Reality+: Virtual Worlds and the Problems of Philosophy

by David J. Chalmers

If we’re in a simulation, the simulated trees we perceive may not be the original trees (simulators may have modeled them after unsimulated trees) and the simulated physical world will not be fundamental (s...

say that virtual objects are not real. Still, these criteria seem marginal as criteria for reality. Dolly the cloned lamb is not original (she’s a clone of another lamb) and not fundamental (sh...

David Deutsch in his 1997 book The Fabric of Reality. In a fascinating discussion of VR, Deutsch advocates a partial sort of virtual realism. He argues that VR environments (including simulations) “pass the test for reality” because they “kick back” at the user. That suggests versions of the second and third criteria: These environments have causal powers and are independent of our minds.

Deutsch does not endorse the first criterion: He says of one scenario, “the simulated aircraft and its surroundings do not really exist.”

Nor does he endorse the fourth and fifth criteria. He says that in another VR scenario you can see rain that’s not there in reality, and that an engine in the scenario is not a real engine. In Deutsch’s view, VR en...

Simulation realism makes a much stronger c...

Of course, if we’re in a perfect simulation, then some things aren’t ...

Some of what we believe wil...

Most people believe that they’re not in a simulation. They believe that flowers are not digital. They may believe that their universe is the ultimate reality. If we’re in a simulation, those beliefs will be wrong. But the undermined beliefs here are mostly scientific or philosophical beliefs about reality. Undermining them ...

If we’re in an imperfect simulation, more of our beliefs will be wrong, but ple...

If the solar system is fully simulated but the rest of the universe is only a sketch, then my beliefs about the Sun may be correct, but my beliefs about Alpha Centauri may be wrong. If 2019 is simulated but 1789 was not, then my beliefs about the French Revolution may be false, but my beliefs about the contemporary United States may be correct. Still, we’ll b...

Assuming the deer I see in my garden is part of the simulation, I’ll still be right that t...

One source of resistance is that trees and flowers certainly don’t seem to be digital objects.

But they don’t seem to be quantum mechanical objects either.

Yet deep down they are, and few people think that the mere fact that trees are grounded in quantum pr...

I think that being digital is just like being quantum...

Science has taught us that there’s much more to reality than initially...

For millennia, we didn’t know that cats and dogs and trees are made of cells, let alone that the cells are made of atoms or that those are fundamentally quantum mechanical. Yet these discoveries about the nature of ...

If I’m right, the discovery that we’re in a simulation should be treated the same way. It will be a discovery about the underlying nature of cats and dogs and trees—that they spring from dig...

I’m not saying in an unqualified way that a simulated tree is the same as a real tree. I’ve said that if we’re in a perfect permanent simulation, then the real trees of our universe are simulated trees—that is, real trees have been digital trees all along. On the other hand, if we’re not in such a simulation and are merely looking at one from the outside, then the trees in the simulation are completely different from the trees in our outside world: Simulated trees are digital; real trees are not. The real trees ...

I will argue that the simulation hypothesis should be seen as a variant of the “it-from-bit” hypothesis that’s b...

This hypothesis posits a digital level underlying physics: roughly speaking, molecules are made of atoms, atoms are made of q...

This is a view in which physical processes are real. There are just underlying levels to reality ...

If the simulation hypothesis is true, the it-from-bit hypothesis is true. Physical reality is perfectly real—there’s just an underlying level consisting of the interaction of bits, and perhaps still further levels still underlying that.

The it-from-bit hypothesis corresponds to one part of the simulation hypothesis—the simulation itself. What about the other part—the simulator who initiated the simulation?

At least, the simulator can be seen as the creator of the it-from-bit universe.

could this allow for demi-god type creatures who could create levels of conscious life/existence under direction of the ultimate Triune God

I’ll argue that the simulation hypothesis is itself a combination of the it-from-bit hypothesis and the creation hypothesis (according to which a creator created the universe).

If I’m right, the simulation hypothesis is not a skeptical hypothesis in which nothing exists. Instead it is a metaphysical hypothesis, a hypothesis about the nature of reality. It’s equivalent to a metaphysical hypothesis (the creation hypothesis) about how our world was created, plus a separate metaphysical ...

If the simulation hypothesis is right, the physical world is made of bits, and a creator created the physical...

My response to Cartesian skepticism rests on a positive answer to the Reality Question. In a perfect simulation, things are perfectly real.

Generalizing simulation realism to these scenarios, we arrive at the no-illusion view of Cartesian scenarios. In these scenarios, things are largely as they seem. Subjects in these situations aren’t deceived; they have largely true beliefs about the world. If the no-illusion view is right, then Cartesian scenarios are no bar to knowing about the external world. If most of our beliefs are correct in a Cartesian scenario, then our inability to rule out that scenario does nothing to cast doubt on our beliefs. Our beliefs are more robust than many Cartesians have thought.

chapter 4, might have held this view. Recall that Berkeley was an idealist who argued that appearance is reality. In an evil-demon scenario, the tables and chairs have the same appearance as tables and chairs in the physical world. If appearance is reality, it follows that people in that scenario will experience real tables and chairs, and will have mostly true beliefs about their reality. Alas, it looks as if Berkeley never endorsed this view. He never discusses Descartes’s evil-demon scenario at all. He would probably have considered the evil demon an impossibility; like Descartes, he thought God alone was perfect enough to produce sensations and perceptions like ours.

Still, he insisted that God is doing this and when he’s doing it, we’re getting things right. You could think of that as a cousin of my view on skeptical scenarios, with God playing the role of the simulator or evil demon. The first really clear statement of the no-illusion view that I know of appears in a 1949 essay by the University of Nebraska philosopher Oets Kolk Bouwsma. Bouwsma was a student of Ludwig Wittgenstein, who inspired Jonathan Harrison’s hallucinating Baby Ludwig

Bouwsma’s article “Descartes’ ...

Like me, Bouwsma thinks that the subject in Descartes’s evil-demon scenario is not undergoing an illusion. His reasons differ from mine, however. Bouwsma thinks that an illusion is an illusion only if it can be discovered. If an illusion cannot be discovered, as in Descartes’s evil-demon scenario, it’s not an illusion at all, and it’s not a deception. More precisely, Bouwsma’s evil demon can discover the illusion, but his subject, Tom, cannot. So when the demon says, “This is an illusion,” he’s correct, but if Tom were to say this, Tom would be wrong. For similar reasons, what Tom experiences counts as a “flower” for Tom but not for the demon. Consequently, when Tom says, “That’s a flower,” he’s correct. Bouwsma’s reasoning here is grounded in a sort of verificationism, which says that all meaningful claims are verifiable. You’ll recall that earlier we discussed the verificationism of Rudolf Carnap and other logical positivists, who held that skeptical hypotheses are unverifiable and therefore meaningless. Bouwsma takes a related line. He thinks that the hypothesis that there’s an illusion is meaningless if it cannot be verified. If we can never verify that the evil-demon scenario is an illusion, it’s not an illusion at all.

I reject verificationism. Many meaningful claims are unverifiable, and there may well be illusions we never discover. So I reject Bouwsma’s analysis of the situation. Nevertheless, I think he’s right about the crucial point. The subject in Descartes’s evil-demon scenario is not undergoing an illusion.

Zhai holds that as long as we have stable and coherent perception of an object, that object is real. In the spirit of Berkeley, we might say that stable and coherent appearance is reality.

reject Zhai’s idealist framework for much the same reasons that I rejected Berkeley’s idealism in chapter 4, but like verificationism, idealism provides one important route to the no-illusion view.

we encountered Putnam’s argument that the brain-in-a-vat hypothesis is contradictory based on externalism, his theory that the meaning of a word depends on its external context. Putnam suggests that the beliefs of a brain in a vat are about the electronic impulses in its environment, and that these beliefs are mostly true. This is a version of the no-illusion view that can potentially drive an entirely different response to skepticism (though Putnam doesn’t make the connection to skepticism). Instead of arguing that the brain-in-a-vat hypothesis is contradictory, one could argue that the beliefs of a brain in a vat are mostly true. In effect, where Bouwsma argues from verificationism to the no-illusion view, Putnam argues from externalism to the no-illusion view.

In my view, the argument doesn’t quite succeed. Just as Bouwsma’s argument requires an implausible verificationism and Zhai’s view requires an implausible idealism, Putnam’s argument requires a strong and implausible version of externalism. Still, like Bouwsma and Zhai, Putnam is on the right track in holding the no-illusion view.

Bouwsma, Putnam, and Zhai offer three different routes to the no-illusion view of simulation scenarios. All three routes can lead to what I think is the correct view of the Cartesian skeptical problem: the subject in Cartesian scenarios has largely true beliefs and is not deceived. However, all three routes rest on strong and implausible philosophical views that I reject. As a result, I think that none of these three philosophers has given a strong argument for the no-illusion view nor a plausible analysis of just why it is true.

I think the best argument for the no-illusion view comes not from verificationism, idealism, or externalism but from a sort of structuralism about the external world. I’ll work up to t...

If the simulation hypothesis is true and we’re in a simulated world, then the creator of the simulation is our god.

The simulator may well be all-knowing and all-powerful. What happens in our world depends on what the simulator wants.

We may respect and fear the simulator. At the same time, our simulator may not re...

The transhumanist philosopher David Pearce has observed that the simulation argument is the most interesting argument for the existence of God in a long time. He may be right.

First, God is the creator of the universe. Second, God is all-powerful. God can do anything. Third, God is all-knowing. God knows everything. Fourth, God is all-good. God is perfectly good, and acts only out of goodness.

At least as a first approximation, we could define a god as a being with these four properties. That is, a god is a being who created the universe and who is all-powerful, all-knowing, and all-good. Later, we can worry about whether all four properties are required, and whether other properties are required.

First, and most important: the simulator is the creator of our universe. She set our universe into motion in a deliberate act of creation—if only by pressing a button in SimUniverse.

Second: the simulator will often be extremely powerful. Many simulations give the simulator huge control over the simulation’s state. Depending on the simulation, there may be limits to this control. For example, someone running Pac-Man can’t rearrange the whole state of the world or change a Pac-Man world into a World of Warcraft world at the press of a button. But many simulators with access to the source code and the data structures involved in their simulation will have near-unlimited powers over the worlds they create.