2014217 NeuroNex: From Odor to Action - Discovering Principles of Olfactory-Guided Natural Behavior
Lead Research Organisation:
University of Hertfordshire
Department Name: School of Computer Science
Abstract
This project addresses a central question of neuroscience: How do animals use information from stimuli in their environment to guide natural behav- iors? Several factors support our plan to address this question by focusing on the olfactory system. Olfaction is an evolutionarily ancient sense. Animals gather odors from their environments to inform survival-critical behaviors such as finding food and mates, or avoiding predators and toxins. These natural behaviors can be largely replicated in lab settings, using stimuli that are natural in composition, and dynamics. Olfactory systems of mammals and insects have independently evolved common structural elements at successive levels of processing in their central nervous systems. Olfactory systems share these structural elements likely due to convergent evolution towards a set of similar solutions to shared olfactory problems. This convergence suggests that the comparative approach that we propose will lead to the discovery of fundamental principles and constraints that could not be discovered by study of a single species-which highlights the importance of the planned theoretical work. Because this project is interdisciplinary, comparative, and theory-driven, it will require a well-integrated team, working synergistically to address overlapping questions. We will coordinate our efforts to achieve convergence and synergy between team members from diverse backgrounds, expertise, and interests, that are key to the transformative potential of this work.
Technical Summary
Fundamental to the work that we propose is a comparative approach to understanding neural representations and behavior in a range of species including mice, flies (adults and larvae), bees and locusts. Each one of these species offers advantages including experimental tractability, behavioral complexity, relevance of the specific ethological niche, etc. Nonetheless we propose to identify common theoretical principles in the ways in which the nervous systems of these organisms represent and transform olfactory stimuli, as well as how their behavioral responses change as a function of odor identity and intensity.
Theories at different scales and levels of abstraction will inform and be tested in this project. To exploit the variety of data sources and species we propose to use will require that we identify common principles and test central theoretical ideas. Generally, neuroscience needs to bring experimentalists and theorists together in a more systematic and productive way to maximize the benefit of novel experimental approaches and make investments into complex and expensive experiments more valuable. However, despite widespread agreement on the general importance of theory and modeling in neuroscience, there is no consensus on which kind/level of theory will be most important for making progress. Traditional dynamical systems models of neurons and circuits have provided insights into the nature and limits of single neuron and local circuit computation. Statistical models provide important information about the information available to neurons for making decisions and generating actions. Therefore in this project we bring together computational/theoretical approaches in areas like dynamical systems, information theory, abstract geometry, machine learning/AI, and machine vision.
Theories at different scales and levels of abstraction will inform and be tested in this project. To exploit the variety of data sources and species we propose to use will require that we identify common principles and test central theoretical ideas. Generally, neuroscience needs to bring experimentalists and theorists together in a more systematic and productive way to maximize the benefit of novel experimental approaches and make investments into complex and expensive experiments more valuable. However, despite widespread agreement on the general importance of theory and modeling in neuroscience, there is no consensus on which kind/level of theory will be most important for making progress. Traditional dynamical systems models of neurons and circuits have provided insights into the nature and limits of single neuron and local circuit computation. Statistical models provide important information about the information available to neurons for making decisions and generating actions. Therefore in this project we bring together computational/theoretical approaches in areas like dynamical systems, information theory, abstract geometry, machine learning/AI, and machine vision.
Planned Impact
This coordinated project on the neuroscience of olfaction across species will have important societal impacts in science, technology, health, and policy. Given the complexity and high dimensionality of chemical space and its primacy in driving behavior among most species, studying how odor leads to action promises to provide insight into optimal biological solutions for encoding complex information about the external world. Biology far outstrips human engineered systems in identifying and processing our chemical world. This is why animal chemical detectors (e.g. scent dogs and explosive-detecting rats) are the best available solutions for applications like explosive and drug detection. Because it is a dominant mode of communication among plants and animals, chemical communication is critical to healthy ecosystems and productive agriculture, as well as for repelling pests. In humans, loss of the sense of smell is associated with loss of a sense of quality of life and the inability to identify danger signals, such as smoke and rotting food. Elucidating biological solutions to olfaction can inform the development of algorithms and engineered devices for detection and identification of chemicals in applications that span the range from homeland security to food safety.
Organisations
- University of Hertfordshire (Lead Research Organisation)
- Francis Crick Institute (Collaboration)
- California Institute of Technology (Collaboration)
- University of Colorado Boulder (Collaboration)
- John B. Pierce Laboratory (Collaboration)
- University of Utah (Collaboration)
- Arizona State University (Collaboration)
- Penn State University (Collaboration)
- Cornell University (Collaboration)
Publications
Crimaldi J
(2022)
Active sensing in a dynamic olfactory world.
in Journal of computational neuroscience
Dennler N
(2022)
Drift in a popular metal oxide sensor dataset reveals limitations for gas classification benchmarks
in Sensors and Actuators B: Chemical
Drix D
(2021)
Resolving Fast Gas Transients with Metal Oxide Sensors.
in ACS sensors
Tayarani-Najaran MH
(2021)
Event-Based Sensing and Signal Processing in the Visual, Auditory, and Olfactory Domain: A Review.
in Frontiers in neural circuits
Description | (HBP SGA3) - Human Brain Project Specific Grant Agreement 3 |
Amount | € 150,000,000 (EUR) |
Funding ID | 945539 |
Organisation | European Commission |
Sector | Public |
Country | European Union (EU) |
Start | 04/2020 |
End | 03/2023 |
Title | Elife '22 Smarts FP |
Description | A collection of SMARTS substructure definitions that work well in odorant space. Published in Burton et al., Elife 2022. This entry describes the code which has been published on github. |
Type Of Material | Data analysis technique |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | I have received a report of another lab using it, but no citable evidence of it being used is available at this point. We frequently use it in our own research. |
URL | https://github.com/BioMachineLearning/elife22_smarts_fp |
Description | Development of Molecular features for odorants |
Organisation | California Institute of Technology |
Country | United States |
Sector | Academic/University |
PI Contribution | We developed molecular fingerprints tailored to a set of odorants used in a physiological study of odorant responses in the olfactory bulb of mice. We shared them with researchers at the University of Utah and Caltech. In a second thread of collaboration we help researchers at Pennsylvania State University explore potential chemotopy in the olfactory bulb of mice. We shared the molecular fingerprints we developed with them, and computed sets of third-party fingerprints for their set of odorants. We help all our collaboration partners interpret the meaning of chemical features. |
Collaborator Contribution | Researchers from the University of Utah have provided recordings of odorant responses obtained in the olfactory bulb of mice. Researchers at Caltech have provided a database of odorants found in natural sources that they have compiled from the literature. Researchers at Pennsylvania State University have contributed algorithms for data modelling and results from data analysis. |
Impact | Mapping odorant sensitivities reveals a sparse but structured representation of olfactory chemical space by sensory input to the mouse olfactory bulb. Burton SD, Brown A, Eiting TP, Youngstrom IA, Rust TC, Schmuker M, Wachowiak M (2022). eLife, 11:e80470. https://doi.org/10.7554/eLife.80470 (Open Access). This is a multi-disciplinary collaboration, involving computer science, machine learning, experimental neuroscience, and mathematics. |
Start Year | 2021 |
Description | Development of Molecular features for odorants |
Organisation | Penn State University |
Country | United States |
Sector | Academic/University |
PI Contribution | We developed molecular fingerprints tailored to a set of odorants used in a physiological study of odorant responses in the olfactory bulb of mice. We shared them with researchers at the University of Utah and Caltech. In a second thread of collaboration we help researchers at Pennsylvania State University explore potential chemotopy in the olfactory bulb of mice. We shared the molecular fingerprints we developed with them, and computed sets of third-party fingerprints for their set of odorants. We help all our collaboration partners interpret the meaning of chemical features. |
Collaborator Contribution | Researchers from the University of Utah have provided recordings of odorant responses obtained in the olfactory bulb of mice. Researchers at Caltech have provided a database of odorants found in natural sources that they have compiled from the literature. Researchers at Pennsylvania State University have contributed algorithms for data modelling and results from data analysis. |
Impact | Mapping odorant sensitivities reveals a sparse but structured representation of olfactory chemical space by sensory input to the mouse olfactory bulb. Burton SD, Brown A, Eiting TP, Youngstrom IA, Rust TC, Schmuker M, Wachowiak M (2022). eLife, 11:e80470. https://doi.org/10.7554/eLife.80470 (Open Access). This is a multi-disciplinary collaboration, involving computer science, machine learning, experimental neuroscience, and mathematics. |
Start Year | 2021 |
Description | Development of Molecular features for odorants |
Organisation | University of Utah |
Country | United States |
Sector | Academic/University |
PI Contribution | We developed molecular fingerprints tailored to a set of odorants used in a physiological study of odorant responses in the olfactory bulb of mice. We shared them with researchers at the University of Utah and Caltech. In a second thread of collaboration we help researchers at Pennsylvania State University explore potential chemotopy in the olfactory bulb of mice. We shared the molecular fingerprints we developed with them, and computed sets of third-party fingerprints for their set of odorants. We help all our collaboration partners interpret the meaning of chemical features. |
Collaborator Contribution | Researchers from the University of Utah have provided recordings of odorant responses obtained in the olfactory bulb of mice. Researchers at Caltech have provided a database of odorants found in natural sources that they have compiled from the literature. Researchers at Pennsylvania State University have contributed algorithms for data modelling and results from data analysis. |
Impact | Mapping odorant sensitivities reveals a sparse but structured representation of olfactory chemical space by sensory input to the mouse olfactory bulb. Burton SD, Brown A, Eiting TP, Youngstrom IA, Rust TC, Schmuker M, Wachowiak M (2022). eLife, 11:e80470. https://doi.org/10.7554/eLife.80470 (Open Access). This is a multi-disciplinary collaboration, involving computer science, machine learning, experimental neuroscience, and mathematics. |
Start Year | 2021 |
Description | NeuroNex Odor2Action IRG 3 |
Organisation | Arizona State University |
Country | United States |
Sector | Academic/University |
PI Contribution | We interact within the Interdisciplinary Research Group (IRG) 3 "Active Sensing". We investigate the physical structure of odour environments and how animals and robotic agents can exploit them for odour-guided navigation. We contributed to a review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 We further contribute by developing analysis algorithms and analysing gas plume dispersion data obtained in a turbulent wind tunnel at the University of Colorado Boulder. We discuss our findings with the other IRG 3 members. We collaborate with researchers at the Francis Crick Institute (London) in record complex odour plume replicates using electronic gas sensors. We regularly communicate with the whole IRG 3, with the researchers mentioned above plus researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University and provide feedback on their work and discuss future avenues and collaborations. |
Collaborator Contribution | Researchers at the University of Colorado Boulder has provided wind tunnel recordings of turbulent gas plume dispersion which we analyse using methods and algorithms we developed. Researchers at the Francis Crick Institute (London) provide laboratory facilities and know-how in delivering complex odour plumes that we use to characterise an electronic gas sensor platform that we developed within the NeuroNex: Odor2Action award. Researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University provide feedback on our work and we discuss future avenues and collaborations during regular IRG 3 meetings. |
Impact | Review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 Further manuscripts in preparation. |
Start Year | 2020 |
Description | NeuroNex Odor2Action IRG 3 |
Organisation | Cornell University |
Department | Weill Cornell Medicine |
Country | United States |
Sector | Academic/University |
PI Contribution | We interact within the Interdisciplinary Research Group (IRG) 3 "Active Sensing". We investigate the physical structure of odour environments and how animals and robotic agents can exploit them for odour-guided navigation. We contributed to a review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 We further contribute by developing analysis algorithms and analysing gas plume dispersion data obtained in a turbulent wind tunnel at the University of Colorado Boulder. We discuss our findings with the other IRG 3 members. We collaborate with researchers at the Francis Crick Institute (London) in record complex odour plume replicates using electronic gas sensors. We regularly communicate with the whole IRG 3, with the researchers mentioned above plus researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University and provide feedback on their work and discuss future avenues and collaborations. |
Collaborator Contribution | Researchers at the University of Colorado Boulder has provided wind tunnel recordings of turbulent gas plume dispersion which we analyse using methods and algorithms we developed. Researchers at the Francis Crick Institute (London) provide laboratory facilities and know-how in delivering complex odour plumes that we use to characterise an electronic gas sensor platform that we developed within the NeuroNex: Odor2Action award. Researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University provide feedback on our work and we discuss future avenues and collaborations during regular IRG 3 meetings. |
Impact | Review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 Further manuscripts in preparation. |
Start Year | 2020 |
Description | NeuroNex Odor2Action IRG 3 |
Organisation | Francis Crick Institute |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | We interact within the Interdisciplinary Research Group (IRG) 3 "Active Sensing". We investigate the physical structure of odour environments and how animals and robotic agents can exploit them for odour-guided navigation. We contributed to a review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 We further contribute by developing analysis algorithms and analysing gas plume dispersion data obtained in a turbulent wind tunnel at the University of Colorado Boulder. We discuss our findings with the other IRG 3 members. We collaborate with researchers at the Francis Crick Institute (London) in record complex odour plume replicates using electronic gas sensors. We regularly communicate with the whole IRG 3, with the researchers mentioned above plus researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University and provide feedback on their work and discuss future avenues and collaborations. |
Collaborator Contribution | Researchers at the University of Colorado Boulder has provided wind tunnel recordings of turbulent gas plume dispersion which we analyse using methods and algorithms we developed. Researchers at the Francis Crick Institute (London) provide laboratory facilities and know-how in delivering complex odour plumes that we use to characterise an electronic gas sensor platform that we developed within the NeuroNex: Odor2Action award. Researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University provide feedback on our work and we discuss future avenues and collaborations during regular IRG 3 meetings. |
Impact | Review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 Further manuscripts in preparation. |
Start Year | 2020 |
Description | NeuroNex Odor2Action IRG 3 |
Organisation | John B. Pierce Laboratory |
Country | United States |
Sector | Charity/Non Profit |
PI Contribution | We interact within the Interdisciplinary Research Group (IRG) 3 "Active Sensing". We investigate the physical structure of odour environments and how animals and robotic agents can exploit them for odour-guided navigation. We contributed to a review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 We further contribute by developing analysis algorithms and analysing gas plume dispersion data obtained in a turbulent wind tunnel at the University of Colorado Boulder. We discuss our findings with the other IRG 3 members. We collaborate with researchers at the Francis Crick Institute (London) in record complex odour plume replicates using electronic gas sensors. We regularly communicate with the whole IRG 3, with the researchers mentioned above plus researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University and provide feedback on their work and discuss future avenues and collaborations. |
Collaborator Contribution | Researchers at the University of Colorado Boulder has provided wind tunnel recordings of turbulent gas plume dispersion which we analyse using methods and algorithms we developed. Researchers at the Francis Crick Institute (London) provide laboratory facilities and know-how in delivering complex odour plumes that we use to characterise an electronic gas sensor platform that we developed within the NeuroNex: Odor2Action award. Researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University provide feedback on our work and we discuss future avenues and collaborations during regular IRG 3 meetings. |
Impact | Review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 Further manuscripts in preparation. |
Start Year | 2020 |
Description | NeuroNex Odor2Action IRG 3 |
Organisation | University of Colorado Boulder |
Department | Civil, Environmental, and Architectural Engineering |
Country | United States |
Sector | Academic/University |
PI Contribution | We interact within the Interdisciplinary Research Group (IRG) 3 "Active Sensing". We investigate the physical structure of odour environments and how animals and robotic agents can exploit them for odour-guided navigation. We contributed to a review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 We further contribute by developing analysis algorithms and analysing gas plume dispersion data obtained in a turbulent wind tunnel at the University of Colorado Boulder. We discuss our findings with the other IRG 3 members. We collaborate with researchers at the Francis Crick Institute (London) in record complex odour plume replicates using electronic gas sensors. We regularly communicate with the whole IRG 3, with the researchers mentioned above plus researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University and provide feedback on their work and discuss future avenues and collaborations. |
Collaborator Contribution | Researchers at the University of Colorado Boulder has provided wind tunnel recordings of turbulent gas plume dispersion which we analyse using methods and algorithms we developed. Researchers at the Francis Crick Institute (London) provide laboratory facilities and know-how in delivering complex odour plumes that we use to characterise an electronic gas sensor platform that we developed within the NeuroNex: Odor2Action award. Researchers at Cornell University, the John B. Pierce Laboratory, and Arizona State University provide feedback on our work and we discuss future avenues and collaborations during regular IRG 3 meetings. |
Impact | Review article published in the Journal of Computational Neuroscience in September 2021: https://doi.org/10.1007/s10827-021-00798-1 Further manuscripts in preparation. |
Start Year | 2020 |
Description | Interview with Journalist |
Form Of Engagement Activity | A press release, press conference or response to a media enquiry/interview |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | I was interviewed by Wynne Parry, science journalist, about a series of preprints describing a novel method to predict scent, on the background of research conducted within the NeuroNex Odor2Action award involving features to efficiently describe odorant space. |
Year(s) Of Engagement Activity | 2022 |
URL | https://www.scientificamerican.com/article/ai-predicts-what-chemicals-will-smell-like-to-a-human/ |