Il Memming Park

A new preprint that revives the theory of continous attractors that may be surprising to you! Despite their mathematical fragility, we show that they are functionally *robust*. No wonder we see approximate continuous attractors in neuroscience often. arxiv.org/abs/2408.00109

If you are at #cosyne2026 come see poster [2-152] Dynamical archetype analysis: Autonomous computation. Ábel Ságodi is presenting how you can compare dynamical systems at behaviorally relevant timescale while keeping the interpretability!

Are you at #cosyne2026 and looking for a postdoc position? We are hiring! Send me a message or find me! catniplab.github.io/postdoc-hiring…

Happy to share that our new paper is out in Nature Communications with Zoe Ashwood, @IntlBrainLab, and @jpillowtime! We study how animals switch between internal decision-making states in non-stationary environments using a GLM-HMM framework. nature.com/articles/s4146…

@paperpile I'm getting "you're offline" state often these days. What could be going on? If I refresh the paperpile page on Chome, it works again.

Joint junior faculty position in Computational Neuroscience, split between Ctr for Computational Neuroscience at @FlatironInst and the CUNY Graduate Center @GC_CUNY. Application deadline: 16 Jan 2026! simonsfoundation.org/flatiron/caree…

How can I accelerate breakdown of caffeine in my body? I will need to increase CYP1A2 (P450) activity (without smoking). Vigorous exercise over 30 days was shown to increase it up to 70%? pubmed.ncbi.nlm.nih.gov/1394840/

Learning a lot while preparing for a lecture on RNNs for neuroscience.

Gonçalo M. Tavares's poetry book, "Mr. Swedenborg and the Geometrical Investigations" is *not* available on amazon.com... ISBN: 9896419981 amzn.to/3YQB4yK

@vinyard_m #ICanHazPDF?

scDiffEq offers a principled framework for studying dynamics in single-cell data. For most datasets, it can run on your laptop (e.g., Apple Silicon) or T4 GPU (free in colab). 📄 Paper: nature.com/articles/s4225… 💻 Code: github.com/scDiffEq/scDif… 📚 Docs: scdiffeq.com With: Anders Rasmussen, @RuitongLi3, @KleinLabHMS, @getz_lab, and @lucapinello. @Harvard @Schmidt_Center @broadinstitute

How does a stem cell "decide" its fate? Development requires both reliability (consistent cell types) AND flexibility (diverse outcomes from identical progenitors). Cells achieve this by dynamically tuning deterministic drift and stochastic diffusion. New in @NatMachIntell: scDiffEq models state-dependent drift AND diffusion, improving fate prediction by ~8% over SOTA. scDiffEq also enables genome-wide in silico perturbation screens and reveals temporal gene dynamics. 🧵nature.com/articles/s4225…

What should the opposite of twisting be called?

The research labs you can join through INDP range from systems neuroscience, computational neuroscience and clinical neuroscience, to neocybernetics, neuroethology, and natural intelligence. To learn more about the culture and value, check out: fchampalimaud.org/about-cr

We seek motivated applicants from all areas of neuroscience, as well as physics, math, computer science, electrical/biomedical engineering, and related quantitative backgrounds. English is the working language. It's an American-style graduate program in Europe.

Applications are now open for the International Neuroscience Doctoral Programme (INDP) at Champalimaud Foundation, Lisbon, Portugal. Deadline for application: Jan 31, 2026 fchampalimaud.org/champalimaud-r… The programme includes an initial year of classes + three lab rotations.

One advantage of monosemantic, sharply-tuned, grandmother-cell, axis-aligned, neuron-centric representation as opposed to polysemantic, mixed-selective, oblique population code is that it can benefit from evolution. Genes are good at operating at the cell level. #neuroscience

Theoretical Insights on Training Instability in Deep Learning TUTORIAL uuujf.github.io/instability/ gradient flow-like regime is slow and can overfit while large (but not too large) step size can trasiently go far, converge faster, and find better solutions #optimization #NeurIPS2025

score/flow matching diffusion models only starts memorizing when trained for long enough Bonnaire, T., Urfin, R., Biroli, G., & Mezard, M. (2025). Why Diffusion Models Don’t Memorize: The Role of Implicit Dynamical Regularization in Training. openreview.net/forum?id=BSZqp…

Some of my favorites from #NeurIPS2025 more neg max Lyapunov exp => faster parallelized RNN convergence Gonzalez, X., Kozachkov, L., Zoltowski, D. M., Clarkson, K. L., & Linderman, S. Predictability Enables Parallelization of Nonlinear State Space Models. openreview.net/forum?id=7AGXS…

Melanie Mitchell's keynote reminds us that it is not easy to evaluate intelligence (AI, babies, animals, etc) and benchmarks can be VERY misleading. #NeurIPS2025