Exploring Spatial Schema Intuitions in Large Language and Vision Models

Abstract: Despite the ubiquity of LLMs in AI research, the question of embodiment in LLMs remains underexplored, distinguishing them from embodied systems in robotics where sensory perception directly informs physical action. Our investigation navigates the intriguing terrain of whether LLMs, despite their non-embodied nature, effectively capture implicit human intuitions about fundamental, spatial building blocks of language. We employ insights from spatial cognitive foundations developed through early sensorimotor experiences, guiding our exploration through the reproduction of three psycholinguistic experiments. Surprisingly, correlations between model outputs and human responses emerge, revealing adaptability without a tangible connection to embodied experiences. Notable distinctions include polarized LLM responses and reduced correlations in vision LLMs. This research contributes to a nuanced understanding of the interplay between language, spatial experiences, and the computations made by LLMs. More at https://cisnlp.github.io/Spatial_Schemas/