| Many product optimization methods, known in general terms as preference mapping, have been described in the literature. However, a direct comparison of some of the approaches has not been undertaken. The objectives of this study were to investigate four approaches of optimization, compare the sensory profiles of the solutions, formulate the ideal solutions for validation with consumers, and determine the drivers of liking for strawberry yogurt.; Strawberry yogurt formulations were chosen to vary sweetness, strawberry flavor intensity and thickness. A D-optimal, three factor experimental design (each factor varying between -1 and 1, where -1 is low and 1 is high) was used to formulate eleven samples. The formulation optimization, based on the optimization of liking, was done using four different methods. EDIPM (Euclidian distance Ideal Point Modelling), an extension of MDPREF allows the determination of ideal points and was used as an example of internal preference map. LSA (Landscape Segment Analysis) was used as an example of probabilistic ideal point model. A method known as an extension of PREFMAP proposed by Danzart was used as an example of external mapping. Finally, a method of optimization based on Just About Scales results was used. All 11 formulations were presented to a consumer panel (N=120) and to a descriptive panel (N=13). Descriptive analysis, using the Spectrum MethodRTM, was used to characterize the flavor and texture profiles of the samples, while consumer testing was performed to determine the acceptability and liking of the samples.; Overall liking scores ranged from 4.2 to 7.3 on a 9-point hedonic scale, a range of acceptability suitable for optimization studies. For sweetness, all four methods identified a high level of sweetness as being ideal (EDIPM=1.0, LSA=0.58, JAR=1.04, Danzart=0.81). Optimization methods also agreed that amid to high level of thickness was optimal (EDIPM=0.62, LSA=0.45, JAR=0.52, Danzart=1.06). However, there was significant disagreement among the methods as to the ideal strawberry level (EDIPM=0.28, LSA=1.09, JAR=-1.1, Danzart=0.17). This study shows that the optimization methods tended to agree on an ideal level for two of the three dimensions investigated. This finding warrants further investigation and comparison of the optimization methods available today. |
