A smile is a typical expression we see in our daily lives. Simple as the expression may be, we know that smiling isn’t always straightforward.

We may smile because we are embarrassed or shy. We may smile because we are being polite. Sometimes we smile even when we are angry or insincere. Not all smiles are made to show that we are genuinely happy.

Telling whether or not a smile is genuine may be challenging for some while others may have a knack for it. Even then, the ability to discern if a smile or real or not is a highly subjective affair for humans.

What is there is an objective way of determining if a smile is real or fake — a smile-detecting machine or algorithm, so to speak? 

This is the matter being explored by the University of Bradford in their released study, A genuine smile is indeed in the eyes – The computer aided non-invasive analysis of the exact weight distribution of human smiles across the face.

What did they discover in the study? Let’s find out.

Methodology

The researchers conducted a face analysis of the subjects in a video recording, mapping out the features of the face — the face, mouth, and cheeks and how the movements of these features are distributed during a smile.

This mapping established the computational model for their AI software used to determine whether a smile is fake or genuine.

In order to measure the accuracy of the model in detecting genuine smiles, they used publicly available footages of smiles. They used the software both for video footages where subjects expressed posed smiles and footages where subjects exhibited spontaneous, genuine smiles.

It’s all in the eyes

The smile-detecting software confirmed what was long concluded in past psychological studies — it’s all in the eyes. In their experiments, they have concluded that there is at least a 10% spike in activity in the eye region for genuine smiles when contrasted to posed smiles.

Right now, that might seem underwhelming. For the most part, this is something we are already fully aware of to begin with. However, the implications in the future will be much more significant.

Think about how machine-human interactions will be improved through software which can detect human emotions — through a non-invasive and fast manner  on top of that.

The researchers are also enthusiastic about the potential applications of their findings in the field of biometrics and person identification, given that the movements created whenever we smile is not easily replicable.