Do we live in a computer simulation?
Here’s a quick and naive thought. We would expect most computer simulations to be of pretty poor quality and limited in scope. If we are in a simulation, the simulation we are in is of extremely high quality and of great scope. That’s not what we would expect on the simulation hypothesis. So, probably, we don’t live in a computer simulation.
But the following argument is pretty convincing: 1. If materialism is true, then probably a computer simulation of a brain can think (since the best materialist theory of mind is functionalism). 2. If a computer simulation of a brain can think, then most thinkers live inside computer simulations.
So, the argument that we don’t live in a computer simulation gives us evidence against materialism.
9 comments:
Interesting thought. Of course, the fact that we're not in a computer simulation wouldn't be evidence against materialism if the best form of dualism is (what we might call) "weakly functionalist," where weak functionalism says that consciousness supervenes with nomological necessity on functional properties. (This is the sort of dualism Chalmers favors.) A bit more specifically, if the epistemic probability of weak functionalism conditional on dualism is the same as the epistemic probability of functionalism conditional on materialism, then the evidence that we're not in a simulation wouldn't confirm dualism. And it's not clear to me why a biological (rather than functional) supervenience base would be more plausible for one who accepts mere nomological supervenience (the typical dualist) than for one who accepts metaphysical supervenience (the materialist).
Another wrinkle here is that functionalist views, as standardly formulated, don't really entail that a simulated brain in a simulated environment would be conscious. Indeed, they typically entail that a simulated brain in a simulated environment *wouldn't* be conscious, or at any rate wouldn't have the same conscious experiences as us. Functionalist views identify mental properties with properties of the form "having some property or other that plays role R," where R involves causal relations to other (intrinsically unspecified) internal states and *specified* physical inputs (e.g. pin pricks, physical pressure, ...) and specified physical outputs (wincing, groaning, moving away from the stimulus, etc.). But (plausibly) none of those physical inputs or outputs exist in a simulation. (At best we'd just have simulations of groans, winces, etc.)
Your second premise (if a computer simulation of a brain can think, then most thinkers live inside computer simulations) could be false for lots of reasons, or some combination of reasons. For example:
- If dolphins were rational, they probably would never invent a significant amount of technology, and therefore no simulations. Most viable planets may be ocean worlds, and therefore most intelligent life might be like rational dolphins.
- A brain might simply be too complicated to realistically simulate, even if it is possible in principle.
- Or even if it is possible, it might simply remain permanently too expensive to be worth it, especially since it is much easier just to have kids.
- Most intelligent beings might refuse to simulate brains due to moral reasons. Many people find the Argument from Evil persuasive or at least a significant challenge to theism, and if it is even a significant challenge, then there are substantial moral reasons not to simulate brains in situations similar to the ones that actually obtain.
- Even if there many brain simulations they might overall be fewer than ordinary thinkers for all of the above reasons.
1. If materialism is true, then probably a computer simulation of a brain can think (since the best materialist theory of mind is functionalism).
If a computer simulation can think, wouldn't this mean that this particular thinking computer simulation is of extreme high quality and of great scope?
2. If a computer simulation of a brain can think, then most thinkers live inside computer simulations.
If we would expect most computer simulations to be of pretty poor quality and limited in scope, I don't think (2) follows,because a computer simulation of poor quality and limited scope is probabaly not able to think. That "privilege" seems to be reserved for the relatively rare computer simulations that are of extreme high quality and scope.
So, I think we could rephrase the argument as follows.
1'. If materialism is true, then probably a computer simulation of high quality and scope of a brain can think (since the best materialist theory of mind is functionalism).
2'. Computer simulations of high quality and scope are extremely rare.
3'. If a computer simulation of high quality and scope of a brain can think, then most thinkers probably don't live live inside computer simulations.
So, the argument that we don't live inside computer simulations doesn't give us evidence against materialism.
Brian:
True about content. But it seems very, very plausible that people could think even if their whole lives are in the Matrix. And I mean, literally like in the Matrix, where their bodies are still functioning, but they have simulated sensory inputs. The best functionalism predicts that the content of most of their non-abstract thoughts would be very different from the content of similar thoughts in us--thus, those of their beliefs that are functionally isomorphic to our veridical beliefs about eating ice cream would be made true by facts about memory states of the computer.
As for weak dualism, that also depends on the probability of theism. Theism + weak dualism might NOT yield the prediction that we're in a simulation. For there is some sort of very serious deceit going on in the simulation (though it is hard to formulate exactly how that deceit works, given that on functionalism, and presumably weak dualism, many of the beliefs of the simulated beings would be true), and God would have reason to prevent or limit that kind of deceit, either by preventing such simulations or by miraculously exempting them from the (alleged) law of nature that makes the mental events nomologically supervene. Theism + functionalism, on reflection, has the same issues. But (a) functionalism is much less plausible given theism (typical functionalist theories involve being in the fray of causal-back-and-forth, and God isn't; moreover, full-blown dualism is more plausible if God exists since then it's guaranteed there is at least one fully immaterial mind), and (b) the functionalist scenario would require God to prevent simulations--he couldn't exempt them from a metaphysically necessary connection between functioning and mental events.
In any case, we can agree the argument is evidence for stronger versions of dualism. :-)
Walter (if I may):
That's a good point. But I think it unlikely that a simulation has to be of very high quality to yield thought, given the most plausible functionalist views. Heavily damaged human brains still manage to think, just not as well. The fine detail of the neural processes does not need to be modeled to get a functional isomorphism. And there is no need to get much simulation quality for stuff outside the brain.
Dr Pruss (if I may): If a simulations doesn't need to be of very high quality to yield thought and there is no need to get much simulation quality for stuff outside the brain, then how does it follow that if we are in a simulation, the simulation we are in is of extremely high quality and of great scope? I mean, thinking doesn't require a very high quality and neither does the stuff outside the brain, so this seems to contradict your argument against us being in a computer simulation.
The argument is that given functionalism, we would expect to find ourselves in a low quality simulation. That does not match empirical observation. So functionalism is probably false.
Professor Pruss, a somehow related question:
All thomistic natural philosophy seems to "trust" one thing: that material beings exist continuously and that we can continuously behold them. But is that certain or evident? I have two objections:
1) God can successively create and annihilate the universe in order to create an illusion of continuity. The universe would exist in the manner of a film. Thus, material beings wouldn't exist continuously.
2) I happen to be writing on my cell phone in my room right now. Nevertheless, there is the possibility that God created 5 rooms in different points of the universe that look exactly like mine. God could be successively moving me from room to room in a very fast way in order to create the illusion that I exist continuously in this very same room. Thus, I wouldn't be beholding my room continuously. I would be seeing different rooms successively.
How does the natural philosopher know with certainty that the material beings he sees exist continuously and that he is seeing continuously the same beings? How does he know that he is witnessing continuity and not succession?
Why does one need certainty here?
Simulation arguments also support the idea that the universe is much less than 14 billion years old and is not nearly as vast as it appears. For in our simulations (video games and tabletop games and whatnot), the simulation will start at a certain arbitrary point in time, although the world it simulates will in some sense go back further. The same goes for the size of the world, mutatis mutandis. I don't know if this should significantly affect the plausibility of simulation arguments, but it's interesting.
Post a Comment