Instrumental single and double compression tests to predict sensory texture characteristics of foods

Twenty four commercial food samples were evaluated by a professionally trained descriptive panel who profiled hardness, springiness, fracturability, cohesiveness, cohesiveness of mass and chewiness in the test samples. Instrumental evaluation was carried out using both single and double compression tests with the TAX-T2 Texture Analyzer and a probe consisting of a set of dentures (B.I.T.E. masterII). Multiple instrumental parameters were extracted from the force-deformation curves of single and double compression rests and used for predicting sensory attributes using Partial Least Squares Regression. Relative Ability of Prediction (RAP), the equivalent of an R-2 taking into account the unexplained variation of the sensory data, were calculated to evaluate the models' predictive quality. Hardness (RAP=0.84), cohesiveness (RAP=0.72), and fracturability (RAP=0.85) were somewhat accurately predicted, while springiness (RAP=0.52), cohesiveness of mass (RAP=0.34), and chewiness (RAP=0.25) were unsatisfactorily predicted using a single compression test. Even though the test methods used differed significantly from traditional TPA testing, a double compression test did not offer significant improvements over the single compression test for the prediction of textural characteristics, except for the attributes springiness (RAP=0.79) and cohesiveness of mass (RAP=0.49).