The Symbol Grounding Problem
Updated 2026-08-08
INTRODUCTION
English translation pending.
CORE DEFINITION
Formulated by Stevan Harnad. A symbol system manipulates tokens according to formal rules, but nothing in those rules connects a token to what it stands for. Harnad argued that meaning cannot be grounded in further symbols alone, since that only defers the problem, and must ultimately connect to non-symbolic representations arising from perception and action. The argument bears directly on whether language models understand anything.
SCAFFOLDING EFFECT
Reduce cognitive load
- Test for grounding: ask whether a system's symbols connect to perception, action, or causal feedback - Distrust fluency: separate the ability to produce plausible text from the ability to represent what it refers to - Anchor meaning: connect abstract labels to concrete sensorimotor or outcome data where understanding matters
Anchor fast decisions
Formal symbol manipulation is defined by relations among tokens, and those relations are independent of anything outside the system. Meaning requires a connection from tokens to things the system can act on or perceive, which purely internal relations cannot supply. Grounding therefore has to come from outside the symbol system, which is why adding more text alone cannot close the gap.
MINIMUM ACTION
In progress 0/1Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- en.wikipedia.orghttps://en.wikipedia.org/wiki/Symbol_grounding_problemverified
PRIVATE NOTES · Only visible to you
SAVED Q&A
ENTRY Q&A · Private saving available
Ask with a clear boundary
thinkingmodels answers from published entry context only.
Your question is sent to thinkingmodels. The answer uses public entry context only.
RELATED MODELS