Improve Language Model and Brain Alignment via Associative Memory

TL;DR

This paper explores enhancing the alignment between language models and human brain activity during speech processing by integrating associative memory, verified through brain mapping and fine-tuning on associative datasets.

Contribution

It introduces a method to improve brain-language model alignment using associative memory and supervised fine-tuning with a new associative story dataset.

Findings

Alignment improves in brain regions related to associative memory

Fine-tuned models better match brain responses

Associative memory integration enhances model-brain correspondence

Abstract

Associative memory engages in the integration of relevant information for comprehension in the human cognition system. In this work, we seek to improve alignment between language models and human brain while processing speech information by integrating associative memory. After verifying the alignment between language model and brain by mapping language model activations to brain activity, the original text stimuli expanded with simulated associative memory are regarded as input to computational language models. We find the alignment between language model and brain is improved in brain regions closely related to associative memory processing. We also demonstrate large language models after specific supervised fine-tuning better align with brain response, by building the \textit{Association} dataset containing 1000 samples of stories, with instructions encouraging associative memory as input and associated content as output.