I’ve been a little lax about updating my blogs lately since I’m spending the summer in Cambridge England and the toils of travel plus the sparse internet access here has made posting more difficult. However, I took up reading Lewis’s “On the Plurality of Worlds” during the trip and it has inspired a great number of questions and thoughts I hope to post when I have the chance. Admittedly I have yet to finish the book and only have a cursory grasp of the literature so please excuse me if I am unknowingly repeating prior work. I always appreciate comments but I would be especially grateful if anyone with more expertise in this area could correct any mistakes or fill in any gaps.

The simplest puzzle that modal realism brings to my mind, aside from its shear counter-intuitiveness, is how it is supposed to be compatible with physical realism. Or to be more precise what grounds could a modal realist have for embracing realism about physical objects like electrons tables and chairs? So while I see no logical tension between modal realism and physical realism I do think there is an epistemic tension. In particular it seems that a modal realist has no reason to find physical realism more probable now than it is prima fascia, i.e., observations about the actual world (it follows laws, appears to have persistent mind-independent objects etc..) can’t serve as justifications for physical realism.

Overview:

The essence of my puzzle is that physical realism is justified on the principle that a single coherent explanation is more probable than postulating many brute facts, i.e., it is more likely that physical objects really exist than we just happen to have experiences as if they exist. However, modal realism would have us believe that all possible worlds are equally real, including those worlds where physical objects behave in bizarre unpredictable ways. Thus it would seem the modal realist is explaining our observations by appeal to the brute fact that this world is actual. Since the specification of which world is actual is equivalent to specifying all the true facts about this world it no longer seems as if physical realism has a preferred status as a simple explanation. Both the modal realist who believes the actual world satisfies idealism and the one who believes the actual world satisfies physical realism are equally explaining our observations by stipulating them as brute facts of existence (which world we inhabit).

Update 6/22/05: Added Overview, Comment about induction Interestingly while modal realists like Lewis do tend to favor physical realism (about the actual world) it is not clear it is essential to the position of modal realism. It might be the case that in this world only subjective experiences have metaphysical substance (i.e. electrons, tables and chairs are defined as correlations in subjective experiences). In fact it seems consistent to be both a modal realist and believe that physical realism isn’t even metaphysically possible. One might believe that existence only makes sense for things with a qualitative subjective experience (what would it be for a tree to make a sound that never affected experience?) but still hold that every possible experiential world (perhaps a collection of subjective viewpoints and their experiences or perhaps just a collection of experiences) exists in the same way that this world exists. Thus modal realism in and of itself does not justify physical realism so the question remains whether the modal realist can support physical realism on other grounds, e.g., the normal grounds for believing in realism.

It appears that the principle reason for favoring physical realism over some kind of idealism is our desire to explain the apparent consistency of the external world. It is tempting to think that taking tables and chairs to be real objects somehow explains why they appear to persist and exhibit predictable behavior regardless of our attention to them or mood. If you are committed to believing that the world we experience is much the same as that experienced by other humans postulating a mind-independent external world seems an attractive explanation (though arguably the inference should go the other way).

Of course we could simply insist that the observed regularities and postulated inter-subjective agreement about the external world was a brute fact about the nature of experiences. It is simply the nature of experiences to come in consistent patterns as if they were caused by independently persisting objects. Clearly this approach to idealism can account for every observation that physical realism can explain but the jury-rigged nature of this response makes it an unattractive choice to many people. Thus the principle argument for physical realism over idealism is that physical realism offers a simple explanation of our experiences while appealing to relatively few seemingly arbitrary primitive facts.

Now if we don’t believe in physical realism for purely a priori considerations this must mean that there are some contingent features of our existence which give us cause to believe in physical realism. That is there are genuinely metaphysically possible worlds which would not warrant belief in physical realism. In modal realism, however, every metaphysically possible world exists in the same sense that our world exists. Thus as a modal realist one must believe physical realism is true on the grounds that you happened to end up in an appropriate sort of world. Yet it seems quite implausible that this sort of fact could rightly support a conclusion of physical realism.

If one didn’t believe idealism was a metaphysical possibility then you would have already ruled it out on a priori grounds. So by my above idealist reply you even believe that there are worlds which appear to be exactly like this one yet are in reality idealist in nature. Thus you must believe something like the following. The probability of ending up in an idealist world given the world looks as it does is higher than the probability of ending up in an idealist world. In and of itself this statement seems troubling as every physical world will have an idealistic counterpart which is observationally identical to the beings in the world. It is particularly hard to see how the features of the probability distribution among multiple worlds which distinguish these two sets of worlds of the same cardinality (physical worlds and their ideal counterparts) could be learned a posteriori.

Additionally since every metaphysically possible world in fact exists it seems one must appeal to brute facts to explain one’s experience regardless of the choice of idealism or physicalism. Since physical worlds where chairs react to your mind and tables turn into snakes when your back is turned exist just as genuinely as our world it no longer seems that appeals to physical realism explain our observations any better than stipulating brute facts about experience explain our observations. In both situations one is now simply stipulating that things happen in the way one observes them.

In the case of idealism one is stipulating that experiences occur in a manner which looks as if there were independent physical objects. In the case of physicalism one is just stipulating that you are a member of a world where the physical objects don’t ‘pretend’ to be idealistic. In other words since modal realism essentially makes the stipulation of brute facts about the actual world equivalent to the stipulation of what world you happen to occupy it seems both theories require and equal amount of arbitrary stipulation. If one is just going to stipulate some parameter which fixes all the observations why should it make a difference if this parameter is a coded up form of the physical events that will occur in that world (i.e. which one of the physical worlds one inhabits) or a coded up form of the experiential events in that world (i.e. which one of the mental worlds one inhabits).

I’m not sure if I find these two objections entirely convincing. In the first case perhaps someone has a clever argument about an a posteriori metaphysical necessity but I am somewhat skeptical about these claims (they seem very epistemicly troublesome). While I reject Krpike’s a posteriori analytic truths for unrelated reasons (I think it is simply a change in definition) even if one believes in them I don’t think they are relevant here. Those examples are about linguistic choices/descriptions while here we are talking about genuine substantive issues (if you think what it means to be a realist is a linguistic matter you have already given up your position).

Also both cases seem to have the potential of being turned into a more general objection to contingent knowledge in modal realism, e.g., how can induction work given there are real worlds which are just like ours up till now but exhibit radical inductive failure tomorrow? It seems like this ought to be the sort of objection previously raised and considered so I’m curious to hear if anyone has any thoughts about the matter.

—-Update 6/22/05—

Kenny argues that this is merely an instance of a general worry about induction. I take him to be saying that this worry just rephrases the general worry that we can never gain contingent evidence for induction as this requires presupposing the truth of induction. While I will quibble about what exactly this means in a latter post I don’t think this is quite what is happening here. If indeed this is the case I think it would be a strong argument against physical realism.

While I can’t justify induction I am willing to give it a high degree of confidence a priori. While I might adjust my credence in some types of induction based on evidence the essential leveraging principle does seem like the sort of thing one just stipulates. Physical realism seems entierly different. The temptation to physical realism stems preciscely from the utility of the hypothesis to explain our observations. I doubt very much that anyone would have confidence in physical realism if the world behaved like a dream or some other manner not well modeled by persistant mind-independent objects.

Quite simply we seem to believe physical realism demands a justification in the way induction does not. Very few people are willing to give up induction for any reason while the same cannot be said for physical realism. If we found that adoption of idealism was essentiall for a compelling new refinement of quantum mechanics I think many people would be swayed (say the theory is most simply expressed in terms of interactions of disembodied mental entities). The same is certainly not true for induction.

If I misunderstood Kenny and he was instead saying that our belief in physical realism is warranted by induction which we hold stipulatively I am more sympathetic to his point. Thus one might interpret my argument as a general objection to induction in modal realism. Namely, how is belief in induction compatible with believing there is an equal (perhaps greater) cardinality of worlds where induction fails as it succeeds? While this is itself an interesting question I think my worry is somewhat more specific.

That is one might hold that induction is valid as a form of predictive inference. Namely that it is a good strategy to approximate the future using the simplest theory which fits all your past data. I think such a view may very well be compatible with modal realism but it falls short of justifying physical realism. After all idealism and physical realism are observationally equivalent. Thus the argument for realism can’t be based on simply choosing the least complex function fitting the past data but must instead rest on a comparison of the number of brute facts/stipulations necessery for a metaphysical model of the world. However, as I argued above once modal realism is assumed it doesn’t appear physical realism has an advantage in this department.