To show this, we modeled gaze in immersive VR scenes using spatial, visual, and conceptual feature spaces — and found that LLMs helped predict each person’s unique gaze patterns beyond visual or spatial models.
TLDR- idiosyncratic conceptual priorities shape how we encode our visual environments.
Very excited to share this paper, led by @ajhaskins.bsky.social !
We find that how people look around the world is stable and idiosyncratic — and that these gaze patterns are shaped in part by the conceptual priorities each person brings to a scene.
1/3
www.pnas.org/doi/10.1073/...
