Predicting Brain Responses To Natural Movies With Multimodal LLMs

TL;DR

This paper introduces a multimodal approach using pretrained models across video, speech, text, and vision-text modalities to predict brain responses to natural movies, achieving competitive results in the Algonauts 2025 challenge.

Contribution

It demonstrates the effectiveness of combining multimodal features and ensemble methods with a simple architecture for improved brain response prediction.

Findings

Achieved a mean Pearson's correlation of 0.2085 on test data.

Ensembling and multimodal feature integration improved generalization.

Method ranked fourth in the Algonauts 2025 challenge.

Abstract

We present MedARC's team solution to the Algonauts 2025 challenge. Our pipeline leveraged rich multimodal representations from various state-of-the-art pretrained models across video (V-JEPA2), speech (Whisper), text (Llama 3.2), vision-text (InternVL3), and vision-text-audio (Qwen2.5-Omni). These features extracted from the models were linearly projected to a latent space, temporally aligned to the fMRI time series, and finally mapped to cortical parcels through a lightweight encoder comprising a shared group head plus subject-specific residual heads. We trained hundreds of model variants across hyperparameter settings, validated them on held-out movies and assembled ensembles targeted to each parcel in each subject. Our final submission achieved a mean Pearson's correlation of 0.2085 on the test split of withheld out-of-distribution movies, placing our team in fourth place for the competition. We further discuss a last-minute optimization that would have raised us to second place. Our results highlight how combining features from models trained in different modalities, using a simple architecture consisting of shared-subject and single-subject components, and conducting comprehensive model selection and ensembling improves generalization of encoding models to novel movie stimuli. All code is available on GitHub.