Skip to content

Reasoning with Neural Cellular Automata

Sep 2026 · arXiv.org · Vol abs/2609.36126 · 1 citation · 36 references
Computer Science Physics

TL;DR

This work test the reasoning capabilities of Neural Cellular Automata, networks of recurrent cells that use strictly local connectivity and asynchronous updates, and shows that NCAs are robust reasoners capable of dynamically modulating compute to recover efficiently from damage, and that they can scale to solve reasoning in raw pixel space.

Abstract

Modern AI architectures used to solve visual reasoning tasks typically rely heavily on global connectivity and synchronization. As biological systems demonstrate, though, sophisticated computation can be performed in a more decentralized fashion. In this work, we test the reasoning capabilities of Neural Cellular Automata (NCAs), networks of recurrent cells that use strictly local connectivity and asynchronous updates. NCAs have been extensively studied in artificial life experiments, but it is unclear whether they can perform complex multi-step reasoning. We show that NCAs produce spatio-temporal dynamics capable of solving challenging visual reasoning tasks, including large mazes, Sudoku, and ARC-AGI-1. Furthermore, we provide evidence that NCAs generalize out-of-distribution when running with larger grids, longer rollouts, or parallel trials; and that the latter can be made more efficient via pruning of redundant trajectories. We find that these generalization capabilities depend on training with sample replay and stochastic perturbations, and that stochasticity remains beneficial at test time. Finally, we show that NCAs are robust reasoners capable of dynamically modulating compute to recover efficiently from damage, and that they can scale to solve reasoning in raw pixel space.

View source

Similar papers

Where Does Randomness Matter in Neural Cellular Automata?

Stochastic cell updates are often used throughout the life of a neural cellular automaton (NCA), from backpropagation through time to final rollout. This leaves two questions entangled: does update randomness help learn a useful rule, and must that randomness remain at execution? We separate training and evaluation upd...

Fei Zuo, Jia-Qi Shi, Yu-Jing Liu · 0 citations

Controlling Collectives of AI Agents in Reasoning Space with Spatial Transformers

This work proposes COMPASS, a scalable, decentralized multi-robot architecture for controlling large collectives of agentic robots with reasoning space feedback control, and shows that reasoning feedback works best when composed with a compact learned token.

Frederic Vatnsdal, Roshan Gopal, Romina García Camargo et al. · 0 citations
Review Open access Nov 2026

A Survey of Agentic AI: Architectures, Reasoning, Planning, Memory, Collaboration, and Real-World Applications

Agentic Artificial Intelligence (AI) is a new paradigm that allows intelligent systems to behave independently, pursue specific objectives, and adapt to changing circumstances. Agentic AI combines reasoning, planning, memory, learning, and tool utilization to do complicated tasks with minimal human involvement, in cont...

Sunita Dixit · 0 citations
#artificial intelligence Preprint Nov 2026

Clocks all the way down: does sequence-processing AI need endogenous temporal structure?

Mainstream sequence-processing AI is temporally reactive: each update is driven by input and learned state transitions, rather than by an independently evolving internal temporal reference. Biological brains, by contrast, continuously generate rhythmic activity that structures perception, memory, and cognition, and thi...

M. Hashemi, Alireza Valizadeh, V. Jirsa · 0 citations
Open access Sep 2026

Reservoir computing based on cellular automata (ReCA) to solve the 20 QA bAbI Tasks from Facebook

Reservoir computing based on cellular automata (ReCA) constructs a bridge between automata computational theory and recurrent neural architectures. Sequence learning is an essential capability for a wide collection of intelligence tasks such as language, continuous vision, and symbolic manipulation in a knowledge base,...

Mrwan Margem · 0 citations

Looped Actor: Depth-Recurrent Reasoning Models for Reinforcement Learning

Looped reasoning models repeatedly apply a shared set of parameters, enabling more computation without increasing the model size. These models also support input-dependent computation by dynamically deciding when to stop looping. Motivated by the recent success of looped transformers in language modeling and reasoning,...

T. Konstantin Rusch, T. Seyde, J. Boyer et al. · 0 citations

Related blog posts

MIT News · Artificial Intelligence Sep 29, 2026

Who we become when we talk to machines

Professor Sherry Turkle’s new book, “Artificial Intimacy,” offers a withering critique of chatbots and the antisocial dynamics she believes they encourage.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.