Full Program »
3. Altered emotion expression in psychosis risk: Determining the promise of thin slicing and comparing human and automated coding methods
Recently, our group employed automated emotion expression analysis methods and observed that clinical high-risk (CHR) participants exhibited alterations in facial expressivity during thicker slices (5-minutes) of structured clinical interviews. However, it is unclear whether thinner slices of time (i.e., 1-minute) can replicate the same results; these data can contribute to methodological questions in this area. Thin slices may be more practical if emotion assessment is to become widely used in clinical settings, allowing the ability to intervene earlier on in treatment. Additionally, as automated approaches are becoming adopted, it will be important to confirm this approach converges with traditional, human coding techniques. Presently, thin slices of video-recorded clinical interviews from 42 CHR youth and 42 controls were analyzed by trained human coders for facial expressions of emotions and these same segments were submitted into automated facial analysis (using 2 different software packages). Results indicated thin slices of time were sufficient to reveal alterations in facial expressivity, specifically blunting in joy expressions, replicating findings from our previous study in which we used thicker slices of time. Furthermore, both automated analysis programs converged in the ability to detect these alterations and were consistent with human coding indicating important strengths in these approaches. Together, these data lay foundation for new studies to build on and suggest that brief segments of time can yield important clues about emotive dysfunction. Further, automated analysis, an approach that does not require lengthy training or interpretation, but does lend well to mobile assessment and computational modeling, shows promise.