Janis Keck

1/7 🧠 My journey into development begins with this work and question: how does the brain's spatial navigation system develop? We found that the neural networks for spatial navigation (tori and rings) are preconfigured and only later anchor gradually to the world with experience! 🧵

1/N: Dear cognitive map fans, I’d like to share a model I’ve been working on for a while (clearing backlog :). I show how a vector navigation architecture (VNA) and a “positional inference network” (PIN) can build Universal Cognitive Maps (UCMs) for abstract spaces.

www.biorxiv.org/content/10.6...

Can whole-brain fMRI responses to naturalistic video stimuli be predicted using only transcripts?

Last year, I gave a tutorial on exactly that. Since the response was very positive, I’ve now released both the tutorial notebook and the trained models publicly.

I am totally pumped about this new work . "Task-trained RNNs" are a powerful and influential framework in neuroscience, but have lacked a firm theoretical footing. This work provides one, and makes direct contact with the classical theory of random RNNs:
www.biorxiv.org/content/10.6...

This shouldn‘t be so subtle, right? Preserved neighborhood relations mean what is mainly preserved is topology, while something like CKA will also be driven by geometry. So there is an ideal platonic manifold, but how it may be embedded/curved can differ.

Matching sounds to shapes: Evidence of the bouba-kiki effect in naïve baby chicks Humans across multiple languages spontaneously associate the nonwords “kiki” and “bouba” with spiky and round shapes, respectively, a phenomenon named the bouba-kiki effect. To explore the origin of t...

“Humans across multiple languages spontaneously associate the nonwords kiki & bouba with spiky & round shapes, respectively...We tested the bouba-kiki effect in baby chickens. Similar to humans, they spontaneously chose a spiky shape when hearing a kiki sound & a round shape when hearing a bouba.”😲🧪

The hippocampal map has its own attentional control signal!
Our new study reveals that theta #sweeps can be instantly biased towards behaviourally relevant locations. See 📹 in post 4/6 and preprint here 👉
www.biorxiv.org/content/10.6...
🧵(1/6)

Introducing DroPE: Extending Context by Dropping Positional Embeddings

We found embeddings like RoPE aid training but bottleneck long-sequence generalization. Our solution’s simple: treat them as a temporary training scaffold, not a permanent necessity.

arxiv.org/abs/2512.12167
pub.sakana.ai/DroPE

If you seek social support, you might be less willing to work towards making genuine connections with other human beings after a while of getting something superficially similar without effort.

I think, just as with fast food, there is a problem of instant gratification, whether you use it for productivity or social support.
If you use it for research this might mean becoming reluctant to think about a problem for a while before consulting a LLM.

Our paper on data constrained RNN that generalize to optogenetic perturbations now citable on eLife:
doi.org/10.7554/eLif...

Meanwhile in Berlin, data showed that air quality improved due to speed restrictions on certain streets, and the government reacted by lifting the restrictions with the reasoning that now that it’s improved the restrictions are not needed anymore…

Arousal as a universal embedding for spatiotemporal brain dynamics Neural activity in awake organisms shows widespread and spatiotemporally diverse correlations with behavioral and physiological measurements. We propose that this covariation reflects in part the dynamics of a unified, multidimensional ...

Would be interesting if this also works if economy isn‘t reduced to one variable, similar to how they related brain-wide activity to a scalar arousal timeseries here:

pmc.ncbi.nlm.nih.gov/articles/PMC...

while of course only the enlightened know that the soul is stored in the ... claustrum

Massive thanks to my co-authors: Lukas Barth, Hannaneh Fahimi, Parvaneh Joharinad and Jürgen Jost. (3/3)

There, weights are obtained by sampling Vietoris-Rips-complexes at random scales. We then show how this interpretation directly allows to build new dimensionality reduction methods, which we illustrate by creating a version of #UMAP based on the Čech filtration. (2/3)

Probabilistic Foundations of Fuzzy Simplicial Sets for Nonlinear Dimensionality Reduction Fuzzy simplicial sets have become an object of interest in dimensionality reduction and manifold learning, most prominently through their role in UMAP. However, their definition through tools from alg...

New preprint! Have you ever wondered, what are these fuzzy simplicial sets, the theoretical framework behind e.g. UMAP? Here we show that you may simply see them as marginal distributions over simplicial sets. This provides a generative model for UMAP. (1/2)

arxiv.org/abs/2512.03899

Three German universities offering post-docs for researchers "who cannot conduct or continue their work in the USA appropriately because of actual political pressure. "
www.uni-konstanz.de/zukunftskoll...

Foraging in conceptual spaces: hippocampal oscillatory dynamics underlying searching for concepts in memory How does the brain access stored knowledge? It has been proposed that conceptual search engages neurocognitive processes similar to foraging in physical space. We tested this idea using intracranial E...

Foraging in conceptual spaces: hippocampal oscillatory dynamics underlying searching for concepts in memory

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

Dynamic updating of cognitive maps via traces of experience in the subiculum In the classical view of hippocampal function, the subiculum is assigned the role as the output layer. In spatial paradigms, some subiculum neurons manifest as so-called boundary vector cells (BVCs), ...

New preprint from the lab and great work by Fei Wang. We show how subiculum trace vector cells can be modeled consistent with known effects in CA1. Traces are driven by a mismatch learning rule to keep associative memories in line with experience.
www.biorxiv.org/content/10.1...

Arousal as a universal embedding for spatiotemporal brain dynamics - Nature Reframing of arousal as a latent dynamical system can reconstruct multidimensional measurements of large-scale spatiotemporal brain dynamics on the timescale of seconds in mice.

Super proud of this collaboration with rockstar Ryan Raut - born out of playing in the sandbox in our last year of grad school! Multi-scale brain activity can be predicted from a simple measure of arousal like pupil diameter. Out with linear causality, in with dynamic systems to explain neurobiology

1/
🚨 New preprint! 🚨

Excited and proud (& a little nervous 😅) to share our latest work on the importance of #theta-timescale spiking during #locomotion in #learning. If you care about how organisms learn, buckle up. 🧵👇

📄 www.biorxiv.org/content/10.1...
💻 code + data 🔗 below 🤩

#neuroskyence

Can a single cell learn? Even without a brain, some microbes show simple forms of cognition. Can this basal cognition be engineered? Check our new paper with @jordiplam.bsky.social on the minimal synthetic circuits & their cognitive limits. @drmichaellevin.bsky.social www.biorxiv.org/content/10.1...

Andrej is a great scientist and teacher, highly recommended!

Congratulations to
@reznikdan.bsky.social
and
@sofievalk.bsky.social
for this tremendous achievement!

In sum, we believe this is the first direct evidence in humans of systems consolidation theory for specific memories: the hypothesized transformation of memories from hippocampus to cortex over sleep.

Take look for yourself on @biorxivpreprint.bsky.social t.co/xFEDpqVXfE 6/6

🚨We believe this is a major step forward in how we study hippocampus function in healthy humans.

Using novel behavioral tasks, fMRI, RL & RNN modeling, and transcranial ultrasound stimulation (TUS), we demonstrate the causal role of hippocampus in relational structure learning.

Now out in @natneuro.nature.com

What happens to the brain’s body map when a body-part is removed?

Scanning patients before and up to 5 yrs after arm amputation, we discovered the brain’s body map is strikingly preserved despite amputation

www.nature.com/articles/s41593-025-02037-7

🧵1/18

The additive formulation is just a rewrite of the multiplicative one: N=-SB, B Bernoulli, and then S+N = (1-B)S

Very Cool! This stimuli-set by @fnkp.bsky.social seems to tackle a related problem, but motivated by memory research: 2025.ccneuro.org/abstract_pdf...