NeurIPS: Neuro-anatomical Inductive Priors for Sphere-based Vision Brain Decoding

ICLR 2026 Conference Submission174 Authors

Keywords: brain decoding, spherical convolution, MoE

Abstract: Generalizable fMRI decoding is hindered by the challenge of aligning signals from anatomically unique brains. Prevailing methods treat this anatomical variation as noise, creating a false performance-fidelity trade-off where efficient 1D encoders outperform geometrically faithful surface-based models. We argue this trade-off is an artifact of two core mismatches: inefficient surface tokenization and the failure to use anatomy as a predictive signal. We present **NeurIPS**, a framework that resolves both by reframing anatomical variation from a nuisance to a powerful inductive prior. NeurIPS unites two innovations: a **Selective ROI Spherical Tokenizer (SRST)** for efficient geometric encoding, and a **Structure-Guided Mixture of Experts (SG-MoE)** that explicitly models individual anatomy using cortical features. On the Natural Scenes Dataset, NeurIPS establishes a new state-of-the-art for surface decoders and achieves performance comparable to strong 1D baselines. This is achieved with unprecedented efficiency, as the model converges dramatically faster (**10 vs. 600 epochs**). This efficiency enables rapid adaptation to new subjects using only **20\%** of their data and ensures robust scalability as the training cohort is expanded. Ablations provide causal evidence that these gains are driven by the model's use of cortical features, not by memorizing subject IDs. By leveraging anatomical priors, NeurIPS provides a principled and scalable path toward robust, generalizable brain decoding.

Primary Area: applications to neuroscience & cognitive science

Submission Number: 174