Large Language Models Align with the Human Brain during Creative Thinking

Researchers found that large language models align with human brain activity during creative thinking tasks, with alignment increasing based on model size and idea originality. Different post-training approaches selectively reshape how LLMs align with creative versus analytical neural patterns in humans.

• →LLMs show stronger alignment with human brain activity in creativity-related networks as model size increases from 270M to 72B parameters.
• →Brain-LLM alignment is strongest during early stages of creative thinking and correlates with idea originality.
• →Creativity-optimized models preserve alignment with high-creativity neural responses while reducing alignment with low-creativity ones.
• →Chain-of-thought training steers LLM representations away from creative neural patterns toward analytical processing.
• →Post-training objectives can selectively reshape LLM representations relative to human creative thought patterns.