Conceptual vs. perceptual wine spaces: Does expertise matter? Jordi Ballester a, * , Bruno Patris a , Ronan Symoneaux c , Dominique Valentin a,b a CSG-UMR5170, 15 rue Hugues Picardet, Dijon, France b ENSBANA, 1, Esplanade Erasme, Dijon, France c ESA Angers, 55 rue Rabelais, Angers, France Received 13 April 2006; received in revised form 2 August 2007; accepted 4 August 2007 Available online 14 August 2007 Abstract This study explores the differences in wine categorization between wine experts and novice wine consumers using 10 Melon de Bourgo- gne (MB) and 10 Chardonnay (CH) wines. Participants performed a free sorting task based on odor similarity followed by a CH and a MB typicality rating task and a liking rating. All tasks were performed orthonasally. We observed a clear agreement between experts concerning typicality scores. Moreover, despite a slight overlap we found a clear differentiation between CH and MB for experts’ typ- icality scores. For novices, no such agreement on typicality scores was observed and we found a complete overlap between both types of wines. These results suggest that experts developed through successive wine tasting separate consensual sensory concepts for CH and MB wines. Multidimensional Scaling analyses of the sorting tasks showed an expertise effect since the expert similarity plot showed a better separation between MB and CH wines than the novices’ one. For experts, significant correlations between both CH and MB typicality scores and the MDS first dimension coordinates suggest that experts’ CH and MB sensory concepts are partially based on perceptual similarities. Additionally, experts’ hedonic scores were significantly correlated with their CH typicality scores as well as with the MDS second dimension. This correlation suggests that liking was involved in the sorting task but was not the main expert’s criteria to sort the wines. Our results suggest that wine expertise may be more of a cognitive expertise rather than a perceptual one. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Perceptual; Cognitive; White wine odor; Chardonnay; Melon de Bourgogne; Expertise; Typicality 1. Introduction Although categorizing our perceptions is one of the most basic ways to organize our knowledge, the classifica- tion of food sensory perceptions into categories reflecting food sensory concepts is an almost unexplored field. It is commonly accepted that natural categories have an inter- nal organization and that all the items belonging to a cat- egory are not equally representative of it. The category is then organized along a typicality gradient from the most to the less representative items where typical items share a lot of features with most of the items in the category and very few with items in other categories. Conversely, non-typical instances share few features with instances of the same category and share features with instances of other categories (Rosch & Mervis, 1975). Additionally, the boundaries between neighboring categories are not clear-cut but rather fuzzy and depend upon the context (Labov, 1973). Interestingly, a few studies have shown that olfactory categories have the same internal organization as natural categories (e.g. Chrea, Valentin, Sulmont-Rosse ´, Nguyen, & Abdi, 2005). Moreover those studies showed that the graded structure of odor categories depends upon culture (Chrea et al., 2005), expertise level (Lawless & Glat- ter, 1990; Solomon, 1997) and context (Lawless, 1989). Some authors have shown that food categories were also organized along a typicality gradient. In these studies par- ticipants were asked to assess the typicality of food prod- ucts such as apricots (Guichard, Schlich, & Issanchou, 1990), Cheddar cheese (Dacremont & Vickers, 1994) and 0950-3293/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodqual.2007.08.001 * Corresponding author. Tel.: +33 3 80 68 16 56; fax: +33 3 80 68 16 01. E-mail address: ballester@cesg.cnrs.fr (J. Ballester). www.elsevier.com/locate/foodqual Available online at www.sciencedirect.com Food Quality and Preference 19 (2008) 267–276