le même processus pour tous - Université de Bourgogne
le même processus pour tous - Université de Bourgogne
le même processus pour tous - Université de Bourgogne
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
m.<strong>de</strong>smas@isa-lil<strong>le</strong>.fr<br />
Is it possib<strong>le</strong> to <strong>le</strong>arn beer sensory categories?<br />
An exploratory study<br />
Maud LELIÈVRE-DESMAS 1,2 , Sylvie CHOLLET 1 & Dominique VALENTIN 2<br />
1 Institut Supérieur d’Agriculture, Laboratoire Qualité <strong>de</strong>s Aliments, Lil<strong>le</strong>, France<br />
2 Centre <strong>de</strong>s Sciences du Goût UMR 5170, CNRS, INRA, <strong>Université</strong> <strong>de</strong> <strong>Bourgogne</strong>, Dijon, France<br />
Introduction<br />
Context<br />
Previous studies have shown that wine experts tend to <strong>de</strong>velop common<br />
mental representations of wines through frequent tasting and<br />
consequently do not categorize wines similarly than novices (Solomon,<br />
1997; Bal<strong>le</strong>ster et al., 2005, 2008).<br />
Objectives<br />
This study proposed to evaluate whether this phenomenon occurs for<br />
beers, i.e. whether:<br />
- novices could <strong>le</strong>arn beer sensory categories through repeated<br />
exposure to different beers,<br />
- they could generalize their <strong>le</strong>arning to i<strong>de</strong>ntify category membership<br />
of new beers.<br />
↓<br />
Two experiments with two different categorization criteria:<br />
- fermentation type<br />
- geographical origin<br />
Figure 2. Percentages of correct answers (means ± standard error) for the<br />
assessor group for top (TF) and bottom (BF) fermentation beers at T 0 and T final .<br />
Conclusion<br />
Products<br />
> Fermentation type:<br />
- 18 top fermentation beers (TF)<br />
- 18 bottom fermentation beers (BF)<br />
> Geographical origin:<br />
- 9 German beers (G)<br />
- 9 Belgian beers (Bg)<br />
- 9 British beers (Br)<br />
General procedure<br />
Assessors<br />
> 19 novices for each experiment<br />
Method<br />
> Training period to i<strong>de</strong>ntify category<br />
membership<br />
> Comparison of correct beer<br />
categorizations at T 0 and T final<br />
> End of session T final : questionnaire<br />
to i<strong>de</strong>ntify possib<strong>le</strong> <strong>le</strong>arning strategies<br />
Figure 1. Schema of general <strong>le</strong>arning procedure<br />
Results<br />
Fermentation type Geographical origin<br />
Learning data<br />
> For <strong>le</strong>arned beers: the percentage of correct answers is significantly higher<br />
at Tfinal than at T0 .<br />
> For un<strong>le</strong>arned beer: no difference between T0 and Tfinal. > 13 out of 20 beers had a higher percentage of correct answers at Tfinal than at T0 (including 5 beers with significant difference).<br />
> No difference between top and bottom fermentation beers <strong>le</strong>arning.<br />
↓<br />
Assessors <strong>le</strong>arned to categorize almost all the beers individually but did not<br />
generalize their <strong>le</strong>arning to new beers.<br />
Questionnaire<br />
> Globally, consensus on the sensory characteristics of each category:<br />
- TF beers: + alcoholised, + persistent, + sweet, + flavour, + heavy.<br />
- BF beers: - alcoholised, - persistent, - flavour.<br />
> Assessors did not used one or two characteristics to i<strong>de</strong>ntify category<br />
membership but more comp<strong>le</strong>x strategies (e.g., “This beer is dark, his<br />
nose is intense, his taste is intense and alcoholised. I infer it is a top<br />
fermentation beer”).<br />
> Half of the assessors <strong>de</strong>clared to have recognized a beer and to have<br />
inferred its category (e.g. “ This beer looks like an Heineken or a 33<br />
Export, I know that these beers are BF, so I infer this beer is BF”).<br />
Learning data<br />
> For <strong>le</strong>arned beers: no difference between T0 and Tfinal but high variability<br />
between beers in each category.<br />
> For un<strong>le</strong>arned beer: no difference between T0 and Tfinal. > 6 out of 15 beers had a higher percentage of correct answers at Tfinal than at T0 (including 4 beers with significant difference).<br />
> No difference between German, Belgian and British beers <strong>le</strong>arning.<br />
> Two British beers (stout beers) obtained higher results than the other<br />
British beers, at T0 and Tfinal → more typical beers?<br />
↓<br />
Assessors <strong>le</strong>arned to categorize only some beers individually and did not<br />
generalize their <strong>le</strong>arning to new beers.<br />
Figure 3. Percentages of correct answers (means ± standard error) for the<br />
assessor group for German (G), Belgian (Bg) and British (Br) beers at T 0 and T final .<br />
Questionnaire<br />
> Globally, <strong>le</strong>ss consensus on the sensory characteristics of each category.<br />
> Assessors were in accordance on only some attributes, and they used a<br />
lot of other attributes which did not <strong>le</strong>ad them to the same conclusion.<br />
↓<br />
It can explain why there is no category <strong>le</strong>arning of geographical origin<br />
categories.<br />
> Assessors seem to be ab<strong>le</strong> to i<strong>de</strong>ntify category membership of <strong>le</strong>arned beers (especially for fermentation categories) but not to generalize to un<strong>le</strong>arned beers.<br />
> These results could be due to a short training period (9 sessions only), or to a high beer heterogeneity insi<strong>de</strong> the categories to be <strong>le</strong>arned, or to a nonadapted<br />
<strong>le</strong>arning context (do not correspond to daily life <strong>le</strong>arning situations).