Automated facial expression analysis for emotional responsivity using an aqueous bitter model

The food industry is seeking ways to understand consumer emotions, using implicit measurements, to differentiate acceptability of products in the marketplace. Automated facial expression analysis (AFEA) is a prospective analysis for product acceptability. This study used aqueous bitter solutions to determine and validate AFEA as an analysis supplement to product liking. Participants (n = 46) evaluated a control (distilled water) and three bitter (caffeine) solutions: low (0.05% w/v); medium (0.08% w/v); and high (0.15% w/v). Individual participant sessions were video-recorded and analyzed (5 s; α = 0.20) for each sample in the default and continuous analysis setting. Participants rated liking and bitter intensity on a 9-point scale. An inverse relationship existed between liking and bitter intensity (rs = −0.90; p < 0.0001). In continuous setting for AFEA analysis of mean emotion intensity, analyzed by ANOVA, only the medium bitter treatment elicited a higher disgust response control (p < 0.20) and no differences were found between treatments in disgust (p > 0.20) evaluations using program default settings. For time series analysis with both the continuous and default settings, disgust was a predominant emotion in the medium and high bitter solutions as well as happy in the high (p < 0.025). Using time series analysis, continuous and default results had similar patterns over 5 s, but continuous data was more intermittent. Time series analysis is a promising tool for interpreting emotional results of a population and is more sensitive to emotional changes than mean comparisons. Future studies should continue to improve the characterization and sensitivity of emotions to food acceptability using AFEA.