Creating a Human-Computer Interactive Learning Environment for Learning to Recognize Facial Emotions, based on Learning Methods used in Art Drawing and Psychology
What is the best way to learn the cognitive task of recognizing and responding to facial emotional expressions? Psychologists, artists, and actors have each created distinctly different training methods to "see the face". The objective of the first phase of this project is to experimentally quantify the relative learning power of the methods developed in the fields of art drawing and psychology. The objective of the second phase of our project is to to create an interactive human-computer learning environment that outperforms the art drawing, actor mimicry, and image-based psychology approaches on the quantitative scales defined in the first phase. We believe that the relative efficacy of the three different methods of training are particularly pronounced in subpopulations with cognitive cognitive deficits, such as autism.
As part of the first phase of our approach, we have examined, in depth, the existing art drawing and image-based psychology training approaches. This was done as part of a course taught at Stanford for the Spring 2008 term, Symbolic Systems 210, titled "Learning Facial Emotions: Art and Psychology". The course was co-taught with Pamela Davis Kivelson, and Artist-in-Residence at Stanford.
The
An artistic approach to facial emotions may be described as immersive involvement in the experience of one-on-one portraiture, wherein a live model is drawing in an art studio setting. So as to allow general statements to be made regarding an art drawing and psychology approach, we we focusing on the most widely used art drawing techniques used for hundreds of years, such as the use of live model in a studio setting, the model holding the poses for extended periods of time. A requirement of a valid experimental design is that the art drawing approach be restricted to a drawing of the approximately 20 Facial Action Units in the face that are associated with the seven universal emotions, as defined by Paul Ekman.
We are collaborating with Prof. Antonio Hardan in the Child and Adolescent Psychiatry Department at Stanford, and we plan to compare an art drawing and psychology image-based approaches using autistic subjects in the experiments.
We are also designing an advanced human-computer interactive training environment, based on techniques such as the use of avatars that can be manipulated to change facial action muscle groups, and automatic facial expression recognition software.