Can the vast range of our facial expressions be recorded for the computer to understand? Well, researchers are trying for sure and they could be nearly there. A Ohio State University research team has apparently managed to teach computers to understand 21 various human emotions from corresponding facial expressions.
Experts have always sought to decode the mechanisms behind facial expressions that denote emotions, even back in the period of the early Babylonian civilization and Classical Greece. The Physiognomonics treatise, which some attribute to Aristotle and others to an author sometime in 300 BC, is a major work on physiognomy which is the attribution of a person’s character from his facial, and also his overall physical, appearance.
The World of Physiognomy
While this branch of ancient study was considered more an art or pseudoscience, it has seen more dedicated and scientific research in the 20th and 21st centuries with cognitive scientists now tracing facial expressions to chemicals, neural pathways – that cause the brain to go through emotion, and genes. However their studies have been restricted to only six primary emotions which are happiness, sadness, anger, fear, disgust and surprise since these facial expressions were thought to clearly indicate the respective emotions behind them.
Twenty-one Facial Expressions Recorded
The Ohio State University experts have, however, managed to document 21 emotions by blending various combinations of emotions. The corresponding facial expressions were then coded resulting in 21 emotions which the computer can now understand. Prof. Aleix Martinez led the research.
For the study, Dr. Martinez and his team had to photograph 230 volunteers of whom 100 were male and the rest female. They were told to provide the appropriate facial expressions for various verbal cues. The resultant 5000 photographs were tagged based on Paul Ekman’s FACS (Facial Action Coding System).
How Facial Action Coding System Helped
The FACS data enabled researchers to carry out cross-referencing of similarities as well as dissimilarities in expressions among the 230 individuals. Martinez and his team were able to record 21 emotions in total, many of which found similar expression in all the people. The expression that was similar in almost all the people was happiness. Around 99% of the individuals studied smiled by stretching their mouths to denote happiness.
According to Prof. Martinez, research such as this has massive therapeutic applications as this model could be used for treating conditions such as post-traumatic stress disorder (PTDS) and autism. PTSD is a really difficult disorder to treat. It can be reflect in an individual in different ways and Psychiatrists who treat these patients can have better ways of documenting. While working with patients with PTSD, billing of the services are difficult and studies such as this with proper documentation via good transcribed notes or psychiatric transcriptions is critical in managing the reimbursements.