"Just a simulation" isn't an argument

On the properties of sim-fire

This piece was inspired by Eric Schwitzgebel’s newest book, The Weirdness of the World, particularly the chapter on transcendental idealism and cyberpunk. Check it out! However, the argument and terminology is original—blame me, not Schwitzgebel, for any mistakes.

---

Gemini 3 Pro Image.

When I asked for arguments why computers can’t think, I’ve gotten comments like “The same reason why I don’t think there are little people in my TV acting out all my shows. Appearance of thought isn’t thought.” But I don’t think this is an argument, I think it’s a premise or conclusion in disguise! One asserts that the difference in substrate matters, without explaining why it matters, just saying that it isn’t “real.” What I want to demonstrate today is that “real” is acting as a linguistic crutch. Objects in a simulation can have properties. So what is different between “thought” in the simulation and thought which interacts with brains that isn’t just trivially implied by substrate differences? And why do you think that? There might be a good argument! But deferring to “real” or “just a simulation” doesn’t explain much. Let’s see why!

In a thought experiment coming to theaters near you, I’ve hooked up millions of brains-in-a-vat to a massive computer network¹ and created a simulated world. It features an immaculate physics engine: all simulated objects behave as though under the real laws of physics to high precision. A brain-computer interface stimulate the right neurons to produce indistinguishable sights, sounds, smells, tastes, feels, proprioceptive orientation, chemoreception, etc. from what you’d have in the real world.² Motion works similarly: signals to the spinal cord are intercepted by the computer, which updates the relevant data on your body position in the physics engine, which sends sensory and spinal feedback signals back to your brain.

In the simulation, I observe what looks, feels, and smells like an unlit fireplace stacked with logs. What is it that I am perceiving?

It is common to say that the sim-fireplace isn’t “real”. And that is a fine use of everyday language, but for philosophical purposes it’s crucial to specify what is meant by the term “real.” When you warm your sim-hands at the sim-fireplace, you are having a sensory experience, causally linked to a physical computer. Those things make the fireplace “real” in one sense, and “unreal” in another.

I don’t mean to cry “it’s all relative!” In fact, I want to do the exact opposite. I want to focus on what is substantively different between a sim-fireplace and a brick-and-wood fireplace—what I’ll call an Earth-fireplace—rather than quibbling about definitions. So I will Taboo the word “real” from here on out, and try to focus on what precisely is different between the sim-fireplace and an equivalent Earth-fireplace.

I get the name “sim-fireplace” by having a sensory experience just like I’d have with an Earth-fireplace. The sensations I get from the Earth-fireplace come from an arrangement of brick and wood, which reflect light, occupy space, and have odorants which trigger my sensory receptors. The sensations I get from the sim-fireplace come from electrons zipping through copper wires and silicon transistors, which trigger my sensory system through direct electrical stimulation. So there is a kind of Earth-substrate to what I am perceiving. But this substrate differs from the substrate of an Earth-fireplace in three primary ways.

1. It is constituted by different matter.

2. It has a different spatial organization. The data associated with the fireplace may be spread all over the computer or localized a million miles away. Even if it is close and well-organized, there is no motherboard in the shape of a fireplace somewhere in the computer.

3. It may have a different temporal organization. If two hammers strike an Earth-fireplace at the same time, their causal effects will occur simultaneously. If two sim-hammers strike a sim-fireplace at the same time, the underlying processing might look like rapid serial calculations: first determining the impact and structural resistance associated with the first hammer-strike, then the second strike, then adding them together, and finally creating a “world update” which updates the stored data on the fireplace and creates sensory perceptions as if both sim-hammers were striking simultaneously. Thus, while the events are simultaneous relative to the sim-world, they may not correspond to simultaneous Earth-world events in the substrate.

NB: All of these are Earth-differences: differences in the properties of the physical computer running the sim vs. a physical brick-and-wood fireplace. They are not sim-differences: the sim-fireplace in the sim-world appears and interacts with other sim-objects just as Earth-objects would interact with each other.

Equipped with this kind of thinking, we can answer more questions about the difference between sim-objects and Earth-objects in more detail. I strike a sim-match, release it over the sim-fireplace, and watch as the logs burst into sim-flame.

Is sim-fire “real”? Well, its Earth-substrate consists in different matter and spatiotemporal organization from Earth-fire. As such, sim-fire does not make the silicon transistors melt. However, sim-fire shares sim versions of Earth-fire’s properties. For instance, on Earth, temperature just consists in the average kinetic energy of atoms moving around in space. Sim-temperature would just consist in the average kinetic sim-energy—a computed quantity obeying the same mathematical relationships as kinetic energy in Earth physics—of sim-particles. So sim-fire does raise the sim-temperature of those things around it; it can make other things sim-combust; it can provide sim-illumination, and so on.

Be maximally precise! What is the difference between Earth-fire and sim-fire? Only the type of matter and spatiotemporal organization of their substrates. Therefore, any other fact which differs between the two must be derivative of facts about that. For instance, Earth-fire (but not sim-fire) can damage Earth-redwoods, because Earth-redwoods are made of carbon-based molecules that break apart in response to sufficient thermal energy. And Earth-fire might be able to damage sim-redwoods by melting transistors. But sim-fire can only damage sim-redwoods. That’s a difference, and it is relevant when we are talking about practical questions of damage prevention: it tells us we need not worry about sim-fires endangering California habitats. However, it is a difference which is directly entailed by microphysical differences in substrates, not by intuitions about which is more “real.”

Now, suppose I take the final step and create sim-brains, extremely detailed physical simulations of Earth-brains, and allow them to control avatars in the simworld just like all the Earth-brains in vats. These sim-brains don’t have Earth-synapses, Earth-neurons, or Earth-hormones. But they do have sim-synapses, sim-neurons, and sim-hormones. Just as in the case of the sim-fireplace, the only difference between Earth-brains and sim-brains are microphysical—what protons and electrons constitute the physical substrate, and how they are spatiotemporally arranged on Earth—and any facts that are strictly entailed by those differences.

So, if we start up the sim-brain, sim-synapses will start firing, the sim-body will start moving, and we get something which is acting a whole lot like a human being.³ So just as it is reasonable to say that sim-fire can sim-burn, sim-spread, and provide sim-heat, it is reasonable to say that the sim-brain is sim-thinking. I am not yet saying that sim-thinking is identical to Earth-thinking, no more than I would say that sim-combustion is identical to Earth-combustion. Different physical events are occurring! But every difference between sim-combustion and Earth-combustion is explicable in terms of microphysical differences, in the details of bond energy between atoms in Earth-molecules which don’t physically interact with sim-heat. All properties which Earth-thinking possesses which sim-thinking lacks have to trace back, in one way or another, to those microphysical details of substrate: the type of matter, their Earth-spatial organization, and their Earth-temporal organization.

So: how does that happen? Explain why it matters that the process occurs in sodium gradients and chemical releases rather than electrons through copper wire. Explain how that creates a difference relevant to philosophy of mind, how carbon is essential for “meaning” or “consciousness.” If there is some kind of chemical or particle reaction which creates those phenomena, explain why there couldn’t be a sim-version of those phenomena which is equally relevant to our philosophical inquiry. Furthermore, explain how you could be justified in believing this. What evidence do you have? Where did you get it?

I am not saying these questions have no answers. People have tried to answer them—John Searle, Antonio Damasio, and Evan Thompson⁴ are a few. What makes these fellows worth considering is that they do try to answer these questions, and offer arguments for their position. But the bare assertion that sim-thinking isn’t “real” is no more meaningful than the assertion that sim-fire isn’t “real”—sure, different particles, everyone agrees on that, but so what? If you can’t explain how they cash out in philosophically relevant properties, then “real” is just a crutch.

1. I am assuming a classical computer here because I tried to write it with a quantum computer and it became muddled very quickly for those not familiar with QC hardware (like me). Replace the computer with a QC and I don’t think anything about the argument changes.

2. Perhaps we’d also need to have heart, intestines, and indeed a full nervous system in the vat to produce this.

3. You might reject this if you think that Earth-brains have non-physical forces operating on them integral to their behavior, such that a sim-brain couldn’t even perform the functions of an Earth-brain. This would produce less of a philosophical zombie and more of a Walking Dead zombie: shambling around on animal instinct, incapable of even functional “reasoning.” Your homework is to explain: (1) why these forces only act on hydrocarbons and proteins organized in the shape of a brain, but not in silicon and copper organized like a computer with the same functional roles, (2) why equivalent sim-forces wouldn’t occur in a simulation, and (3) what evidence you have for your position without begging the question that the sim-brain isn’t “real.”

4. No relation.