How the nose knows

In their paper published in PlosOne today, IBEC researchers have tackled a current obstacle to a better understanding of the mammalian olfactory system, namely how chemical information is coded and processed. To do this, they looked at the performance of the early rat olfactory system in identifying the quality of the incoming stimuli – that is, its ability to detect and discriminate different odours – and made an analysis of their results by quantifying the number of smells that could be coded by a particular set of odour receptors (ORs) in the system.

“There’s a complex arrangement of OR neurons in the system which are distributed over the nasal epithelium to detect airborne chemicals, and the number of types depends on the species; for example, there are 387 different kinds of ORs in humans and 1,035 in mice,” explains Santiago Marco, group leader of IBEC’s Artificial Olfaction group. “We looked at the role played by this diversity and these varying quantities in encoding chemical information.”

Previous work has shown that a particular olfactory system adapts to the statistical properties of the set of chemicals to which it is exposed – for example, a dog’s ORs will be more attuned to the smell of meat than a rabbit’s. Since different neurons respond to a different set of odours, this forms the basis of the ‘code’ by which smells are identified, and so the researchers classified rats’ olfactory receptors according to their receptive range (RR) when exposing them to a large number of odours – an immense feat of systematic mapping, due to the large number of receptors found in rats and the huge amount of potential ligands. They also looked at the capacity to discriminate depending on distribution, and the correlation among receptors. In addition, instead of relying on simplified theoretical models, they used actual olfactory bulb data from an extensive dataset made available by the University of California Irvine..

“We found that optimal performance corresponds to a set of sensors with a receptive range of 50%, so the ORs are not particularly selective,” says Santiago. “Nevertheless, the biological system has a remarkably low correlation, or overlap, of sensors, and good coverage of the odour space. For sensors with low correlation, adding more to the set maximizes the coding capacity of the system.”

From this, the researchers surmise that biology has evolved toward a combination of more selective sensors for critical odours and a collection of less selective sensors to cover larger areas of the odour space. “An extreme case of this evolutionary drive is the presence of highly specific sensors for pheromone detection,” adds Santiago. “This better understanding of odour coding in olfaction may provide valuable insights for the design of general-purpose artificial olfaction systems; for example, it was previously thought that chemical sensors are not selective enough, but our study shows that selectivity may not be the most relevant parameter.”

J. Fonollosa, A. Gutierrez-Galvez, S. Marco (2012). Quality Coding by Neural Populations in the Early Olfactory Pathway: Analysis using Information Theory and lessons for Artificial Olfactory Systems. PLoS ONE, in press.



>> IBEC in the Media: Coverage on Nano Patents and Innovations blog: