BraiNCA: brain-inspired neural cellular automata and applications to morphogenesis and motor control
This work addresses the need for more biologically-realistic and efficient self-organizing systems in AI, though it is incremental as it builds on existing NCA frameworks.
The paper tackled the problem of Neural Cellular Automata (NCAs) lacking long-range connections and complex topologies by introducing BraiNCA, a brain-inspired NCA with attention and long-range edges, which showed better robustness and learning speed compared to Vanilla NCAs on morphogenesis and motor control tasks.
Most of the Neural Cellular Automata (NCAs) defined in the literature have a common theme: they are based on regular grids with a Moore neighborhood (one-hop neighbour). They do not take into account long-range connections and more complex topologies as we can find in the brain. In this paper, we introduce BraiNCA, a brain-inspired NCA with an attention layer, long-range connections and complex topology. BraiNCAs shows better results in terms of robustness and speed of learning on the two tasks compared to Vanilla NCAs establishing that incorporating attention-based message selection together with explicit long-range edges can yield more sample-efficient and damage-tolerant self-organization than purely local, grid-based update rules. These results support the hypothesis that, for tasks requiring distributed coordination over extended spatial and temporal scales, the choice of interaction topology and the ability to dynamically route information will impact the robustness and speed of learning of an NCA. More broadly, BraiNCA provides brain-inspired NCA formulation that preserves the decentralized local update principle while better reflecting non-local connectivity patterns, making it a promising substrate for studying collective computation under biologically-realistic network structure and evolving cognitive substrates.