Modal Realism

Brief Summary of the View

Davd Lewis’ Modal Realism is misleadingly named. A better name for his view would be ‘Possible Worlds Realism.’ This is because while many people are realists about modality, or modal facts, not many people are realists about the entities that the modal facts are analyzed in terms of—viz., possible worlds. Here’s why: it has been widely agreed that it is extremely theoretically useful to analyze modal talk in terms of possible worlds as follows: possibly p iff in some possible world, p; necessarily p iff in all possible worlds, p. Further, without such a reduction, the notions of ‘possibly’ and ‘necessarily’ would have to be taken as primitive, which is seen as an ontological mystery (and, thus, a theoretical burden). However, if the possible worlds analysis of modal terms is adopted, we now have a new problem: we are seemingly quantifying over possible worlds and, if we are good Quineans, we are now committed to having such worlds in our ontology. Moreover, we are now pressed to say just what these possible worlds are.

Lewis boldly claims that possible worlds are of the same ontological kind as the actual world—where the actual world, he claims, is intuitively anything that is spatio-temporally and causally connected with us, here. More technically, the actual world—like other possible worlds—is a maximal, consistent set of spatio-temporally and causally connected stuff. Anything that is any spatial or temporal distance from us is part of our world; anything that’s not, isn’t. Other possible worlds, Leiws claims, are just like ours. They are maximal, consistent sets of spatio-temporally and causally connected stuff. Anything that is any spatial or temporal distance from (world-bound) people in other possible worlds is part of their world; anything that’s not, isn’t. Individuals in these possible worlds—including our own—are world-bound; they cannot move from one world to the next. By definition, any world or place they could ‘get to’ would be somewhere in their own world.

Lewis thinks that the main reason to be a modal realist is because doing so is so theoretically beneficial. In other words, if analyzing modal terms in terms of possible worlds is so useful, and such an analysis honestly requires quantification over possible worlds, then (as any good Quinean knows) this is enough to posit possible worlds in our ontology. Put yet another way, Lewis is committed to the following general, Quinean principle: if our best theory about the world quantifies over certain entities, then we are committed to having such entities in our ontology. Our best theory of modality, Lewis claims, is committed to possible worlds, so we mustn’t deny that there really are such things.

Lewis parallels this move about ontological commitment to the debates in the ontology in mathematics. At least on the surface, mathematics quantifies over abstract entities such as sets, numbers, functions, etc. However, even though having such entities in our ontology might be ontologically costly, doing so provides us with so much theoretical benefits that the price seems worth it. Put more generally, there is always a trade-off between theoretical benefits and ontological costs. If the benefits received outweigh the costs, then it is seen as a worthwhile trade. Lewis thinks that not only is the trade worthwhile when it comes to modal realism, but what’s more, no other theory about modality will fare near as well. [More on this last point below.] 

To help massage our (likely) resistance to his view, he makes a distinction between qualitative and quantitative ontological costs. Since, according to Lewis, possible worlds are not ontologically different kinds of things, his modal realism is only demanding a quantitative ontological cost. He claims that his view simply posits more of what we already think there is. All (or most) of us are already committed to the actual world, no matter what one’s view of modality or possible worlds. So, in committing oneself to modal realism, Lewis thinks, you are simply committing yourself to more of what is already in your ontology. Instead of one world, there’s many. But it’s not as if the many worlds are anything spooky or new or are a kind of entity that we should be suspicious of, on ontological grounds. Modal Realism, then, isn’t claiming that there are totally different kinds of things (like, say, dualism would if one is a materialist). Thus, Lewis thinks that (i) the theoretical benefits are worth the ontological costs of modal realism and (ii) that the ontological costs aren’t as great as they might initially seem anyway, since it is only a quantitative, not a qualitative cost.

Moreover, with modal realism, Lewis thinks, we get a nice way to reduce or analyze our somewhat mysterious, intensional modal talk, with non-mysterious, extensional talk of possible worlds. If we had difficulty understanding what it meant for something to be possible or necessary, in other words, possible worlds will help us, conceptually, to understand modal truths. We know what it is for someone to be lounging on the beach, for example. And we seemingly know what it would be for someone very much like me (in the relevant ways) to be lounging around on the beach. This is enough, then, Lewis argues, to understand what it is for it to be possible that I am lounging on a beach. For, so long as there is a possible world in which a counterpart of me (i.e., someone very much like me in relevant ways) is lounging around on a beach, this is what makes it true that it’s possible that I could be lounging around on a beach. This gets us into issues about his counterpart theory, which I will address below.

Further Details of the View

Counterpart Theory: One of the interesting features of Lewis’ modal realism is his adoption of counterpart theory. Lewis thinks that possible worlds are concrete, like the actual world, and that many of the individuals that live in these possible worlds are concrete as well, just as individuals in the actual world are. However, because individuals are world-bound—because individuals cannot travel or get to a different possible world than the one they are in—the relation between individuals in different possible worlds cannot be one of identity. For, if it were, then we would be able to have someone in more than one possible world.

But then, how does Lewis evaluate counterfactuals? Recall that for Kripke, to say that (1)--
(1) It is possible that I am on a beach in Tahiti tomorrow.

--is true, is to say that there is a possible world where I—me, myself—am on a beach in Tahiti tomorrow. For Lewis, this can’t be how counterfactuals work, since he claims that individuals are world-bound. We know ahead of time, in other words, that on his account, there is no possible world where I am on a beach in Tahiti tomorrow (given that in the actual world I will not be on a beach in Tahiti tomorrow). Because the only world where I am anywhere, on his view, is this one, the actual one. So Lewis suggests that what makes a counterfactual like (1) true, is not that there is a possible world where I am on a beach in Tahiti tomorrow, but rather that there is a possible world where someone very much like me—and very much like me in the relevant sorts of ways—is on a beach on Tahiti (or, more carefully, on a beach very much like Tahiti). Individuals in different possible worlds cannot be identical to one another, on Lewis’ view, but they can bear relations of similarity to one another. If certain individuals in different possible worlds are similar enough in the right respects, then they become counterparts of one another. Facts about me and what it is possible and impossible for me are made true by what my counterparts are doing in different possible worlds.

Counterparts are made by the similarities held between individuals in possible worlds. Since, for any two things, there can be an infinite number of similarities and dissimilarities between them, what ultimately counts as a counterpart of something else is a highly contextual and flexible matter. Depending on the attributes that we are interested in, it may be that an individual in another possible world will qualify as my counterpart in one context, but not in another. If we are in a context where we are being fairly strict with the similarity relations, for instance, then we won’t allow for a monkey wrench in another possible world to count as a counterpart of me. And this is what will make it the case, in this context, that it is not possible that I could have been a monkey wrench (i.e., there is no possible world in which a counterpart of me is a monkey wrench).

However, in some other contexts, where we are being much more loose about similarity relations, we might allow that a monkey wrench could be a counterpart of me. In this case, this is what would make it true that I could have been a monkey wrench. Lewis leaves it open, then, just how restricted (or unrestricted) the counterpart relation is, and admits that it can change from context to context. The flexibility of the counterpart relation is part of what accounts for the fact that the truth value of our modal claims from context to context. To borrow one of Lewis’ example, it is not possible for me to speak Finnish right now, since I don’t know the language. But it could have been that I learned Finnish when I was younger, or that I had just finished Intensive Finnish for Philosophers-in-Training, in which case it is possible that for me to speak Finnish right now. Part of what contributes to the fact that our modal intuitions are so flexible from context to context in this way, Lewis thinks, is the fact that the counterpart relation is based on similarity relations, and such relations are highly flexible and malleable sorts of things. (The other (and arguably more substantial) part has to do with the fact that Lewis thinks that the quantifiers that  range over various possible worlds are contextually shifty as well. More on this below.) 

Counterparts also help us account for de re, as opposed to de dicto possibilities. [More on this later if I have time.]

Trans-World Identity: There are two sorts of problems that people seem to have in mind when they talk about the problem of trans-world identity. One is epistemological, the other, metaphysical. The epistemological worry runs as follows: if we are inclined to think that possible worlds are like distant countries or planets, then we might think that we will have a problem finding or picking out certain individuals. For if we only have the external, physical properties to go by, we will have a difficult time discerning certain objects from the qualitative duplicates of those objects. But intuitively, there is a difference between, say, Nixon, and someone who merely looks like Nixon, and has all the external, descriptive features of Nixon. So the epistemological worry involved in the problem of trans-world identity is that we will not be able to tell an individual from his look-a-like when we are searching through all of the possible worlds. (By the way, this is the worry that Kripke seems most concerned with when he’s discussing trans-world identity in Naming and Necessity. It also seems that this is related to one of Lewis’ worries about linguistic ersatzism concerning the (seeming) lack of descriptive power or descriptive resources that the ersatzer has at his disposal.)

The metaphysical worry, however, involves whether the relation that’s held between people in different possible worlds can ever be one of identity. For example, if we thought that possible worlds were concrete things, that they really were something like very far way and distant planets, then there seems to be a difficulty in saying that one and the same individual can occupy two or more different possible worlds. We say that it’s possible that I am on a beach in Tahiti tomorrow. We say that what makes this true or explains this fact is that there is a possible world where I am on a beach in Tahiti tomorrow. Assuming that this possible world is not the actual world, then there is a possible world (the actual world) where I am not on a beach in Tahiti tomorrow, and there is a possible world where I am on a beach in Tahiti tomorrow. But then it looks like we have a violation of Leibniz’s Law: I—the very same, numerically identical individual in both worlds—is both on a beach and not on a beach in Tahiti tomorrow. How can an individual, x, in world, w1, be identical to individual, y, in world, w2, when x and y have contradictory properties? Or never mind contradictory properties—that they have different properties at all is a violation of the indiscernability of identicals. This is the metaphysical worry of the problem of trans-world identity (also known as the problem of accidental intrinsics).

Lewis thinks that there is not a problem of trans-world identity because of his counterpart theory. As explained above, the only relation individuals in different possible worlds can have with one another is that of similarity, but never of identity. So the metaphysical worry of trans-world identity clearly isn’t a problem on Lewis’ view because no individual can be in more than one world; all of the individuals and objects in any possible world whatsoever are world-bound. So there is no problem about whether one and the same individual has different properties in different possible worlds. 

And similarly for the epistemological problem. Recall that this was a worry about how we could know or recognize whether an individual in a different possible world was the same individual as one we might be concerned with when analyzing certain counterfactuals. We know ahead of time, on Lewis’ view, that no individual is in two different possible worlds, so there’s no need to try to figure out or recognize whether someone in one possible world is really the one we’re looking for, or whether they’re just a very good look-a-like. The best we can hope for, according to Lewis, is that two people in different worlds are very good look-a-likes. For then these people have a good shot at being counterparts (assuming that whatever properties or similarities they share are the ones we’re interested in at the time), and then whatever one of them does will make it the case that the other could have done those things. (I do not mean to imply here that the counterpart relation is symmetric; it isn’t. Nor is it transitive. But more on that in the Lewis handout.) Since Lewis has counterpart theory, then, neither the metaphysical worry nor the epistemological worry of the problem of trans-world identity will be worries for Lewis. 

Quantifier Restriction: An initial worry some might have with Lewis’ modal realism is the following: “when we say there are things, we mean to be committed to them, ontologically, but we also mean that these things actually are. Whenever we are ontologically committed to things, in other words, we are ontologically committed to them existing in the actual world. So it doesn’t make sense to say that purple gnomes and magic beans and innumerably many fat men in doorways are, but that these thing are not actual, like Lewis wants us to do.”

In response to this sort of worry, Lewis argues that our use of ‘there are’ is contextually sensitive. Imagine, for example, that we were at a party and after looking at the sorry contents (or lack of contents) of the fridge, someone exclaims: “There’s no more beer!” Surely, we will think that what this someone is saying is true; after all, his exclamation is what motivates a bunch of us to pile in a car and head out on a beer run. But we do not thereby think that he meant that there is no more beer anywhere. That is part of the reason we left for the store to get more beer; we thought there was more to be had. It’s just that there wasn’t anymore beer at the party. Examples like these abound, Lewis thinks. What we are doing in such cases, he explains, is trading on contextually sensitive quantifiers. When we use quantifier words such as ‘everything’, ‘nothing’, ‘there are’, ‘most’, etc., we can restrict our quantifiers in each case to include just the things around here, or those things a bit further out, or we can use them completely unrestrictedly and talk about everything, everywhere. This is what is happening, Lewis thinks, when we talk about what there is. When we say that there are no purple gnomes or magic beans or innumerably many fat men in the doorway, our quantifiers are restricted to the actual world, and all of what we think is in it. When we force our quantifiers wide open, however, and want to account not only for what there actually is, but what there is tout court, then we will, Lewis thinks, say that there are purple gnomes and magic beans and whatever else we think is possible as well as what’s actual. Or at least, this is what we will say if Lewis is right.

So, Lewis thinks that he has given a fairly plausible explanation for why we have the intuition we do regarding the apparent equivalence of claims about what we think exists and what we think actually exists. As examples such as the beer case show, quantifier restriction is a subtle matter and can often go unnoticed. Moreover, he thinks that his explanation partly explains why it is that our modal claims in general are so flexible. In the section above on counterparts, I explained how it could be that in one context we think that it is possible that I speak Finnish, and in another context we think it is not possible that I speak Finnish. Endorsing quantifier restriction nicely accounts for the fact that we think that there’s different grades of modality—e.g., epistemic, physical, metaphysical, logical modality, etc.                                             

‘Actual’ is an Indexical: Another distinguishing feature of Lewis’ modal realism is his claim that ‘actual’ is an indexical. To aid our intuitions here, Lewis considers parallel indexicals such as ‘here’ or ‘now.’ If I say in North Carolina in winter, “It’s cold here,” and you say in Southern California in winter, “It’s not cold here,” we do not thereby disagree. This is because ‘here’ picks out a different location at each utterance, depending on context. In the case above, “here” said by me in NC picks out NC; “here” said by you in So Cal picks out So Cal. This indexical feature of the word “here” does not mean that our locations are significantly metaphysically distinct. As far as space goes, on the contrary, our locations are metaphysically on a par. Lewis compares this fact about location with facts about modality. We call our world actual because we are in it. Other individuals in other possible worlds, however, could call their world actual as well. Neither of us would be wrong, just as neither of us would be wrong in calling our locations in space ‘here’, even if we were on different coasts. In this way, for Lewis, what counts as the actual world is going to be something that is determined by context. Notice the contrast here with ersatzers, who think that only one world is actual, and that the word ‘actual’ is not an indexical at all.

So what makes our world so special on Lewis’ view? Nothing, really, except that we’re in it. It’s just as if one was a four-dimensionalist about time. The only thing special about ‘now’ is that it is the time that we’re in. But, metaphysically, it’s not any different than ancient Rome. Similarly, Lewis thinks, there is nothing metaphysically distinct about our world as opposed to other possible worlds. All worlds are metaphysically on a par, in other words.

Why Modal Realism is Better than the Alternatives: Lewis thinks that modal realism is better than any of its competitors for several reasons. Concentrating just on the realist alternatives for now, Lewis thinks that his view is better than any of the ersatz views mostly because his view seems to have fewer descriptive limitations and more descriptive resources than the alternatives.

Briefly, an ‘ersatzer’ (as Lewis himself calls them), or an ‘abstract modal realist’ (as they tend to call themselves), claims that there are possible worlds, but that these worlds are not concrete in the way that Lewis thinks. Rather, they are abstract sorts of things—although, exactly what kind of abstract thing varies according to the abstract modal realist view under consideration. However, Lewis thinks that there are two main problems for ersatzers in general, stemming from the fact that they only have so many descriptive resources at their disposal.

To see his worry, imagine that there is a world just like this one, except that every billion years it descriptively repeats itself, where this repetition eternally recurs both forwards and backwards (i.e., the world has always been repeating itself, and always will). Then, in every cycle there will presumably be someone very much like me who will be typing these words on a computer very much like this one. Intuitively, though, despite the descriptive similarities, there could be one of them that is me, not just someone very much like me. It seems, anyway, that I could exclaim in one of the eras, “Hey! It’s me!”. But, of course, on the linguistic ersatzer’s picture, all of the other duplicates of me could exclaim the same thing, or utter a similar sentence, and so there would be no way to descriptively distinguish me from any of my counterparts. In other words, it looks like me and all of my duplicates in this world are descriptively indiscernible. Nonetheless, there’s an intuition that despite the indiscernibility, one of them is me, whereas the others are not. But an abstract modal realist is going to have difficulty describing this difference, because without difference in description, a (linguistic) ersatzer has no way to account for the (intuitively) ontological difference. Since Lewis doesn’t rely on sentences or descriptions in this way—since he can have individuals, de re—he can make a distinction between an individual and a descriptively indiscernible duplicate. This shows, he thinks, a sizable advantage of his view over competing views.    

Another problem has to do with ‘alien properties.’ We might think, for example, that there could have been properties totally different from ones that there actually are—so different, in fact, that they are not built out of anything in the actual world. But if this is right, then the abstract modal realist is seemingly going to have difficulty accounting for such a possibility, since his worlds are presumably made up out of descriptions (or pictures or something else) of things that actually exist. The abstract modal realist, in other words, has a sort of combinatorial approach to possible worlds in that all of the possibilities are built out of actual elements or properties here in the actual world. But if this is right, then it looks like he will not have the descriptive resources to allow for the possibility of truly alien properties.


Too Ontologically Costly:
One of the biggest complaints about Lewis’ modal realism is that it is too ontologically costly. He wants us to posit the existence of continuum many possible worlds—all of which are spatio-temporally and causally isolated from each other. In each of these worlds, are numerous possible, world-bound individuals, none of whom can exist in more than one possible world. When our existential quantifiers are wide open, then we are ontologically committed not just to all of these possible worlds, but to all of the possible individuals in them. We are committed to purple gnomes and magic beans and innumerably many fat men in doorways, etc. This is too ontologically costly, some say.

Lewis would simply claim that the theoretical benefits outweigh the ontological costs. 

Too Epistemically Mysterious: One concern we might have with Lewis’ Modal Realism is that it is too epistemically mysterious. Lewis claims that what ground our modal truths are the things that happen in spatio-temporally and causally isolated possible worlds. But, how, we might wonder, can something that happens in a world that is spatio-temporally and causally isolated from us have any effect whatsoever in the modal properties we have here and now. What’s more, even if this view of modality were true, how could we ever know what was happening in other possible worlds? If other possible worlds are spatio-temporally and causally isolated from us, then how could I ever find out what other people are really doing in other possible words? And if we have no way of finding out what they are doing, then how can we ever find out what the modal truth are, and so how can we know whether anything is possible or necessary?

Lewis’ response here is to invoke what’s known as Benacerraff’s Dilemma. In the philosophy of mathematics, for example, we can either deny that our talk of mathematics is useful (or deny that it really is quantifying over mathematical entities), or we could deny that  the only way we can come to know things is causally. In the one case, we are letting our metaphysics be guided by an empiricist epistemology, in the other, we are letting our epistemology be guided by our metaphysics, or at least letting our epistemology be guided by the sorts of entities that our language quantifies over. If we find mathematics and mathematical talk indispensable, and irreducible, however—i.e., if we can’t find a way to reduce such talk to something more metaphysically friendly—then we might be inclined to plunk down for mathematical entities in our ontology rather than an empiricist-friendly epistemology. It’s all a matter of trade-offs and theoretical benefits, Lewis thinks. And like philosophy of mathematics, Lewis argues, we should decide in favor of metaphysics over epistemology in the case of modality and possible worlds.  


The Argument from Concern: This objection can be rather simply stated, and I think originated with Rosen (in his Modal Fictionalism article). It goes like this: “Look, I care about whether I could have been a millionaire, or could have gone out with that one guy if I would have just let him beat me at pool, or could have passed the metaphysics exam if I would have studied just a bit more diligently, etc. I care, in other words, about certain modal facts about me, about what I could and could not have done, or what is possible or not possible for me to do in the future. But I do not care about what someone very much like me in some possible world way too far away does! I don’t even really care about what someone very much like me one state over is doing. So why should I care what someone very much like me in a spatio-temporally and causally isolated possible world is doing? That I care for one thing so very much (e.g., my own modal facts), and care about the other so very, very little (e.g., some stranger very much like me in another possible world) shows that the ‘two’ facts cannot be the same thing.

        I don’t know whether Lewis addresses this sort of worry; I don’t think he does in Plurality of Worlds, at least. But there’s an obvious problem with the above line of reasoning: it commits the
        Intensional Fallacy. Lois Lane might claim the following: “Look, I care very much for that hunk, Superman, but I care very little for that milquetoast reporter, Clark Kent. So Superman can’t
        be identical to Clark Kent.” Tsk, tsk. Shame on you, Lois. You should've taken a philosophy class.

The Incredulous Stare: There’s not much to be said here, since it’s not really an argument. Nonetheless, the ‘argument’ of the incredulous stare basically goes: “You believe in concrete possible worlds that are spatio-temporally and causally isolated from us, and this is what is supposed to ground modal facts? WTF?!?!” Followed by an incredulous stare...

       I suppose Lewis can shrug his shoulders, scratch his head, or simply whistle dixie in response.

Page Last Updated: Aug. 30, 2009

Back to Meg's Teaching Page
Back to Meg's Main Page