Appropriateness is defined as the perceived fit between a food and a given situation. It was originally developed as a metric intended to capture the gap between liking and choice behavior. This review assesses (i) the theoretical foundations of situational appropriateness as well as parallel concepts that aid in its implementation, (ii) the diversity of techniques that have been used to operationalize the metric and the consequences of this diversity for generalizing results, and (iii) critical gaps in its discriminant validity with relation to the liking construct. Appropriateness research has focused on well-liked, familiar, everyday foods that fit reasonably well across many situations. As a result, appropriateness and liking ratings often strongly correlate, a pattern we argue reflects a narrow sampling of products and situations rather than an inherent overlap between the two constructs. We suggest that appropriateness may be most informative when products have a clear situational function, social meaning, or functional value that can be dissociated from hedonic response.
Food odorant perception is increasingly understood in the context of dynamic olfactory receptors (ORs), which are G protein-coupled receptors, sampling conformational ensembles rather than fixed binding pockets. This review translates and integrates this established dynamic-receptor paradigm into food olfaction by linking odorant structural motifs, receptor conformational reweighting, matrix-dependent delivery, and mixture-level outcomes such as masking and synergy. Recent cryo-EM structures and mutagenesis studies, including OR52-family receptors and the reversibly covalent aldehyde receptor OR6A2, illustrate how ligand binding can redistribute receptor states without, by itself, distinguishing conformational selection from induced fit. Current artificial intelligence models are discussed as useful structure-to-receptor or structure-to-perception predictors that still lack explicit receptor dynamics and realistic food-matrix exposure. A multiscale framework is proposed to connect these layers quantitatively and identify the data required for future predictive models.
