VLMs May Not Globally Enhance Human Alignment over LLMs During Natural Reading

ArXi:2605.28818v1 Announce Type: new Large language models (LLMs) have become increasingly useful computational models of human language processing, but it remains unclear whether vision-language learning makes text representations human-like during natural reading. Here, we address this question by comparing tightly matched LLM and vision-language model (VLM) pairs under a strictly text-only setting, allowing us to isolate the effect of multimodal