Brief

Researchers have developed a new framework called NeurIPS to improve brain decoding using fMRI data. This approach reframes anatomical variation as a predictive signal, moving beyond the typical performance-fidelity trade-off seen in current decoders. NeurIPS incorporates a novel spherical tokenizer for efficient geometric encoding and a structure-guided mixture of experts that models individual anatomy. The framework achieves state-of-the-art performance for surface-based decoders, matching efficient 1D baselines with significantly faster convergence and requiring less data for subject adaptation. AI

Researchers have developed a novel method to improve brain activity prediction by fine-tuning language encoding models using fMRI data. Despite fMRI's significantly lower temporal resolution compared to ECoG, models trained on fMRI showed enhanced prediction performance for ECoG data. This approach successfully generalized even when fMRI data was temporally downsampled, demonstrating that slower brain recording methods can be valuable for building better models of faster brain signals. AI

Researchers have developed FPED, a novel Mixture-of-Experts (MoE) framework designed for interpretable brain decoding using fMRI data. This approach explicitly models different functional brain networks as specialized experts, utilizing adaptive routing to capture their combined contributions to visual semantic understanding. FPED aims to overcome limitations of current methods that flatten fMRI signals, thereby disrupting the brain's natural network topology and reducing neuroscientific interpretability. The framework demonstrates competitive performance with a small parameter count and offers transparent insights into the correspondence between brain networks and semantic processing. AI