Integrated Calendar

Olfaction researchers at all levels are ultimately trying to solve the same problem, namely a transform across three spaces: from the physicochemical space of odor molecules, through the brain space of neural activity, and on to the space of odor perception and its ensuing behavioral decisions. To solve these transforms, one has to be able to measure each one of these spaces independently. As each of these three spaces is apparently of very high dimensionality, we applied principal components analysis (PCA) to data in each of these three domains. We observed that the functional dimensionality of these spaces was significantly lower than their apparent dimensionality. Moreover, the key axis (PC1) was correlated across domains. In other words, the key axis of olfactory perception was correlated with the key axis of odorant structure, and both of these were correlated with the key axis of neural activity in the olfactory system across species. These correlations allowed us to construct a modest but significant predictive framework across domains. In other words, we could now look at the structure of a novel molecule, and predict modest but significant aspects of its perception and ensuing neural activity across species. Beyond this predictive framework, our approach has several implications regarding sensory phenomena within a metric space. For example, it implies a point of sensory convergence where all olfactory mixtures should smell the same. We call this point "olfactory white". Our metric approach also implies points (odors) that are at the upper and lower boundaries of this metric space, and should therefore be odorless. We call these points "infra smell" and "ultra smell". In this talk I will consider the implications of this approach, as well as its potential shortcomings, and their possible solutions.