*4.3. Facial Expression Recognition Responses (FER Method)*

To overcome the bias associated with self-reporting sensory terms, an unconscious method was developed that involves the recording of the facial expressions of consumers, without informing them (intrinsically). In the yogurt study, the most liked product (cookies) was related to the 'surprised' facial expression emotion. The least liked product (berry) was related to the 'disgusted' facial emotion by both cultures. 'Happy' was not an effective discriminator for the products based on facial expressions, as it was related to 'disgusted' in the case of Western participants and to 'neutral' for Asian participants. Previously, in a study tasting orange juices, the 'neutral', 'disgusted' and 'angry' facial expressions explained liking towards the tasted samples, whereas the 'happy' emotion was not a useful discriminator during the implicit measurement of samples. It was also shown in this implicit study that participants displayed more negative emotions, and the least liked samples were easier to differentiate [34], consistent with observations made here.

The FaceReaderTM emotions detected clear cultural differences here among Asian consumers, who showed a significant difference in means for 'surprised' and 'disgusted', whereas no differences were observed for Western consumers. A similar outcome was previously observed for chocolate samples, where no significant differences were observed by FaceReaderTM emotions, although the consumer panel was not differentiated by culture [21].

The facial expression recognition method also has some limitations. The mean value for the term 'neutral' was the highest for both cultures. This could be attributed to the fact that participants performed tastings in an isolated booth, which was a socially distant and neutral environment. Hence, the consumers did not show much variation in positive emotions, but negative emotions were more pronounced. In a related study, it was seen that liking was not well correlated with emotional effects but a shift in liking for the yogurts tasted was well distinguished, showing the importance of emotions experienced [35]. Negative emotions were also better displayed for disliked samples in another study of

different juices [36]. This suggests that it is easier to distinguish a disliked product using facial recognition. Alvarez-Pato et al. [37] also found that facial recognition could not be used as a stand-alone method for predicting consumer acceptance of food products and this method is better applied in combination with other techniques, which is consistent with the findings here.
