Semantic-First Spatial Cognition: A Functional Affordance Architecture for Visual Understanding

Abstract

We propose that the standard vision pipeline inverts the actual structure of biological spatial cognition. Spatial awareness is grounded in functional semantics: geometric structure becomes accessible to reasoning only through prior contextual and affordance encoding. We formalize this as the Semantic-First Spatial (SFS) architecture and derive testable predictions distinguishing it from geometry-first alternatives.