Sensory evaluation is essential in food development, and has been widely used to assess the palatability of food. However, simultaneous evaluation of multiple senses (taste, aroma, and texture) and their interactions make it difficult. This method relies heavily on the subject's verbal expression ability, making it difficult to apply to infants. Furthermore, subjective influences limit reproducibility and objectivity.

It is suggested that biological responses could be an effective method to overcome these challenges. It is known that characteristic facial changes occur in response to taste stimuli such as sourness, and that changes in saliva secretion resulting in taste and/or aroma affected food bolus formation. Differences in food bolus formation are also thought to reflect oral movements, i.e., chewing movements, reflecting each food texture.

We developed a smartphone application to measure temporal changes in facial surface shape during mastication. This application uses a smartphone-mounted 3D scanner to capture a 3D coordinate of the face along three axes over time. Using this method, changes in facial surface shape over time were recorded during chewing of food alone and chewing of food containing added sugar, salt, and vanilla flavoring, several hundred times. Deep learning was employed for texture analysis. Deep learning enables automatic extraction of characteristic values even from huge and complex dataset.

The results showed that the presence or absence of flavor could be determined only from chewing behavior with approx.. 70% accuracy for sweetness and 80% accuracy for saltiness and aroma, demonstrating that chewing behavior changes depending on the flavor. Furthermore, when estimating the flavor concentration perceived by subjects while chewing, it was possible to estimate the concentration for taste/flavor from chewing movements alone.

This study revealed that chewing movements change when flavors are added, and that this reflects the flavor intensity perceived by people.

Conference Theme: Physical properties of food hydrocolloids for enhanced product development