On March 11th, Venkatesh Murthy from the Department of Molecular and Cellular Biology of Harvard University, US, gave a conference at the PRBB invited by the CRG. He explained his study “An olfactory cocktail party: figure-ground segregation of odorants in rodents”, which was the cover of Nature Neuroscience in September 2014. After a brief introduction to the anatomy of the olfactory system of rodents, he explained that many odours are complex mixtures: different chemicals combine and then we can smell a particular object. His main question was: how well can a mouse pick out an individual odorant from a mixture?
They wanted the mice to pick a single ingredient within an odour cocktail. In order to do that, mice were trained to recognize target odorants embedded in unpredictable and variable background mixtures. They used 14 different chemicals, so there were more than 16.000 possible mixtures. It was impossible for the mice to memorize the combination; they had to recognize the single odorant. The test used was the go/no go, in which stimuli, in this case smell, are presented in a continuous stream and mice perform a binary decision on each stimulus. One of the outcomes (the correct smell) requires mice to make a motor response (go) in order to receive the reward, whereas the other requires mice to withhold a response (no-go). Accuracy and reaction time are measured for each event.
Mice could learn this task in a few days, they performed it well, but performance dropped with increasing number of background odours. To understand why, the researchers first had to overcome a problem particular to olfaction.
While the relationship among different visual stimuli is relatively simple – differences in colour can be described as differences in wavelength of light – there is not a simple explanation for describing how odours relate to each other. Instead, the researchers tried to describe scents according to how they activate neurons in the brain. They used optical imaging and computational models to relate behavioural performance to the combinatorial neural representation of odorants in odour receptors.
Using fluorescent proteins, they created images that showed how each of 14 different odours stimulated neurons in the olfactory bulb. Each odour gave rise to a particular spatial pattern of neural responses. When the spatial pattern of the background odours overlapped with the target odour, the ability of mice to identify the target was diminished. Therefore, the difficulty of picking out a particular smell among a cocktail of other odours depends on how much the background interferes with the target smell.
All in all, it was a very interactive session, with the public discussing several issues all through the talk, especially about methodology, so both the speaker and the public got new ideas!
A report by Mari Carmen Cebrián