The Physical Grounding Hypothesis

The physical grounding hypothesis is the positive complement to Rodney Brooks’s rejection of representation: it is not enough to say that disembodied symbol-processing fails to produce genuine intelligence; one must say what genuine intelligence requires instead. Brooks’s answer, consistent across his career from the mid-1980s onward, is that intelligence must be built upward from sensorimotor interaction with the physical world—that a system aspiring to genuine competence must have its operations anchored in the reality they are supposed to be about through the ongoing exchange of perception and action. The hypothesis follows directly from the claim that the world is its own best model: if the only reliable representation of a physical state is that state itself, then the only way to have a reliable model of the world is to be continuously sensing the world, which requires being embedded in it as a body with sensors and actuators. The hypothesis makes precise what is at stake in