Development and implementation of novel sensory evaluation procedures of consumer acceptability towards chocolate based on emotions and biometric responses

Abstract

Chocolate is a condensed suspension of different particles, with a continuous phase containing cocoa butter and milk fat, and a diffused phase containing cocoa particles, sugar, and non-fat milk solids. Chocolate is the most commonly consumed confectionery product worldwide. Sensory evaluation is a scientific discipline which uses human senses for evaluating consumer products with the application of experimental design and statistical analysis. The application of novel procedures in sensory analysis is needed to prevent the high failure rate of new market launches based on testing with conventional sensory protocols. Hence, the objective of this study was to use novel sensory evaluation procedures including subconscious biometrics [skin temperature (ST), heart rate (HR) and facial expressions (FE)] and self-reported responses to determine consumer acceptance to food, using chocolate. Initially, in Chapter 3, a cross cultural study using Asians and Westerners was conducted to develop lexicons of emotions for milk and dark chocolate using online surveys (N = 206) and by conducting tasting sessions of the chocolates (N = 75). As per the responses of the survey and according to previous research findings, the main reasons for chocolate consumption were emotional satisfaction and indulgence. The main factor that consumers considered in making their purchase decision was the taste/flavour of the chocolate. Three separate emotion lexicons were developed using the results of this experiment, and they were used for further studies conducted using chocolate. In Chapter 4, chocolate with five basic tastes (bitter, salty, sour, sweet and umami) were developed to obtain sensory and physiological responses of consumers to different tastes using 45 participants. An integrated camera system coupled with a tablet-PC using Android OS and containing a BioSensory application was employed to capture infrared images, videos and sensory responses. Inputs from this application were used to determine ST, HR and FE. Sensory responses were gathered using hedonic scales and emotions elicited were obtained using lexicons. Results revealed that the most liked sample was the sweet chocolate, while the least liked was the salty chocolate. Findings of this study can be used to assess novel tastes of chocolate in the industry. Furthermore, in Chapter 5, flavour was added to chocolate directly to produce different flavoured chocolate (caramel, cinnamon chai, mandarin, strawberry and peppermint). A sample without any flavour was considered as the control. These six samples were used for sensory evaluation with 113 participants. ST, HR, FE, sensory responses and emotions elicited were determined like the previous experiment. Results revealed that the most liked sample was the strawberry and the least liked was cinnamon chai chocolate. Findings of this study can be used to determine the acceptability for different flavoured chocolate based on self-reported and subconscious responses. In Chapter 6, encapsulated flavours were developed to be added to chocolate and release the flavour in a different way compared to the experiment reported in Chapter 5. Samples [normal strawberry (NS), normal mint (NM), encapsulated strawberry (ES), encapsulated mint (EM) and control (with no flavour)] were developed and tested by 52 participants to obtain biometric (ST, HR and FE) as well as sensory and emotional responses. Encapsulated chocolate samples showed higher liking compared to normal flavoured chocolate. Findings of this study can be used to determine the acceptability of differently flavoured chocolate in future studies using chocolate. As the next step, in Chapter 7, artificial intelligence was used to develop machine learning models with near-infrared spectroscopy (NIRS) to assess the quality of chocolate based on chemical fingerprinting. Chocolate samples with basic tastes (bitter, salty, sour, sweet and umami) were used for this experiment with 45 respondents and their chemometrics (pH, brix and viscosity), colour (CieLab scale) and sensory properties (basic taste intensities) were determined. Data were used to develop two machine learning models to predict the chemical parameters (Model 1) and sensory parameters (Model 2). These models showed high accuracies of R = 0.99 and R = 0.93 respectively. The developed models can be used as a substitute method to determine sensory properties of chocolate with low cost more accurately. In this study, emotion lexicons were developed and used for sensory sessions conducted using chocolate. Conventional and novel sensory techniques were used for analysis. Results revealed significant differences in emotional terms selected based on gender and culture for different samples. Furthermore, there were significant correlations between conscious and subconscious responses of chocolate. A combination of implicit, explicit and emotional responses may help to better understand the acceptability to different food products. Moreover, models with high accuracy were developed to predict sensory properties of chocolate using chemometrics. Findings of this study can be used for future acceptability research on chocolate.