Large Language Models Align with the Human Brain during Creative Thinking

2026-04-03

AI Summary

This study addresses a critical gap in current research by investigating how large language models (LLMs) align with human brain activity during creative thought, an area largely overlooked in favor of passive tasks. Leveraging fMRI data from 170 participants performing the Alternate Uses Task (AUT), the authors employ representational similarity analysis (RSA) to systematically evaluate neural alignment across LLMs ranging from 270M to 72B parameters, subjected to distinct post-training strategies—creativity optimization, behavioral fine-tuning, and reasoning training—within the default mode network (DMN) and frontoparietal network. The findings reveal, for the first time, that alignment between LLM representations and creative brain activity increases with model scale exclusively in the DMN and positively correlates with originality. Moreover, creativity-optimized training selectively enhances alignment with high-creativity neural responses while suppressing low-creativity alignment, demonstrating that post-training objectives reshape the geometric structure of LLM representations in a functionally specific manner relative to creative cognition.

Abstract

Creative thinking is a fundamental aspect of human cognition, and divergent thinking-the capacity to generate novel and varied ideas-is widely regarded as its core generative engine. Large language models (LLMs) have recently demonstrated impressive performance on divergent thinking tests and prior work has shown that models with higher task performance tend to be more aligned to human brain activity. However, existing brain-LLM alignment studies have focused on passive, non-creative tasks. Here, we explore brain alignment during creative thinking using fMRI data from 170 participants performing the Alternate Uses Task (AUT). We extract representations from LLMs varying in size (270M-72B) and measure alignment to brain responses via Representational Similarity Analysis (RSA), targeting the creativity-related default mode and frontoparietal networks. We find that brain-LLM alignment scales with model size (default mode network only) and idea originality (both networks), with effects strongest early in the creative process. We further show that post-training objectives shape alignment in functionally selective ways: a creativity-optimized \texttt{Llama-3.1-8B-Instruct} preserves alignment with high-creativity neural responses while reducing alignment with low-creativity ones; a human behavior fine-tuned model elevates alignment with both; and a reasoning-trained variant shows the opposite pattern, suggesting chain-of-thought training steers representations away from creative neural geometry toward analytical processing. These results demonstrate that post-training objectives selectively reshape LLM representations relative to the neural geometry of human creative thought.

Problem

• creative thinking
• large language models
• brain alignment
• divergent thinking
• neural representations

Innovation

• brain-LLM alignment
• divergent thinking
• representational similarity analysis
• post-training objectives
• creative neural geometry