Sensory discrimination by consumers of multiple stimuli from a reference: Stimulus configuration in A-Not AR and constant-ref. duo-trio superior to triangle and unspecified tetrad?

Abstract

In the food industry, overall discrimination tests are used with untrained/naïve consumer subjects to compare multiple test stimuli against a fixed reference, such as a company’s gold standard or a stimulus familiar to the consumer. Such tests are used for various objectives, including reformulation and cost reduction. Yet, studies on relative discrimination power and efficiency have been limited to experimental designs with a fixed pair of stimuli and method comparisons based on the same numbers of tests. In the present study, two reminder methods, A-Not A with Reminder (A-Not AR) and 2-AFC with Reminder (2-AFCR), were investigated as potentially better methods for experimental designs including comparisons of multiple pairs of stimuli for consumer discrimination. 2-AFCR is procedurally equivalent to a constant-reference duo-trio test with the reference presented first (DTF) and thus this test is referred to as the constant-ref. DTF/2-AFCR test in this paper. The practical efficiency of these two reminder methods, attributed to their effective stimulus configurations in replicated tests (i.e. using a fixed reference and lower number of different stimuli required in a test), was tested in comparison with the two most commonly used balanced reference classification methods, the triangle test and the unspecified tetrad test, by equalizing the number of stimuli required for the different methods. Namely the relative operational discrimination power was studied based on the same number of stimuli rather than the same number of tests. 180 naïve consumers performed a set of 12 replicated triangle tests and, based on the results, were divided into one of three equally-performing groups. A related-samples design was implemented for comparison between the triangle and the other three methods. An independent-samples design was implemented across the three groups to compare the A-Not AR, constant-ref. DTF/2-AFCR, and unspecified tetrad methods. Statistical ratio comparisons of d′ estimates obtained from different methods revealed that discrimination performance in the reminder methods was better than in both the tetrad and triangle methods. No discrimination difference was found between the triangle and tetrad tests having all possible test sequences, although the triangle test considering only the optimal test sequences, which were the same as those in the constant-ref. DTF/2-AFCR, resulted in superior discrimination than the tetrad test. Collectively, these results suggest that when assessing the discriminability of multiple stimuli from a fixed reference, the reminder scheme is the superior research design.