17th May 2012

Woroni

The Australian National University student paper since 1948

Micro-Expressions & Depression

 

Ever since I was little, people were able to read my face and tell whether I was lying, embarrassed, jealous, angry, in pain, and so on. While I was amazed by their ability, the real reason for their perceptiveness was that my face was so easy to read, and my eyes revealed unspoken words. To this day, I cannot lie; I’ll embarrass myself if I attempt to. It is too obvious: my voice changes, my eyes role, and my lying smile appears!  Since then I have started to read books on how to translate peoples emotions. I was 15 and you can imagine how difficult it was for me at that age to understand scientific words! Besides, it is difficult to distinguish all these combinations of muscle movements if you don’t have the talent for it.

A few years later (actually more than a few), I saw the TV series Lie to Me, which is based on Dr. Ekman’s studies on emotions. In the show, a professional doctor, Dr. Lightman, was doing exactly what I was trying very hard to do: read emotions from people’s faces. Being a computer engineer, that show inspired me and I thought it would be a breakthrough if we could implement those studies in a computer program to detect the muscle movements that reveal emotions. Googling this I found several researchers who have been doing the same thing for decades which made me feel way behind. However, I decided to be involved in emotion-related research, in particular detecting depression from face, voice, body posture and eye gaze. 

For any emotion-based research, data collection is the most difficult and time consuming part. I consider myself lucky to have the Black Dog institute to collect the data from depressed and non-depressed people. Then the fun part comes in, where we teach the computer to differentiate between depressed and non-depressed subjects based on the data collection process. For example, it has been proven that depressed people have monotonic voices and less facial activity; they lack energy which may reduce the movements in their body, and they avoid eyes contact. So, can the computer understand this? That is my research. In the future, the computer will be able to help depressed people by calling their psychiatrist, or even calling Lifeline for suicidal cases.