A Reexamination of the Two‐Stage Triangle Test for the Perception of Sensory Differences^a

Abstract

SUMMARY: A review of the theory and practice of the two‐stage triangle test for the sensory perception of small differences leads to the conclusion that its disadvantages usually outweigh its advantages. When the test is used its information content is difficult to assess; and a “least objectional” method, based on a scoring scheme and statistical significance procedures, is here proposed.

Variations of the triangular method and the relationship of its unidimensional probabilistic models to three‐alternative forced‐choice signal detection theory models

“…These differences underline once more the distinction between the two types of models and show the importance of the experimental design and of the instruction as a task variable. Now this distinction has been recognized it can be used to resolve the long-standing 'paradox of discriminatory non-discriminators' (Gridgeman, 1970) identified by Byer & Abrams in 1953. The nature of this paradox can be exemplified by one of their experiments, in which these authors investigated the bitterness difference between quinine sulphate solutions, For this purpose they used the normal triangular method design, i.e.…”

“…When in a particular case the test shows that the samples are different, no information has been obtained about the nature and magnitude of the perceived difference between the stimuli. In order to rectify this state of affairs, several modifications and elaborations of the original procedure have been proposed, Those encompass the additional assignment of the direction of the difference perceived (Davis & Hanson, 1064) or a subsequent rating of its magnitude (Bradley, Gridgeman, , 1970). …”

“…Finally, by analysis of experimental data, it is shown that tho theoretical distinction botweon tho Thurstone-Ura model and the 3-AFC SDT model is useful bocause it appeared the key for the resolution of the 'paradox of discriminatory non-discriminators' (Gridgeman, 1970).…”

The evolution of a model: A review of Thurstonian and conditional stimulus effects on difference testing

“…The approach provides a useful tool for understanding and modeling the decision processes or cognitive strategies involved in difference testing. For example, the paradox of discriminatory non-discriminators (Byer & Abrams, 1953;Gridgeman, 1970), whereby judges perform better on a triadic difference test when the difference between the two stimuli is specified (3-AFC method) than when it is not (triangle method), cannot be explained in terms of traditional binomial statistics. For years, it remained an unexplained paradox.…”

Comparison of the triangle and a self-defined two alternative forced choice test

“…Part of the difficulty with the triangle procedure arises from the inherent variability of the three implied paired comparisons that must be performed in order to select the ''odd" or outlying (most different) sample. This difference in the test methods is evident in the paradox of the discriminatory nondiscriminators (Byer & Abrams, 1953;Gridgeman, 1970), a finding that some panelists who fail to find the correct answer in a triangle test will nonetheless answer correctly given a 3-AFC task with the same items (e.g. ''choose the strongest sample").…”