ROC analysis in olive oil tasting: A Signal Detection Theory approach to tasting tasks

摘要

Sensory evaluation of olive oil has been conducted by using the Classical Threshold Theory perspective. This approach does not allow the separation of purely sensory processes from other factors that may affect performance such as tasters cognitive biases. Two experiments were conducted with the goal of developing a task that, using the logic of Signal Detection Theory (SDT), would allow sensory and decision processes to be separately assessed within an olive oil tasting task, overcoming the shortcomings of the traditional Classical Threshold Theory approach. Experiment 1a used classic psychophysics theory to establish the olive oil concentrations that were used later in Experiment 1b. Experiment 1b presents a taste task based on SDT (by using signal and noise stimuli) that allowed establishing ROC curves and the separate calculation of sensory and decision indexes. Non-parametric detection and sensory indexes, and robust analyses of variance are proposed to overcome the problems associated with the use of a limited number of trials involved. The advantages of the SDT approach in olive oil tasting are discussed.