Research conducted by scientists at the Italian Institute of Technology (IIT) and Harvard University explains for the first time the mechanisms used by our brain to recognize specific smells. The study, published in Nature, sheds new light on the brain processes involved in the continuous flow of information arriving from our senses, in particular from the sense of smell. Thanks to this result, researchers will be able to think about the realization of an artificial sense of smell, to be transferred to robots and other intelligent machines in the future.
If we talk about sight or hearing, scientists are well-versed in the mechanisms that lead us to distinguish two colors or two notes. This knowledge was translated into relatively well-established theories: we know which wavelength a neon must emit to appear red and which frequency an electronic keyboard must produce to make us hear a G. The same does not happen with odors: we are not able to say how a molecule smells just by looking at its chemical structure.
Imagine that you are at a restaurant, and you are having a sorbet with a pleasant citrus scent. If you are a chef, you could use the olfactory information to understand whether they used lemon or lime, two fruits producing odorous molecules perceptively similar, but chemically different. If you are not interested in the recipe, you may simply notice that it is a citrus sorbet instead of a coffee sorbet. This means you will discard the small chemical differences between lemon and lime smells to generalize using the unique category of citrus.
It is not easy to flexibly choose between olfactory discrimination and generalization according to our experience and goals. Until now, scientists did not know the code used by our brain to smell a specific odor.
The new study answers this question for the first time. The team of Italian American scientists, coordinated by Bob Datta of Harvard Medical School, identified the tricks used by the brain to discriminate and generalize odorous molecules possessing chemical structures with various levels of similarity. The team includes two Italian research groups from the Center for Neuroscience and Cognitive Systems at IIT in Rovereto: the team of Giuliano Iurilli, who returned to Italy thanks to the Armenise Harvard Foundation, and the team of Stefano Panzeri, coordinator of the Center.
Iurilli, one of the project’s main authors, acted as a “bridge” between Italy and the United States. After a PhD in Neuroscience and Robotics at IIT, he moved to Harvard in 2012 where he studied the brain mechanisms underlying the perception of odors and their effects on our daily behavior. In 2018, he won the Career Development Award from the Armenise Harvard Foundation, a million-dollar grant for 5 years that promotes basic biomedical research. Thanks to these funds, Iurilli returned to Italy and started his neurophysiology laboratory at IIT in Rovereto. He brought with him a fundamental piece of research realized in Boston, and contributed to complete it with the Armenise Harvard grant. Bob Datta, lead author of the study, was also an Armenise Harvard awardee, being the Foundation’s HMS Junior Faculty Grant winner in 2010 and 2016.
“We developed ad hoc analysis methods,” Giuliano Iurilli explains. “And we saw that the sensory neurons in the nose capture the odorous molecules, analyzing them almost like chromatographs, machines that precisely describe the chemical differences between molecules.”
However, this precision only concerns the first “door” for the odors’ recognition, namely the nose. But something changes when the information collected and processed by the nose arrives in a more central structure of the brain, called the olfactory cortex. At this stage, the odors’ descriptions begin to “get personal”, and they no longer respect chemical differences.
This is where memory comes into action. The new similarities described in the olfactory cortex are less about the chemistry and more about the subject’s previous experiences. For example, if our nose is scenting an ethanol molecule and an octene-3-yl acetate molecule, our olfactory cortex simply registers the scent of a gentian schnapps if we have drunk one before.
“We found out that this happens because past experiences modify the way the neurons of the olfactory cortex exchange the chemical information they received from the nose – Giuliano Iurilli concludes – Now we can start thinking concretely about how to build an artificial brain that does the same thing in a robot“.
In fact, this discovery could have a great impact to the field of robotics and artificial intelligence, through the realization of an artificial olfactory system. This intelligent system might be able to evaluate the safety of an environment or to recognize an object by quickly scenting the volatile molecules, just like a human being.
Pashkovski et al. (2020). Structure and flexibility in cortical representations of odour space. Nature. DOI: https://doi.org/10.1038/s41586-020-2451-1
This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source.