Presentation Details
| A Quantitative Perceptual Framework for Reconstructing Complex Food Odors Xuebo Song1, Christiane Danilo1, Robert Pellegrino1, Joel Mainland1, 2. 11Monell Chemical Senses Center, Philadelphia, PA, USA.22Department of Neuroscience, University of Pennsylvania, Philadelphia, PA, USA |
Abstract
Odor perception plays a central role in food preferences, yet we lack a quantitative framework for understanding how complex food odors emerge from molecular mixtures. Odor mixtures have been reported to exhibit qualities distinct from their individual components, suggesting that interactions between odorants dominate mixture perception. However, linear models that assume independent and additive perceptual contributions from component molecules have performed surprisingly well at predicting odor mixture character in human behavioral studies using mixtures of up to 10 components. To test how well linear models can reproduce complex food percepts, we collected descriptive ratings for 24 foods alongside their published Sensomics-based reconstructions (GC-MS analysis followed by sensory testing). Using greedy search within a linear perceptual framework, we iteratively selected components from a database of ~700 individual stimuli to identify component mixtures minimizing perceptual distance to each target food and best approximating each food’s sensory profile. For 18 of 24 foods, these perceptually-optimized mixtures matched the target food more closely than the Sensomics reconstructions, despite using no information about the foods' chemical composition. These results demonstrate that odor components combine predictably even in complex foods, extending the validity of linear mixture models from simple laboratory mixtures to real food systems. A perceptual framework for olfactory mixture design offers a complementary approach to analytical methods, enabling food odor reconstruction from any available ingredient library.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.
No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.