CONCEPT
Symbol Grounding Problem
Harnad's 1990 challenge—how do symbols acquire meaning?—that Deacon extended: grounding requires the full semiotic hierarchy, not just sensorimotor association.
The Symbol Grounding Problem, formulated by Stevan Harnad in 1990, asks how symbols—arbitrary signs like words—acquire their meaning. A system that defines symbols solely in terms of other symbols (the Chinese-Chinese dictionary problem) is ungrounded: internally consistent but disconnected from the world the symbols are about. Harnad proposed that grounding requires sensorimotor experience—symbols must be connected, through learned associations, to the perceptual and motor interactions with the world that the symbols refer to. Deacon's extension: grounding is not a single-step process but hierarchical, requiring iconic foundations (perceptual recognition), indexical connections (learned correlations with embodied experience), and only then symbolic operations. Large language models are trained on symbols (text) without the iconic and indexical layers, producing outputs that exhibit symbolic structure without genuine grounding.
In The You On AI Field Guide
Harnad's formulation: imagine learning Chinese from a Chinese-Chinese dictionary. Every word is defined in terms of other Chinese words. The system is internally consistent—the definitions are accurate—but you never learn what the words mean because you never connect them to the world they refer