Whole-brain simulation seems to support mind uploading if computational functionalism about consciousness is true. Implementationists deny this: a digital brain model running on a computer may represent rather than implement consciousness-relevant computation. Dung and Kersten (2025) argue that such constraints cannot be general conditions on computational implementation, since mainstream theories imply that ordinary computers implement computations, whereas implementationist constraints would rule out such systems. I argue that this response moves too quickly. Implementationist constraints are, in fact, substantively equivalent to constraints in mainstream theories of physical computation: computational structure must be borne by objective mechanistic, causal, or dynamical processes. Strictly applied, these theories may not license high-level software implementation. They distinguish low-level physical/digital computations from the looser sense in which computer science says machines “implement” programs. On this reading, implementationists can deny that ordinary computers implement the high-level computation described by a brain model, without denying that they implement some lower-level computation. This does not refute mind uploading. Rather, it clarifies its hardest challenge: computationalist defenders of simulated consciousness must explain how that possibility remains open while taking seriously the prima facie appeal of implementationist theories that privilege objective physical processes.
