Inspired by a seminal experimental work in olfaction (SNAC-K test [1]), we are performing a computational study to investigate the interaction of olfaction and language information during retrieval. After the presentation of each odor stimulus, the SNAC-K participants were asked to freely identify and name the presented odor. Interestingly, the data suggest that odors with low language variability lead to lower omission rates (blank responses). To mechanistically explain this observation, we built a computational model consisting of two reciprocally connected networks that stored overlapping odor and language representations as distributed memory patterns. We implement a fuzzy learning paradigm at which odors form up to four associations with different word-label descriptors using Bayesian-Hebbian plasticity, and evaluate semantic olfaction-name omission rates. Our model aims to reproduce quantitatively SNAC-K omission data, and based on preliminary results, we found higher blank responses with an increase in the number of olfaction-name associations. Encoding olfactory items along with multiple name-descriptors leads to odor-language network decoupling. Therefore, in this study, we plan to propose and evaluated a novel hypothesis about Bayesian-Hebbian synaptic plasticity mechanism which may cause decoupling of olfaction and language representations. [1] M. Larsson, M. Hedner, G. Papenberg, J. Seubert, L. Bäckman, E.J. Laukka. Olfactory memory in the old and very old: relations to episodic and semantic memory and APOE genotype. Neurobiology of Aging, 38: 118–126, 2016. https://doi.org/10.1016/j.neurobiolaging.2015.11.012.