Wednesday, May 29, 2013

Unexperienced Virtual Realities

I've long thought that a shared virtual world could, in principle, be just as "real" (in any sense that matters) as the material world.  But Helen recently got me thinking about what we metaphysical anti-chauvinists should say about "virtual worlds" that, though capable of supporting interactions between conscious minds, never actually do support any such experiences or interactions.  (Many of the following thoughts I owe to her.)

On the one hand, it seems weird to say that every simulation (assuming the information it contains could, in principle, be "experienced" in some way, however simple and low-resolution the resulting "perceptions" would be) is thereby a new reality in its own right.  That would seem to stretch the concept beyond all recognition.  On the other hand, however, it doesn't seem that the metaphysical status of a world as "real" or not should depend on its happening to actually support conscious experiences.  The material world could have existed without ever giving rise to sentient beings, so why should the same not be said of a rich Matrix-style virtual world to which nobody ever plugs in?

In Section 6 of 'The Matrix as Metaphysics', Chalmers writes:
[G]iven an abstract computation that could underlie physical reality, it does not matter how the computation is implemented.... In particular, it is irrelevant whether or not these implementing processes were artificially created, and it is irrelevant whether they were intended as a simulation. What matters is the intrinsic nature of the processes, not their origin. And what matters about this intrinsic nature is simply that they are arranged in such a way to implement the right sort of computation. If so, the fact that the implementation originated as a simulation is irrelevant to whether it can constitute physical reality.

There is one further constraint on the implementing processes: they must be connected to our experiences in the right sort of way. That is when we have an experience of an object, the processes underlying the simulation of that object must be causally connected in the right sort of way to our experiences. If this is not the case, then there will be no reason to think that these computational processes underlie the physical processes that we perceive. If there is an isolated computer simulation to which nobody is connected in this way, we should say that it is simply a simulation.

But why should we say this?  The preceding sentence suggests that Chalmers is focused on the specific question of whether the processes constitute our reality.  But can't we also ask the more general question of whether they constitute a genuine reality at all (regardless of whether it is ours)?

At this point one might reasonably question what hangs on whether we call something a "reality" or not.  Is this merely a verbal dispute?  I think the best way to avoid this danger is (as usual) to focus on the potential normative implications.  Some people think that natural beauty (say) can have intrinsic value, even in a world with no conscious beings.  So, the question becomes, should such people also value any similar natural beauty that is to be found within a never-experienced virtual world?  Would it be worthwhile to create such worlds, just to increase the amount of natural beauty "out there"?

We might also consider less extreme versions, say where a world has been experienced in the past but never will be again.  Often we care about what will happen after our deaths.  Could an inhabitant of such a "dying world" reasonably care about what happens after the last conscious person leaves it?

In either case, I'm (tentatively) inclined to think that anyone who would say "yes" to the parallel question regarding a material world should likewise answer "yes" to the case involving a virtual world.  But it's a little tricky, because I'm dubious about attributing value to any states of affairs that lack conscious beings or agents.  (In which case, it would no longer matter whether an unexperienced world counts as "real" or not, because it lacks a crucial precondition for genuine value regardless.)

What do you think?


  1. I think you are right in speculating that this is a mostly verbal dispute. If we define "real" as "the objects one perceives are made of quarks" or something like that, then no virtual reality program is real. If we define "real" as "capable of satisfying all the desires a typical human would have if the human plugged into them" then a sufficiently advanced multiplayer virtual world is "real," even if no human ever actually plugs into it.

    As for your question about the value of states of affairs containing no conscious beings, I admit that I'm confused about that too. I think that, if I do value beautiful things that one ever perceives them, my values relating to conscious beings are lexically prior to such values. So I might value a lifeless planet full of beautiful things over a lifeless planet full of ugly things. But a planet with one conscious being living a life worth living upon it would be more valuable than a googolplux of planets full of unperceived beauty.

    So I suppose my answer is "no" to whether or not we should create beautiful unperceived natural worlds. As long as benefiting other creatures is an option instead, we should devote our resources to that.

  2. You ask whether the virtual world simulator should be regarded as a "new reality in its own right". However, the emphasis here should be on "new" rather than "reality", since of course the simulation is real. Everything that exists is real.

    We know that the virtual world is real, because it is composed of entities in the "real" world. That is, there really is a physical computer there, which really is executing a certain program. Surely this is enough to give the computer program the same normative status as any other inanimate process occuring in the "real world". Since it is an example of an inanimate process occuring in the real world!

    1. The simluation is real enough, but the simulated "world" might not be -- it might be merely simulated. A simulated tree is not generally thought to be a (real) tree, after all, and so a simulated scene of natural beauty may not actually be beautiful. (Suppose that the computational underpinnings are not particularly elegant, considered merely as a pattern of zeros and ones.)


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)