Andrea Brovelli

Higher-order links and broken standards The balance in the Force has been restored

I have also written for you a short semi-technical piece on #ComplexityThoughts about this:

open.substack.com/pub/manlius/...

Enjoy!

New preprint out, first timid experiments on foundation models for #neuroimaging and Ornstein-Uhlenbeck processes (checking inhibitory/excitatory causality):
Prediction and Causality of functional #MRI and synthetic signal using a Zero-Shot #TimeSeries Foundation Model
arxiv.org/abs/2509.12497
#LLM

⚠️JOB ANNOUNCEMENT⚠️
Postdoctoral Position in Computational Cognitive Neuroscience (Marseille) in collaboration with @alainbarrat.bsky.social. Join us to study higher-order brain interactions in causal learning using information theory and network science. More info
www.fens.org/careers/job-...

Sensory robustness through top-down feedback and neural stochasticity in recurrent vision models Biological systems leverage top-down feedback for visual processing, yet most artificial vision models succeed in image classification using purely feedforward or recurrent architectures, calling into...

Our findings pave the way for biologically-inspired vision architectures that move beyond purely feedforward or locally recurrent designs by incorporating explicit top-down pathways alongside stochastic regularization

To read the full story, check out the preprint 👇
arxiv.org/abs/2508.07115

Preprint: Equations/generalizations for TC, DTC, RSI, O-information, and TSE-complexity for multivar real/cmplx data. How “connections” contribute to system inf measures.
Helpful comments and reports on errors are appreciated.
arXiv: 2025-07-11
doi.org/10.48550/arXiv.2507.08773

A big thanks to all collaborators!
@ecombrisson.bsky.social @danielemarinazzo.bsky.social @lordgrilo.bsky.social Ruggero Basanisini, Matteo Neri, Guillaume Auzias and Stefano Panzeri

Finally out on Nat Comms 🚀
We show that an intrinsic motivational learning signal (information gain) is encoded through synergistic and higher-order functional brain interactions and is broadcast to prefrontal reward circuits.

When your statistical model "fits the data well" ...

Researchers in France are working on creating a french network of researchers to organize interaction, communication and training in #Computational_Neuroscience. If you are a CompNeuro working in France, consider joining, and registering to our mailing list: listes.services.cnrs.fr/wws/subscrib...

Adversarial testing of global neuronal workspace and integrated information theories of consciousness - Nature Multimodal results (iEEG, fMRI and MEG) of predictions from integrated information theory and global neuronal workspace theory align with some predictions of both theories on visual consciou...

Congratulations to the whole team. Impressive work! www.nature.com/articles/s41...

#ChooseCNRS

New Call for International Researchers interested in working in France. From postdoc to senior PIs.
carrieres.cnrs.fr/actualites/i...

Great work!

Shannon invariants: A scalable approach to information decomposition Distributed systems, such as biological and artificial neural networks, process information via complex interactions engaging multiple subsystems, resulting in high-order patterns with distinct proper...

Preprint time:
“Shannon invariants: A scalable approach to information decomposition”
arxiv.org/abs/2504.15779

Studying information in complex systems is challenging due to difficulties in defining multivariate metrics and ensuring their scalability. This framework addressed both challenges!

Functional organization derived from network-driven processes offers clear advantages wrt traditional methods when studying human brain networks, outperforming SoTA communication models in explaining functional communities from structural data.

#Neuroscience 🧪🧠

www.pnas.org/doi/10.1073/...

Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems
arxiv.org/abs/2504.01990

Foundation Models на новом уровне

github.com/FoundationAg...

Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems (264 pages)

This survey provides a comprehensive overview, framing intelligent agents within a modular, brain-inspired architecture

Trump’s new tariff math looks a lot like ChatGPT’s ChatGPT may be the White House’s latest economic advisor.

Reading this hypothesis the trade deficit was figured out by AI made me wonder if it's the same reason why uninhabited islands (Heard and McDonald) wound up in the tariffs?

Were they hallucinated? Nope!

(h/t @dompreston.com)

www.theverge.com/news/642620/...

New Deep Learning Framework Reveals Hidden Structure in Neural Activity - Harvard University - Department of Molecular & Cellular Biology Understanding how neurons encode information is one of the most pressing challenges in neuroscience. A new study from a multidisciplinary team including MCB researchers and those from the […]

New Deep Learning Framework Reveals Hidden Structure in Neural Activity 🧠 #neuro #science
www.mcb.harvard.edu/department/n... @naoshigeuchida.bsky.social @saramatias.bsky.social @neurovenki.bsky.social @btolooshams.bsky.social @kempnerinstitute.bsky.social @harvardbrainsci.bsky.social

Non-equilibrium whole-brain dynamics arise from pairwise interactions The human brain operates far from thermodynamic equilibrium, with complex interactions between its neural elements driving non-equilibrium dynamics. Geli et al. reveal that pairwise interactions betwe...

Very cool paper on how interactions of different orders contribute to patterns of brain activity.
I do have some questions about whether losing the distinction between redundancy and synergy might be confounding the results at all though.
www.cell.com/cell-reports...

Drop me an email Thomas. My group could be a soft landing. And I could suggest you other labs in France/EU, in case

Geometric influences on the regional organization of the mammalian brain (!)

Great new work lead by James Pang,
@alexfornito.bsky.social & star collaborators!

www.biorxiv.org/content/10.1...

Great news, everyone! Cross Roads is BACK for episode #50! 😱🎉

Join us on Jan 21 @ 2pm JST to hear exciting ideas from Dr. Fernando Rosas about "The Many Faces of Emergence."

We'll see you next week on YouTube Live! www.youtube.com/live/bgB8e3G...

🚨 THOI: Revolutionizing Higher-Order Interaction Analysis 🚨

Complex systems are more than the sum of their parts! Our latest study introduces THOI a groundbreaking Python library designed to unveil emergent collective behaviors.

Discover more: arxiv.org/pdf/2501.03381 github.com/Laouen/THOI

MAIN Conference Montreal AI and Neuroscience (MAIN) is an international conference organized by the UNIQUE Centre in collaboration with the Centre de Recherches Mathématiques (CRM) at the University of Montreal since...

I see the talks from Montreal AI and Neuroscience (MAIN) 2024 are now online:

www.youtube.com/@MAINConfere...

#neuroai #neuroskyence

Oscillations-in-measurement and process. Oscillations are characterized by rhythmic temporal structure in any of the entities in our framework, manifesting variously in both measurements and physical processes. Oscillations-in-measurements are rhythmic fluctuations in the signal of electrophysiology recordings. At the level of processes, oscillations can instantiate as fluctuations in the electric field, or in other neurobiological events, such the release of neurotransmitters and action potentials across neuronal masses. At a behavioral level, oscillations can manifest, for example, as a waxing and waning of reaction time, stimulus detection rates, and arousal levels. Each of these characterizes one type of oscillation.

Processes and measurements: a framework for understanding neural oscillations in field potentials
👇
www.sciencedirect.com/science/arti...
By @rdgao.bsky.social and stellar co-authors

Predictive learning shapes the representational geometry of the human brain - Nature Communications This study shows that the human brain aligns its neural representations to the statistical structure of sensory inputs, and that the magnitude of this representational shift correlates with the synerg...

🔵 New Paper 🔵

Happy to announce that our paper linking predictive learning and representational geometry in the human 🧠 is out today in #NatureComms!

with @drjuliamoser.bsky.social, Hubert Preissl and Markus Siegel

📜👉 nature.com/articles/s41467-024-54032-4

Episode #23 in #TheoreticalNeurosciencePodcast: On human whole-brain models - with Viktor Jirsa

theoreticalneuroscience.no/thn23

The guest builds personalized whole-brain network models, partially to aid clinicians in treating brain ailments. @ebrains.bsky.social

Human brain dynamics are shaped by rare long-range connections over and above cortical geometry.

Some new results to inform this debate.

www.pnas.org/doi/10.1073/...