Project

Back to overview

Synaptic trading: energy savings versus information

English title Synaptic trading: energy savings versus information
Applicant Jolivet Renaud
Number 170079
Funding scheme Project funding (Div. I-III)
Research institution Département de physique nucléaire et corpusculaire Université de Genève
Institution of higher education University of Geneva - GE
Main discipline Neurophysiology and Brain Research
Start/End 01.11.2016 - 31.08.2020
Approved amount 429'000.00
Show all

All Disciplines (2)

Discipline
Neurophysiology and Brain Research
Biophysics

Keywords (5)

energetic optimality; multicompartment neuron models; synaptic energy consumption; information theory; brain energy

Lay Summary (French)

Lead
Bien que le cerveau ne consomme que peu d’énergie en comparaison d’autres systèmes d’information comme un ordinateur personnel, il représente une fraction significative du métabolisme énergétique chez l’homme. Ne représentant que 2% de la masse du corps chez l’adulte, le cerveau est responsable pour 20% de sa consommation énergétique. Cela pose la question de l’existence de contraintes énergétiques dans le fonctionnement du cerveau.
Lay summary

Lors d’études préalables, nous avons démontré que l’essentiel de la consommation énergétique prend place aux synapses, les structures cellulaires qui connectent les neurones entre eux et qui servent de substrat à la communication neuronale. Plus récemment, nous avons démontré que certaines synapses dans le cortex et dans le système visuel opèrent un compromis entre le transfert d’information et leur consommation énergétique. Afin de limiter leur budget énergétique, ces synapses sont configurées non pas pour maximiser le transfert d’information mais pour maximiser le ratio entre le transfert d’information et la consommation d’énergie concomitante. Ce compromis entre énergie et information n’en est probablement pas pour autant un principe universel. Ce projet explorera dans quelles conditions ce compromis apparaît et quels sont les mécanismes qui le contrôlent. Cela nous permettra d’une part de mieux comprendre les principes dictant l’organisation fonctionnelle du cerveau et d’autre part de fournir des pistes à l’industrie qui explore aujourd’hui le design de processeurs à basse consommation inspirés du fonctionnement du cerveau.

Direct link to Lay Summary Last update: 08.11.2016

Responsible applicant and co-applicants

Employees

Publications

Publication
Periaxonal and nodal plasticities modulate action potential conduction in the adult mouse brain
Cullen Carlie L., Pepper Renee E., Clutterbuck Mackenzie T., Pitman Kimberley A., Oorschot Viola, Auderset Loic, Tang Alexander D., Ramm Georg, Emery Ben, Rodger Jennifer, Jolivet Renaud B., Young Kaylene M. (2021), Periaxonal and nodal plasticities modulate action potential conduction in the adult mouse brain, in Cell Reports, 34(3), 108641-108641.
Modelling Neuromodulated Information Flow and Energetic Consumption at Thalamic Relay Synapses
Conrad Mireille, Jolivet Renaud B. (2020), Modelling Neuromodulated Information Flow and Energetic Consumption at Thalamic Relay Synapses, in Artificial Neural Networks and Machine Learning – ICANN 202029th International Conference on Artific, Springer International Publishing, Cham.
Analysis of Signaling Mechanisms Regulating Microglial Process Movement
Kyrargyri Vasiliki, Attwell David, Jolivet Renaud Blaise, Madry Christian (2019), Analysis of Signaling Mechanisms Regulating Microglial Process Movement, Springer New York, New York, NY.
Energy-efficient information transfer at thalamocortical synapses
Harris Julia Jade, Engl Elisabeth, Attwell David, Jolivet Renaud Blaise (2019), Energy-efficient information transfer at thalamocortical synapses, in PLOS Computational Biology, 15(8), e1007226-e1007226.
Pivotal role of carnosine in the modulation of brain cells activity: Multimodal mechanism of action and therapeutic potential in neurodegenerative disorders
Caruso Giuseppe, Caraci Filippo, Jolivet Renaud B. (2019), Pivotal role of carnosine in the modulation of brain cells activity: Multimodal mechanism of action and therapeutic potential in neurodegenerative disorders, in Progress in Neurobiology, 175, 35-53.
A Process for Digitizing and Simulating Biologically Realistic Oligocellular Networks Demonstrated for the Neuro-Glio-Vascular Ensemble
Coggan Jay S., Calì Corrado, Keller Daniel, Agus Marco, Boges Daniya, Abdellah Marwan, Kare Kalpana, Lehväslaiho Heikki, Eilemann Stefan, Jolivet Renaud Blaise, Hadwiger Markus, Markram Henry, Schürmann Felix, Magistretti Pierre J. (2018), A Process for Digitizing and Simulating Biologically Realistic Oligocellular Networks Demonstrated for the Neuro-Glio-Vascular Ensemble, in Frontiers in Neuroscience, 12, 664.
Microglial Ramification, Surveillance, and Interleukin-1β Release Are Regulated by the Two-Pore Domain K+ Channel THIK-1
Madry Christian, Kyrargyri Vasiliki, Arancibia-Cárcamo I. Lorena, Jolivet Renaud, Kohsaka Shinichi, Bryan Robert M., Attwell David (2018), Microglial Ramification, Surveillance, and Interleukin-1β Release Are Regulated by the Two-Pore Domain K+ Channel THIK-1, in Neuron, 97(2), 299-312.e6.
Energy use constrains brain information processing
Conrad Mireille, Engl Elisabeth, Jolivet Renaud B. (2017), Energy use constrains brain information processing, in 2017 IEEE International Electron Devices Meeting (IEDM), San Francisco, CA, USAIEEE, San Francisco CA, USA.

Datasets

Code for simulations performed during the course of the project

Author Jolivet, Renaud
Persistent Identifier (PID) https://github.com/JolivetLab
Repository GitHub
Abstract
The code developed to perform simulations within the frame of the project is being progressively deposited on our GitHub lab page (https://github.com/JolivetLab) after curation and test.

Collaboration

Group / person Country
Types of collaboration
University of Tasmania, Prof. Kaylene Young Australia (Oceania)
- Publication
University of Torino, Prof. Corrado Cali Italy (Europe)
- Publication
Khalifa University, Prof. Dimitris Goussis United Arab Emirates (Asia)
- Publication
University of Catania, Dr. Giuseppe Caruso Italy (Europe)
- Publication
KAUST, Prof. Pierre Magistretti Saudi Arabia (Asia)
- Publication
CERN, Dr. Magdalena Kowalska Switzerland (Europe)
- Publication

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
Bernstein Conference 2019 Poster Plasticity rules for learning unreliable noisy inputs under energetic constraints 18.09.2020 Berlin, Germany Grytskyy Dmytro;
International Conference on Synergy of Sciences (ICSS - 2020) Talk given at a conference The brain’s heterocellular complexity through the lens of brain energetics 27.02.2020 Thanjavur, India Jolivet Renaud;
10th Luxembourg Symposium, Systems Neuroscience: From molecule to brain dynamics Talk given at a conference Energy-efficient information transfer in neural networks 05.11.2019 Belval, Luxembourg Jolivet Renaud;
Methods of Information Theory in Computational Neuroscience Talk given at a conference Combining information theory and energetics into a coherent framework to study the brain's heterocellular circuits 16.07.2019 Barcelona, Spain Jolivet Renaud;
CNS*2019 Poster Mutual Information vs. Transfer Entropy in Spike-Based Neuroscience 13.07.2019 Barcelona, Spain Conrad Mireille;
CNS*2019 Poster Plasticity rules for learning sequential inputs under energetic constraints 13.07.2019 Barcelona, Spain Grytskyy Dmytro;
Colloque Annuel GDR BioComp Talk given at a conference Energy-efficient information transfer in neural networks 14.05.2019 Lille, France Jolivet Renaud;
Lemanic Neuroscience Annual Meeting 2019 Poster Mutual Information vs. Transfer Entropy in Spike-Based Neuroscience 15.04.2019 Les Diablerêts, Switzerland Conrad Mireille;
Royal Society Theo Murphy Discussion Meeting: Integrated control of cerebral blood flow Talk given at a conference Building a mathematical model of brain energy metabolism 04.12.2018 Newport Pagnell, Great Britain and Northern Ireland Jolivet Renaud;
Invited colloquium at Maastricht Centre for Systems Biology (invited by Prof. Ilja Arts) Individual talk Modelling brain energy metabolism 27.11.2018 Maastricht, Netherlands Jolivet Renaud;
Nanoscale modelling of synaptic transmission, calcium dynamics and transduction, cell sensing and chemotaxis Talk given at a conference Energy-efficient information transfer at synapses 09.10.2018 Pisa, Italy Jolivet Renaud;
Bernstein Conference 2018 Poster Learning rules balancing information inference and energy costs 26.09.2018 Berlin, Germany Grytskyy Dmytro;
Lemanic Neuroscience Annual Meeting 2018 Poster Energy consumption and information transfer at synapses 15.09.2018 Les Diablerêts, Switzerland Conrad Mireille;
Neural Coding 2018 Talk given at a conference Energy-efficient information transfer at synapses 13.09.2018 Torino, Italy Jolivet Renaud;
Methods of Information Theory in Computational Neuroscience Talk given at a conference Mutual information vs. Transfer entropy in spike-based neuroscience 15.07.2018 Seattle, United States of America Conrad Mireille;
Innovative technologies to study brain energy metabolism with high spatial and temporal resolution Talk given at a conference Modelling the metabolic cost of information at synapses and brain energy metabolism 10.04.2018 KAUST, Saudi Arabia Jolivet Renaud;
Royal Society Theo Murphy Discussion Meeting: Toward a Computational Theory of Life Talk given at a conference Energy-efficient information transfer at synapses 21.03.2018 Newport Pagnell, Great Britain and Northern Ireland Jolivet Renaud;
AMCOS Analysis and Modeling of Complex Oscillatory Systems Poster Plastic neural networks with oscillating dynamics inferring context from the input 19.03.2018 Barcelona, Spain Grytskyy Dmytro;
Invited colloquium at Ludwig Maximilian University of Munich (invited by Prof. Nikolaus Plesnila) Individual talk Energy-efficient information transfer at synapses 06.02.2018 Munich, Germany Jolivet Renaud;
IEEE IEDM 2017 Talk given at a conference Energy use constrains brain information processing 05.12.2017 San Francisco CA, United States of America Jolivet Renaud;
Methods of Information Theory in Computational Neuroscience Talk given at a conference Energy-efficient information transfer at synapses 19.07.2017 Anvers, Belgium Jolivet Renaud;
RIKEN Brain Science Institute Summer Program Poster Energy consumption and information transfer at synapses 07.06.2017 Tokyo, Japan Conrad Mireille;
Invited colloquium at UCSC (invited by Prof. Alan Litke) Individual talk Neuroenergetics: How energy constraints shape brain function 18.11.2016 University of California Santa Cruz, United States of America Jolivet Renaud;


Communication with the public

Communication Title Media Place Year
New media (web, blogs, podcasts, news feeds etc.) At the interface of AI, Neuroscience and Policy YouTube International 2020
New media (web, blogs, podcasts, news feeds etc.) Future of Research YouTube International 2019
Other activities Taste of Science, San Antonio TX, USA, 5 Feb 2019 International 2019
Talks/events/exhibitions World Conference of Science Journalists, University of Geneva, Geneva, Switzerland, 5 July 2019 Western Switzerland 2019
Talks/events/exhibitions Nuit de la Science, University of Geneva, Geneva, Switzerland, 7-8 July 2018 Western Switzerland 2018

Awards

Title Year
Elected to the Board of Directors of the Organization for Computational Neurosciences (https://www.cnsorg.org/board-of-directors) for a first term 2019-2021. 2018

Abstract

Information transmission in the brain is energetically expensive, yet has to satisfy demands of speed and signal-to-noise reliability. We have recently shown that the strong retinogeniculate synapse relaying information from the retina to the thalamus resolves these competing constraints by maximizing energetic efficiency when transferring information. In their physiological state, these synapses are not set to transmit the maximum amount of information possible: information transmission increases when larger excitatory postsynaptic currents (EPSCs) are injected into the postsynaptic thalamic neuron. However, EPSCs that are larger or smaller than physiological EPSCs decrease the information transmitted per energy used. The physiological EPSC size therefore maximizes energy efficiency rather than pure information transfer across the synapse. In other words, the retinogeniculate synapse trades information for energy savings.This finding echoes our earlier finding that, for information being transmitted at a synapse in the presence of noise, a low presynaptic release probability is the optimal solution to maximize the information transmitted per energy expended on the transmission, providing an exciting explanation for the previously not understood fact that synapses are extremely unreliable, often only releasing neurotransmitter a quarter of the times that an action potential is fired.Together, these findings suggest maximization of information transmission per energy used as a design principle in the brain. However, it is unclear how broadly this principle applies. Whether energy efficiency at excitatory synapses is a special property of strong relay synapses, or a more general principle, also governing synaptic inputs that contribute more weakly to determining the output of the postsynaptic cell is an open question. These findings also raise the question of what mechanisms are in effect in order to achieve energetic efficiency of information transfer at synapses.This project will address these questions using information theory and simulations of biologically validated neuron models. In particular, we will investigate under which conditions energetic efficiency of information transfer arises at weak synapses, what biophysical properties of neurons give rise to energetic efficiency of information transfer and how it can be modulated by local network activity. Finally, we will test the hypothesis that energetic efficiency of information transfer is an emergent property of Hebbian learning, more specifically of spike-timing dependent plasticity. In addressing these questions, this project will open a novel avenue to study neural coding by clarifying the natural physical energetic constraints the brain has to do with. This approach eventually holds the potential to identify some of the mechanisms which link energy consumption at synapses to pathologies. Additionally, this approach might open new avenues in the design of artificial computing systems where energy reduction techniques are now being introduced.
-