Improving the prediction for sensory texture attributes for multicomponent snack bars by optimizing instrumental test conditions

Abstract

Ten instrumental variables derived from a variety of five test probes (three-point bending, needle, 2-mm/6-mm cylinder compression and small blade shear) and five deformation rates ranging from 5 to 100 cm/min were used to predict the sensory texture attributes for 16 commercial snack bars. Hardness, crispness, brittleness and hand-evaluated hardness were selected as relevant texture attributes and evaluated from a descriptive analysis by 10 trained panelists. To determine the optimal instrumental test conditions (test probes and deformation rate) for the prediction of each sensory attribute, various model statistics based on partial least squares regression were used. The blade exhibited high predictive abilities, showing validation coefficient determination (Rval) greater than 0.8 for all the sensory attributes evaluated. The results display that, provided the correct instrumental conditions are set, a valuable improvement in the prediction of sensory attributes by instrumental variables can be achieved.© 2010 Wiley Periodicals, Inc.