Exploring sources of variation in inter-observer reliability scoring of facial expressions using the ChimpFACS

The Facial Action Coding System (FACS) is currently the most popular systematic and standardized method of studying facial expressions in mammals. Since its original release for humans, the FACS has expanded to include primates such as chimpanzees. The FACS trains its users to analyze facial expressions by identifying individual muscle movements (called action units or AUs) that are combined to create facial expressions (or AU combinations). All researchers must pass a species‐specific certification test and are then considered reliable coders. After certification, interobserver reliability is rarely performed. However, reliability may be affected by variables like video length, apex identification, AU combination composition, and sociality effects. We analyzed the effects of these variables on reliability and propose that researchers should predetermine action units of interest for apex choice to reduce variation between observers. Two ChimpFACS certified researchers independently coded facial expressions from 30 hours of video of 17 chimpanzees ( Pan troglodytes ) housed at the Los Angeles Zoo. Coding reliability was then evaluated using Wexler's ratio (the method of calculating FACS inter‐observer reliability on the certification test). Preliminary results indicate that AU combination composition and sociality effects could have significant effects on coding reliability. We argue that researchers should predetermine which AUs to code as notable to facial expressions of interest and only code one individual during a time of high sociality to improve reliability in studies involving FACS coding. Employing these additional steps would improve replicability in facial expression studies, thereby strengthening research on the evolution of facial communication in primates. Support or Funding Information This research was supported by the National Institute of General Medical Sciences of the National Institute of Health under award number R25GM055052 awarded to T. Hasson. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This project could also not have occurred without the support of the Los Angeles Zoo. This abstract is from the Experimental Biology 2019 Meeting. There is no full text article associated with this abstract published in The FASEB Journal .