Probabilistic unfolding models for sensory data

Abstract Unfolding models are conceptually appealing for the analysis of consumers’ hedonic evaluations of food products. The appeal of the unfolding model is three fold; its conceptual simplicity, spatial character, and assumption of satiety — more is not always better. Unfortunately, the success of unfolding models does not always match their appeal. Reformulating unfolding models in a probabilistic framework is shown to improve their success, extend their application and further enhance their conceptual attraction. Data provided by the organizers of The Fifth Sensometrics Meeting are used to illustrate the proposed reformulation.