Facial expressions are not enough.
MIT CSAIL team. Image: Jason Dorfman
If you've ever dated an introvert you know that guessing someone's emotions can be very risky. But soon, you may have more clues: A new device can detect people's emotions using wireless signals. Designed by MIT's Computer Science and Artificial Intelligence Lab, "EQ-Radio" measures heartbeat and breath to determine if someone is excited, happy, angry, or sad.
"Our work shows that wireless signals can capture information about human behavior that is not always visible to the naked eye," said MIT professor and project lead Dina Katabi. "We believe that our results could pave the way for future technologies that could help monitor and diagnose conditions like depression and anxiety."
EQ-Radio measures emotions with 87 percent accuracy. The device uses wireless signals that reflect off people's bodies in order to measure their heartbeats with the same accuracy as an ECG monitor. An algorithm breaks the reflections into individual heart beats to study slight variations. It analyzes waveforms from within each heartbeat to match a person's behavior to how they acted previously in one of the four emotional states (joy, anger, excitement, sadness).
"Just by generally knowing what human heartbeats look like in different emotional states, we can look at a random person's heartbeat and reliably detect their emotions," said PhD student Mingmin Zhao, who co-wrote the paper on EQ-Radio.
For example, a person who signals "low arousal and negative affect" would likely be tagged by the device as sad, while "high arousal and positive affect" would be tagged as excited. To test and train the device, subjects listened to certain music or watched certain videos to evoke memories of the four emotions.
The team had to ensure the device didn't conflate heartbeat movements with breathing movements, since the latter is much more pronounced physically than the former. To make sure they don't get confused, the wireless signals are based on acceleration instead of distance traveled by the chest (rise and fall). According to researchers, this method could also be applied to non-invasive heart monitoring and diagnostics.
"By recovering measurements of the heart valves actually opening and closing at a millisecond time-scale, this system can literally detect if someone's heart skips a beat," said PhD student Fadel Adib, who co-wrote the paper. "This opens up the possibility of learning more about conditions like arrhythmia, and potentially exploring other medical applications that we haven't even thought of yet."
The device could also extend beyond health. It could be used to measure consumer behavior, or for entertainment, according to Katabi. Ad agencies or film studios could use the device to test viewers' emotional reactions in real time, or smart homes could use it to detect information about mood and make or suggest adjustments in heating, lighting, or air quality. A "smart" environment could detect emotional states like depression, for instance, and inform us or encourage us to get help.While only four emotions don't cover the range of human feeling, it's a start for detecting emotions and other spin-off applications useful for mental and physical health, as well as entertainment.
Get six of our favorite Motherboard stories every day by signing up for our newsletter.