Tuesday, April 25, 2006

The Actual World is not a Possible World

As I understand the standard picture of modality accepted by contemporary philosophers, there's a space of possible worlds, and this - the actual world - is one of them. It has some special properties, being concrete or realized in a way that the other, merely possible worlds aren't. But this picture seems to run into trouble, for reasons that I've previously mentioned in passing.

In brief: the modal "multiverse" is seen as static and necessary. Contingent facts vary from world to world, but the worlds themselves remain constant, and hence world-indexed facts (e.g. "P is true at w") hold necessarily. But then, if it's a property of our world '@' that it is actual, then it seems that "@ is the actual world" comes out as a necessary truth. We're stuck with narrow fatalism. That's bad. We should be able to make sense of the idea that our world's actuality is a merely contingent fact, and that other possibilities could have been actualized instead. But to achieve this, we must deny that actuality is an intrinsic property of any possible world. That is: there's no red flag built into modal space to specify actuality.

That's not to deny the existence of the actual world, of course. I simply think we should deny that the actual world is a possible world. Instead, I think it is a fundamentally different kind of thing, existing quite separately from modal space. This allows it to have genuine contingency, by escaping the bounds of the static and necessary "modalverse".

We can motivate this idea from another direction too. Possible worlds are typically characterized as "ways a world might be" (see, e.g., Stalnaker). There's a possible world to represent each way, including the way the world actually is. There's a possible world representing that. But the representation is clearly not identical to the thing itself. There's a possible world representing "the way the world is", and then there is the actual world, that concrete thing which contains us - flesh and blood - and not mere abstract representations of "the way we are". As someone (van Inwagen?) noted, we wouldn't dream of identifying "Socrates" with "the way Socrates is". It doesn't even make grammatical sense. So why make the same mistake with the world?

Despite the misleading name, "possible worlds" aren't really worlds. (We're not Lewisian realists here.) They're just abstract representations. Maybe they're primitive entities, or maximal properties or states of affairs, or sets of sentences in an idealized language; the details don't much matter. In any case, the actual world is clearly a very different kind of thing. It really is a world -- a concrete thing, filled with other stuff, real entities, and not mere representations of those entities. Since possible worlds aren't really worlds at all, it follows that the actual world is not numerically identical to any possible world. Rather, it corresponds to that possible world which represents "the way things actually are". There's a representational relation between them, but representation is not identity. (Compare: a photograph doesn't really contain you - the concrete person - as a part. I could cut it up without thereby decapitating you.)

So, I see three major advantages to denying that the actual world is numerically identical to a possible world:

1) Consistency. All possible worlds remain on an ontological par. You don't have one special one made out of concrete stuff while all the others are abstract things. Instead, we can hold that all possible worlds are the same kind of thing, and the actual world is simply a different kind of thing.

2) Grammar. "Ways things are" are distinct from the things themselves.

3) Genuine contingency. We can accept that the "modalverse" is static whilst avoiding fatalism. The actual world is contingent, and can be so because it is outside and separate from the modalverse. No possible world has the property of actuality intrinsically. Instead, it is a relation that holds between the static possible world and the contingent actual world. This makes the relation itself contingent, of course. If the actual world had been different, then it would correspond to a different possible world. None of this requires any change in the modalverse itself. The changes all occur in the fundamentally separate space of concrete actuality.

Update: Ah, it turns out Kripke beat me to it.



  1. I think this is a very plausible view.

    I think I may not be catching something in your argument here, though:

    Contingent facts vary from world to world, but the worlds themselves remain constant, and hence world-indexed facts (e.g. "P is true at w") hold necessarily. But then, if it's a property of our world '@' that it is actual, then it seems that "@ is the actual world" comes out as a necessary truth.

    Surely from this it follows that what would come out as a necessary truth wouldn't be "@ is the actual world" but "@ is the actual world at @". Or to put it another way, "This world is the actual world, on the condition that what is actual is at this world." And that wouldn't be a problem. "@ is the actual world" can only be necessary if it is true at every possible world -- something we have no reason to believe.

  2. Seems too easy, isn't this stuff supposed to be hard? ;)

    Seriously, I like your approach here. But I feel still feel somewhat cowed by Lewis' critique of the "ersatz" worlds approach in ch. 3 of OPP. He demanded an account of how this abstract representation works. Even though the merits of moving from identifying the concrete world with one of the possible worlds to a representation relationship seems like an improvement for the reasons you give, the problem of how an abstract world (aro maximal state of affairs) can represent the world as it actually is (or how it might have been) was considered fatally problematic. Lewis seemed to take the representation issue to be equivalent to the posing of the problem of how one abstract possible world gets "actualized", which you have also discussed.

    Anyway, I'm OK with invoking abstract entities to help here, but just was recalling this Lewisian opposition when I read this post.

  3. Has anyone ever suggested that it might be possible that while not all possible worlds are actual more than one might be? A limited pluarlity of worlds?

  4. Timothy - I've discussed that before, but introduced a "that's all" clause into the specification of a possible world, to ensure that they are genuinely maximal compossibilities. (Though a plurality of spatiotemporal regions could still be found within a single possible world.)

    Steve - For now I'm taking it for granted that the standard ersatzist views can be made to work. But that is something I'll need to look into more.

    Brandon - this might relate to our previous discussion on the ambiguity of 'actually'. Here I use the term 'actual' in its absolute sense (to mean something like "concrete"), not the indexical sense. Note that if a world w1 has some intrinsic property P, then this fact is true no matter where you say it from. Even in other worlds w2, it's still true that w1 has the property P. The intrinsic properties of w1 don't change depending on where you look at it from -- no more than the capital of New Zealand changes depending on which country you ask about it from. (This idea draws heavily on the spatial metaphor of "modal space", of course. Perhaps you have qualms about that.) So if w1 is concrete, then this is true in every possible world, i.e. necessarily true.

  5. Thanks, Richard; that clarifies things.

  6. Hi Richard
    I agree with your argument entirely. I believe this distinction between the actual world and its corresponding possible world lends itself particularly well to the idea of possible worlds as maximally coherent sets of propositions. I think that in that case the "representation" relation that holds between the actual world an its corresponding possible world can be explained in terms of an aggregate of individual relations of the sort that hold between any true proposition and the world (outside of modal contexts). Im interested to know what kind of ersatz worlds you favor.

  7. Yeah, I've always intuitively assumed something like the proposition-based account, but I don't have any settled position here yet. That's a nice point you make about how well it fits with my distinction. (Though one worry is that many philosophers want to analyze propositions in terms of possible worlds, which is obviously precluded on your account.)

  8. Yeah that is a worry. Lycan thinks that this is a small price to pay if we consider the strong objections to alternative theories, and I would lean that way at this stage, but maybe further reading on this subject will change my mind (for example reading someone other that Lycan!)

  9. Maybe I've misunderstood, but it seems to me either one is a modal realist, so must take the idea of the idexicality of 'actual' seriously so that it isn't an intrinsic property, and so you argument fails; or, we aren't moral realists in which case of course the actual world is something different - it isn't a fiction or an abtract object.

  10. Sure, ersatzers all agree the actual world is different in that it is concrete while other worlds aren't. Yet most (at least in my experience) still talk as though the actual world is one among many possible worlds, albeit a special one. They hold that it just is (numerically identical to) the possible world that is actual. I, by contrast, am arguing that the relation here is one of correspondance, not identity. Hence my claim that "The Actual World is not a Possible World".

    (While it would be nice to say that "of course" I'm correct here, as I understand it there's a long tradition of ersatzers who would disagree, as they take the actual world to be itself a possible world, albeit not an "abstract" one like all the others. In general, philosophers tend to conflate actuality with the possible world @ which represents it. See, for example, Soames' claim that we can demonstratively identify our possible world @.)

  11. Richard,
    You claim above that "if a world w1 has some intrinsic property P, then this fact is true no matter where you say it from."
    That seems right, but I wonder why you restrict the claim to intrinsic properties. Isn't it true that if a world, w1, stands in some relation (an accessibility relation, say) to some other world, w2, then it is true that w1 stands in that relation to w2 no matter where you say it from?
    Also, while I haven't read the paper from which you extract the Soames quote, it might be that he had in mind only demonstrating a certain way that the world is. If you think that possible worlds are ways for the world to be, then one possible world is the way the actual world (the concrete thing) in fact is. And, one way to pick out that world is by demonstrating that way by flapping ones arms around and gesticulating.

  12. "I wonder why you restrict the claim to intrinsic properties."

    Fair point, the same holds for all properties that are internal to the "modalverse" as a whole, i.e. including relations between worlds. What I want to exclude is the relation between a possible world and actuality. Because the latter is merely contingent, the fact that (say) w42 corresponds to actuality is a merely contingent fact. (I should add that this modal property cannot be analyzed in the usual 'true at some possible worlds' way, because that only works for properties that are wholly internal to the modalverse. This allows us to get around the fatalistic problem that there doesn't seem to be any place from which the relation fails to hold.)

    Can we point to the way the world is, as opposed to the concrete world itself? That's not entirely clear to me.

  13. Richard,
    I think you've anticipated my worry, but let me briefly state it anyway.
    Let some version of ersatzism be correct. One and only one ersatz world, w1, stands in some relation (the actualization relation) to the concrete actual world. So, it is true in every possible world that w1 stands in the actualization relation to the actual world. So, it is a necessary truth that it does so. So, we get some kind of fatalism.
    I take it your response is to deny that it is true in each world that w1 stands in the actualization relation to the actual world. I'm not sure why you get to deny this. I take it that it has something to do with this particular relation not being "internal to the modalverse". That is, we need to, and get to, exclude certain properties from inclusion amongst the modal facts. We don't want to let facts about the actualization relation to get in, because that results in fatalism. But, why can't the proponent of the view that the actual world is a possible world say something similar? In particular, she can say that what is true at any possible world, w, that the actual world represents itself as being concrete. But, it is not true at w that the actual world is concrete. I take it that this is the right result. Every world represents itself as being concrete. That preserves contingency. To get the result we just have to exclude a certain intrinsic fact about the actual world from being true at every other world. But, the ersatzer has to do something similar.
    On a different note, we demonstrate ways a thing can be pretty frequently, it seems to me. Suppose I hire a maid to clean my apartment. The maid does a terrible job, so I call her supervisor and ask her to come look. I say to the supervisor "Look at the way your employee left my apartment" (waving my arm around and pointing). I take it I've demonstrated a certain way that my apartment is. I don't see any reason to suppose we can't do something similar with a total way the world is. The only significant difference, it seems to me, is the "magnitude" of the way. But, if you think that we can demonstrate a certain very large concrete thing, the actual world itself, then I don't see why we should suppose that we cannot demonstrate a way that it is.

  14. Hi Will, nice to hear from you! I like your new blog...

    Anonymous - if every world represents itself as being concrete in just the same way, then what sets ours apart as special? What makes the representation accurate? I say it must correspond to the extra-modal reality. So we need to go beyond just the space of possible worlds. What we're in is quite a different kind of thing (and shouldn't be confused with the representation thereof).

  15. Richard,
    Our world is different in this respect - it alone is concrete. Like all worlds, it represents itself as being concrete, but it and it alone does so accurately.
    Of course, we don't want it to be true at all of those other possible worlds that the actual world is the only world that accurately represents the actual world. So, despite the fact that it is an intrinsic property of the actual world that it alone represents itself accurately, it is not true at other possible worlds that the actual world alone represents itself accurately. From the perspective of every other possible world, it accurately represents the actual world. The pure ersatzer (the philosopher who thinks that all worlds are ersatz world) has to say something similar. She has to say that, despite the fact that only one world, w1, in fact stands in the actualization relation to the actual world, it is not true that, from the perspective of every possible world, w1 alone stands in the actualization relation to the actual world. From the perspective of every other possible world, w, w stands in the actualization relation to the actual world. That is, every other possible world mis-represents.
    Am I missing the problem?

  16. Oh, I see! Nice point. So your mixed ersatzer can (claim to) escape the narrow fatalist argument. Though it sounds a little more ad hoc, at least to my ear. That is, I think the 'pure' ersatzist picture gives a better reason for holding the 'actualization' relation to be contingent and not accessible from other worlds.

    My main concern is that if you build actuality right into the metaphysics of the modalverse, then the necessary and static nature of this entity would seem to rob it of any contingency. I mean, you can say that "according to w2" our world w1 is not actual. But if the fact is that the modalverse couldn't have been other than what it is, and that each possible world has its constitution necessarily, then fatalism would still follow. Our possible worlds talk would fail to track these modal truths, but that wouldn't make them go away. w1 is what it is, and there's no real sense in which it could have been something else, say made of abstract rather than concrete stuff. (By contrast, I get to sneak in genuine contingency by appealing to an extra-modal concrete realm. Any "changes" in modal space are mere 'Cambridge changes' to the relations between the static possible worlds and the dynamic extra-modal concrete world. The real change all occurs outside of modal space.)

    Here's another concern (and it can be reinforced by the other arguments presented in my main post): while I can plausibly hold that all possible worlds ('ways a world could be') purport to represent the one extra-modal reality, it isn't so clear on your picture that they're each trying to represent one of themselves. And which one? A world of dragons surely doesn't purport to represent our dragonless world @ (the world that is actual). Perhaps they aim to represent whatever world turns out to be concrete. In that case they're right about it being concrete, but mistaken about it containing dragons. Or perhaps they simply aim to represent themselves -- in which case they're plausibly right about everything except the concreteness? It all seems a little muddled (though perhaps that's just me).

  17. Richard,
    I certainly agree that the mixed ersatzer has a strange view. I also haven't any idea what would motivate someone to adopt it. I find your first and second advantages for rejecting mixed ersatzism reasonable (the first more so than the second). But, it does seem false to me that mixed ersatzism entails some kind of fatalism.
    Regarding your main concern; its an odd complaint from someone with sympathy for possible world semantics. After, possible world semantics are supposed to give us the meanings of the modal operators and modal sentences. How could it be, then, that our "possible worlds talk" failed to track the modal truths? There is a real sense in which w1 could have been non-actual; it is non-actual according to (or at) some possible world. That's just want it is for it to be the case that it could have been that p. I'm not sure what else could be wanted here.
    As to your second concern; why can't the mixed ersatzer say that every possible world represents (some more accurately than others; the actual world itself being a limiting case of accuracy) the actual world? A world at which there are dragons does represent the actual world, it just does so less accurately than the actual world itself. Of course, if it is problematic we could get rid of "representation" talk altogether. A Plantinga type ersatzist doesn't have to, I think, talk about "representation". Instead we would say that w1 could have been non-actual because there is a possible world which properly includes the state of affairs w1-not-being-actual.

  18. You absolutely lost me in the jargon and citations that I am wholly unfamiliar with, but will that stop me from making a uneducated comment? Hell no! I'm an American! Anyway, I've always viewed the worlds as a never eneding amalgamation of possibilities ended and created by choice. I'm a very visual person so a lot of what I think doesn't convey well to others. It's as if there are thousands of strings (I know it's a horrible analogy,but press on) coming from the point you occupy. With every choice you make, a string is terminated or a number of other strings branch from the first. As I stated in a previous comment, I have done quasi-law enforcement work for years. One of my duties for eight years was to perform building searches during burglar alarms. I was nick named Kevorkian Kommando. When I would find a subject in the building, I could feel the focus of my life narrow. As the confrontation progressed, the more choices were eliminated. At the final conclusion, there were only two choices to be made, and that was to be decided by the perpetrator. They could either flee, in which case I would pursue them or they could bring it all into the "physical realm", and I would defend myself. At these points, I was the calmest I have ever felt in my life. I had no choices to make anymore, the chaos of my life became calm until another person's free will chose the course to be taken. Sounds kind of psychotic now that I'm sitting here reading it. Oh well. Poor Man's Philosophy at it's best.
    The Writhing Of Something Nailed Down In Torment

  19. Ditto to this being out of my field, but when you say "The Actual World is not a Possible World", it seems to me to turn the idea of possible worlds on it's head. Any actual world would need to be a possible world, otherwise it is an impossible world. And there's nothing impossible about the actual world. Quite the contrary, it's one of the most likely worlds I've ever heard of.

    It seems to me you're just caught up on actuality being a predicate that applies to only one world, which seems to say that the world could not be any other way. To me the escape from your feared conclusion (and fearing fatalism is not a sufficient reason alone to deny it), is to say "This Actual world is not the only Possible Actual World". And that seems quite true to me. It is merely our lack of ability to actually experience any other possible world that makes us feel our one is more 'Actual'. It is quite possible that there are actually other actual worlds, selected from the domain of possible worlds.

    Unless you *equate/define* 'actual' to 'we can experience it', in which case the feared statement is that 'there is only one possible world we can experience'. Again, I see no proof at all for that claim. We, ourselves, are merely objects of the actual world, and we might be actual objects in other possible actual worlds. In which case we would be actually experiencing them too. You could further refine your fear by saying 'We' means only the instance of us in this possible actual world we are experiencing. In which case you are really stating a tautology: "The only possible world we can experience is this one we can experience, which we will label the 'actual' world".

    Amusing though. Modal logic is so intricate despite being so utterly useless. Lots of squares and triangles, as I remember.

  20. Yeah, I mean the title in a less paradoxical way than it initially sounds. I certainly don't mean to imply that this is an impossible world. Rather, I'm suggesting that the word "world" is ambiguous, and that actuality isn't really any kind of "world" (in the possible/impossible world sense) at all. I later found that Kripke has said something similar. The core idea is that we should distinguish possible world-states (they might better be called "possible histories") from the giant physical object that we inhabit. The actual world-state (our actual history) is certainly among the possible such states. But it shouldn't be conflated with the physical universe (what I call the "actual world") itself. There's the way the world is, and then there's the world itself, and these are two different things.

    My worry with saying "This Actual world is not the only Possible Actual World" is that this might be interpreted (according to possible worlds semantics) as saying that ours is not the only actual world in modal space. That's okay if you're a Lewisian realist, who thinks that actuality is merely indexical, and that objectively speaking all possible worlds are on an ontological par, existing in exactly the same way. But not many people want to hold such a view, and I think we should want to at least be able to make sense of the idea that there's only one actual world in modal space, and that it's a contingent matter which one it is.

  21. Hi Richard. I just stumbled over this new article, which might be of interest to you:


    It is a short paper with the title `Propositions, Sets, and Worlds', and it poses some interesting comments on `the actual world' and ersatzism.

  22. Van Inwagen in "Two concepts of possible worlds" claims that you cannont say that our world is actual, or that you and me are actual. Propositions (or states of affairs, or what have you) are actual.


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.)

Note: only a member of this blog may post a comment.