Still Not Living in a Computer Simulation

A couple of commentators on a previous post pointed me to an Arc Digital article by Thomas Metcalf which contends that the Simulation Argument (SA) ought to be taken more seriously. (Metcalf’s article wasn’t written in response to mine, although it appeared a week or so afterward: post hoc sed non propter hoc.)

I don’t think there’s anything in the article that poses any problems for the arguments I gave in my post. Rather than respond to every point, I’ll just quote a few sections and make some comments.

Brain in a Vat

Metcalf observes that the SA has two key premises:

The Empirical Premise: Most of the “people” who think they’re real, flesh-and-blood humans are actually conscious computer programs.

The Indifference Premise: If most people are simulated, then you are probably simulated.

As I explained in my earlier post, I think the first premise is false, and necessarily so. It’s not metaphysically possible for a computer program to be conscious, assuming that the computer in question is a purely physical mechanism. (I think Metcalf actually commits a category error in his statement of the Empirical Premise. A computer program is abstract in nature; it’s a set of instructions that can be run by one or more physical computers. So a computer program wouldn’t be conscious; it would be the computer running the program, if it were possible for a computer to be conscious.)

The second premise looks problematic too. Metcalf elaborates:

The idea behind the Indifference Premise is simple: If most people have some feature, then absent other evidence, you should guess that you probably have that feature.

Most people have the following feature: not being me. Should I therefore guess that I’m probably not me too? Perhaps the “absent other evidence” clause is supposed to foreclose such trivial counterexamples. But what kind of evidence is in view here? Observational evidence? Surely that’s not the kind of evidence that would confirm my self-identity. Self-identity is known a priori. Couldn’t I also know a priori that I’m not a computer simulation? Well, if I can know a priori that no purely material object can be conscious, then I can know a priori that I’m not a computer simulation. All this to say, both premises of Metcalf’s SA seem to hang on the controversial notion that a computer can be conscious.

Metcalf later considers “some better objections” to the SA, the first of which is this:

1. “Is it even possible for a computer program to be conscious?”

Some experts say yes; others say no. Imagine a single atom of silicon or gold. How could that tiny object be conscious? Now make a big pile of such atoms. How could a big pile of unconscious silicon and metal (i.e., a computer) be conscious?

This objection may go too far. Compare: “Imagine a single atom of carbon, hydrogen, oxygen, or nitrogen. Surely that tiny object is unconscious. Now how could a big pile of such unconscious things be conscious?” Of course, another name for a certain pile of carbon, hydrogen, oxygen, and nitrogen is “a human brain.” Yet we probably agree that many human brains are conscious.

The fact that philosophers disagree over whether a computer program (or a computer running a program) can be conscious shouldn’t move us one way or the other. The question is, which side has the better arguments?

Metcalf’s brief defense of the ‘yes’ side is question begging. I don’t agree that “many human brains are conscious,” precisely because I don’t think that any physical object is conscious. I think the human mind is conscious (or perhaps better: the human person has a conscious mind) and the mind is not the same thing as the brain. That’s why I’m a dualist about the mind-body issue. Any dualist is simply going to deny that “many human brains are conscious.” Metcalf doesn’t give us any non-question-begging reason to think that a computer can be conscious. (There are some serious arguments for physicalism about the mind; I commented on a common one here.)

2. “The simulation hypothesis is more complex than the ‘real world’ hypothesis, because it requires the existence of a real world plus the existence of simulators. And Ockham’s Razor tells us to prefer simpler hypotheses.”

This objection depends on the assumption that more-complex hypotheses are less likely to be true. That’s debatable. I’ve criticized a version of it.

I haven’t read Metcalf’s paper, but based on the abstract it looks like the “common principle of parsimony” he criticizes in that paper is not the principle of parsimony at work in the objection here. The principle he criticizes has to do with ‘ontologies’, i.e., theories about what types of entities there are. But the simulation hypothesis isn’t that kind of theory. It’s a theory about (1) what kind of beings we are, (2) what is causing our conscious experiences, and (3) whether our experiences are veridical. And clearly there’s a simpler theory that can equally well explain the data in question: the theory that we are not living in a computer simulation and the world we experience just is the real world.

3. “It doesn’t feel as if we’re in a simulation. The world around me seems real.”

Yes; it’s just very implausible that we are living in a simulation right now. The people around us seem conscious and human. Most of us have the strong intuition that we live in the real world, maybe an intuition even stronger than the intuition that the Empirical Premise or that the Indifference Premise is true.

This objection depends on trusting such intuitions. Many experts are skeptical. Still, a case can be made that denying the reliability of intuitions is a big mistake. It might leave us in a position of global skepticism: the position that we aren’t justified in believing anything. If so, then we’d also have to be skeptical of the beliefs that led us to take the Simulation Argument seriously in the first place.

Exactly right. In fact, that’s quite close to my argument in the original post that “the simulation hypothesis has a deeply self-defeating character to it.”

Finally, on the implications of the simulation hypothesis for how we should live our lives:

In turn, you should want to be an entertaining, likable, and morally good person. Of course, whether you think we’re in a simulation or not, that’s good advice.

Well, at least we can agree on that!

3 Responses to Still Not Living in a Computer Simulation

  1. Jonathan

    “In turn, you should want to be an entertaining, likable, and morally good person. Of course, whether you think we’re in a simulation or not, that’s good advice.”

    On the original article I pushed back on this some, but it occurs to me now that there might be a further issue.

    If one thinks the SA is good, then they should think the Simulation within a Simulation with a Simulation Argument is good:

    The Empirical Premise: Most of the “people” who think they’re real, flesh-and-blood humans are actually conscious computer programs within conscious computer programs within conscious computer programs.

    (1) Conscious computer programs will likely have access to the computing power necessary to run conscious computer programs within conscious computer programs who believe they’re real humans, and who populate simulated environments that appear real

    (2) These conscious computer programs within conscious computer programs will actually want to run at least two instances of such programs, either for historical, recreational, or scientific purposes…

    etc.

    It seems impossible to extrapolate what the aims and desires of a conscious computer program running a conscious computer program within a conscious computer program might be. Thus, how we should behave in order to entertain or endear ourselves to the Ur Conscious Computer also seems inscrutable (and, per (1), reason to think there wouldn’t be only one).

    • Yes, maybe it’s computer programs all the way down!

      Seriously, that’s a good observation. If the SA works for us, then it (or an analogous version) ought to work for the people who created the simulation we inhabit, and so on.

  2. Pingback: End of May 2018 Presuppositional Apologetics’ Links | The Domain for Truth