The One Where They Brain-Tune for Social Cognition: Multi-Modal Brain-Tuning on Friends

TL;DR

This study extends brain-tuning of multi-modal models to enhance social cognition by aligning with the Superior Temporal Sulcus during video watching, improving both brain activity prediction and sarcasm detection in sitcoms.

Contribution

It introduces multi-modal brain-tuning targeting social brain regions, demonstrating improved brain alignment and social cognition task performance.

Findings

Increased brain alignment to STS during tuning

Enhanced sarcasm detection accuracy in sitcom context

Demonstrated effectiveness of multi-modal brain-tuning

Abstract

Recent studies on audio models show brain-tuning - fine-tuning models to better predict corresponding fMRI activity - improves brain alignment and increases performance on downstream semantic and audio tasks. We extend this approach to a multimodal audio-video model to enhance social cognition, targeting the Superior Temporal Sulcus (STS), a key region for social processing, while subjects watch Friends. We find significant increases in brain alignment to the STS and an adjacent ROI, as well as improvements to a social cognition task related to the training data - sarcasm detection in sitcoms. In summary, our study extends brain-tuning to the multi-modal domain, demonstrating improvements to a downstream task after tuning to a relevant functional region.