Nima Dehghani
@neurovium.bsky.social
Computational neuroscience, Physics of Complex Systems, Bio-Inspired Intelligence, Foundations of Physical Computing https://neurovium.science/ https://compneuro.mit.edu/home
Ten years. Three fields. One long stare. A decade forging a unified framework across physics, computation & neuro.I’ve released a few pieces along the way — foundations of physical computing. Ten more theoretical papers done and coming. Just tightening the screws. Stay tuned — don’t blink.
ALT:
static.klipy.com
The paper might be hard to penetrate for some that are not familiar with the mathematics/physics machinery. But what this means for neuroscience (and why NeuroAI needs it)? See the companion that explains and unwraps it for computational/systems neuroscientists. 👇 neurovium.science/posts/pblog-...
When Does Matter Compute? Wave Memory, Closure, and the Architecture of Autonomous Physical Computation
Companion post to: Autonomous Physical Computation: A Categorical Closure Criterion for Physical and Neuromorphic Reservoirs. Nima Dehghani arXiv 2026. DOI: https://doi.org/10.48550/arXiv.2607.23902
neurovium.science
Nima Dehghani@neurovium.bsky.social · last mo.Week 4 of summer of papers! Here we go! 📣📷 📜 The central question: when does memory in matter become autonomous computation? 1/n 👇🧵 arxiv.org/abs/2607.23902
New additions TO-READ ! Tommy's (Poggio) "Brains, Minds and Machines" & Max Welling "GenAI and Stochastic Thermodynamics"
Cortical Machinery as a Design Principle for RNNs. Functional connectomics as architectural resource for AI Read here 👇📜 neurovium.science/posts/pblog-...
When Prediction Shapes the Machine: Evolutionary Reservoirs, Chaos, and Structural Constraints "What is under a reservoir's hood?" Read here 👇📜 neurovium.science/posts/pblog-...
New arrival....time to "re-learn" probability from none but Giorgio parisi
The paper might be hard to penetrate for some that are not familiar with the mathematics/physics machinery. But what this means for neuroscience (and why NeuroAI needs it)? See the companion that explains and unwraps it for computational/systems neuroscientists. 👇 neurovium.science/posts/pblog-...
When Does Matter Compute? Wave Memory, Closure, and the Architecture of Autonomous Physical Computation
Companion post to: Autonomous Physical Computation: A Categorical Closure Criterion for Physical and Neuromorphic Reservoirs. Nima Dehghani arXiv 2026. DOI: https://doi.org/10.48550/arXiv.2607.23902
neurovium.science
Nima Dehghani@neurovium.bsky.social · last mo.Week 4 of summer of papers! Here we go! 📣📷 📜 The central question: when does memory in matter become autonomous computation? 1/n 👇🧵 arxiv.org/abs/2607.23902
I have decided to add the story behind papers to technical companion posts. Here is a starter: "When Prediction Shapes the Machine: Evolutionary Reservoirs, Chaos, and Structural Constraints What is under a reservoir's hood?" 👇 neurovium.science/posts/pblog-... Let me know if you want to see more📜
When Prediction Shapes the Machine: Evolutionary Reservoirs, Chaos, and Structural Constraints
Companion post to: Evolutionary Optimization Reveals Structural Constraints on Reservoir Architecture for Spatiotemporal Chaos Nima Dehghani arXiv 2026. DOI: https://doi.org/10.48550/arXiv.2606.22765
neurovium.science
Nima Dehghani@neurovium.bsky.social · 2mo agoWeek 3 of summer of papers! Here we go! 📣🎥📜 If a recurrent dynamical system is selected only for prediction, what structure does it evolve? Not just: can it predict complex timeseries (like Kuramoto-Sivashinsky)? But: what does prediction do to the machine? 1/n 🧵👇 arxiv.org/abs/2606.22765
How does nature find Turing patterns? Check in @pnas.org this beautful paper led by @imperialcollegeldn.bsky.social Robert Endres on how evolution can search in high-dimensional parameter spaces using oscillations: www.pnas.org/doi/epdf/10.... And here's my comment: www.pnas.org/doi/epub/10....
Looking forward to this
Kempner Institute at Harvard University@kempnerinstitute.bsky.social · 2mo ago📢 Just announced! Join us for the #KempnerInstitute workshop “Learning Dynamics in Natural and Artificial Intelligence: Evolution, Adaptation, and the Foundations of Efficient Learning.” Learn more, register, or submit an abstract 👉 bit.ly/3QwJlHR
Week 4 of summer of papers! Here we go! 📣📷 📜 The central question: when does memory in matter become autonomous computation? 1/n 👇🧵 arxiv.org/abs/2607.23902
Autonomous Physical Computation: A Categorical Closure Criterion for Physical and Neuromorphic Reservoirs
Physical reservoirs, neuromorphic devices, and wave-mediated systems often possess memory, feedback, and rich state-dependent dynamics, but these properties do not by themselves establish autonomous c...
arxiv.org
Week 3 of summer of papers! Here we go! 📣🎥📜 If a recurrent dynamical system is selected only for prediction, what structure does it evolve? Not just: can it predict complex timeseries (like Kuramoto-Sivashinsky)? But: what does prediction do to the machine? 1/n 🧵👇 arxiv.org/abs/2606.22765
Evolutionary Optimization Reveals Structural Constraints on Reservoir Architecture for Spatiotemporal Chaos
Biological systems maintain function in fluctuating environments by transforming past stimulation into internal dynamical states that support future-oriented responses. Reservoir computing provides a ...
arxiv.org
Perfect open sci + ed. Fueled by the curiosity of great students, backed by the generous support of @neuromatch.bsky.social (1.5 yr of serious compute beyond their Impact Scholar Program), MICrONS data funded by IARPA generated by @alleninstitute.org @jhuapl.bsky.social hosted by @dandiarchive.org
Neuromatch@neuromatch.bsky.social · 2mo agoFrom Neuromatch Academy ➡️ Impact Scholars Program ➡️ published paper 🎉 Just published: Harnessing cortical geometry, wiring, and function as inductive biases for recurrent neural networks 🤓 Full story and paper link: www.linkedin.com/feed/update/... #ComputationalNeuroscience #DeepLearning
A key point of "Open Science" is that it enables the sci community to make advances that otherwise wouldn't have been possible. Thanks to @bingbrunton.bsky.social group who made the efforts to make the AJILE12 data available on @dandiarchive.org , we were able to develop a cool method. Try it!🫵
Nima Dehghani@neurovium.bsky.social · 3mo agoHow local is a “local” field potential? In our new paper 📜, we introduce Spatially Masked Regression (SMR), a reconstruction-based framework for asking how much of an electrode signal is locally redundant, and how much is embedded in broader distributed dynamics. 1/n 🧵👇 arxiv.org/abs/2606.11415
Cool paper from Danyal @danakarca.bsky.social & co👇 www.nature.com/articles/s42...
Algorithm–hardware co-design of neuromorphic networks with dual memory pathways - Nature Machine Intelligence
Pengfei Sun et al. develop a spiking neural network with a dual memory pathway, co-designed with a custom neuromorphic chip. The approach delivers over 4× throughput and 5x energy efficiency gains whi...
nature.com
new paper, #NeuroAI 📣📜 Can measured cortical organization be used as an inductive bias for artificial recurrent neural networks? In this work, we ask whether cortical geometry, wiring, and function can push RNNs learn. Not as metaphor, but as measurable structure! 1/n🧵👇 arxiv.org/abs/2606.14975
Harnessing cortical geometry, wiring, and function as inductive biases for recurrent neural networks
How the wiring and functional organization of cortex shape recurrent computation remains a central question in both neuroscience and machine learning. Here, we leverage data released through the Machi...
arxiv.org
We're thrilled to announce the 2026 #KavliPrize Laureates, scientists pushing boundaries of what humanity knows in #astrophysics #nanoscience #neuroscience. They have revealed new chapters in the story of our universe, the smallest materials, and how the brain works. 1/2
Eukaryotic evolution via massive HGT "Our relative timing estimates suggest early interactions between an Asgard archaea and a Planctomycete (bacterium) ... Thus, the prokaryotic-to-eukaryotic transition was probably a gradual and complex process, with a constant flow of HGT from diverse sources."
Ricard Solé@ricardsole.bsky.social · 3mo agoHow did complex cells emerge? In this groundbreaking paper, @tonigabaldon.bsky.social & co show that eukaryotes arose through multiple waves of ancient genetic exchange involving diverse bacteria (& giant viruses!) revealing a far more complex origin than we imagined. www.nature.com/articles/s41...
How local is a “local” field potential? In our new paper 📜, we introduce Spatially Masked Regression (SMR), a reconstruction-based framework for asking how much of an electrode signal is locally redundant, and how much is embedded in broader distributed dynamics. 1/n 🧵👇 arxiv.org/abs/2606.11415
Spatially Masked Regression Reveals Local and Distributed Predictability in Electrophysiological Recordings
Neural recordings are often interpreted as local measurements, yet the signal at any one sensor can also reflect structured activity distributed across the broader network. This raises a basic questio...
arxiv.org
Recent work from the lab: big contrast between the synaptic dynamics of excitatory and inhibitory synapses!
bioRxiv Neuroscience@biorxiv-neursci.bsky.social · 3mo agoFunctional Continuum of GABAergic Synaptic Dynamics Reflects Genetic Identities https://www.biorxiv.org/content/10.64898/2026.06.09.731181v1
See the paper card (for link and the blog) on the recent manuscript on inhibitory neurons across different layers of the cortex! ps...the universality paper (quoted tweet) will be a follow up on this one. neurovium.science/papers/Inhib...
Morphoelectric properties of inhibitory neurons shift gradually and regardless of cell type along the depth of the cerebral cortex
This work shows that cortical interneuron morphoelectric properties are organized by two separable axes: molecularly related, depth-independent cell-type structure and gradual, depth-dependent shifts ...
neurovium.science
Nima Dehghani@neurovium.bsky.social · last yr.Chihuly gardens is my "pilgrimage" whenever I come to Seattle. The sacred beauty of colors and forms, this time in the "Chandelier" room resonated w the manuscript that Felipe Yanez & I are working on together: "The universality of form/function of cortical inhibitory neurons"...
@macshine.bsky.social talks with @satra.bsky.social about the #OHBM2026 Abstract Atlas—a community-built tool that places every 2026 abstract inside the broader neuroscience landscape. Where does your work sit in the field? #neuroskyence www.thetransmitter.org/open-source-...
Finding your place in neuroscience
A new tool places every OHBM 2026 abstract inside a semantic map of the neuroscience literature, so you can find your neighbors.
thetransmitter.org
Mistakes happen. That is part of the sci progress. What is most important is reproducibility, verification & honesty. That the authors shared their data and in essence agreed to be criticized is itself necessary. That they opted to retract is honorable. Use @dandiarchive.org to amplify this! 🧵👇1/5
The Transmitter @thetransmitter.bsky.social · 5mo agoThe authors of a Nature paper outlining a mechanism for multisensory memories in Drosophila melongaster have retracted the work after they were unable to replicate a set of imaging experiments. By @callimcflurry.bsky.social #neuroskyence www.thetransmitter.org/retraction/r...
A must-needed benchmarking👇🏼 How much is methodological and how much is conceptual? A reason for us developing #rippl-AI github.com/PridaLab/rip... Don’t miss the next hackathon avant-FENS pre-fens-brainhack.github.io/brainhack2026/ with @acnavasolive.bsky.social & @mattiachini.bsky.social
The Transmitter @thetransmitter.bsky.social · 3mo agoDifferences stemmed from the way researchers defined concepts, to the algorithms they used to analyze the data, to the parameters they employed when implementing the algorithms, note @mattiachini.bsky.social and Gaelle Chapuis. #neuroskyence www.thetransmitter.org/reproducibil...
Perspective piece from Blake @tyrellturing.bsky.social @mandanas.bsky.social 👇
Blake Richards@tyrellturing.bsky.social · 3mo agoNew perspective piece with @mandanas.bsky.social: We argue, based on LLMs and old connectionist theories, that schemas shouldn't be viewed as distinct from semantic or episodic memories. They're just one end of a detailed-to-abstract memory spectrum: www.cell.com/neuron/fullt... #NeuroAI 🧠📈 🧪
