Triadic discrimination testing: refinement of Thurstonian and sequential sensitivity analysis approaches.

The effect of cognitive search strategies and variations in the oral environment on discrimination test performance were investigated. Subjects were required to discriminate between low concentration NaCl solutions and water using the 3-AFC and triangle test protocols. As predicted by Thurstonian modeling, subjects obtained a higher proportion of correct tests for the 3-AFC protocol than for the triangle protocol. The d' values obtained from both protocols corresponded. As predicted by the Sequential Sensitivity Analysis Model, which is largely based on changes in the oral environment, subjects obtained a higher proportion of correct tests for triads containing one NaCl stimulus than for triads containing one water stimulus. Measurement of physical signal strengths of the stimuli, by analysing the Na cation concentration change in the oral environment on tasting, indicated that the classical Thurstone-Ura two-distribution model was insufficient. The strong carry-over effects in the chemical senses require a model based on more than two distributions. It was also noted that subjects did not always use the search strategy required for a given test protocol. The artifactual effects of strategy change during an experiment are discussed.