Reality+ : Virtual Worlds and the Problems of Philosophy (9780393635812), page 39
Like the TARDIS, simulations are bigger on the inside. If I look at Sim Universe from the outside, it doesn’t look impressive. All I’ll see is a computer—perhaps with some people hooked up to it, depending on how the simulation is arranged. The computer might be running on a device as small as a smartphone. When I look at Sim Universe from the inside, however, it’s enormous. I’ll experience an immersive environment with all sorts of varied content. Like the TARDIS, it might go on forever. From the inside, Sim Universe is a whole world.
Much about simulations depends on whether you consider them from the inside or the outside. I’ve argued that if I’m in Sim Universe—that is, if it’s been my lifelong environment—then the objects in it are completely real. It’s a giant world of trees and mountains and animals. But if I didn’t grow up in a simulation, then Sim Universe does not contain real trees or mountains or animals; it contains simulated trees and simulated mountains and simulated animals. The simulated trees may be real digital objects inside the computer, but they aren’t real trees.
How can this be? The simulation itself is part of objective reality. How could its nature depend on me? How could whether it contains trees or mountains depend on my perspective?
My answer is that the difference between a simulation-from-the-inside and a simulation-from-the-outside is not a difference in reality. It is a difference in language, along with associated differences in thought and perception. If I grew up inside Sim Universe, then I’ve been applying the word “tree” to digital trees my whole life. Digital trees are what I mean by “tree.” If I grew up outside all simulations, then I’ve been applying the word “tree” to nondigital trees my whole life. Nondigital trees are what I mean by “tree.”
So I’ll describe Sim Universe differently depending on whether I’ve grown up inside it or not. If I’ve grown up in Sim Universe, I’ll say it contains trees, because “tree” for me means “digital tree.” If I’ve grown up outside all simulations, I’ll say that Sim Universe doesn’t contain trees, since “tree” for me means “nondigital tree.”
In objective reality, Sim Universe is not affected by our perspective. It contains digital processes running on a computer. These processes involve objective algorithms supporting objective digital objects. What varies depending on the perspective is how we experience things and how we describe them. We can understand this idea better by thinking a little about language.
Philosophy of language
There are many traditions in philosophy. In this book, I have most often followed the European tradition that was passed down from the ancient Greeks and Romans to the medievals and then to 17th- and 18th-century figures we have encountered such as Descartes and Kant.
In the 19th and especially the 20th century, the European philosophical tradition split in two. One branch has come to be known as continental philosophy because of an early association with the European continent. Its key figures include the German philosophers Hannah Arendt, Martin Heidegger, and Edmund Husserl, and the French philosophers Simone de Beauvoir, Maurice Merleau-Ponty, and Jean-Paul Sartre. The other branch has come to be known as analytic philosophy, originally because of its use of linguistic analysis. Its key early figures include British philosophers we’ve encountered already—Bertrand Russell and G. E. Moore—as well as Germans and Austrians such as Rudolf Carnap, Ludwig Wittgenstein, and Gottlob Frege.
Much of the philosophy in this book falls under the rubric of analytic philosophy. One feature of analytic philosophy, especially in the early days, is an intense focus on logic and language. The analytic philosophers of the Vienna Circle (see chapter 4) held that once a philosophical problem had been clarified enough via logic and language, this would either dissolve the problem or break it down enough that it could be settled by science. A century later, analytic philosophy has become a much broader church, but a concern for clarity and a focus on logic and language remain among its distinctive elements.
Perhaps the founding figure in analytic philosophy was the German philosopher Gottlob Frege, who founded the field of logic as we know it today in the late 19th century. Outside philosophy, he is perhaps best known for developing a theory of the foundations of mathematics—a theory that turned out to be inconsistent when Russell pointed out that it led to a paradox about “the set of all sets that do not contain themselves.” (Would that set contain itself? The answer can’t be “yes” and it can’t be “no.”) Nevertheless, Frege’s theory was a monumental achievement, as was his clarification of the tools of modern logic. Frege was also a pioneer in the philosophy of language, setting out one of the first major theories of what our words mean.
Sadly, Frege was seriously anti-Semitic, as was Martin Heidegger decades later. Like great artists, great philosophers are not always great people. Aristotle and Immanuel Kant have writings that are laden with a now-shocking racism. We can try to separate their core philosophy from their awful views. This is not always straightforward, but in Frege’s case, it’s arguable that his philosophy of logic and language has little connection to his anti-Semitism.
Frege’s best-known contribution to the philosophy of language was to distinguish two aspects of what a word means: its sense and its reference. Reference is easiest to explain. The referent of a word is what it refers to in the world. “Plato” refers to Plato (the person). “Sydney” refers to Sydney (the city). “Groundhog” refers to groundhogs (the animals). “Seventeen” refers to 17 (the number). And so on.
Sometimes two words refer to the same thing. The classic example is the names “Hesperus” and “Phosphorus” for the evening and morning star, respectively. Both refer to the same object, the planet Venus. But they seem to have a different meaning. In his 1892 article “On Sense and Reference,” Frege used this example to argue that there’s more to meaning than reference. Although “Hesperus” and “Phosphorus” refer to the same thing, they have different senses. The sense of a term is roughly the way it presents the referent to the speaker. “Hesperus” presents Venus as the evening star, so its sense is tied to being visible in the evening. “Phosphorus” presents Venus as the morning star, so its sense is tied to being visible in the morning.
Later, Russell offered a twist on Frege’s idea with a suggestive picture of just how words can refer to things in the world. In his landmark theory of names and descriptions, he argued that every ordinary name (such as “Hesperus”) is equivalent to a description—say, “the star that is visible in the evening at a certain location.” This description would denote whatever object satisfied it—in this case, the planet Venus. His theory allowed an analysis of ordinary language using the tools of logic.
The Frege-Russell theory of meaning was popular for many years, but in the 1970s there was a small revolution. Two American philosophers, Saul Kripke and Hilary Putnam, building on earlier work by the philosopher and logician Ruth Barcan Marcus, argued that the Frege-Russell picture builds on numerous false assumptions. In Kripke’s book Naming and Necessity, his main target was descriptivism—the idea that the meaning of a word is something akin to a description. In Putnam’s article “The Meaning of ‘Meaning,’” the main target was internalism—the idea that the meaning of a word is internal to the speaker and does not involve the speaker’s environment.
Putnam’s famous slogan was “Meanings just ain’t in the head!” In their theories of meaning, he and Kripke favored externalism, which says that the meaning of a word depends partly on the speaker’s environment. They replaced Russell’s descriptivism with the causal theory of reference. Putnam’s version of the causal theory says roughly that a word refers to whatever entity in the environment causes the word to be used.
Putnam argued for externalism using a thought experiment: the story of Twin Earth. Twin Earth is a faraway planet that’s just like Earth, except that all the H2O on Earth is replaced by a superficially identical substance, XYZ (a molecular structure that’s different from H2O). XYZ looks and tastes just like water. XYZ falls from the skies, it fills the rivers and oceans, it runs through pipes and comes out of faucets, and all the creatures on Twin Earth drink it.
Is XYZ water? Putnam makes a strong case that it isn’t. Water is H2O, a natural substance found on Earth. XYZ is a different substance, with a similar appearance. We don’t call fool’s gold “gold,” even though it resembles gold. Similarly, we shouldn’t call XYZ “water.” Earth is largely covered with water, but Twin Earth isn’t. It’s largely covered with what we might call twin water.
Figure 49 Do Hypatia (studying H2O) and Twin Hypatia (studying XYZ) mean different things by “water”?
On Twin Earth, there are language users very much like on Earth. Consider Hypatia, the brilliant 4th-century Alexandrian philosopher and mathematician who built hydrometers for measuring the specific gravity of water and other liquids. Hypatia has a near-duplicate on Twin Earth, who is studying XYZ where Hypatia is studying H2O. Let’s suppose the two liquids have behaved just the same in all experiments so far, and no one has discovered the chemical makeup of these liquids yet. Both Hypatia and Twin Hypatia call their liquids “water.” Suppose that Hypatia says, “I’m measuring water.” She’s talking about H2O. But when Twin Hypatia says, “I’m measuring water,” she’s talking about XYZ.
This is enough for Putnam to mount his argument that meaning isn’t “in the head.” Hypatia and Twin Hypatia are near-duplicates of each other, but their words mean different things. Putnam further argues that this was so even before anyone discovered the chemical makeup of water and twin water. It follows that the meaning of a word like “water” doesn’t depend just on what’s intrinsic to the speaker but also on the speaker’s environment.
One way to think about this is that for both Hypatia and Twin Hypatia, “water” picks out whatever plays the water role in their environment: roughly, it picks out whatever is the clear liquid found in the oceans and lakes that people drink and bathe in. For Hypatia, H2O plays that role, so “water” refers to H2O. For Twin Hypatia, XYZ plays that role, so “water” refers to XYZ.
You can construct Twin Earth cases like this for all sorts of words. There’s a Twin Earth with no trees but with non-DNA-based counterparts of trees. When my twin says “tree,” he refers not to trees but to these counterparts that play the tree role on Twin Earth. There’s a Twin Earth with a robot counterpart of Obama who plays the Obama role on Twin Earth. When my twin says “Obama,” he refers to this counterpart and not to Obama himself. And so on.
For all these words, it looks like their meaning is not in the head. We might call these externalist words: their meaning is anchored to certain things in their environment. Hypatia’s word “water” is anchored to H2O, and Twin Hypatia’s is anchored to XYZ.
There are some limits to externalism. One limit arises from logic and mathematics. When my twin on Twin Earth says “seven,” he will refer to the number 7. So perhaps the meaning of “seven” is “in the head.” The same might go for a term like “and.” Logical and mathematical words like these don’t need to be anchored in the environment. They might be seen as internalist words.
Externalism also doesn’t seem so apt for words like “consciousness,” “causation,” and “computer.” These words aren’t anchored in specific things in my environment. I have a general conception of what sort of thing a computer is, in broadly structural terms. Anything I count as a computer will also count as a computer for my twin on Twin Earth. Even if computers on Twin Earth are made of graphene where Earth computers are made of silicon, both count as computers all the same. As a result, my twin plausibly means the same thing by “computer” as I do. That suggests that “computer” is an internalist word.
In my own work on this topic, I’ve argued for a two-dimensional view of meaning, with both internal and external aspects of meaning. It is roughly as if Frege and Russell were right about the internal dimension of meaning, and Kripke and Putnam were right about the external dimension. For the purposes of this chapter, though, we mainly need the external dimension. What matters is that we can construct Twin Earth cases for many ordinary words. Putnam and Kripke have convinced most philosophers that at least this much externalism about meaning is correct.
Twin Earth and Sim Earth
Putnam’s Twin Earth provides a great model for thinking about language use inside and outside a simulation. Putnam himself used it to think about the brain-in-a-vat scenario, as I’ll discuss later in this chapter. Here I’ll use it to think about simulations.
Here’s the idea. Suppose there’s an original unsimulated Earth as well as a cosmic simulation containing a simulated Earth. Language use on Earth and Sim Earth is a lot like language use on Earth and Twin Earth.
For someone who has grown up on Earth, the word “hurricane” refers to unsimulated hurricanes: storms made up of giant, fast-moving patterns of air and water, which themselves are made of atoms at a deeper level. This fits our intuitive sense of the meaning of “hurricane.” It also fits with the causal theory of reference. When our community uses the word “hurricane,” this is triggered by unsimulated hurricanes in our environment.
For someone who has grown up on Sim Earth, the word “hurricane” refers to virtual hurricanes: simulated storms made up of patterns of simulated air and water. Virtual hurricanes have played the hurricane role in Sim Earth all along. Following the virtual digitalism I argued for earlier, at a deeper level virtual hurricanes are made of bits. The causal theory of reference helps explain how the word “hurricane” works: When members of a simulated community use the word, it’s triggered by virtual hurricanes in their virtual environment.
The same goes for “water”: It refers to water on Earth and to virtual water on Sim Earth. H2O (a chemical kind) plays the water role on Earth. Virtual water (a digital kind) plays the water role on Sim Earth. The same goes for wetness: It refers to wetness on Earth and to virtual wetness on Sim Earth. And so on.
Now we can analyze Dennett’s objection that simulated hurricanes don’t make you wet. If we’re on Earth, hurricanes certainly make you wet. For Earthlings, the words “hurricane,” “water,” and “wet” refer to nondigital things. Simulated hurricanes contain only virtual water, which doesn’t make anything wet—although it makes virtual things virtually wet!
If we’re on Sim Earth, then “hurricane,” “water,” and “wetness” refer to digital entities: entities that people on Earth call virtual hurricanes, virtual water, and virtual wetness. If people on Sim Earth say, “A simulated hurricane doesn’t make you wet,” they’re saying that a virtual hurricane doesn’t make anyone virtually wet. That’s a false statement. Virtual hurricanes make them virtually wet. If we are inside a simulation, then our hurricanes are simulated hurricanes, and they make us wet.
You might object that a creature in a simulation would have many false beliefs. For example, a sim might think, “I’m in New York,” when in fact the simulation is running on a server in Silicon Valley. Is the sim’s belief false this time? No! When the sim says, “New York,” the name doesn’t refer to the unsimulated New York on Earth. It refers to a place on Sim Earth: Sim New York. The sim is indeed in Sim New York. Or at least he’s virtually in Sim New York while he’s physically in Silicon Valley. And when a sim says “in,” that word means “virtually in,” which means that the sim’s virtual body is in the virtual location. So when the sim thinks “I am in New York,” this means that the sim is virtually in Sim New York, which is true.
Traveling between Earth and Sim Earth
What happens to language when people move back and forth between simulated and nonsimulated environments? Much depends on whether people move knowingly or unknowingly. If they know their environment has changed, the meanings of their words may instantly change. If they don’t know, the meanings may change more slowly.
Let’s see how it works for Twin Earth, starting with unknowing subjects. Suppose astronauts from Earth land their capsule in the ocean on Twin Earth. They have no idea that the ocean is made of XYZ. They say, “Hey! There’s water here!” Are they right or wrong? Putnam thought that the astronauts would be wrong. Their word “water” refers to H2O, and there’s no H2O on Twin Earth. The word “water” hasn’t suddenly changed its meaning just because they’ve encountered some XYZ.
By analogy: Let’s suppose that some unsimulated people from Earth unknowingly enter the Sim Earth simulation. Perhaps they’d been traveling in Africa, on safari with a tour group, and the company running the tour decided it could save money by putting the tourists in Sim Earth’s Africa instead. The tourists don’t catch on. When they see a herd of giraffes, they say, “Hey! There are giraffes over there!” Are they right or wrong? Following the astronauts’ case, we’d have to say they’re wrong. Their word “giraffe” refers to biological giraffes, but what they’re seeing are digital giraffes.
Now suppose the tourists stay in Sim Earth for years. They still don’t know it’s digital (the tour company is monstrously unethical), but they like new places, and simulation technology has gotten so good they can’t tell they’re not on Earth anymore. At some point, they’ll have encountered more virtual giraffes than the biological ones they may have seen in Earth’s zoos. By this time, the causal theory of reference suggests that the word “giraffe,” for them, will come to include virtual giraffes as at least part of its meaning. In effect, that meaning will slowly shift to include digital giraffes, and when they say, “There are giraffes here,” they’ll be right.
There are trickier cases. For example, what if someone from Sim Earth escapes the simulation without realizing it, and encounters a bio-tree for the first time. Is she right when she says, “There’s a tree”? Intuitions differ here. If digital trees are based on bio-trees and not vice versa, there’s a case that bio-trees are part of what caused her use of the word “trees” in the first place. Still, I’m most inclined to think that her “tree” refers to digital trees (which she has interacted with directly) and not bio-trees. So she’s wrong.
