Statistical physics of cognition
Lead Research Organisation:
IMPERIAL COLLEGE LONDON
Department Name: Bioengineering
Abstract
This project asks how the statistical mechanics of critical processes such as avalanches underpin mammalian cognition. This is important because self-organising criticality has been proposed as a general explanation of the architecture and operating point of brain circuitry, however until now it has not been possible to relate it to cognitive function itself. We bring an inter-disciplinary approach to bear upon this problem, simulating neuronal mechanisms at scales from sub-cellular to brain-wide, and employing a new mesoscopic neuroimaging technology to detect avalanches and measure distance to criticality in very large (~10,000) populations of neurons during a cognitive task. This will bring new insight into system-wide brain function during health and disease states.
Publications
Arnaudon A
(2022)
Connecting Hodge and Sakaguchi-Kuramoto through a mathematical framework for coupled oscillators on simplicial complexes
in Communications Physics
Arnaudon A
(2024)
Algorithm 1044: PyGenStability, a Multiscale Community Detection Framework with Generalized Markov Stability
in ACM Transactions on Mathematical Software
August E
(2023)
Finding positively invariant sets and proving exponential stability of limit cycles using Sum-of-Squares decompositions
in Journal of Computational Dynamics
Dimmock S
(2024)
Hierarchical Bayesian modeling of multi-region brain cell count data
Dimmock S
(2024)
Hierarchical Bayesian modeling of multi-region brain cell count data
Dimmock S
(2025)
Hierarchical Bayesian modeling of multiregion brain cell count data.
in eLife
Garcia-Font N
(2022)
Ca2+ imaging of self and other in medial prefrontal cortex during social dominance interactions in a tube test.
in Proceedings of the National Academy of Sciences of the United States of America
Georgiev D
(2024)
Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders
in Proceedings of the National Academy of Sciences
Go MA
(2025)
Advantages of three-photon live imaging for deep tissue analysis.
in Expert review of medical devices
| Title | Light-field deep learning enables high-throughput, scattering-mitigated calcium imaging |
| Description | Raw jGCaMP8f 100-Hz light-field videos Weights for the LISTA-based network for estimating volumes from light fields Weights for the light-field microscope wave-optics forward model CNN (used in the "LNet Matrix" workflow) Described by "Light-field deep learning enables high-throughput, scattering-mitigated calcium imaging" Abstract: Light field microscopy (LFM) enables volumetric, high throughput functional imaging. However, the computational burden and vulnerability to scattering limit light field's application to neuroscience. We present a strategy for volumetric, scattering-mitigated neural circuit activity monitoring. A physics-based deep neural network, LNet, is trained with two-photon volumes and one-photon light fields. A processing pipeline uses LNet to extract calcium activity from light-field videos of jGCaMP8f-expressing neurons in acute cortical slices. The extracted time series have high signal-to-noise ratios and reduced optical crosstalk compared to conventional volume reconstruction. Imaging 100 volumes per second, we observed putative spikes fired at up to 10 Hz and the spatial intermingling of putative ensembles throughout 530 x 530 x 100-micron volumes. Compared to iterative algorithms, LNet LFM cuts light-field video processing time from hours to minutes and hence advances the goal of real-time, scattering-robust volumetric neural circuit imaging for closed-loop and adaptive experimental paradigms. |
| Type Of Art | Film/Video/Animation |
| Year Produced | 2025 |
| URL | https://zenodo.org/doi/10.5281/zenodo.14900716 |
| Title | Light-field deep learning enables high-throughput, scattering-mitigated calcium imaging |
| Description | Raw jGCaMP8f 100-Hz light-field videos Weights for the LISTA-based network for estimating volumes from light fields Weights for the light-field microscope wave-optics forward model CNN (used in the "LNet Matrix" workflow) Described by "Light-field deep learning enables high-throughput, scattering-mitigated calcium imaging" Abstract: Light field microscopy (LFM) enables volumetric, high throughput functional imaging. However, the computational burden and vulnerability to scattering limit light field's application to neuroscience. We present a strategy for volumetric, scattering-mitigated neural circuit activity monitoring. A physics-based deep neural network, LNet, is trained with two-photon volumes and one-photon light fields. A processing pipeline uses LNet to extract calcium activity from light-field videos of jGCaMP8f-expressing neurons in acute cortical slices. The extracted time series have high signal-to-noise ratios and reduced optical crosstalk compared to conventional volume reconstruction. Imaging 100 volumes per second, we observed putative spikes fired at up to 10 Hz and the spatial intermingling of putative ensembles throughout 530 x 530 x 100-micron volumes. Compared to iterative algorithms, LNet LFM cuts light-field video processing time from hours to minutes and hence advances the goal of real-time, scattering-robust volumetric neural circuit imaging for closed-loop and adaptive experimental paradigms. |
| Type Of Art | Film/Video/Animation |
| Year Produced | 2025 |
| URL | https://zenodo.org/doi/10.5281/zenodo.14900715 |
| Description | By using advanced imaging to track thousands of neurons in mice, we discovered that the brain's ability to coordinate across long distances relies on a sophisticated "teamwork" mechanism called synergy. While neurons do sometimes repeat the same information (redundancy), visual stimulation specifically triggers a massive boost in synergistic interactions where cells combine their efforts to create complex, unique signals. These two styles of interaction actually complement one another, forming an efficient network that allows the brain to process information at multiple scales simultaneously. We also developed and tested a novel mesoscopic two-photon microscope that can image activity from thousands of neurons over 7 millimetres apart in the brain. |
| Exploitation Route | The analysis tools we developed in this project have broad applicability beyond the calcium imaging data we applied them to, and there is strong interest from a startup company Connectome Health, founded by one of the postdocs who worked on the early phase of this project, to apply these tools to new human brain imaging datasets with application to tracking brain health and cognitive capacity. |
| Sectors | Creative Economy Healthcare Pharmaceuticals and Medical Biotechnology |
| Description | Imperial College strategic seed funding for international partnerships in biosciences and biotechnology: Imperial-RIKEN collaboration on the dynamic engram |
| Amount | £49,008 (GBP) |
| Organisation | Biotechnology and Biological Sciences Research Council (BBSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 02/2024 |
| End | 09/2025 |
| Description | International centre to centre research collaborations: UK Japan collaboration on mesoscopic multiphoton neuroimaging technology |
| Amount | £50,142 (GBP) |
| Funding ID | EP/Y020316/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2024 |
| End | 12/2025 |
| Description | International collaboration on mesoscopic multiphoton brain imaging technology |
| Amount | £12,000 (GBP) |
| Funding ID | IEC\NSFC\242498 |
| Organisation | The Royal Society |
| Sector | Charity/Non Profit |
| Country | United Kingdom |
| Start | 02/2025 |
| End | 01/2027 |
| Description | Neurodegeneration Challenge Network (NDCN) Collaborative Pairs Pilot Project Awards |
| Amount | $200,000 (USD) |
| Organisation | Chan Zuckerberg Initiative |
| Sector | Private |
| Country | United States |
| Start | 03/2024 |
| End | 09/2025 |
| Title | Moore_Schultz_Sox14_expressing_neurons |
| Description | Sox14-expressing cell count data from ~50 brain regions in two groups of mice: Sox14/GFP heterozygotes and Sox14 null. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| URL | https://zenodo.org/doi/10.5281/zenodo.12787286 |
| Title | Moore_Schultz_Sox14_expressing_neurons |
| Description | Sox14-expressing cell count data from ~50 brain regions in two groups of mice: Sox14/GFP heterozygotes and Sox14 null. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| URL | https://zenodo.org/doi/10.5281/zenodo.12787287 |
| Description | Imperial-RIKEN collaboration on the dynamic engram |
| Organisation | RIKEN |
| Country | Japan |
| Sector | Public |
| PI Contribution | This is a new collaboration between Imperial College and the RIKEN Center for Brain Science in Tokyo, triggered in part by results of the EPSRC/Wellcome project "Statistical Physics of Cognition" (SPC) as well as the NEURMOD+ project, and which provides one avenue for follow-on work from that grant. At Imperial we are collecting data based on a two-photon mesoscale that has been constructed as part of the SPC grant. |
| Collaborator Contribution | The RIKEN team (headed by Tom McHugh) are providing data and insight into how to apply the techniques developed in SPC to questions relating to the consolidation of memory from the hippocampus into the neocortex. |
| Impact | A new collaboration, so not yet. |
| Start Year | 2024 |
| Description | International collaboration on mesoscopic multiphoton brain imaging technology |
| Organisation | Suzhou Institute of Biomedical Engineering and Technology |
| Country | China |
| Sector | Public |
| PI Contribution | As a result of our work on two photon mesoscopic brain imaging, we were connected with an Institute in China (SIBET) which specialises in the design and fabrication of advanced optical components. They had constructed an instrument similar to the one we built, but in some respects better - with a wider field of view. However, they were not able to use it to perform neuroscience experiments. We brought a small team over there and set up to perform a variant of the same experiment we have been carrying out in London, with great success. This has lead to the generation of data that we are writing up for a paper to a high impact journal, and the award of a small grant from the Royal Society (£12000) for ongoing UK-China collaboration in this area. |
| Collaborator Contribution | SIBET provide expertise in optical design and manufacture of specific large lenses which we do not otherwise have access to. The lens employed in our construction of the Diesel2p mesoscale design was manufactured by a company in the US for $US 200,000, is inferior in design spec, and it was very difficult to persuade the company to manufacture it (at one point they tried to raise the price on us by $100,000 - after the award of our grant, when we had fixed funds). The Chinese design is much cheaper, and with overall better optical quality. This makes it possible for us to construct a next generation mesoscope that will image even larger neural populations at higher signal to noise ratio. The Chinese Academy of Sciences contributed a CAS PIFI Fellowship to Prof Schultz which covered the costs of a trip in 2024 to launch the collaboration. |
| Impact | A paper is currently being written. |
| Start Year | 2024 |
| Description | Sharp-wave ripple triggered neuromodulation for memory enhancement. Simon Schultz, Hayriye Cagnan, Ashwini Oswal, David Dupret |
| Organisation | Imperial College London |
| Department | Department of Bioengineering |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | Neuromod+ feasibility study funding award of £78,890.00 granted |
| Collaborator Contribution | The team will conduct a feasibility study on Sharp-wave ripple triggered neuromodulation for memory enhancement, using funds from the network. Simon Schultz will lead the overall project, coordinating the individual research actions and leading efforts to leverage results into a larger-scale externally funded research programme. His group will develop the optogenetic mouse model of SWR-triggered neuromodulation. Hayriye Cagnan will provide expertise on human MEG imaging. Ashwini Oswal will provide clinical input on the treatment of memory disorders in human patients, as well as collaborating on the MEG imaging to take place in Oxford. David Dupret is an expert on the use of electrophysiological techniques to study memory in mice. He will provide specific guidance on the measurement of sharp-wave ripples in the optogenetic model. We will consult with potential users (patients, lay-people, clinicians, carers) to gather feedback on what type of noninvasive neuromodulation system subjects will be most comfortable with, and how studies using the equipment should be designed. |
| Impact | The collaboration is multidisciplinary and combines bioengineering, electrophysiology, clinical expertise. This project makes several impactful contributions: 1. Physiological Parameterisation of SWR Control. The calibration data establish clear dose response relationships and demonstrate that bilateral optogenetic stimulation can reliably and safely prolong SWRs. These findings provide an empirically grounded framework for designing ripple modulation paradigms, filling an existing gap in standardised stimulation parameters. 2. Identification of Critical Failure Modes in Behaviour. The discovery that identical parameters can prolong SWRs during rest but truncate them during behaviour reveals crucial constraints on real time neuromodulation. Factors including elevated detection latency, reduced early window coverage (<30 ms), and theta dominated locomotor dynamics all contribute to inconsistent effects. This insight helps the field move beyond simple parameter tuning toward a deeper understanding of state dependent neural responsiveness. 3. Design Rules for Future Closed Loop Systems. The work generates actionable guidelines that directly inform the next generation of closed loop neuromodulatory experiments. 4. Highlighting the Importance of Network State Alignment. The project frames optimal stimulation not as a fixed set of parameters, but as alignment between stimulation and appropriate physiological states. Overall, the project significantly advances our understanding of the constraints and opportunities inherent to real time manipulation of memory related neural events. Outputs 1. RN Wang, J Hu, N Dundov Muñoz, N Zabouri, M merkler, S Sakata and SR Schultz (2025). Close-loop optogenetic manipulation of hippocampal sharp-wave ripples. Poster presentation at UK Symposium on Neuromodulation and Neurotechnology 2025 (UKSNN 2025), Newcastle, UK, November 2025. 2. RN Wang, J Hu, N Dundov Muñoz, N Zabouri, M merkler, S Sakata and SR Schultz (2025). Close-loop optogenetic manipulation of hippocampal sharp-wave ripples. To be presented at International Neuromodulation Society Conference, Lisbon, Portugal, May 9-14, 2026. |
| Start Year | 2024 |
| Description | Sharp-wave ripple triggered neuromodulation for memory enhancement. Simon Schultz, Hayriye Cagnan, Ashwini Oswal, David Dupret |
| Organisation | Medical Research Council (MRC) |
| Department | MRC Brain Network Dynamics Unit at the University of Oxford (BNDU) |
| Country | United Kingdom |
| Sector | Public |
| PI Contribution | Neuromod+ feasibility study funding award of £78,890.00 granted |
| Collaborator Contribution | The team will conduct a feasibility study on Sharp-wave ripple triggered neuromodulation for memory enhancement, using funds from the network. Simon Schultz will lead the overall project, coordinating the individual research actions and leading efforts to leverage results into a larger-scale externally funded research programme. His group will develop the optogenetic mouse model of SWR-triggered neuromodulation. Hayriye Cagnan will provide expertise on human MEG imaging. Ashwini Oswal will provide clinical input on the treatment of memory disorders in human patients, as well as collaborating on the MEG imaging to take place in Oxford. David Dupret is an expert on the use of electrophysiological techniques to study memory in mice. He will provide specific guidance on the measurement of sharp-wave ripples in the optogenetic model. We will consult with potential users (patients, lay-people, clinicians, carers) to gather feedback on what type of noninvasive neuromodulation system subjects will be most comfortable with, and how studies using the equipment should be designed. |
| Impact | The collaboration is multidisciplinary and combines bioengineering, electrophysiology, clinical expertise. This project makes several impactful contributions: 1. Physiological Parameterisation of SWR Control. The calibration data establish clear dose response relationships and demonstrate that bilateral optogenetic stimulation can reliably and safely prolong SWRs. These findings provide an empirically grounded framework for designing ripple modulation paradigms, filling an existing gap in standardised stimulation parameters. 2. Identification of Critical Failure Modes in Behaviour. The discovery that identical parameters can prolong SWRs during rest but truncate them during behaviour reveals crucial constraints on real time neuromodulation. Factors including elevated detection latency, reduced early window coverage (<30 ms), and theta dominated locomotor dynamics all contribute to inconsistent effects. This insight helps the field move beyond simple parameter tuning toward a deeper understanding of state dependent neural responsiveness. 3. Design Rules for Future Closed Loop Systems. The work generates actionable guidelines that directly inform the next generation of closed loop neuromodulatory experiments. 4. Highlighting the Importance of Network State Alignment. The project frames optimal stimulation not as a fixed set of parameters, but as alignment between stimulation and appropriate physiological states. Overall, the project significantly advances our understanding of the constraints and opportunities inherent to real time manipulation of memory related neural events. Outputs 1. RN Wang, J Hu, N Dundov Muñoz, N Zabouri, M merkler, S Sakata and SR Schultz (2025). Close-loop optogenetic manipulation of hippocampal sharp-wave ripples. Poster presentation at UK Symposium on Neuromodulation and Neurotechnology 2025 (UKSNN 2025), Newcastle, UK, November 2025. 2. RN Wang, J Hu, N Dundov Muñoz, N Zabouri, M merkler, S Sakata and SR Schultz (2025). Close-loop optogenetic manipulation of hippocampal sharp-wave ripples. To be presented at International Neuromodulation Society Conference, Lisbon, Portugal, May 9-14, 2026. |
| Start Year | 2024 |
| Description | Sharp-wave ripple triggered neuromodulation for memory enhancement. Simon Schultz, Hayriye Cagnan, Ashwini Oswal, David Dupret |
| Organisation | University of Strathclyde |
| Department | Strathclyde Institute of Pharmacy & Biomedical Sciences |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | Neuromod+ feasibility study funding award of £78,890.00 granted |
| Collaborator Contribution | The team will conduct a feasibility study on Sharp-wave ripple triggered neuromodulation for memory enhancement, using funds from the network. Simon Schultz will lead the overall project, coordinating the individual research actions and leading efforts to leverage results into a larger-scale externally funded research programme. His group will develop the optogenetic mouse model of SWR-triggered neuromodulation. Hayriye Cagnan will provide expertise on human MEG imaging. Ashwini Oswal will provide clinical input on the treatment of memory disorders in human patients, as well as collaborating on the MEG imaging to take place in Oxford. David Dupret is an expert on the use of electrophysiological techniques to study memory in mice. He will provide specific guidance on the measurement of sharp-wave ripples in the optogenetic model. We will consult with potential users (patients, lay-people, clinicians, carers) to gather feedback on what type of noninvasive neuromodulation system subjects will be most comfortable with, and how studies using the equipment should be designed. |
| Impact | The collaboration is multidisciplinary and combines bioengineering, electrophysiology, clinical expertise. This project makes several impactful contributions: 1. Physiological Parameterisation of SWR Control. The calibration data establish clear dose response relationships and demonstrate that bilateral optogenetic stimulation can reliably and safely prolong SWRs. These findings provide an empirically grounded framework for designing ripple modulation paradigms, filling an existing gap in standardised stimulation parameters. 2. Identification of Critical Failure Modes in Behaviour. The discovery that identical parameters can prolong SWRs during rest but truncate them during behaviour reveals crucial constraints on real time neuromodulation. Factors including elevated detection latency, reduced early window coverage (<30 ms), and theta dominated locomotor dynamics all contribute to inconsistent effects. This insight helps the field move beyond simple parameter tuning toward a deeper understanding of state dependent neural responsiveness. 3. Design Rules for Future Closed Loop Systems. The work generates actionable guidelines that directly inform the next generation of closed loop neuromodulatory experiments. 4. Highlighting the Importance of Network State Alignment. The project frames optimal stimulation not as a fixed set of parameters, but as alignment between stimulation and appropriate physiological states. Overall, the project significantly advances our understanding of the constraints and opportunities inherent to real time manipulation of memory related neural events. Outputs 1. RN Wang, J Hu, N Dundov Muñoz, N Zabouri, M merkler, S Sakata and SR Schultz (2025). Close-loop optogenetic manipulation of hippocampal sharp-wave ripples. Poster presentation at UK Symposium on Neuromodulation and Neurotechnology 2025 (UKSNN 2025), Newcastle, UK, November 2025. 2. RN Wang, J Hu, N Dundov Muñoz, N Zabouri, M merkler, S Sakata and SR Schultz (2025). Close-loop optogenetic manipulation of hippocampal sharp-wave ripples. To be presented at International Neuromodulation Society Conference, Lisbon, Portugal, May 9-14, 2026. |
| Start Year | 2024 |
| Description | Public debate "can reductionism explain the mind" after the Statistical Physics of Cognition workshop |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Public/other audiences |
| Results and Impact | We organised and ran a 2-day workshop at the Institute of Physics, with the same title as our grant, "Statistical Physics of Cognition". This was very successful, with numerous invited international speakers, and over 100 people attending, including students and academics from London, the UK and elsewhere. After the event, we held a public debate "Can reductionism explain the mind" between Karl Friston and John Krakauer. This was well attended by the general public as well as by conference attendees. |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://iop.eventsair.com/spc2024/ |