Turtles or Axiom: Two Structural Barriers to Inferring Beyond Experience

Skeptical scenarios about brains in vats, simulations, and related outsides are usually framed as disputes between realists and skeptics over which external world is real. This framing overlooks that both sides share a commitment to an inference pattern that moves from what is experienced to what lies beyond experience. This paper argues that the shared inference pattern faces two structural barriers that operate independently of any content. The first is that any inference form strong enough to license the move from inside to outside is also valid, by its own operation, applied to whatever it reached. Iteration produces either endless regress, which empties the target of content, or an arbitrary halt not licensed by the argument. The second is that inferences use rules, and those rules had to be learned through use, which is experiential. For the inference to cross the boundary between the experienced and the claimed unexperienceable, its rules must either transport, in which case uniformity across the boundary is itself an externalist claim requiring the kind of justification the externalist cannot provide, or they do not, in which case no inference reaches across. The two barriers are independent and mutually reinforcing. What remains when both are honored is a binary: commit to endless regress and empty the outside of content, or accept that some things are axiomatic and face them directly. The argument is structural rather than metaphysical, and it bears on brain-in-a-vat scenarios, simulation arguments, Moore-style realism, and any other attempt to infer an outside from an inside.