The biological sense of smell is an impressive piece of engineering designed by evolution. In our olfactory epithelium, we can find millions of sensory neurons whose cilia extend in a mucous film. In the membranes of these cilia, the neurons specialise in expressing a single type of olfactory receptor (OR). Humans have approximately 400 types of olfactory receptors, some of which are highly selective to odorants of high biological significance, while others are more general and respond to a wide variety of volatile compounds. Odours are encoded by a pattern of response across these 400 receptors, allowing us to discriminate millions of different odours. Axons from neurons expressing the same type of olfactory receptor converge in glomeruli located in the olfactory bulb and the lead neurons send the information to the olfactory cortex for recognition. 

Already four decades ago, the pioneers of artificial olfaction proposed the grand vision of a biomimetic system that could emulate the biological sense of smell. Thus, the combination of a multiplicity of solid-state chemical sensors and an appropriate processing of the response pattern by artificial intelligence techniques made it possible to discriminate odorants. This concept has given rise to a rich literature and important practical applications in environment, health and food. 

However, by the end of the 1990s, it was already clear that the electronic noses of the time were very different from the biological system and offered far inferior performance. New avenues had to be explored by going deeper into biological inspiration: both the chemical sense and the processing of information had to be similar to the biological sense of smell. At that time, the use of olfactory receptors as the basis for biosensors was a dream. 

Slowly but surely, significant progress has been made in the last 20 years. Biosensors based on olfactory sensory neurons have been successfully integrated, but also olfactory receptors expressed in cell membranes, nanovesicles and, more recently, lipid nanodiscs have been explored. On a parallel track, researchers developed neuromorphic computational models inspired by the neural circuits of the olfactory bulb and olfactory cortex. Other groups have previously demonstrated the integration of a limited number of mouse and fruit fly olfactory receptors for odorant recognition on carbon nanotubes. 

In contrast, this work describes the integration of three human olfactory receptors into nanodiscs on a graphene sheet and an artificial synapse for the differentiation of short-chain fatty acids. 

The researchers characterise the dose-response characteristic of the three biosensors and demonstrate a differential response to short-chain fatty acids. The results are very interesting, although two of the receptors used are shown to be highly correlated with each other. The authors use the transient response of the biosensors for fatty acid recognition using a conventional neural network and obtain recognition rates of 90 %. 

This work is another step towards the dream of artificial olfaction: to technologically emulate the sense of smell and get closer to the enigma of human chemical perception. However, it is still necessary to increase the number of integrated receptors, which, for the moment, is limited to a few units. And, above all, to encourage their use in practice, it is necessary to extend the lifetime of these devices, which at present is typically only one or two weeks. In the future we can envisage the use of olfactory prostheses based on these technologies for people with anosmia, but there is still some way to go.