Automatic analysis of facial expressions in an advertising test with Chinese respondents

An automatic system designed to infer valence (positive or negative feeling) from facial expressions, EMO Scan, was employed and evaluated in an advertising test in China. Its validity has already been demonstrated with Western participants (Garbas et al., 2013), and the purpose of the present study was to examine to what extent it generalizes to Asian participants. In a computer-based interview in a studio setting, participants’ faces were recorded while watching TV commercials as well as positive and negative affective pictures. Recordings were analyzed with EMO Scan to infer emotional valence from facial expressions. Inferred valence proved to be a valid predictor of picture valence. Regarding the TV commercials we found that results from facial expression analysis yield results that are similar to explicit ratings. This holds for both time series analysis and overall, time-aggregated results. The conclusion is that EMO Scan is a reliable tool to assess the valence of responses to emotional stimuli also in Chinese respondents.