iiinova — where your digital self keeps evolving

Reposted by @mikkellepperod.bsky.social

Mia-Katrin Kvalsund @mia-katrin.bsky.social · 12/03/2026

What happens when you give a Neural Cellular Automata whiskers? 🐭🤖 In our new paper, we introduce the ANCA: a model where attention emerges through active movement and sensing. Full Paper (open access): doi.org/10.1016/j.ne... #DeepLearning #CellularAutomata #ComputationalNeuroscience #EmbodiedAI

doi.org

Redirecting

051

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 14/08/2025

➡️ The model learns place cells from border cells, with no grid cells needed. ➡️ These artificial place cells "remap" just like real ones when the environment changes. ➡️ A strange finding: The place cell centers form a hexagonal grid! 🤯 We even found preliminary evidence for this in mouse brains.

010

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 31/03/2025

For more info, see our video description at youtu.be/-ERmvXo0XTs And the entire manuscript at www.biorxiv.org/content/10.1... 4/6

youtu.be

Presenting the Active Neural Cellular Automata (ANCA)

YouTube video by Mikkel Lepperød

100

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 31/03/2025

Our research bridges neuroscience and AI by showing how distributed neural architectures can interact with movement, offering new insights for adaptive, embodied AI systems. This could be particularly valuable for robotics applications where sensor movement and fault tolerance are crucial. 3/6

youtu.be

Presenting the Active Neural Cellular Automata (ANCA)

YouTube video by Mikkel Lepperød

110

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 31/03/2025

1️⃣ Active sensing emerges without explicit attention mechanisms 2️⃣ Movement allows the system to focus on salient information, simplifying classification tasks 3️⃣ The architecture becomes highly scalable when paired with movement 4️⃣ These systems show remarkable fault tolerance to damaged sensors 2/6

youtu.be

Presenting the Active Neural Cellular Automata (ANCA)

YouTube video by Mikkel Lepperød

100

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 31/03/2025

🔬 Excited to share our new paper: "Sensor Movement Drives Emergent Attention and Scalability in Active Neural Cellular Automata" We found that when neural systems can move their sensors (like animals do!), they develop attention-like behaviors without being explicitly programmed to do so. Post 1/6

youtu.be

Presenting the Active Neural Cellular Automata (ANCA)

YouTube video by Mikkel Lepperød

145

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 15/10/2024

You can find them both at BrainInspired and in the Theoretical Neurocience Podcast: braininspired.co/podcast/195/ braininspired.co/podcast/196/ theoreticalneuroscience.no/thn19/ theoreticalneuroscience.no/thn20/

braininspired.co

BI 195 Ken Harris and Andreas Tolias with Gaute Einevoll and Mikkel Lepperød | Brain Inspired

020

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 15/10/2024

I recently participated in two podcast episodes by Paul Middlebrooks (Brain Inspired) and Gaute Einevoll (Theoretical Neuroscience). These conversations were recorded aboard the Hurtigruten cruise ship during a workshop on NeuroAI that I had the pleasure of organizing.

330

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 14/10/2024

Proud to see this research build on concepts from our own paper, "Inferring Causal Connectivity from Pairwise Recordings and Optogenetics,". It’s amazing to see how these tools continue to push the boundaries of understanding brain connectivity and function! 💡🔬 journals.plos.org/ploscompbiol...

journals.plos.org

Inferring causal connectivity from pairwise recordings and optogenetics

Author summary Understanding the interactions between neurons is essential for grasping brain function. Neuroscientists often face a challenge when stimulating multiple neurons simultaneously: determi...

011

mikkellepperod.bsky.social @mikkellepperod.bsky.social · 14/10/2024

🧠🚀 Exciting new developments in neuroscience! The recent paper "The Fly Connectome Reveals a Path to the Effectome" takes an innovative step forward by combining instrumental variables (IV) with Bayesian priors to efficiently estimate effective connectivity, and how neural circuits drive behavior. 🦟

291