How are you feeling right now?
Soon, your Wi-Fi router may be able to tell.
Computer scientists have developed a new technique to measure human emotions wirelessly. CBC Radio technology columnist Dan Misener explains how.
How can emotion be measured using wireless signals?
It's all about measuring the human body, and the physiological signals that change along with our emotions. In the past, this has been done with electrocardiography (ECG) monitors — electrodes strapped to the body.
But a team at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) has built a device that can do this without wires. They call it EQ-Radio, and in a lot of ways, the device is similar to a Wi-Fi router.
It sends out a wireless signal. When the signal reaches a human body, it bounces back into the device. By measuring these reflections — the way the human body interacts with the wireless signals — the device can tell if you're excited, happy, angry or sad.
That's the major breakthrough here (and possibly the creepy part) — the ability for a computer to recognize emotion at a distance, without wires — and potentially, without you knowing about it.
What exactly is the device measuring?
MIT professor Dina Katabi led the project. She and her team looked at two factors in particular.
"The heartbeat and the breathing. This kind of vital sign, as you would expect, is related to our emotion — he's now angry, he's now happy, he's sad. And all of that without touching the person's body," she said.
Small variations in our heart rate and breathing can reveal emotions that aren't obvious on the outside, Katabi said. For example, we might smile even if we're not happy.
But the team at MIT believes their device can measure "inner emotions," because it taps into the autonomic nervous system.
How accurate is it?
The team reported they could correctly classify emotions 87 per cent of the time. That means their wireless system is on par, accuracy-wise, with existing ECG-based systems, which require electrodes on the skin. They also say it's more accurate than image-based systems, which use cameras to examine the emotions on your face.
The high degree of accuracy is significant, because there are many situations where you might want a computer to recognize human emotion, but you don't want to strap a bunch of sensors to your subject.
For instance, Katabi imagines her technology being used in video games,when testing movies, in mental healthcare, or in the advertising industry.
Imagine you're walking through a supermarket that can wirelessly measure your emotional state. If it knew you were having a bad day, it could advertise ice cream or chocolate.
What are the privacy concerns around this?
Because it's wireless, this technology could be used to recognize your emotional state without your knowledge. That's why Katabi says the device her team designed has a built-in consent mechanism, based on the fact the system understands motion.
"It will ask the person to do certain motions and movements. And if that person doesn't do them, it knows that that person did not give consent. So it would not give you the information," she said.
"Like any technology, people should not use it without asking for consent."
That said, there are probably all sorts of situations where a business or organization might want to perform emotion recognition without consent. And that's the big privacy concern — surreptitious emotion recognition.
What's next for emotion-sensing computers?
The MIT team plans to present their work on EQ-Radio at a mobile computing conference next month.
But wireless emotion recognition is just one of many ways computer scientists are trying to teach machines to identify human emotions. This is sometimes called "affective computing."
The image-based systems mentioned earlier are an example. Or audio-based systems which use microphones to hear emotional cues in your voice.
Now, wireless measurement of heart rate and breathing has been added to the mix.
Katabi says the next step will likely mean combining these technologies for even greater accuracy. With enough miniaturization, you could even build a very robust emotion detection system into a smartphone.
Most phones already have cameras, microphones, and wireless capabilities. It's a matter of refining the algorithms, and getting the measurement electronics small enough.
The question then becomes: when is it appropriate (or polite) to pull out your phone to see how someone really feels?