|Abstract:||The food industry faces many challenges when attempting to formulate new or reformulate existing products to reduce sugar content. This problem is particularly pronounced in confectionary systems, such as caramels, in which sugar contributes the majority of sensory (i.e., sweetness, flavor, browning, texture) and physical (i.e., adhesiveness, structure) properties of the system. Sugar replacers can mimic some functional properties of sugar, but costly trial and error in reformulation is often required to achieve desired sensory and physical end-product properties. Matching product texture may pose a particularly steep challenge when replacing sugar in confectionary systems. Reliable instrumental prediction of sensory texture can supplement insight from sensory testing and streamline the reformulation process. Based on literature findings, accurate instrumental prediction of sensory texture has proven to be challenging. In order to meet this challenge, the goal of this research was to provide simple, reliable tools for the control of product texture through strategic formulation.
Caramel coating was identified as a promising model system for the development of instrumental predictors of sensory changes due to sugar replacement. In order to characterize the key categories of commercially available caramel coatings, a Napping-Ultra Flash Profile (Napping-UFP) study was performed with 12 commercial caramel popcorn samples. Hierarchical clustering analysis of the Napping-UFP data set for global, aroma-by-mouth, texture, and taste attributes resulted in the identification of 3 categories of caramel coatings: Small-scale Dark (SD), Large-scale Dark (LD), and Large-scale Light (LL). Compositional information from representative samples in each category was used to develop 3 model systems that matched the sensory properties of each category. The caramel coating model systems were then manipulated by varying sugar replacer (isomalt, maltitol, mannitol, or sorbitol) and replacement level to achieve a wide range of textures.
Next, a descriptive analysis panel was utilized to generate a complete sensory profile of full- and reduced-sugar caramel coating samples. Principal components analysis (PCA) showed that aroma and aroma-by-mouth attributes of the samples were most strongly influenced by the model system (SD, LD, or LL), while texture attributes were most strongly influenced by the sugar replacer and replacement level used. Texture attributes of the caramel coating samples were then compared against a range of common instrumental metrics to test the usefulness of these metrics in predicting sensory effects of sugar reduction. Modest correlations were found between moisture content, water activity (aw), and texture profile analysis (TPA) parameters and select texture attributes; however, glass transition temperature (Tg) showed the strongest correlations to sensory evaluations of texture attributes.
Evolution of full- and reduced-sugar caramel coating texture throughout mastication was then studied using the Temporal Dominance of Sensations (TDS) method and compared to both trained and consumer evaluations of stickiness in order to deepen understanding of stickiness perception and the effects of sugar replacement on texture. By correlating TDS dominance parameters and check-all-that-apply (CATA) selection rates with sample stickiness ratings, two tiers of stickiness-contributing attributes were identified. The texture attributes stringy, tacky, and enveloping comprised the first tier, showing significant positive correlations to stickiness by CATA and TDS, while attributes toothpacking, cohesive, and deformable comprised the second tier, showing significant positive correlations to stickiness only when multiple attribute selections were allowed (CATA). Consumer and trained panelist evaluations of tactile and oral stickiness were congruent and highly, inversely correlated to Tg. Further, Tg proved to be a good predictor of textural trajectory, with samples for which Tg < room temperature (RT) following a trajectory from deformable to enveloping, and samples for which Tg > RT following a trajectory from brittle to toothpacking.
Given the power of Tg as an instrumental predictor of sensory texture attributes, the final phase of this research aimed to explore how thermal processing and Tg of ingredients relate to the Tg of a sample. The original and modified Couchman-Karasz equations were used to calculate an expected Tg (TCK) for a range carbohydrate blends cooked to 120, 130, 140, and 150C, and TCK was compared to measured Tg. The TCK calculated using the original and modified Couchman-Karasz equations deviated from measured Tg by an average of 20.1C and 11.3C, respectively. To improve the predictive power of the Couchman-Karasz equation, empirical corrections based on sample moisture content or final cook temperature were developed. Application of these corrections reduced the averaged difference between predicted and measured Tg to <5.6C. Integration of the empirically modified Couchman-Karasz model with previously developed models for sample texture by Tg should enable prediction and strategic design of product texture. Further, this model should assist in the efficient formulation and process design of reduced sugar confectionary products through the use of Tg as a control parameter to minimize negative changes to product texture.