Feasibility of reformulating flavours between food products using in vivo aroma comparisons

Abstract The feasibility of reformulating a commercial flavouring using a simple predictive model was studied in two confectionery (candy) systems. Initially, the variation in the different stages of the reformulation process was quantified using a model flavour. MS‐Nose analysis showed variation between 1% and 3% for static headspace analyses (ethyl nonanoate showed 9% variation). Flavour content in the finished candies varied from 4% to 22%. Despite variation between panellists' aroma release (3–60%), data analysis indicated that representative aroma release could be obtained using five panellists consuming three replicates. For reformulation studies, a simple commercial strawberry flavouring was chosen that delivered a highly acceptable flavour in a pectin–sucrose gel, but did not perform so well in a chewy candy containing sugar, protein and fat. By measuring in vivo aroma profiles in people eating the gel and the chewy candy, the relative change in aroma release could be determined and used to reformulate the strawberry flavouring so its release in both candies was similar. The sensory performance of candies with the original and reformulated flavourings was measured using difference testing ( n = 100) and descriptive analysis ( n = 5). Both analyses indicated that the reformulated flavour performed significantly better than the original flavour. The work shows the translation of a laboratory‐based concept into the commercial world and provides another tool to assist flavourists in reformulating flavourings to deliver food products with the desired sensory properties. Copyright © 2010 John Wiley & Sons, Ltd.