Substrate Independence

Substrate Independence

Wooden Computer

Computers can be implemented as a mechanical apparatuses, made of say wooden components. A wooden computer can use spheres and wooden ports. Spheres passing through two consecutive ports implement AND operation. Spheres passing through parallel ports implement OR operation.

Just like in the case of electronic computers, also the state of wooden computer can be serialized to a static set of spheres - the execution trace, describing all the state transitions that took place during the execution.

Thermostat as a Computer

Thermostat has only two possible states: it can either be open or closed. Nevertheless, it can be used as a computer. Its operation would directly correspond to the execution trace (a sequence of ones and zeros). Compared to its more advanced counterpart - modern computer - thermostat would just require many more instructions to describe simulated Alice.

Substrate Independence

Because one can construct computers from wood, electronic circuits, waves, or even with pencil and paper, there is no single common substrate to which the simulated consciousness could be traced back to. The only common factor is the information those different implementations and substrates describe.

The physical substrate with which such a structure is described—silicon, neurons, magnetic domains, or a static bitstring—does not therefore determine whether the structure describes conscious Alice.

Proposition: Simulated Consciousness

A simulation $H'$ that is physically and behaviorally indistinguishable from a conscious human $H$ must itself instantiate subjective experience.

Assumption for contradiction. Suppose $H'$ exists such that $H' \equiv H$ (physical and behavioral equivalence) but $S(H') = \emptyset$ (lacks subjective experience):

$$ \mathrm{Behaviour}(H') = \mathrm{Behaviour}(H) \quad \text{and} \quad S(H') \neq S(H). $$

From $A_4$. The behaviour of $H$ is a function of its physical inputs \emph{and} its subjective experience:

$$ \mathrm{Behaviour}(H) = f(\mathrm{Inputs},, S). $$

Contradiction. If the behaviors are identical despite the difference in $S$, then $S$ is not a necessary input to $f$. Subjective experience is causally inert.

Violation of $A_4$. Causal inertness of $S$ directly contradicts Axiom~4. The assumption is therefore false.

Therefore $S(H') \neq \emptyset$: the simulation instantiates subjective experience. $\square$

Two identical axiomatic systems must behave the same.

Axiomatic Alice with pain and Alice without pain are two different systems.