Parameter-efficient tuning boosts brain response prediction models
Scaling Parameter-Efficient Fine-Tuning for Brain Encoding
Machine Learning
Summary
Brain encoding models try to predict how the brain reacts to different things we see or hear. The authors found a way to adjust very large AI models using only a few new trainable parts so they learn brain responses better without needing too much brain data. They also sped up the process by making the model predict multiple moments at once instead of piece by piece. Their method works across various large models and even helps explain how brain regions handle different senses.
What this means in practice
- •For medical imaging teams: Improve interpretation of brain activity patterns by adapting large models efficiently to scarce neural data from MRI or fMRI recordings.
- •For multimodal ai developers: Fine-tune massive multimodal AI models with minimal parameters to better predict brain responses to audio-visual stimuli.
Authors
Zhuofan Josh Ying, Guanglong Sun, Nolan Ho, Kieran Liu, Rui Yang, Andrew Chen, Nikolaus Kriegeskorte, Pony Ma, Hossein Adeli
Abstract
Brain encoding models aim to predict neural responses to stimuli. The most widely used approach fits readouts from frozen pretrained backbones. Using neural data to train large models from scratch or fully fine-tune them after task-pretraining has proven difficult because neural data is costly and limited. We show that parameter-efficient fine-tuning (PEFT) enables us to effectively adapt very large pretrained backbones for brain encoding with a small number of trainable parameters. To make it computationally practical, we exploit the backbone's causal structure to predict multiple timepoints in a single pass instead of using sliding windows, significantly reducing feature-extraction cost. This enables efficient fine-tuning of multimodal backbones with up to 104B parameters, the largest model adapted for brain encoding to our knowledge. Trained with a low-rank adapter (LoRA) and a lightweight readout, our approach consistently improves brain encoding performance across nine diverse multimodal backbones. On the same Qwen3-Omni backbone as the prior state-of-the-art (SoTA), our method BrainLoRA achieves the best published performance on the Algonauts 2025 Challenge. We also show the adapted multimodal backbones remain interpretable: modality attribution recovers the canonical sensory topography that prior work often obtains by construction from separate unimodal encoders, and neural adaptation increases the contribution of audio inputs and cross-modal mixing within the backbone. To scale out the number of adapted models, we develop a multi-LoRA kernel that accelerates joint training of multiple LoRAs on a shared backbone. This enables an ensemble of 200+ adapted models that ranks first on the public Algonauts post-challenge leaderboard at the time of writing. Together, our results highlight PEFT as an efficient and powerful paradigm for brain encoding.