logging data, captured computer screens, physiological signals such as heart rate, and more. For example, an eye-tracker could be used to find out which picture you were looking at when showing the emotion ‘disgusted’. Of course, a human observer would do all this too, but not as reliably. Besides, we think our creation would be far more inconspicuous.”

“Hmm, that sounds creepy, my friend,” you say. “Shades of Big Brother equipped with omnipresence, watching over the crowds, looking for signs of potential agitation?”

“And what if I have a poker face, Gozo?” says Danny, putting on a sombre look. “What if I knew FaceLook was looking and decide to conceal my feelings, eh?”

“Ah, at Face Reality Solutions we have an answer to such situations, too. Do you know that our new research shows your face will betray your true emotion, but not in the stereotypical ways you might think?”

“And what does that mean?” you ask, sceptically.

“It is much more than the shifty eyes or sweaty cheeks or an elongated nose (à la Pinocchio) that FaceLook traps and captures. Other elements of a liar’s look will give them away—‘cracking’ briefly and allowing displays of true emotion to leak on to the face.”

“Hey, isn’t that like a good mother? One who can see right though your angelic look and see that you did the nasty thing that you want to cover up for?”

“That’s right, Danny,” says Gill, beaming. “The human face and its musculature are so complex—so much more complex than anywhere else in our external bodies. There are some muscles in the face you can’t control…and those muscles will not be activated in the absence of genuine emotion—you just cannot fake it. In ‘The Expression of the Emotions in Man and Animals,’ Darwin noted: ‘A man when moderately angry, or even when enraged, may command the movements of his body, but … those muscles of the face which are least obedient to the will, will sometimes alone betray a slight and passing emotion.’ ”

“Woo, that is something, Bozo. So if a human someone is telling a really important lie in which the consequences are dire—say life imprisonment—then the lie will be revealed anyway on FaceLook?” says DeVito.

You look at your CTO with a little concern. A ‘human someone’ is what this humanoid is learning about, and it seems that the lesson is going a little too well. Is it wise putting ideas into DeVito’s program that you may later regret?

“Exactly,” coos Gill, happy to find a rapt listener. “That’s the beauty of FaceLook’s algorithms. Unlike body language, you cannot monitor or completely control what’s happening on your face—even if that face, by virtue of belonging to you, would seem to be totally under your control. Our research has packed into the software the secrets revealed when people put on a ‘false face,’ faking or inhibiting various universal emotions.”

“Amazing, Gozo,” you confess. “And all very educative and enlightening. However, our usual excuse for considering facial recognition in these cases—the application of systems in Baffle that can help us with facial authentication—is to keep the corporate fortress impregnable.”

“But of course,” booms Gozo, without missing a beat. “FaceLook is made for the corporate insecurity mavens. In addition to recognizing the gender, age, ethnicity, wearing of glasses, and ... um, facial hair of a subject, the product can also recognize the humans themselves—a feature made possible by the inclusion of a basic human recognition algorithm. FaceLook can spot an individual following a single, previous input of the original facial image—meaning it not only knows who you are—it also becomes a butler that knows exactly what you’re like before your first cup of coffee.”

“Well, that’s reassuring, Gozo,” you say. “I can see how such a product might be used in the IT world—the study of user reactions to some of our user interfaces, especially when we have planted some horrendous bugs in there.”

“And for those of you working on the next (or first) HAL, FaceLook’s detection capabilities can additionally be accessed by other systems in real-time—so you, too, can program your application to respond immediately to a user’s emotional state,” says Gill, smugly.

“Wow, you guys think of everything,” says DeVito.