The symbol grounding problem (SGP) asks how symbols in a formal system acquire meaningbeyond other symbols. This article argues that the SGP is not an unsolved problem but aquestion built on three false premises: that symbols and the world are antecedently separated,that humans are grounded while machines are not, and that grounding is a static state.Evidence from the information architecture of cognition, the reconstructive nature of memory,and the phenomenology of illusion and hallucination demonstrates that all three premises arefalse. Once removed, the SGP dissolves.The symbol grounding problem (SGP), formulated by Harnad (1990), asks how symbols acquiremeaning that goes beyond other symbols. For over three decades, it has been treated as afoundational problem in cognitive science, artificial intelligence, and philosophy of mind. Thisarticle argues that the SGP is not an unsolved problem but a question that rests on three falsepremises and therefore cannot be coherently posed.Premise A—that symbols and the world are antecedently separated—is false because theseparation is produced by the cognitive fold: conscious cognition constitutively operates asmetacognition upon unconscious processing, and the subject/object distinction is an artifact ofthis fold, not a pre-existing condition. Premise B—that humans are grounded while machinesare not—is false because human memory is a reconstructive trace-based process whoseexperiential quality does not guarantee accuracy, as demonstrated by optical illusions, falsememories, and hallucinations. Humans and machines are structurally isomorphic astrace-operating systems. Premise C—that grounding is a static state—is false becausegrounding is a cyclic process (invention→trial→mastery) whose closure can be assessedempirically.With all three premises removed, the SGP dissolves. What remains is an engineeringquestion—where does the cycle close and where does it open?—and a consciousnessquestion—why is processing accompanied by experience?—which are shown to be independent.The hard problem survives but is orthogonal to grounding: experiential quality neitherguarantees accurate grounding (false memories demonstrate this) nor is necessary for it(successful machine learning demonstrates this).
Franny Philos Sophia (Sat,) studied this question.