View all journals Search Login

Explore content About the journal Publish with us Subscribe Sign up for alerts RSS feed

nature outlook article

OUTLOOK 22 June 2022 Related Articles

Building neural networks that smell The science of smell steps into the
spotlight

like a brain
Computational neuroscientist Guangyu Robert Yang lifts the lid on the use of machine Unpicking the link between smell and
memories
learning to detect and process odours, and the wider implications for neuroscience.

Conor Purcell
The science behind COVID’s assault
on smell

How to bring back the sense of smell

The dogs learning to sniff out disease

Restoring smell with an electronic

nose

Olfactory receptors are not unique to

the nose

Credit: Guangyu Robert Yang Sniffing out smell’s effects on human

behaviour

Advances in artificial intelligence (AI) have given machines the ability to see and speak.
Training an AI system to smell, however, is a nascent endeavour. Last year, researchers at Sponsor feature: From molecules to perception: 126
Columbia University in New York City developed a neural network that evolved an olfactory years at the forefront of olfactory science
system markedly similar to that of the fruit fly Drosophila melanogaster (P. Y. Wang et al.
Neuron 109, 3879–3892; 2021). The study’s leader, Guangyu Robert Yang, is now principal
investigator of the MetaConscious Group at the McGovern Institute for Brain Research at the Subjects
Massachusetts Institute of Technology in Cambridge. He spoke to Nature about the use of
Computer science Machine learning Brain
machine learning to process odour information, and the wider implications for the study of
the brain. Neuroscience

How do you make a machine that can smell?

For the most part, it is a matter of building good sensors. There are many types in the
olfactory system; the fruit fly, for example, has around 50. People have even more. They give
us a rich sense of smell, and enable us to tell the difference between different types of Sign up to Nature Briefing
chocolate, and check whether food is fresh. These abilities rely on having good, accurate An essential round-up of science news,
sensors, and incorporating those into machines is hard. opinion and analysis, delivered to your inbox
every weekday.

Building vision into machines has the opposite issue:

RELATED
researchers have the sensors, but the circuitry — the network — Email address

has been really difficult to construct. We don’t fully understand e.g. [email protected]

the circuitry involved in olfaction, but we think it is a simpler Yes! Sign me up to receive the daily Nature Briefing
problem than for vision. email. I agree my information will be processed in
accordance with the Nature and Springer Nature
Limited Privacy Policy.

Where do you start?

Sign up
Generally speaking, we have a good understanding of the
architecture of the fruit fly’s olfactory system: input neurons
project to a second layer of neurons, then a third. And we know
Part of Nature Outlook: Smell the number of neurons in each layer. This happens to resemble
an artificial neural network, which means we were able to build a
network that is structured in a similar way to a fruit fly’s olfactory system using standard
machine-learning tools.

After we had that basic network in place, we needed to train it to do something — and that was
trickier. Engineers working on machine vision have a large library of images that they use to
train neural networks. We didn’t have anything like that for olfaction, so we had to develop an
artificial data set. These data sets were simple compared with natural odours, and were
designed around one essential property: that mixing multiple odours from a single category
— apple, for instance — results in a combined smell that is still recognizably apple.

Once we trained the network using this data set, we found that the system evolved the neural
connectivity that we had already observed in the fruit fly — it was almost the same system of
signal processing at work in the animal and the machine. We were surprised by that, given
how we trained the network. Such a simplified artificial data set would probably not work for
vision.

What does the similarity between these networks tell us about

olfaction?
One point of criticism that has been directed at neural networks is that the way in which they
are trained — by being fed lots of data — does not reflect the biology. And therefore, even
though they produce reasonable results, people say that the system can’t be relied on for
understanding biological systems. But our neural network evolved an olfactory system in a
different way to biological evolution, and yet we end up with the same result. Our model
prioritizes efficiency, just as natural selection does, and reaches the same conclusion on how
best to perform the task.

What’s the next step for modelling the brain?

Computational models have been developed for many different systems in the brain over the
past ten years. The holy grail now is putting them together; these kinds of multi-system model
are the main focus of our laboratory. It is a huge challenge — we don’t really know how these
systems work together in the human brain, and finding out involves a lot of collaboration with
neuroscientists in what are currently considered separate fields. But there has been a clear
path of improvement in developing neural networks that behave like the brain, and progress
is only going to accelerate. I think that in 20 years we will have made tremendous progress
towards understanding how the brain functions as a whole, and these tools will help us to do
that.

doi: https://doi.org/10.1038/d41586-022-01633-y

This interview has been edited for length and clarity.

This article is part of Nature Outlook: Smell, an editorially independent supplement produced
with the financial support of third parties. About this content.

Latest on:
Computer science Machine learning Brain

DeepMind AI learns These ‘quantum-proof’ Open-source language

simple physics like a algorithms could AI challenges big tech’s
baby safeguard against models
NEWS | 11 JUL 22
future cyberattacks NEWS | 22 JUN 22

NEWS | 08 JUL 22

Jobs

Multiple Positions Open in Stomatology Hospital, School of Stomatology, Zhejiang

University School of Medicine(ZJUSS)
Stomatology Hospital, School of Stomatology, Zhejiang University School of Medicine(ZJUSS)
Hangzhou, China

Postdoctoral position: Quantum mechanical descriptors for improved prediction of

ADMET molecular properties
University of Luxembourg
Luxembourg, Luxembourg

Research & Development Specialist for Neurometabolic Disease Modeling in Zebrafish

University of Luxembourg
Luxembourg, Luxembourg

R&D specialist in a research-to-market project based on recent materials physics

breakthroughs
University of Luxembourg
Luxembourg, Luxembourg

Nature (Nature) ISSN 1476-4687 (online) ISSN 0028-0836 (print)

Nature portfolio
About us Press releases Press office Contact us

Discover content Publishing policies Author & Researcher services Libraries & institutions
Journals A-Z Nature portfolio policies Reprints & permissions Librarian service & tools
Articles by subject Open access Research data Librarian portal
Nano Language editing Open research
Protocol Exchange Scientific editing Recommend to library
Nature Index Nature Masterclasses
Nature Research Academies
Research Solutions

Advertising & partnerships Career development Regional websites Legal & Privacy
Advertising Nature Careers Nature Africa Privacy Policy
Partnerships & Services Nature Conferences Nature China Use of cookies
Media kits Nature events Nature India Manage cookies/Do not sell my data
Branded content Nature Italy Legal notice
Nature Japan Accessibility statement
Nature Korea Terms & Conditions
Nature Middle East California Privacy Statement

© 2022 Springer Nature Limited