A critical test of the overlap hypothesis for odor mixture perception.

Abstract

The overlap hypothesis of mixture perception is based on the observation that mixtures of perceptually similar odorants tend to smell different from their components (configural), whereas mixtures of dissimilar odorants smell like their components (elemental). Because input patterns of perceptually similar odorants tend to overlap more than dissimilar ones, it has been hypothesized that component pattern overlap can predict a mixture's perceptual quality, with high overlap predicting a configural response and low overlap an elemental response. The authors used 7 pairs of odorants chosen for different degrees of overlap in their monomolecular 2-deoxyglucose activation patterns to test the theory in a go/no-go behavioral assay that measured generalization from binary mixtures to components. The authors show that individual component odorant input patterns are not sufficient to predict mixture quality, falsifying the overlap hypothesis. An important finding is that different odorant pairs with similar glomerular overlap showed opposite behavioral-perceptual responses, suggesting nonlinear effects at the receptor or glomerular level or the critical involvement of higher order areas. Thus, the authors posit that imaging the mixtures themselves may provide additional information needed to reliably predict mixture quality.