판 이력 — Do vision-language models search like humans? Reasoning tokens as a reaction-time analog in classic visual-search paradigms | AIChainDay