Lewis, David (1941–2001)
David Lewis was born in Oberlin, Ohio. He studied as an undergraduate at Swarthmore College before gaining a PhD in philosophy from Harvard University in 1967 where he studied with Willard van Orman Quine. His first job was at the University of California at Los Angeles, where he worked from 1966–1970, before moving to Princeton University where he worked for the rest of his career. Lewis published four monographs and more than one hundred papers, most of which have been gathered into five volumes of his collected papers. Lewis made contributions to virtually every area of contemporary Anglo-American philosophy but is probably best known for his contributions to metaphysics, in particular, his work on modality (necessity and possibility) and possible worlds and also his theories of laws of nature, causation, and chance. His work in the philosophy of mind has also been influential, as has his work on conventions and language.
Some of Lewis's best known work is in the metaphysics of modality: that is, his account of the nature of necessity and possibility. Lewis thought it was important to make sense of what we are doing when we talk about different possibilities that seem open, or when we say that certain facts (such that 2+2=4) are necessary, or that it is impossible for them to be otherwise. Lewis held, along with others whose claims about possibility and necessity were to be understood as implicitly generalizing over possible worlds, complete ways things could be: To say something was possible was to say that it occurred in at least one possible world, and for something to be necessary was for it to obtain in all possible worlds. Where Lewis was nearly unique was his account of what these other possible worlds were.
According to Lewis, possible worlds were large spatiotemporal regions filled with objects and events of the same kind as those in our world, except, of course, that every possible sort of thing is found in one world or other. So Lewis's worlds contain people and trees and galaxies and tables; but also dragons, extra-spatiotemporal dimensions, ghosts, and so on. This construal of possible worlds became known as modal realism Despite the counterintuitive nature of this theory, Lewis showed that it brought with it many advantages, and he argued that attempts to construe possibilities as some sort of abstract object (ersatzism about possible worlds, in Lewis's vocabulary), failed to provide an analysis of modality, and many varieties suffered crippling internal problems.
Lewis also suggested a novel way of dealing with de re necessities and possibilities (possibilities or necessities for an object rather than as concerning the status of a proposition). Lewis argued that these were best analyzed using counterpart theory: where what is possible for me is what happens for one of my counterparts in another world. Since Lewis held that, strictly speaking, each possible individual was part of only one possible world, he could avoid some of the puzzles about trans-world criteria for identity. In addition, counterpart theory allowed more flexibility than literal identity would. Lewis argued, for instance, that the counterpart relation need not be transitive (so something that could happen to one of my counterparts need not be something that is possible for me) though a failure of transitivity is harder to understand if it is literal identity across possible worlds that is required for de re possibility (i.e., if something has to be literally happening to me in some other possible world in order for it to be possible for me). Lewis also allowed that there were multiple counterpart relations, which might give rise to multiple kinds of de re possibility for an object. So, for example, what the counterparts of an object are when that object is considered as a statue might be different from what the counterparts of that object are when the object is considered to be a lump of bronze. Lewis could thus allow that what we appropriately say is possible for the statue is different from what we appropriately say is possible for the piece of bronze even though the two objects might nevertheless be identical.
Counterfactuals, Laws, Causation, and Chance
Issues about contrary-to-fact-conditionals, laws of nature, causation, and chance are often thought to be connected, and Lewis's contributions to these topics formed a unified neo-Humean system. Lewis's book Counterfactuals (1973a) offered an analysis of conditionals of the form if p, it would have been the case that q in terms of possible worlds: a conditional such as if dolphins had had legs, they would have walked on land is true if and only if the nearest possible world where dolphins have legs is one where dolphins walk on land. Nearness, in turn, is analyzed as overall similarity in salient respects: it is thus context-dependent, and Lewis had more to say about what sort of similarity is significant for particular sorts of these so-called counterfactual conditionals, for example the ones employed in causal reasoning.
This analysis of counterfactual claims has several advantages. It is formally tractable, yielding a logic of counterfactual judgments with some initially surprising features that do seem to correspond to features of our ordinary counterfactual judgments. For example, Lewis's system delivers the result that strengthening the antecedent is invalid: that is, the inference (if p then q ), therefore (if p and r then q ) is invalid. But consider this argument: If I leave now, I will catch the train; therefore, if I leave now and am assassinated on the way I will catch the train. The premise might well be true and the conclusion false if I run no real risk of being assassinated.
In addition, since the analysis of these conditionals does not itself appeal to, for example, dispositions or causation, it leaves the way free for counterfactual analyses of other puzzling parts of metaphysics. And, indeed, Lewis championed a counterfactual analysis of causation: At a first pass, an event C causes an event E if and only if both C and E occur and had C not occurred, E would not have occurred either. A lot more than this first pass is required for an adequate counterfactual account of causation: Sometimes E would have happened in any case, even without C, for example, if E is overdetermined. Lewis experimented with a number of counterfactual theories of causation: Their development can be seen in Lewis 1973b (and see especially the postscripts in Lewis 1986b), Lewis 1979a, and most recently Lewis 2000 and Lewis 2004.
The connection between counterfactuals and causation, on the one hand, and laws of nature, on the other hand, is slightly circuitous in Lewis, but it is another key connection in his overall system. Lewis defended a regularity theory of laws of nature: Following Ramsey, Lewis held that the laws of nature were given by the set of generalizations that provided the best tradeoff of simplicity and strength in capturing the goings-on of the world. Since the laws supervene on the patterns of particular matters of fact, at this point, at least, Lewis's metaphysical posits are minimal.
Even though the laws are only descriptions of certain privileged regularities, they make a difference to which counterfactuals are true, in Lewis's system, because similarity with respect to whether our laws hold is one of the most important components in the kind of similarity relevant for the nearness relation between possible worlds central to Lewis's analysis of counterfactuals. So when some event A would follow as a matter of law from another event B, the nearest world where A occurs will be one where B also does. Thus mere patterns of particular occurrences give rise to laws of nature, counterfactual dependencies, and so to causation—at least, if Lewis is right. Lewis extended his regularity framework to handle objective chances as well: Another member of the nomic family was explained, ultimately, in terms of regularities in particular events.
Lewis made contributions to several areas of the philosophy of mind. First in importance is his defense of an identity theory of the mind. Lewis characterized mental states according to the role attributed to them in our ordinary folk understanding of the mind: A belief, for example, is a state that tends to go together with desires with certain contents to produce certain sorts of actions. Folk psychology, when articulated, describes causal roles for each different sort of mental state (beliefs, desires, pains, emotions), and these roles are interdefined so that the typical causal profile of a belief is specified partly in terms of other beliefs it tends to cause, partly in terms of perceptions that tend to cause it, how it interacts with desires, and so on.
Armed with this role statement of the typical causes and effects of mental states, Lewis then argued that mental states are identical to those physical states in us that play these causal roles: So Tom's belief that it is raining is identical to the brain state of his that is typically caused by the sight and sound of rain, and typically goes together with other brain states to yield umbrella-grabbing behavior, and so on. This may well mean that which type of physical state is identical to which type of mental state may vary from subject to subject: Lewis says that which physical state is identical to a given mental state depends on what causal roles that state typically plays in the kind of thing that has it. So in humans pain will be a certain sort of brain state while in advanced robots it may be some electronic state, and in extraterrestrials it might be a matter of how gases are distributed within internal bladders.
This typical-for-the-species criterion allows both for multiple realizability : pain-in-aliens or pain-in-invertebrates may not be the same physical state as pain-in-humans; and we loosen the behaviorist insistence that pain must be the state, whatever it is, that produces pain behavior since, for example, an atypical human may have the state that typically causes pain reactions but makes no outward show of it, or even engages in some nonstandard behavior (imagine a madman who whistles, but shows no discomfort, whenever he is in the state that produces pain behavior in normal humans). The view is still a type–type identity theory, according to Lewis, because, for example, the type pain-in-humans can be identified with a particular physical property (e.g. C-fiber-firing-in-humans) even though there is no unified physical type corresponding to pain or belief simpliciter.
Language and Convention
In Lewis's first book, Convention (1969), he developed a theory of conventions as patterns of mutual expectation and conditional intentions. Roughly, according to Lewis, there is a convention in a population to act in a certain way in certain circumstances if everyone does tend to behave in that way, everyone has the conditional intention to continue behaving that way, conditional on everyone else so acting, and this is common knowledge. Finally, there must also be some other alternative action that people are deciding against: We all breathe oxygen, intend to continue and know that we intend to continue, but this does not make our practice of oxygen-breathing conventional since we all have no choice. He claimed these patterns could arise fairly spontaneously (certainly without the existence of an explicit agreement) and that they tended to arise to solve coordination problems: common cases of collective action where everyone does better by coordinating their activities than if everyone does their own thing. (A decision about which side of the road to drive on is an example: The most important thing is not whether people drive on the left or right hand side but, rather, that either everyone does the one or everyone does the other.)
Lewis argued that we could understand what it was for a population to use a language as a matter of convention, in his sense. In "Language and Languages" (1975), Lewis explained how we could integrate the formal, abstract theories of languages as functions from expressions to truth-conditions, on the one hand, with theories of language that concentrate on practices of language usage. Lewis also made significant contributions to the formal theory of language and philosophical semantics—his "General Semantics" (1970) is a prime example. Lewis was also responsible for a lot of work exploring the role of context in language: His "Scorekeeping in a Language Game" (1979b) is a classic in this area.
monographs by david lewis
Convention: A Philosophical Study. Cambridge MA: Harvard University Press, 1969.
Counterfactuals. Oxford: Basil Blackwell, 1973a.
On the Plurality of Worlds. Oxford: Basil Blackwell, 1986a.
Parts of Classes. Oxford: Basil Blackwell, 1990.
volumes of collected papers
Philosophical Papers. Vol I. Oxford: Oxford University Press, 1983.
Philosophical Papers. Vol. II. Oxford: Oxford University Press 1986b.
Papers in Philosophical Logic. Cambridge. U.K.: Cambridge University Press, 1998.
Papers in Metaphysics and Epistemology. Cambridge, U.K.: Cambridge University Press, 1999.
Papers in Ethics and Social Philosophy. Cambridge, U.K.: Cambridge University Press, 2000.
other works cited
"General Semantics." Synthese 22 (1970): 18–67.
"Causation." Journal of Philosophy 70 (1973b): 556–467.
"Languages and Language." Minnesota Studies in the Philosophy of Science 7 (1975): 3–35.
"Counterfactual Dependence and Time's Arrow." Nôus 13 (1979a): 455–476.
"Scorekeeping in a Language Game." Journal of Philosophical Logic 8 (1979b): 339–359.
"Mad Pain and Martian Pain." In Readings in the Philosophy of Psychology, Vol. 1, 216–232. Cambridge MA: Harvard University Press, 1980.
"Humean Supervenience Debugged." Mind 103 (1994): 473–90.
"Causation as Influence." Journal of Philosophy 97 (2000): 182–97.
"Causation as Influence" (extended version). In Causation and Counterfactuals, edited by J. Collins, N. Hall, and L. Paul, 75–106. Cambridge, MA: MIT Press, 2004.
Daniel Nolan (2005)