Relevance of omission tests to determine flavour-active compounds in food: application to cheese taste Erwan Engel a, *, Sophie Nicklaus b , Christian Salles b , Jean-Luc Le Que´ re´ b a Institut National de la Recherche Agronomique, Unite ´ Mixte de Recherches en Ge ´nie et Microbiologie des Proce ´de ´s Alimentaires INRA-INA-PG, 78850 Thiverval-Grignon, France b Institut National de la Recherche Agronomique, Unite ´ Mixte de Recherches sur les Aro ˆmes INRA-ENESAD, 17 rue Sully, 21065 Dijon, France Received 6 July 2001; received in revised form 26 August 2002; accepted 26 August 2002 Abstract One of the main problems to overcome in determining the flavour-active compounds of a food product is to point out a causative linkage which may exist between its composition obtained by physico-chemical analysis, and its flavour properties determined by a trained sensory panel. Starting with an overview of the strategies classically used to solve the problem and their limitations, this paper presents an alternative methodology based on omission tests. Through different applications concerning cheese taste studies, the main interests of this methodology are shown and discussed. # 2002 Published by Elsevier Science Ltd. 1. Introduction Among the criteria allowing the consumer to judge the overall quality of food, flavour plays a key role by contributing to the rejection or the acceptance of the product. One of the main questions to answer if one is to understand why a food product is preferred to another, is to determine which food components are responsible for its flavour. Whatever the strategy used to identify such flavour- active compounds, at least two parallel steps are neces- sary: (1) The determination of the flavour properties of the studied food using appropriate sensory technique, (2) The characterisation of the food composition using adapted methods of physico-chemical analysis. The determination of flavour-active compounds consists in finding a causative linkage between the composition of food and its studied flavour property. This paper mainly deals with the determination of taste-active compounds. The first part focuses on the principle and limitations of strategies classically used to relate sensory and physico-chemical data. The second part presents the key steps of an alternative methodol- ogy based on omission tests and highlights some of their critical points. In a third part, two applications of this technique to the determination of cheese taste-active compounds are presented and discussed. 2. Strategies classically used to determine taste-active compounds: principle and limitations Several studies have shown that the taste of food was mainly due to compounds recovered in their water- extract (McSweeney, 1997; Petro-Turza & Teleky- Vamossy, 1989; Schlichterle-Cerny & Grosch, 1998). For that reason most studies investigating the com- pounds responsible for food taste have been performed on their water-extract. 2.1. Statistical correlations This is the most classical approach used to determine flavour-active compounds of food. It involves studying different variants of a given product and looking for a mathematical correlation between the concentration of one or several food components and the intensity of a studied sensory property. One of the main limiting fea- tures of statistical correlation is linked to the mathema- tical nature of the relationship, which need not necessarily involve causative stimuli (Williams, 1994). One of the main dangers is to confuse statistical risk which allows the significance of a correlation to be judged with the real risk of being wrong in accepting the model (Piggott, 1990). Other factors may explain the difficulties and limitations linked to the use of statistical correlations: numerous parameters varying among the products (Williams, 1994); pertinence of physico-che- mical variables in order to properly explain a sensory 0950-3293/02/$ - see front matter # 2002 Published by Elsevier Science Ltd. PII: S0950-3293(02)00136-2 Food Quality and Preference 13 (2002) 505–513 www.elsevier.com/locate/foodqual * Corresponding author. Tel.: +33-1-3081-5495; fax: +33-1-3081- 5597. E-mail address: engel@grignon.inra.fr (E. Engel).