This action might not be possible to undo. Are you sure you want to continue?

. 85, No. 3 (Dec., 1990), pp. 355-389 Published by: Springer Stable URL: http://www.jstor.org/stable/20116853 . Accessed: 01/10/2012 02:21

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Synthese.

http://www.jstor.org

CHRISTOPHER

MENZEL

ONTOLOGICAL ACTUALISM, AND POSSIBLE WORLD

COMMITMENT, SEMANTICS

ABSTRACT.

is the doctrine that the only things there are, that have being Actualism are the things that actually exist. In particular, actualism eschews possibilism, It is widely that there are merely held that one cannot both objects. possible be an actualist and at the same time take possible world that is, semantics seriously or look to it for of truth for modal take it as the basis for a genuine languages, theory in any sense, the doctrine structure of reality. For possible world semantics, insight into the modal one to possibilism. commits In this paper I take issue with this view. To and yet remain semantics argue that one can take possible world seriously ance with actualist scruples. it is supposed, the contrary, in full compli I

Here

[Possible possible

**are some quotes
**

world semantics] individuals.1

representative

carries a commitment

of a widely

to the

shared belief.

of possible worlds and

reality

of a set of possible includes a function which maps each member semantics] [Kripke-style to the set of objects in that world, worlds need not exist in our objects which existing are given in terms of a totality of all possible world. truth-conditions Thus, objects, including nonactual possibles.2 A tions. semantical . . .The to the reality of the entities it uses in its explana is committed theory or other possible worlds to is committed theorist, Montague grammarian, to tell us what they are if we are to take their theory seriously. and needs worlds is not a responsible they are "just indices" theory/

possible Saying

I am an actualist. I think that whatever exists - whatever has being in - is actual. Thus, I do not think that there are any merely any sense I am Furthermore, objects or any merely possible possible worlds. actualist reconstructions of these notions. None skeptical of proposed a lot. I think it reflects - in a theless, I like possible world semantics, certain sense - the modal structure of reality, that it yields correct truth conditions for much of our modal discourse, and indeed, in the guise a powerful of Montague that it provides basis for a full Grammar, to the belief blown semantics for much natural language. Contrary represented commitment counterparts. I think it does all this without in the quotes above, any mere or their actualist to possible worlds, possibilia, In this paper, I explain why.

?

85: 355-389, 1990. Synthese 1990 Kluwer Academic Publishers.

Printed

in the Netherlands.

356

1.

CHRISTOPHER

MENZEL

PLAIN

VANILLA

SEMANTICS

My

semantics for first-order starting point is plain vanilla, nonmodal for the time languages. The usual story, in which we can acquiesce like this. A language is an uninterpreted formal being, goes something are the result of recursively elements system whose applying certain to an initial set of basic elements. rules of construction Thus, more we can think of a first-order exactly, language ?? as a pair (Lex, G) whose first element is a (first-order) lexicon - terms (variables and and the like - and constants), logical constants, n-place predicates, whose is a grammar that generates second element all the well-formed formulas of the language from the lexicon in the standard fashion. Hand in glove with the notion of such a language goes the idea of a first-order

relative to which such semantical notions as truth and model, can be defined. can be Such a model (more ambitiously) meaning, defined as a pair (D7 R), where D is an arbitrary set of objects, and R = {Rn:?GiV,n>0}, where each Rn is the set of all subsets of Dn.4 The members of D are often called the "individuals" of the model, and each set Rn is thought of as the set of (extensional) n-place relations in D. among the individuals The connection a language and a model between is provided by means of a valuation function. Thus, given a language if and a model M = (D, R), a valuation function V assigns elements of D to the terms of i?, and elements of Rn to the rc-place predicates. The pair (M, V ) is called an interpretation I of the language. So armed we can define for formulas of ifthe notion of truth in, or relative to, the model M under / in the usual way in terms of an extension V of V that maps formulas into a set of "truth values" (e.g., the set {0, l}).5 a semantic theory is supposed to Now, among other things perhaps, tell us something about truth. Indeed, our plain vanilla semantics above, or something like it, is often (though somewhat referred inaccurately) to as "Tarski's theory of truth". But so far all we have is the notion of

truth-in-a-model, or more exactly, truth-of-formulas-in-an-uninterpre How does truth plain ted-language-in-a-model-under-an-interpretation.

fit into the picture?6 Something along these lines, I think. Truth and falsity attach (perhaps to declarative sentences that we use in natural indirectly) language. a sentence Which of these attaches to (on a given occasion of use) is a function of two parameters: its linguistic meaning, and the world.7 In

and simple

POSSIBLE WORLD SEMANTICS 357 other words. albeit only-when we turn our gaze to models of a very special sort: those which the world itself provides. as far as it goes. Let TVL(Lex) be those sentences of natural language whose logical form is represented by some formula in G(Lex). for any given chunk of ordinary language like NL(Lex) there is typically some piece of the world that it is about. If either had been or the world . in G(Lex) there is exactly one formula LF(S) that represents NL(Lex) its logical form. we must be able to consider the theory. "logical form" is plausibly formulas of the sort by well-formed represented is one way. Since not every natural language (specifically. For only then can it be taken to characterize (modulo some correctly to the deter the world's contribution of detail or "granularity") degree mination of a sentence's truth value. Begin by keying on some generated by our grammar. More exactly. G) is thus a language in our formally defined sense. . say. what 'Stalin signed The Declaration of Indepen dence' means) and in virtue of how the world is.S. truth plain and simple we are after. Here names and verb phrases of English (perhaps conveniently to serve as the constants and predicates of a lexicon Lex.it might well have been untrue.. natural numbers. our notion of truth-in-a-model will still do the job for us. items in the congressional representatives. fragment. !?= (Lex.planets. abbreviated) a set G(Lex) A corresponding G will then generate of well grammar formed formulas from this lexicon in the usual way. the world then determines whether or not it is true. Thus. different its meaning If a semantic theory is going to provide a genuine theory of truth. English) construction our first-order semantics to natu has a first-order counterpart. 'Russell wrote The Principles is true both in virtue of the fact that it means what it of Mathematics' does (and not. is then a fragment of natural language to TVL(Lex) which our theory can apply in virtue of its formal "interpreted" counter collection I will assume that for each sentence S in (For simplicity. first. U. applying ral language is a matter of isolating an appropriate natural language some set of natural language sentences whose i. When part ??.e. then.) How then does our semantics function as a reliable picture of reality? How is the world's contribution to the truth value of the sentences of our fragment of natural language characterized? it is Directly. to be a reliable guide to the structure of the world. given that a sentence means what it does. some set of objects . and second. there must be some way of applying it to (at least a fragment of) natural language.

exactly. the set of all exten second element R contains. relative in the actual world. of other possible worlds that these are the key to that exist in them. of ways it could be. is true inM under /. the many possible ways the world might have been.9 and the corresponding model theory. To this extent there is acknowledgement of some sort which will have to be faced commitments of metaphysical to provide a genuine theory of truth. whose over D. we ways not just a can get a semantical grip on necessary truth by considering. of a logic cum semantics cum metaphysics involved in the development as manifested in intuitions of modality. and the values of our chosen chosen are their actual extensions. though. single model. Necessary truth. We begin with our modal . by means of a particular model that represents the world as it is (albeit in a rather coarse fashion). one of each representing but an entire cluster of models. and what the relationship not at this point the main focus. There is talk of the and perhaps the world. The major concern the right sort of semantical apparatus for necessity.358 fridge to which CHRISTOPHER MENZEL its the terms of the chunk refer. We can then define the intended inter sional rc-place relations I = (M. Thus. and it is suggested things an adequate modal semantics. POSSIBLE WORLD SEMANTICS a matter of how things are Truth for nonmodal languages is essentially our formal semantics.of its !? whose domain D intended model counterpart interpreted about and quantified over and consists of the very things being talked for each n > 0. is There is a certain amount of bootstrapping This is entirely appropriate. and over which the This set can then form the basis of a model M range. the presentation This is the sort of pep talk that often accompanies semantics. But what those if the semantics is them is between commitments are. but of all the other it might have been as well. We captured this in to a given language. all other possible worlds. At this stage of the game of a system of possible world is yyith getting a fix on things are quite indefinite. V ) for 5E in its intended model M to be the result of pretation our valuation function V such that the semantic values of our defining names are their actual denotations.8 Truth plain and simple is now predicates a sentence S of TVL(Lex) is true just in case its formal straightforward: counterpart LF(S) 2. is not just a matter of how the world actually is. quantifiers . one of the many other possible worlds.

Finer discriminations ?-place to some background made relative whose only predicates language in different ways in the same model. To see this more clearly. rigorous be held in abeyance. (some might the intuitions further.POSSIBLE WORLD SEMANTICS 359 and are corroborated corroborate. R(S)). On the reflect the structure of modal whatever this reality. On we want it to the one hand. we are interested only in changes on the world's we want to depict. comes to. Intuitions and imagery together "perverts"!) drive the construction of a formal logical and semantical apparatus. we want it to characterize the relation between adequately as they stand. a cluster of such models would provide a dim Regarding vision of modal reality indeed. Once the apparatus is in place. the first. where R(S) partitioned) can be of all extensional relations on S. below) serve as the focal point of the surrounding metaphysical Until of those issues can (and will. Clusters of plain vanilla models language are in need of supplementation on both counts. or the individuals there are could have had to one another. side of things: fixed. then. our modal discourse. we hold meaning means. since the models represent different ways the world intuitively to reflect could have gone. Our modal semantics has (at least) two jobs to do. and that reality. rather than having to differences in interpretation to represent in order adequately appeal a language-indepen different ways the world could have gone. however. for now) discussion then. by. The idea. we should prefer them in and of themselves to and explain the differences in interpretation. the second. since for any given the model is the set (appropriately (S. it can (and will. the turn pumps in which say imagery of possible worlds. note that things in one of two ways: there could have been could have been different more or fewer individuals. In modal semantics. Or at least it is a good start.10 How could then be interpreted ever. at least tentatively. different and stood in different relations properties can capture the first sort of difference well Distinct plain vanilla models but not the enough domains). But it has a couple of deficiencies good as things stand. how its truth value dent solution . to this shortcoming is much preferred. Hence. It is a one. given the meaning of a sentence. other hand. we noted above that there are two central Regarding a sentence's that determine truth value: what the sentence parameters and what the world is like. (simply in virtue of having different set S we get only one plain vanilla model: second. is to get at necessity via clusters of models. as we have unspecifically acknowledged. These issues.

nothing across models that pins stable the fixity of meaning. By the same token. tics. different extensions two ways mentioned can be distinguished and by assigning properties the second in the models. hidden. in short. of modal vanilla models fail in their characterization reality. and add .e. different ways But how is this nailed down in P'ishness could have been exhibited. changes in certain respects. predicates) same property having different extensions in different since situations on this approach. us make all this prim and of a plain vanilla model definition in intension. down the meaning In regard to the first point. ties and relations (not implausibly) we can fix their meanings across models despite assigning them different extensions. ambiguity. make room for entities that play such a role in our are removed. are supposed to reflect to the two models in P's extension relative in which different ways the world could have been. in the cluster and suppose a predicate P is assigned extension E relative value would these changes and extension E' {+ E) relative toM'. are extensional. (in general) accompanies common set of properties and relations m intension to our models (or more accurately. Intuitively. will be lost. representing two models M and M' Consider different way things could have been. the fixity of meaning a each model suppose we have a given cluster of models.n>0} Let our proper. But there is nothing answering to this intention in is not in the semantics that itself. However. There the semantics is. To see this. plain The diagnosis here is not difficult.. the extensions just are properties a change in reason a cluster of plain vanilla models cannot distinguish in extension is that difference change inmeaning if we add a both sorts of change. from a case of mere is this to be distinguished the semantics? How mere change in the linguistic meaning that of P? Granted. nothing captures of the predicate. our intention. or more what its truth accurately. thus capturing in which above things could have been from a mere toM the second point. Models definition of a model). First let us augment room for properties and to make matter: we just replace the set R an pairwise disjoint sets. by taking proper Furthermore. have been had the world been different in our modal seman ifwe use clusters of plain vanilla models However. of course. and relations the semantic values of their properties because (i. the for example. represent they cannot.360 varies with CHRISTOPHER MENZEL in the world. or at best. then both shortcomings the world with of the same domain the relations different. relations It is a simple of with a set P = {Pn:nE N. regarding to be the meanings of predicates.

and Ext will in Dn to each such the right extensions assign language relation. a possible worlds model as was the idea in the of as a cluster of apv-models (indexed by W). and Ext assigns subsets of D to elements takes pairs (w. for a given language (on this aug But in the intended model please. Now we can turn to full-blown modal semantics.the very ones. elements is ised) (Note the assignment made of any particular the "meaning" of a index. and an element of Pn to every n In an arbitrary model. we . Dom. So in essence. (where Ext^ can be thought with our emendations above. On the syntactic our notion of a language (for these purposes) side. where V is a valuation interpretation to terms12 and (as prom function which assigns elements of Dom(@) of Pn to ?-place predicates. Ext). the question itself: again presents What of truth plain and simple? Formal theory up and running. 3. V ) for if. @ E W. (Dom(w). Ext). an for if will again be a pair (M. for example.11 For each index w. where W is a set of indices. @. P. P is as above. D.though of predicate course its extension can (and typically will) vary. P.pn) from W x Pn (for all n > 0) to subsets of Dom(w)". Pn will be a set of bona fide ?-place conception relations . independent V(P) P is thus fixed across apv-models in the cluster .) an element of D to each term of if. is a triple (So an augmented (apv-) model plain vanilla A valuation function V for a language if would then assign (D. pep talk. V).POSSIBLE WORLD SEMANTICS 361 "extension" function Ext that assigns a subset of Dn to each element of Pn. a possible worlds model M and a modal Given language if. mented of a model). of course. each Pn can be any set you place predicate. D is an arbitrary set. a formula of the form Pa will be true just in case V(a) G Ext(F(P)). at least. A formula D<p in particular.13) Given an interpreta tion (M. expressed of the by the predicates . truth for a formula of if inM relative to an index w will be defined in the obvious way. will be true inM just in case cp is true relative to all indices w. FROM POSSIBLE WORLDS SEMANTICS TO POSSIBILIA As in our plain vanilla semantics. Truth in a model will work in the obvious way. Ext^) is an augmented plain vanilla model is the restriction of Ext to {w} x U P). P. Dom of W. side. On the semantic we define a possible worlds model to be a 6-tuple (W. and truth in M simpliciter will be defined as truth inM relative to @. just needs a small modal upgrade: the addition of an operator D to our logical apparatus and the appropriate formation rule to our grammar.

@?. possible But what exactly is the argument here? The idea seems to be this. M' not just a piece of the actual world. @? is the actual world. i. of course. then. intuitive talk of possible worlds and possi ble individuals literally as the "sober metaphysical truth" about modal so ity. the actual world. for each world w and ? place relation pn E P?.e. of course. D?. exhausted A correct between theory of truth lays bare the connection language and reality. truth in the actual world. we have to tell an open and honest story about how itmaps the structure of the world. but the actual world itself in its In virtue of this. signs to each element of P' its actual extension. and perhaps perhaps the only member D? = D'.the set of all possible worlds. is that there must be at least as many possible worlds and possible . taking our informal.14 P' contains all ?-place relations and Ext' as partitioned).. where the domain D' consists of all the objects that actually exist.e. every possible world w has its corresponding apv-model Mw. This is the point at which Kripke-style semantic theories (of which ours is in essential a species) are taken to entail respects certainly to possible worlds and unactualized commitment individuals.362 have CHRISTOPHER MENZEL our credit of imagery and indefiniteness. The cluster of all such models forms the basis of a distinguished possible worlds model M? = (W?. really the set of all possible D? actual and otherwise. Dom0 to each world w the set assigns of all objects in D that exist in w. those objects that would have been actual had been w. and Ext? assigns. truth for interpreted nonmodal entirety. just as the actual world has its corresponding apv-model M'. is Ext?>. W? and truly . What reason is there to think otherwise? The answer. P'. can therefore be de Truth for interpreted modal languages generally as truth inM?. Dom0. P? = P' is objects the set of all ?-place relations. nothing follows concerning of W? and D?. i. Now. languages can be defined generally simply as truth in M'. P?. for all that has been said in the previous paragraph. or more informally.. of W? is@?. Insofar as we want our formal semantics to provide such a theory. All the intended models for interpreted nonmodal languages can be thought of as submodels of a single all-encompassing model M' = (D'. (appropriately thus represents. a genuine representation ifM? is to be considered of the modal facts. Thus in M?. fined simply the contents Strictly speaking. perhaps the only possible objects there are are the actual ones. Ext'). the correct extension of the latter in the former.

15 Hence. for the proponents of M?. Consider. is rather badly out of proportion with the ubiquity of strength.e. that model. iff there is a w E W? and a Dom^?0). The pro to the existence of of any semantic ponents theory are committed are appealed to in its account of truth. In brief. PLANTINGA'S HAECCEITIST ALTERNATIVE Thus to possibilia. Hence. for example. then. another model. The mandate of a semantic is only that the proponents theory tell some semantics between their formal story or other about the connection and reality. for (1) to be true in our distinguished there must both be other possible model M? than those that are worlds and other possible individuals than @?. languages requires a distinguished out must contain mere possibilia. But naturalness is one thing. But clearly. In many discussions it is difficult to separate the intuitive trappings semantics. here is the sort of argument we are after. So if (1) and its ilk are to be a<EDom?(w) true. for modal and second. Kripke-style from the formal semantics proper. considered 4. Possible world whatever entities It semantics appeals to possible worlds and (merely) possible objects. (1) is true iff there is a some object exists that does not exist in the world in which possible actual world. of) natural language. is that could have been different things (1) There could have thing. are committed to the exis follows that the proponents of the former tence of the latter. inevitability quite an other. might have had a third child. Its the argument semantics from possible world I think. as noted. specifically. it is not hard to appreciate the naturalness of the distinguished model M? (which we can call the standard model). and an a E D? such that i. been something distinct from every actu ally existing I. with another underlying . that can be challenged: that a theory of truth the tacit assumption first. there must be other possible worlds and merely pos sible individuals.. actual.POSSIBLE WORLD SEMANTICS 363 to yield an intuitively correct distribution of individuals as are needed sentences of (any interpreted truth values among the modal fragment that one of the ways then. in the argument There are two important assumptions its conclusion. the distinguished model singled Let us take up the second assumption of first.

essences. worlds are understood conception. i.. can speak Plantinga object having a property possible instead of an essence being coexemplified with r in a world. essences can exist unexemplified. entails that i.e.. Plantinga guished he identifies possible worlds with (in effect) maximally Specifically.18 The actual world @t that is true. worlds are identified with abstract concrete possibilia. if x has e. in his notion of an essence. or haecceity. This was essentially might Plantinga's metaphysics. then it is not possible for anything but x to have necessarily. According e is an essence in case (i) it is possible to for something property just have e. (ii) necessarily. however. D? = Dom?(@<>). "how can the actualist understand [(l)]"?16 as hinted at.17 On one Two conceptions of possible worlds dominate to be alternate physical universes. Plantinga's one that both satisfies the actualist proscription on model. Its truth conditions in the standard model. or another. it fol lows that for all w. Dom?(w) so. The real difference is found haps a to Plantinga. The standard model Since for the poses a problem are no mere possibilia. where a proposition/? is maximally just possible possible propositions. require the and an a E D? existence of a w E W? such that a E Dom?(w) and a ? Dom?(@<>). actualist there the set D? of all possible objects in the standard model is the very same set as the set of all actually = But as we saw.364 CHRISTOPHER MENZEL do equally well. insight. D? existing objects. Plantinga can speak instead of an essence being exemplified in in a world. being the essence in question it. two features These Furthermore. and (iii) whatever has e has it essentially. ? Dom?(@<>). like (1). like all properties. p entails q or in particular is the maximally p entails not-q. for Plantinga. for actualists. a world. in case it is possible and such that for any proposition q. essences to play the role accorded to merely possible objects in enable In place of the notion of a possible object existing the standard model.e. was to supply a different distin answer. possible proposition This does not in and of itself separate Plantinga's from conception the standard model is per of worlds conception (though the concrete better suited to the standard model). r in a world. Plantinga wonders. being such that the world (qua proposition) and in place of the notion of a is exemplified. on the other. Dom?(@?).. But since for the actualist. Now. true. (1) is intuitively i. objects of one sort opts for the latter view.e. and yields the correct truth values for sentences possibilia current lore. . exist necessarily.

properties by than one thing or. and others are not. haecceitist model shows that the argument for com Plantinga's to possibilia is not the only mitment is unsound.are not model To see this. properties like living somewhere and happy. in the standard model.19 its formal theoretical foundations and the entire framework. we have such properties Thus. and the function Extf that are exemplified a given world w and ?-place relation pn(EUPf (= the set of to the set of all ?-tuples of essences that are coexemplified lations) pn in w.20 is unduly baroque. actual and to the takes all re with such that the world entails that r and the essence coexemplified. conditions are identical with the truth conditions in its of the difference Unlike the standard model. married being being or other. such that a E Domf(w) the truth fl?Df Formally. i. purely qualitative and are. in the case of relations. Df. route from our formal possible world semantics One might wonder has been gained.. and being taller than every other man. demonstratives. or other referential devices. . for some tastes. note first that some proper unproblematic. being and on the other. the haecceitist model entails no commitment content. and e is is a world w and an essence e such that e is exemplified not exemplified in fact. more than one group of on . ties are most naturally picked out by means of expressions involving names. are still rather ened by paradox. the one hand.21 Intuitively. because to possibilia. as tall as that man. the set Df is to contain. or are logical "compounds" of. Call those married of the latter sort purely qualitative. While the though how much to be preferred abstract propositional of worlds is perhaps conception to the concrete conception. the standard model to a theory of truth.e. The truth conditions for (1) in Plantinga's actualist model (call it the are now straightforward: is true just in case there haecceitist model) (1) in w. though. It is threat it is not without its problems.POSSIBLE WORLD SEMANTICS 365 in question are Ext+). just in case there is a w E Wf and an and a ? Dom^?1"). The function Domf set of all essences in w. as living in Texas. seriously underdeveloped. general properties properties more and relations that could be exemplified relations. not the set of all possible objects then. In Plantinga's intended model M1" = (W\ @f. P\ Dom1. essences themselves the heart of the haecceitist Furthermore.but rather the set of all essences and otherwise exemplified then takes each world w E Wf otherwise. . to Xantippe. and being Reagan.

Here an essence earns its abstract status by way of comprising one or more qualitative compo If this compo component? the view runs afoul of the an there is an alternative. are universals. problems or being identical with Reagan. being identical with Reagan might be thought of as a nents. But what of the purely nonqualitative nent is supposed to be a property. but it is not at all obvious what that might be. it does not seem possible to some object without could be essential it being possible properties that they be exemplified is by some other.is that. and'haecceitism into there is another understanding of properties that Perhaps possibilism. of this variety collapses that too is stripped away. at the most basic level. they properties this understanding there seems no justification for purely nonqualitative essences nor logical at all. There is little enough compounds essences to distinguish from concrete possibilia purely nonqualitative save a thin actualist veneer. as Adams I tist framework has argued. at any rate. in the first instance. opposed. In light of this understanding of properties. or they have a purely nonqualitative Neither choice is especially happy. A traditional platonic understanding one . But that any collection of purely qualitative think. However.366 CHRISTOPHER MENZEL that are not purely qualitative whereas properties involve some things. Thus. properties argue the dominant are what diverse but similar particulars in common. a term If we whose semantic value (in a context) is exhausted by its referent. component. an appealing to be a property. Now. then . The first option seems to me subject to a simple conceptual short . have That is.24 is kinder to haecceities. The name essence like being Reagan. So maybe the second possibility fares better. the assumption doubted. Consider just noted. But on are. of general properties and relations.23 convincingly requires. 'Reagan' is (in a widely held view) a directly referential term. nonqualitative if every object has a purely qualitative then since essence. general. take the semantic value of a gerund is that a gerund contain metaphysical corollary to this view of reference a name expresses a property in which the referent of the name (as ing to a sense) is itself a component.I of properties would coming. returning to the say. This leaves two options for haecceitism: either reasonably essences are purely ?^?qualitative. component. issue at hand. since they are neither general themselves. exist necessarily. there is rea purely qualitative properties presumably son to think that there are necessarily as the haeccei existing essences.

.e. the claim that there are unexemplified essence to be an essence which. exist necessarily.25 If so. exist either necessarily not. requires either I or haecceities. then the of an essence had not existed. necessarily.POSSIBLE WORLD SEMANTICS 367 the relation of identity as its qualitative component. So although it is plausible just what the view as essences in this sense.. purely qualitative .e. McMichael troublesome find his own alternative semantics of what it is to say that an suggests that we alter our understanding individual might have had a certain property. on the individual that exemp if an essence is ontologically dependent sense essences makes lifies it. exist necessarily only if all individuals do. on his semantics. i. Similarly. a theory of truth for the semantics world semantics. it follows that essences all individuals have that. complex comprising as its nonqualitative and Reagan himself component. so it must be rejected. either (1) or there are things that are not actual. is not exemplified are But mere possibilia by a mere possibile. at least. it would of essences. simply possibilia as well. that. Thus. or role. is false. are like Reagan is that if contingent The trouble.. that Kripke alone in "maximal" property. or could have been. seem to or otherwise.e. (as it must be). i. objects though. it is by enough that there are such properties no means that that there are. individual components Assuming. if we wish to does in haecceitist model shun possibilia.26 exemplified was designed to avoid. actual components logical follow that essences are ontologically i. while we take an unexemplified only if by anything actual. dependent on those objects. But in that case. essences obvious or unexemplified. Kripke. MODAL REPRESENTATION What other options does the actualist have? Alan McMichael argues the confines of possible there are none within for the actualist. a fact about is not ultimately that Kripke might have been a carpenter not directly. exemplified an essence 5. its inevitable. So even though Plantinga's is not deed show that the standard model with its possibilist ontology own costs are too high for it to be a tenable alternative. the only way to is to hold that their salvage the view that all essences exist necessarily as Plantinga does. that if an individual component essence itself would not have existed either. Rather it is a fact about the at least that guy.27 However. possibilia. an essence.

for intuitions about de re modality is also unpalatable.2. {1. 4)}. non-modal case. in the right sort of way. P1? Ext^. I might draw a picture with four characters. a further option McMichael's remains for the pessimism. that some role "accessible" cludes the property of being a carpenter. This assumption I will in the semantics of modality.29 a homey example. Despite we will explore in this section. argue. or what I will call an for the model that maps the members of Di onto the ?jl.368 CHRISTOPHER MENZEL to Kripke's fact exemplifies. Pl the sets {10} (= (Pi)i) where Dx consists of the numbers = = and where and Exti(ll) and {11} (=(P1)2). 'Liisa'. figures labeled 'Chris' and 'Liisa' somewhat I could then label the entire picture 'The Menzels'. That is. and indeed rather better. (2. embedding members of my family. larger than the other two. and the members of UPi onto the property H of being happy and the relation P of being a parent of. as in the must be defined relative to a distinguished. and label them 'Chris'. stick figures and 'Galen'. and put a frowning face on the figure labeled 'Galen'. I might take a a model Mi = (Di. (1. Corresponding I could define instead a representation function. success is of course the existence The key to Mx's representational to reality. or pictures.28 my tastes anyway. not happy. To make this more precise. and so. and the like. which Iwill call the extensionalistfallacy: that. 'Annie'. to "the way things of embeddings like ?ii that are "faithful" are". Mi and i?i together represent in question no less. in general. that I have exactly two children.. is (at the moment) that my son Galen. we define an embedding ?x for an . sense: representations. for example. by way of counterexample.3} Ext^lO) to the labeling in the drawing {(1. 4). modal truth seems to me to be the crux of the intended model. 3). I wish to represent a few simple Consider Suppose facts about my family. with the say. 3). In that figure. I think that a lot of the unclarity semantics has surrounding modal are just that . role in viz. issue of ontological commitment that it is false.models arisen by not taking the idea that models are models in semantic models seriously enough.30 We can picture Mi along with an appropriate /xx embedding as in Figure situation the familial 1. more and construct abstract approach 1 to 4. the ordinary of a certain sort. This option challenges actualist that the first of the two assumptions in the argument from possible world seman tics to possibilia. than the crude drawing. (2. This move abandons strong and the semantics of names. and no one else. On the other hand.

is just the in tended* model that has the identity function as a faithful ^-compatible be embedding. (Of course. in our original sense. = function from ele (D.) M is an intended* model for ifjust in case it has a faithful ^-compatible embedding. M!.3} . . . The intended model for SE.3). . Thus. .4). Ext) to be a total one-to-one apv-model M ments of D to objects in the world. there might not be such an embedding. relations expressed by the predicates of ??. then. and from elements pn E UP to in intension. . for a given interpreted groundwork language SE and model M = (D.4)} \JPi: 11 Dn 12 3 4 A*i Hi H P The Menzels: Fig. P.POSSI BLE WORLD SEMANTICS Ext i?10 i ~ {1..2. To drive the point home (and to lay some for further argument).dn) E iff fx(di).3). {(1.(2.(2. 1. Ext). P. . let SEY an interpreted language designed to express well . the intended (apv-) model for an interpreted lan is just one of infinitely many isomorphic models that serve equally guage for defining truth. fJi(dn) stand in the relation /?(p?).(1. We relations say that ?x is faithful ?-place just in case . .e. say that an embedding for M is^-com fi to D) is patible just in case rng{?x\D) (i. {dx. Ext(pn) On this approach. the range of fi restricted the set of things over which the quantifiers of SE range in ordinary and rag(/i|UP) is a superset of the set of properties and discourse.

enough. we can avail ourselves to define of intended models where pointless Now . 'is a parent of.e.A} {(C. For a given model M = (D. 'Liisa'.370 CHRISTOPHER MENZEL ExU H {C. Similarly. 'Annie'. for languages and models in gen even all this is obvious enough.G)} UP2: P Do: C L A G ?2 A*2 H P The Menzels : G Fig. Vcomposed with fi) maps each term and predicate of SEi to its actual semantic value. A formula of SEY is true. Mx of course that distinction.A)..P.(L.G).L.A). V) let the intended* ?? for M. obviously to the model M2 of Figure 2. then. 2. interpretation such that Vo ?i (i. M2 the facts captured in our model M1# So SEXconsists of the unary predi cate 'is happy'. for plain vanilla semantics. and 'Galen'.(C. belongs intended* But there is nothing privileged about M2. and the names the binary predicate is an intended* 'Chris'. and.(L. but not the intended model. eral. just in case it is true in an intended* model under the appropriate intended* interpretation. Ext) of SEXand any faithful ^-compatible embedding I of SEXbe the pair (M. model for SEX. of course.

e. and was represented correctly fact of the is. in Figure 3. intended* models. by some figure as being in a certain state in the drawing.3). except that in I could have had a third child. I now want to represent not simple facts about my semantics. an X. this time with five figures rather than only in what way does this four. (2.4). modal fact about the entire drawing: the drawing could have been such as to represent how things stand with my family.5} {(1. no deeper modal (viz. for example. Suppose but simple possibilities. are no different when we move to more abstract pictures. there is no need to muddy limpid waters with the added com of embeddings. in virtue of the represent the possibility some merely possible fifth figure representing the way family member the other four represent actual members. As with at hand not in virtue the drawing.. 3. I want to depict not suppose.3) (2.5).4>. Once again.POSSIBLE WORLD SEMANTICS 371 10 UP3 11 Ext3 {1. Consider the model M3 as depicted i.3. I could draw a addition picture like the one above. the fifth labeled with.(1. and intended* plexity interpreta I think. Now truth. M3 represents the possibility of there actually being an embedding from D3 onto possible and actual or essences there could thereof.5)} Dz Fig. could have been much as depicted above.e.2. There matter than this. family. it could have some family been the case that each figure in the drawing represented and that each family member was represented member. I claim. at hand? Not. happy or not). Things.(2. I want to insist. Things formal models..(1. say. My family situation. but rather because family members. i. are utterly different when we turn to modal tions. that is. but how they could have been. have been a faithful embedding for M3 that mapped elements of D3 onto . how things are. it is in virtue of a Rather. M3..

P4. In virtue of the isomorphism. where ?x is completely completely faithful embedding case it is faithful.372 the members CHRISTOPHER MENZEL question of the representational This simple conception is. D4. We might well rest content with possible play. that there really is a standard model M? replete with inhabitants. as M? truth for modal for defining languages. is the set of all individuals. or family members. and McMichael. as well as in Plantinga's us a full-blown world. allow that like Plantinga the not analyzable in terms of primitive worlds. of course. dality is how to specify a possible The problem with this image. pure sets. First a preliminary the modal facts pertaining only to certain that represent worlds models there are subclasses of all the individuals stars. however. of course. At possible possible worlds and their (in general) merely same time. @4. for an apv-model. An initial image might be useful for setting the modal pictures for the moment the course of our argument. modality an equivalent statement statement about thus yields at most modal is to be considered but no such statement worlds and their denizens. M4 repre sents modal reality no less than M?. how possible world semantics the key facts. Consider the modal of M4 to M?. imagine. but constructed (W4. and that the domain at each index is the gives in (exemplified in) that existing (or haecceities) we will use the notion of a an analogous world.32 and rng(i?\D) set of all individuals of my family been actual. Ext) just in set of all ?-place relations. of modality. Every is primitive. The standard model drops away. Now. but M4 endures with only pure sets as accurate a representation of mo in place of worlds and possibilia. vision of modal and take the possibilist's reality that is to say. M4 would do just as well say. Ext4) isomorphic only out of unproblematic (relatively speaking!) necessary beings. while retaining your belief in the primitiveness reinvoke your skepticism about worlds. Suspend your skepticism at face value.31 To accommodate idea. clear how worlds model like the M4 of our story in a way that makes to the standard model the model represents modality without appealing This is where the ideas above come into and its untoward constituents. structurally. rng(ii\ UP) is the faithful for M=(D. = a model an analysis statement. that would have existed had the possibility in . note. as before. The idea in the standard model. P. is that each index haecceitist model.and with to only certain of the properties items in the fridge respect and relations that hold among them. Dom4. though. powers of models to understanding to my mind. .

if things had been at all different. that is to say (and that is only to say). individual would have existed. there is an index w W such that A(w) is faithful. In such a case. in the haecceitist model it is by postula can be exemplified that the same haecceity in distinct propositional ting worlds. rors the idea that W? is the set of all possible in the standard model worlds . P.something like the following: no matter how things could have turned out. and that the same element of our model would necessarily have represented the same entity. had it existed. Ext(a) of A must obviously be faithful. To the contrary. we want to existing or relation) property. Ext) have to meet? To begin with. if things had been otherwise. represents as such that. (unless otherwise I will mean faithful'. necessarily. (Ci) has a faithful embedding \x such that. D. @. This does not picture the facts in the right sort of way . y~l{b) ?x'~l{b). we require (at least) that: precisely. w' E W. P. tie). stated) by 'faithful' 'completely accurate possible worlds model So what conditions would a modally A = (W. since A could meet the condition = even if Dom(w) D Dom(vv') 0 for all w. A(@) there is a w E W such that (i) A(w) has a faithful embedding = yJ such that for all b E rng{p) Pi rng(ii'). and so we need y(e) = to convince . condition. necessarily. Now. then no actually i.such a model individuals as "modally fragile". Dom. We can achieve the same "transworld" effect and at the same time avoid these arcana by appealing More instead to embeddings. the "base" = (Dom(@). apv-model A(@) a condition we want a "comprehensiveness" that mir Next. would be distinct from any element of Dom(@) that represents him in fact. require that the very same entity (individual. element of D that would have represented say. and (ii) for all e E UP U (Dom(@) H Dom(w)). But this will not suffice as it stands. In the standard model this is accomplished that the same individual can exist in by positing more than one possible world. (Ci) is essential to my account. any Pete Rose.e.POSSIBLE WORLD SEMANTICS 373 faithful if there is a completely Also say that M itself is completely That noted. no would have been represented the same element of our model by matter what. henceforth faithful embedding for it.. there must always be some index w EW such that the apv model A(w) would have been an accurate representation of the world.

.contains lots of contingent the pairs that make beings. Rather. though.e. of the condition is.e. elements objects. We have supposed A to be constructed sets or the like. however. ix~\b) = fx'~l(b)). we can simply rephrase (Ci) as a slightly less condition: formal (though..sets functions of ordered pairs. not easily generalizable) has a faithful embedding fx such that. just to be extensional embeddings ?x for A(@) of the sort required by (Cx) is going Thus.374 CHRISTOPHER MENZEL ourselves that A's satisfying it involves us in no unacceptable metaphys out of pure ical commitments. as we shall see. you and for instance. is referred to Note. wholly transparent and cannot yet rest content. every to it by D Dom(w) the same element of Dom(@) mapped as is mapped to it by ?x in fact (i. for things are more complicated than this. no matter how some apv-model have gone (i. (Ci) seems to fly in the face of this. he would have been spoiled' if the antecedent had been true. I take presumably ..e. y' A(@) have been mapped to in fact. This above and (ii) every element of UP U (Dom(@) D Dom(vv)) would to the same object by ?x' that /xmaps it rendering reproach. so the elements of W and the apv-models A(w) all exist necessarily. necessarily). I think. Yet me. What might give one pause is this. in the former case. things might would have had a faithful embedding yJ such A(w) (w GW) have that (i) every actually existing object b that would existed b E rng{y) Pi rng(fji')) would have had (i. the terms 'mg(/i)' and '/a-1' in both cases are parts of more complex terms used to describe. for the occurrence of the terms 'rag(tt)' and '/x-1' within the scope of the necessity operator seems to imply that rng(y) and yS~l (hence ?x itself) must exist necessarily in order for (Ci) to hold. a subset of and in the rng{y) that would have existed had things been otherwise. to those in our model A that would have been mapped latter.the set of second elements in it up . that neither rng(jx) nor tt-1 simpliciter in (Ci). entails that John would have existed To see this more clearly. no embedding to be a necessary being. The use of these more complex terms thus no more entails that have existed even if members of themselves would rng(/Ji) and y~l 'If of 'John' in the statement rng{y) had not than does the occurrence John's brother had been an only child. Condition that every individual is necessarily (Ci) ensures We . since its range .

POSSIBLE WORLD SEMANTICS 375 tied to exactly one element of the domain D of our model A (relative to a given embedding But to get things right. (= tt'_1(being and the very same element y'~l{c) of D to c.that is to say (and that is only to say). By (Ci). 03x(Txm sopher. and. if things had been different. y' would have mapped the same element p of UP to the property of being a philosopher that does in fact. 03x(Txm has a faithful embedding ?x such that if I had had a third child c who was not a philosopher. That is. of another model A(w") that could have had an (Ci) is the existence for A(vv') above except that y!' similar to the one described embedding to my third child the element of D that y!f would have been mapped would have been in the extension of p in A(w") after all. That is. Ext(/t'_1 same element e of D to me that y does in fact. . ensure that the same is true (to speak with the vulgar) for all possible . we must for A(@)). y'~l y! (being = a philosopher) and such that y'~l(c)?? ?x~l(being a philosopher). But all this is the truth of the conjunctive that I could have guarantees proposition and I could have had one had a third child who was not a philosopher a Px). tt_1(being a philosopher) have mapped (= y'~l(me)) /t_1(me) a philosopher)) to the property of being a philosopher. OPx). particular. had there been things other than those that exist in fact. A(@) child who was not a philosopher. have been represented by the A(w').the fact that I could have had a non-philosophical . y' would have mapped the (being a philosopher)). since my third child would not in this scenario have ?x been a philosopher. we get to representing this fact given the closest However. the element of D that would have been mapped to her by ?x'would not have been a member of the extension of p in true that if I had a third child who was not a philo a ?\Px a she might have been one nonetheless. For A to represent the a ?\Px) a OBx(Txm who was. we must ensure that objects and in the same would have been the case had things been otherwise. what. I could have had a third a ?\Px). just as we require that I could have been represented by the same element of D no matter we also must require that. 03x(Txm above fact correctly . any object that would have existed could necessarily same element of D as well.we third child such that she might have been a philosopher should it is also have to insist in addition that the embedding still ?x" for A(w") would to me. one of our apv-models A(vv') would have had a faithful embedding such that /t'_I(me) = ?A_1(me). Now. An example will make the matter clearer.

first. l?~l{b) yi2l{b). What this form depicts is the possibility of an object which is itself the subject de re of a further possibility. (i) for all b E rng{y) fl rng(/xi). so to speak. that. fx?\b) = fx2\b). ogous to.it involves a second modal operator (Ci) is. and such that.e.34 The best we can do is something like the somewhat stilted reconstrual. and for y(e) y2(e). But itmight well be. that there is no such object in fact. a faithful embedding such that A(w2) has tt2 such that = and for for all b E rng(fx) D rng(y2). W (/') all all all as a subjunctive paraphrase (C2) does not afford as straightforward in natural language The reason for this is that there is no operator (Ci). = e E UP U (Dom(vvi) fl Dom(w2)). . the following more general version must hold: of (Ci) (C2) tt such that. b E r?g(tti) H rng{?x2). and (ii) for all e E UP U (Dom(@) H Dom(wi)). for our model A to capture possibil ities that involve nested modalities. fx?\b). whose is anal semantic effect on sentences within a nested modality but more delicate the effect of 'actually' or 'in fact'. no guarantee There is thus that an apv-model contains that represents the relation between the two that make up the larger possibility correctly. and (ii') for = e E UP U (Dom(@) n Dom(w2)). no matter has a faithful embedding might have gone (i.. correspond to the two levels of nesting: tt such that. The indentations following. necessarily. no than. necessarily). bumps us up only one level of modal nesting operator rather than all the way out. as in our example.and nested within the scope of another the fact that a variable that is quantified within the scope of the wider operator into is quantified the scope of the narrower. y~l(b) = there is a w2 E y(e) tti(e). necessarily. has a faithful embedding A(@) a wY E W such that A(n>i) has a faithful embedding there is = tii such that.376 CHRISTOPHER MENZEL When we look at the logical form of the above possibility we see that what most saliently distinguishes it from the sort of case that motivates its modal degree33 . our model A. tti(e) y2(e). possibilities To make the above precise. in satisfying only (Ci). (1) some apv-model A(wx) would have had a faithful A(@) how things embedding tii such that (i) every actually existing object b that would fl have existed would have had the same element of Dom(@) . then.

This calls for the following generalization of (Ci) and (C2). of course. . = y?e) yfn(e). . and for that.e. above example of higher degree . necessarily. for instance. and such ti0(e) yi(e). .I could have had a third child who was not a philosopher and but might have been both a philosopher the mother of a child who was not a violinist but might have been. necessarily. A(wn) has a faithful for all yn such that for all i<n. and every element of by ti2 that y maps Pi Dom(w2)) to would have been mapped UP U (Dom(vvi) same object by ti2 that tii would have mapped to it in the been the above case. is. straightforward extensions of the possibilities. every object b that would have existed b E rng(y) Pi mg(tt2)) would have had the same element n Dom(w2) to it by y2 as is mapped of Dom(@) mapped to it by tt in fact.. can be given into the scope of yet a third.. embedding = b E rng(fXi) PI rng(yn). yT\b) and for all e E UP U y~\b). (Dom(w/) H Dom(wn)). no finite bound on the modal degree of such There are. there is a wn E W such that that. ti2 for (2) there still would have been a faithful embedding some apv-model A(w2) such that (i) every actually existing in this case (i. .e. Pirag(ti2)) E rag(tii) b would have had the same element fl Dom(w2) mapped to it by ti2 as would have of Dom(vvi) case to it in the above by mapped Pi and element of UP U (Dom(@) tii every (ii) to the same object would have been mapped Dom(w2)) it to in fact. a ?\Px a 0(Px a 3y(Mxy a ?\Vy a OVy))). A(w>o) has a faithful embedding is a Wi E W such that A(wx) has a faithful embedding yx such = for all b E mg(/x0) PI rng(yi). . and every object b that would also have case in existed the above every (i. y?l(b) yS[l(b). . a vari Here. 03x(Txm There is quantified within is the scopes of two modal operators. Similar extensions quantified for any finite degree. and (iii) no matter how else things might have gone. = all e E UP U (Dom(@) Pi Dom(wi)). where @ is w0: able that (C?) there ii0 such that. Dom(wi) mapped Pi Dom(wi)) would and (ii) every element of UP U (Dom(@) to the same object by tii that y maps it to have been mapped in fact.POSSIBLE WORLD SEMANTICS 377 to it by tii as is mapped to it by y in fact.

mapping dom) to the property being a prime number. that there be a specific faithful embedding such that (C?) /x* for A(@) in for fx0. A might satisfy (C?) for all ? relative to y*. that A(w') would have been incoherent relative to y!. there is nonetheless model A(w) that ismodally incoherent relative to ?x. and this (so far as I can see) is not guaran teed by (C?). for all and eEUPU (Dom(@) y~\b). Such models would be in a sense modally incoherent relative to /t*. or that. Condition (C?) ensures that there will necessarily be apv-models with faithful embeddings that (loosely speaking!) would have agreed with other possible where their domains and embeddings For A to be a fully adequate representa ranges would have overlapped.This specter needs to be banished. tion of modality. that is. there would have been another model e to things other A(vv') that could only have been faithful by mapping than c. could be faithfully embedded only by or /t*_1(wis say. (CO) For all w E W. deeply nested the modalities One last condition. suppose. This suggests the prospect of A itself being modally in the sense that.378 CHRISTOPHER MENZEL For our model A to represent modality then. y*~l(b) H Dom(w)). /x*~x (Pete Rose). assuming A satisfies (C?) (for all ?) relative to y*. to a garden vegetable.We can holds for all natural numbers ? when ?x* is plugged then be sure that A pictures the modal facts correctly regardless of how within those facts might be. there would have been an apv-model A(w) that had a faithful some element e of D to c. for example. for example. even though it may well contain apv-models that could not be faithfully embedded at all. relative to /t* (or some similar embedding). But y! that would have mapped embedding suppose. Once again. By (Ci). found with nested As we since possibilities (Ci). though. potential modal had a third child c. we need to be sure that each of its apv-models in a similarly congenial way relative to could be faithfully embedded other possible embeddings. Suppose there would have been a similarly incoherent model for . for any embedding incoherent ?x an apv for A(@) that validates (Cn) for all ?. what is required here is that. we need to say more. we require accurately. How so? At the least complex level. had that been the case. to more modalities of involving give rise complex examples I could have incoherence. it is possible that there is a faithful embedding = ?xfor A(w) such that for all b E rng(/x*) Pi rng(y). though. = ix*(e) ix(e).

= and (ii) for all vx. . which one might find this sort of incoherence. = e E UP U (Dom(w/) Pi Dom(vi)). and for all = e E UP U (Dom(@) H Dom(vvi)). for all b E mg(y?) D rng(yn). the generalization of (CO) is within (Note that (ii) below the scope of all prior modal operators): any ation than even (Cnm) A(wo) there yx has a faithful embedding ?xqsuch that. By H Dom(vm)). (b) = y-\b). there y!e\ fx?(e) is a faithful embedding yx for A(vx) such that. for all i< ?. . Accordingly. = for all b E rag(/i?) Pi rag(yi). ?xil yil(b). . Integrating this generalization (CO) with (C?). . possibly. has to be generalized for all ?. and second.e. relative to some faithful embedding of degree ? correctly. for all e E UP U (Dom(^) and for all e E UP U (Dom(v7) H Dom(vm)). for all ym\b). first. operators. . embedding that for all /< ?.POSSIBLE WORLD SEMANTICS 379 for A(w). y*. necessarily. y[l(b) e E UP U (Dom(w/) Pi Dom(w?)).. fx?(e) . . and pos yx(e). and suppose also that a similar situ faithful embedding for A(@) would have arisen for any faithful embedding other still have to be counted modally Then A would incoherent. y?l(b) . for any ? and m. possibilities course similar examples could be generated by yet more deeply nested no finite bound on the depth at there is. we arrive at the following (admittedly hairy) overarching schema. Hence. and such that. To deal with this case we should have to generalize (CO) to a . ?x. b E rng(yj) D rng(ym). ym(e). there is a ?xi(e). say that A is (modally) adequate relative to y . j < m for all b E rng(?x?) D rng(ym). there will be no modal incoherence of degree m. that.A will represent any possibility relative to that same embedding. there is a faithful ym for A(vm) such sibly. and for all yil(b) y^l(b). vm E W. more generally. of degree two. . as with (Ci). then. if it had satisfied (CO). necessarily. we ensure that if our model A satisfies (C?m). yj\b) = = ?x?(e) y-(e) integrating (C?) with (COm). . (i) is a wx E W such that A(w:) has a faithful embedding = such for all b E rng(/x0) H rng(yi). and for all ix?\b). . . ?x{e) wn E W such that A(wn) has a faithful embedding yn such = that for all i< ?. But of i.that rules out incoherence from possibilit (C02) say arising principle ies involving a nested modality. Jmie). from possibilities for any faithful arising embedding /x for A(@).

and rng(y\UP) is a superset of and relations expressed of SE. Unlike models the representational of an adequate discussed above. (Since we have been considering only completely are unrestricted. The intended* interpretation relative to A and ?x. represent stract objects of some ilk . We have laid worlds model A = (W. its "worlds" world" @ an arbitrary member of W.35 its "actual W a mere out the conditions for a possible to be modally to Ext) adequate. under that there the appropriate intended* are intended* models I assume hence interpretation.380 CHRISTOPHER MENZEL just in case (Cnm) holds for all natural numbers ?. assume that the quantifiers will of SE and hence that is the set of all individuals. appropriateness in any intrinsic properties of its constituents. one of the chief goals of modal semantics For Plantinga and McMichael. ADEQUATE MODELS AND TRUTH CONDITIONS Let us now take stock. but only in a certain rather as a whole. then. lie not model. a language. not just an exten is to provide for modal truth conditions statements. and A = (W. @. is the pair (A. m when ?x is plugged in for ?xq. and say that A itself is (modally) adequate just in case it is relative to some embedding. A formal theory of complex modal property of the model modal truth demands nothing more. the set of properties by the predicates we faithful models. Dom. then.D. V) such that Vo ?x takes each term and predicate to its actual semantic value.) A is an intended* model rng(?x|Dom(@)) in case there is an ^-compatible for ifjust ?x for A(@) such embedding I for SE that A is adequate relative to /x. As in the nonmodal case.we have set of indices. is SE-compatible ?x for A(@) say that an embedding case rng(fx\Dom(@)) is the set of things over which the quantifi just in ers of Grange in ordinary discourse. the modal facts accurately. As before. @. A proper semantics will describe each sentence cp in its turn what the world must be like in order for ? But there is perhaps more . for any reasonable modal lan 6. for sionally correct theory of truth. be such Dom. are familiar ab Its constituents supposed pure sets. P.P. Ext) a possible worlds model. to a theory of truth than truth simpliciter. capacity and its corresponding for defining truth. adequate Truth for interpreted modal Let SE languages is now straightforward. its "possible the intended individuals" D an arbitrary set. D. a formula of i^is true just in case it is true in an intended* model forth guage.

37 rather. models the truth conditions only intended* models. existing (2) is true.38 model-independent is correct. @. Ext) be such a model. truth conditions. adequate models.POSSIBLE WORLD SEMANTICS 381 to be true in terms of the constituents of the semantics' intended of an arbitrary adequate model the constituents model. Consider the genuine. (Ci. something distinct from every actually Possibly.. correct whole. just in case it is true in some intended* a model (hence adequate) (for an interpreted language containing of (2)) under the appropriate intended* So counterpart interpretation.e. would have had to have mapped ?x hence it would have mapped it onto something that does not exist in fact. Hence. then there e E Dom(vv). A that figure into the truth value of a given sentence cp in A in and of on its truth or falsity simplic themselves have no bearing whatsoever as a is a function of the model iter.i)).. D.. Thus. is a w E W such that forborne i. its truth value. total functions. But since embeddings are mapped e onto something.0)) that if there could have existed something distinct from every actually existing thing. statement (1) (1) again. Since there are no intended this objection Strictly speaking. By similar reasoning. But that just means we have to look a little more carefully. By (CO) (i. Dom. P. (C0.36 By contrast. have been objects distinct from every actually could object. and let y* be an relative to which A is adequate. this is true only if there could e?Dom(@). on our theory. then with y* .e. could not have e onto any actually existing thing. precisely.. y.e. as noted. have been a faithful embedding for A(w) that would have "agreed" ?x as far as it would have gone. while providing extensionally are unable to provide truth value distributions for modal statements. for the sentences of a language generally cannot be identified with the truth conditions in some distinguished for the sentences model.e. e (?Dom(@). on our approach. let A = (W. There existing or perhaps more (2) i. there exists thing. Then (2) is true in A (when embedding all is said and done) iff there is a w E W such that for some e E Dom(vv). is true only if there could have existed (2) something distinct from every actually existing thing. we can show that it follows from (Ci) (i. in particular.

Had this occurred. . from every actually existing thing.e. Quine. ne') E Ext(F('human'). G? Ext(F('human').382 CHRISTOPHER MENZEL (2) is true in A. ? Ext(F('exists'). Again. necessarily. conditions. (3) is true in A only either Quine does not exist or he is human. Thus. = = such that. then he is human. possibly. it would have agreed with ?x* on the values of F('exists') (= /^"^(exis and. But by assumption. w) or F('Qui either F('Quine') Thus. necessarily. viz. relative to which A is ad be such a model. (3) is true just in case it is true in some intended* model under the right intended* So let A = (W. (3) is true iff. Dom. if Quine exists. truth con The point is that. Thus. @. nonrelative truth there exists something distinct from every actually there exists something thing' is true iff. F('human') (= ?x*~^humanity)). w) V('Quine') either Quine But if that is true. contradiction. (4) 'Possibly. to be explicit. if Quine (1) and (2) are true iff it is possible from every actually existing thing. (3) is true in A iff for all w E W. where existence. does not exist or he is human. reasoning. V) be the appropriate viz. In particular. w). since Quine would have tence)). w) and V('Quine') for all wElW. Fo/t('exists') F%i('Quine') = and Vo /t('human') humanity. using both (Ci) and (CO). hence true. D. iff. By similar if. then he is human. F('Quine')) E Ext(V('ex would have been it follows that F('Quine') faithful. then it follows that. interpretation.. i. existing distinct and straightfoward.. Then Quine could have existed without human. when we move beyond model-relative we have imposed on ditions and think about the further conditions we do in fact get adequate models. let ?x* be an embedding and let (A. (=f y*~l(Quine)). ists'). being there would have been a faithful embedding y! for some w E W that would have agreed with ?x* as far as it would have gone. P. ? Ext(F ('exists'). Ext) interpretation. that there exists something distinct a second example: Consider (3) Necessarily. w). necessarily.. Since y! existed (albeit as a nonhuman). necessarily. either Quine does not exist or he is human. either or F('Quine') E Ext(V('human'). then by (Cx). the one equate. For suppose not. we show the converse. exists. w).

x has e. in its guise as (2): Plantinga's (6) there exists something distinct from every actually existing thing' is true iff there is a world w and an essence e such that e is exemplified in w and e is not exemplified in 'Possibly. been lost. for (2). whatever and (c) neces necessarily. Modal to extensional But is operator gives way dutifully quantifier. (5) 'Necessarily. that the resulting truth conditions "literal explanation and articulation of our modal provide notions" But it is hard to see the advantage. p. if w is true. that is otherwise lacking. if such that . there any gain. Both Plantinga and McMichael suggest that their semantic theories yield truth conditions that are somehow more "illuminating" than the rather mundane truth conditions above. then so is not-p. these truth conditions .39 there is likely to be unease. either. 'every' as primitive is simply to accord the same status to 'possibly' and dality 'necessarily'. then he is human.the could things be otherwise? Modality of intensional very paragon in my story all has been an irreducible and essential primitive ity in the same semantical com Thus.40 Consider analysis of (1) once again. are not extensional. of course. we need to understand what all is said and done. But how Now. (b) for something (a) it is possible has e has it essentially. necessarily. or. and (ii) for all propositions (i) if w is true. I count the modal operators along. we offer no deeper analysis of and connectives: pany with quantifiers or 'not' than the ordinary meanings to take mo of the words. when ditions look like this: (7) 'Possibly. then he is human' if Quine exists. if Quine exists. of an analysis of the modal might in terms of a quantifier over worlds. fact. the truth con there exists something distinct from every actually w such that existing thing' is true iff there is a proposition it is possible that w be true.POSSIBLE WORLD SEMANTICS 383 is true iff. necessarily. then it is not possible for anything but x to sarily. any increased illumination of the modal facts? The lack on the right-hand of any overt modalities in the truth conditions side foster an illusion of explanation. and (iii) there is a property e to have e. a sense that something has Nonetheless. then so is/?. But in order genuinely operator to understand the truth conditions worlds and haecceities are. necessarily.

My claim has only been that. rather. the haecceitist no less overtly account. itmight be felt that something is still missing. are true or false. either haecceitist there is perhaps positive Indeed." i. Rather. conditions of the sort we have given do not really account for the truth statements like (2). then something have it. in what exists... in a "good" to mere semantical account of truth. The pretheoretic same applies no less to the role semantical truth conditions for (2). there must be objects to serve as the "ground" of modal in virtue of whose properties entities truth. there is no deeper illumination. It seems quite reasonable that what is is something or necessary for that matter.e. introduce new and unfamiliar into our loss. whatever modal facility it takes to grasp (2) itself.41 to the contrary. they to this. and (v) nothing has e in fact. be it haecceities combinatorial and their properties. of the relations that could obtain between the most basic elements or the power of God. we have no other use for them). McMichael seems statements and relations our modal to have something like this in mind when he claims that. ontology of course. Nor intensional than on our alternative is in the structure of the truth conditions that might there any difference some sort of conceptual constitute oper gain. that truth Nonetheless. if we are going to take modality in the broadly logical sense at face no reason to ask for any more than a homophonic value. no nested modal we need at a minimum to grasp the truth conditions ators. e. or role semantics. the structure abstractly characterised in the formal semantics proper. then there is . has truth conditions for (2) are When the smoke clears. e.42 A similar thought seems to lie behind Plantinga's insistence that the actual ist "must appeal to essences" in order to give acceptable truth con ditions for (2). since we must objects just to get those semantic theories off the ground (assuming. or articulation to be gained from of our modal notions explanation. must in some sense be grounded possible. that modal truth so understood is value of modal not sufficiently rooted in reality.g. But I have not been so am physical universe. and (iv) necessarily.384 CHRISTOPHER MENZEL if w is true. "there are real entities [as opposed formal constructs] which exhibit the given semantic structure. First impressions then. the way There bitious as to try to answer that question.43 We are then able to say that it is because some possible world are related and some role or haecceity conceived) (appropriately are that (2) is true.

is for things to be as the statement as in the nonmodal just truth this answer to the question of the form of modal says. to truth conditions provide a rich theory of truth and model-independent I have tried commitments. are otiose. But no surprise. whatever ground of modal a different question. framework in which in full these issues can be fruitfully addressed This compliance with actualist scruples.silent as it is on of the ground modal truth conditions questions . Granted.45 without . What the above promises is a Montague though.POSSIBLE WORLD SEMANTICS 385 to be true statement for a modal truth conditions: theory of modal case . a model cut straight from the world Since M is an intended model. it is truth. and but only at great ontological theory of truth for our modal discourse. for instance. itself. for the sake atively naive picture of the nature of formal representation of brevity and ease of exposition. that it can yield a genuine it has only heuristic value. any untoward metaphysical I have. of of 7. what more direct and revealing account could one hope to find of the ground of its truth value? But modal truth needn't be so cashed. or of the representation of less than at all was said about the issue total alternative Nothing possibilities. Freed from their grip in modal intended models semantics.44 or about problems of "granularity" in representational of systems. then the truth conditions for a sentence S must be the truth conditions generated directly inM. or the specific nature of representation in and change. To the contrary. adopted a rel is of course just a beginning. idea that anything more // modal be cashed in terms of an intended model M. Little was said about the problems of "total" models like those above.is to be modal truth expected. not answer about the metaphysical does conditions any questions that might be. rightly construed. one can take the semantics cost. temporality I believe. Grammar. account less deep) than a homophonic nothing deeper (and nothing . CONCLUSION two views of possible world to steer a course between are often taken to be exhaustive: on the one hand. that semantics that on the other. Blame must be laid at the feet of the extensionalist fallacy for the truth must should seem necessary.

15 (1986). to the actualism without that undergirds the view I will be contrary property existing. Abstract D. forthcoming 8 J. can be distinguished with the same domain. of Philosophical Logic 2 A. It is dispensible. 116. Individual Actualism and Three-valued Modal Part I: Model Logics. p. J. (1959). 10. pp. of Philosophical Logic The Concept of Logical Consequence. . III. Logic (1983). 147-8. ch. to do. See. 1-15. esp. This assignments 6 Hodes. Hintikka. 369.. Reidel. Oxford. 259-60. pp. Linsky: Attitudes'. Reference simplifies Journal matters for our purposes here. p. and Modality.)" 24 for Propositional in L. 97 (henceforth NAMS). pp. 14 To avoid problems with (and perhaps there objects nought even abstract Dordrecht. of Richard Montague. objects cardinality to be among all the cannot be considered objects generally) are. 12 too is an actualism-inspired that This restriction since it seems reasonable constraint. For those who it is perhaps best to insist on mathematical platonism. clusters Note also developing. the informal analysis which motivated is these definitions if and only if it is true in all 'possible worlds'. p. p. 13 R. we cannot name objects that do not exist. mathematical and nonwellfoundedness. only relative 11 Most modal own. Philosophy 4 of D with itself. ch.): 1974. Perry. But. Thomason (ed. Barwise and Shaken Attitudes'. and Press. J. 'A New Actualist 12 Modal Journal of Philosophical Semantics'. pp.. Gupta. facts the modal logic that best gets the modal will I am not arguments. 'Modal Logic and Truth'. Kripke. is not (It to analyze for our present the concept of a 'possible world' any purposes 'A Completeness S. 441-472. Hawthorne. Notre Dame Languages: 26 (1985). 3 and J. New Haven. Zalta: 1983. 3.e. e. p. from Harvard Press. that for convenience structures so these an accessibility relation including to my This is not essential be S5 models. / to be defined to bring Since I have stipulated there is no need variable into the picture. McMichael. forthcoming. for those with qualms. are essentially in Kripke structures his quantificational model semantics. theoretic Journal Semantics'.g. including Kripke's to subsets of Dn. Hanson University in Intensional A New Approach'. and Linguistics 'Shifting Situations 8 (1985). also J. of Quantification Selected esp. on W in this definition. models "qualitatively" to a language and corresponding valuation function. 'The Proper Treatment Compare in R.e. cit. right. Such models i. Montague. the nth Cartesian product 5 on terms generally. Hodes. 369. See my Logic'. Section 2. Compare 7 See A. Theorem in Modal Journal of Symbolic Logic'.. though I think S5 is probably 'The True Modal Journal of Philosophical Logic. pp. Objects. 1971. Etchemendy. in Ordinary English'. stipulate only that Ext map such pairs semantics. Papers and E. happens that permits of plain vanilla models indexed in a way "indiscernible" models. University 'Semantics Oxford 7 (1978). W. "Validity Journal of Formal Logic 9 Thus Kripke: "The basis of that a proposition is necessary necessary Logic 10 As further.386 CHRISTOPHER MENZEL NOTES 1 H. this is to allow that something could have had a intuitively.. I. op. (as is the restriction though on Ext). as objects of higher think of sets and numbers with which our first-order theory has type. Formal Philosophy: Yale University Press. 247-70.

7. Philosophical see esp. and verbs (such as 'Leibnizian' and 'pegasizes'). APW). Journal of Philosophy the property being taller than every other man (i. Reidel. see K. p. Bringsjord. Not Forbes: that there Cornell 1985. van Philosophy. Zalta: that does not appear is found 1988. without sufficiently expressed. Pollock a rather it makes elaborate is a set of that every state of affairs has a complement. University of Possible Worlds'. Analysis P. though [AjcMAN(x) a x ? y ^TALLER(x. McMichael. in Plantinga's framework at that one can show there are any propositional worlds possible a proof. 'Plantinga in J. of Possible Worlds'.g. 23 49 (1981). actually Plantinga possible at hand. 28-38. first. Fine and A. Loux and Possible Worlds'. the idea of complex see. pp. 18 See A. The Possible (ed. a Vy(MAN(>>) tive. 3-41. 60-1. N. pp. proper adjectives uses of definite and referential R. and his 'Actualism 'A Problem Review 92 for Actualism about Possible Worlds'. 598-603. See also A. Press. 121-4. of Necessity. I take it to be purely qualita y))}) of the property the relation being taller it is a logical compound being a man. 46 (1986). identity relation. since and 76 (1979). esp. Prior: the . and second. 186-191. Oxford. cit. C. 16 WORLD SEMANTICS 387 See A. Bealer: being logical compounds properties and Concept. e. p. 1979. Grim's A Reply 'On Sets and Worlds: Grim. An alternative world theory to suffer from these shortcomings in E. Possible Worlds'. (1983). See P. Tomberlin and P.e. if and if it could be is purely property only qualitative as in a language the aid of such referential devices rich.. MIT Press/Bradford Books. pp. of Intentionality. 206 (henceforth Press. Oxford. the Actual. Clarendon 1982. between isomorphism propositions all. in 'Two Concepts Midwest Studies 1986. For more on the view that some intensional Compare Reagan's singleton properties entities are ontologically dependent on contingent objects.): Alvin Plantinga. Menzel. pp. offers a proof for a proposition that entails recht. Analysis B of the Principles. Notre Dame Journal of Formal Logic 30 (1989). Menzel." and Primitive 22 even Thus. van Inwagen D.. but be an 19 45 (1985). than. 4.): Ithaca. Compare 25 set {Reagan}. and that assumptions: see his 1985: on Possible states of affairs that obtain. and Thisness'. 21 Compare Logic and Massachusetts. G. p. 24 PAPW. Quality Press. Oxford. Identity'. of Minnesota Volume XL Studies Press. The Metaphysics is anything particularly of Modality. it depends upon is not proved. 46 (1986). two dubious there all see C. pp. 'On Set Theoretic 64. indexical proper names. since he takes there to for the matters is unimportant the difference and states of affairs.g. pp. Synthese Op. in M. is in Appendix which is akin to the paradox Russell reports paradox also 20 easily reconstructed It is not even clear offers as it stands. Dord Worlds'. See S. 4 Press. states of with maximally identifies worlds NN). pp. Inwagen: in Essentialism. and 17 Plantinga. wrong with Clarendon this. 'Actualism Plantinga (henceforth affairs. 57-61. Adams: "[A] the Metaphysics ch. 'On an Unsound Proof of the Existence flawed. 'Primitive Thisness expressions descriptions. (eds. For more on of other and relations... Adams. Intensional Cambridge. . 'Are There Set Theoretic Possible Worlds?'. G. is not general. e. 49-66 (henceforth PAPW). 121-44. theory of states of affairs whose consistency 1974: The Nature Clarendon ch. Plantinga is the existence of possible worlds in his reply to Pollock but the proof (pp. 68-72. pp. pp. since. Minnea University 185-213.. Analysis to Menzel'.POSSIBLE 15 pp. but it is hard to find it convincing. 328-9). polis. which are all general. pp.

62. Dordrecht. Foris. University .. esp. e. Schuurman: 1980. 63. Ockam's Theory of Propositions: of Notre Dame Press. Indiana. though. Except perhaps 38 in PAPW. and Objectivity. Plantinga 27 PAPW. London. 75. Realism. This is the typical medieval See. (eds. MIT/Brad approach Massachusetts. Ltd. degree(0(p) degree(cp) tive). reveals the form of its truth conditions. Gettier-style. op. Cambridge. Logic J. See also PAPW. 44 (1983). Not 1968.. 1987. semantics criticism of nonrealist McMichael's pp. Freddoso approach Part II of the Summa Logicae. does not deal 11 (1988). cautious For the more 1984. 26 that would like being the person I omit consideration of essences spring from A and that in fact never got together. 96.): D. ch. Section 5 of the Introduction pp. and H. See also C. ford Books.g. 32 be dubious about the idea for those who might It should be borne in mind. we could make all the same since with minor of these notions. be excepted of essences requires generally. 37 by accident. which Enrich First-order S5\ Journal of Philosophical esp. 8.): in J. Stalnaker: to worlds. 1 dijk et al. degree((p = = + 1. pp. p. 63-4. (ed. Kamp. Dordrecht. 1974.g. E. Measurement. Swoyer. Interpretation. nesting involving mere and Perry in the introduction. Plantinga. Compare and Information. and A. these alone are not sufficient from this claim. cit. R. with here.388 CHRISTOPHER MENZEL Thisness'. " 425-6. 423-454. 2. Adams. 1977. London. with appropriately " = de = = 0 for atomic Where formulas. See 36 NAMS. H. of course. that gives rise to the sorts of cases we are concerned modal operators 34 to this idea. Notre Dame. Philosophical 1-20. An Introduction all nesting. Duckworth. Times and Selves. Hughes and M.. pp." for any given first-order modal statement. 41... 28 See NAMS p. A. p. degree(ijj)) if/) gree(3xcp) degree(?>).g. and degree(n<p) and Co. Compare 39 to modal truth conditions. p. Creswell: 50. or n-place follows hangs on the that nothing which of a set of all individuals relations. pp. specifically in ch. Studies 'On Existentialism'. Groenen 'A Theory of Truth and Semantic Representation'. e. complications legitimacy moves restricted domains. to do the job In any case. Inquiry. 91-106. See his 'Modal Logics formal expression Hodes' i gives precise operator 13 (1984). as well as his 'Models. pp. to Modal Logic. the quote from Barwise the valid the semantics "I have not constructed Thus McMichael: simply to segregate to give a semantics I have tried in addition and invalid formulas of a first-order language. 235-90. and Philosophy Etchemendy Though Linguistics that follows was strongly the account of modal with the semantics languages. 29 and Logical Truth'. and APW. G.. Truth. where A and B are particular gametes actually existing then of course If there are such essences they are to (and this is not uncontroversial). which. e. See. 1 see R. Semantics See Etchemendy. Reidel. degree^'x(p) degree(~\(p) degree((p) = max = * any binary connec (* (degree ((f). pp. 1 of his book. Conditions'. gives is S5. and Truth 5: 'Essences Section p. Forge 'The Metaphysics of Measurement'. Worlds. where one's underlying and is in effect eliminable iteration of modal operators logic into the scope of nested It is quantification conditions below the added unnecessary. B. Methuen rise to similar difficulties. remarks influenced by several suggestive 30 in J. and 'Actualism ch.

'Weak Necessity and Truth Davies. 77843-4237 . Michael Tom Morris. my emphasis. p. p. like for and on early drafts that significantly for incisive comments the paper. cit. also M. Perry. being to a primitive second and so also must appeal of the modal understanding operators. 42 NAMS. 99. p. generally. The definition of a role (p.e. Theories' 40 See PAPW. E. also in terms of pp. to Situations'.S. necessarily. they will be formulated roles. then it has the i. 'Possible Worlds and Situations'.A. extensive discussion of these issues over the years. 415-39 and pp. Most of it was during written summer there and during the following in College Station under a grant from the Texas A&M College of Liberal Arts. and Bob Burch. fn. those truth conditions are. 'From Worlds of the same (1986). my emphasis. I would in particular to thank Fred Freddoso. But no matter what they turn out to be. and NN. 97. 77-9. A&M of Philosophy University Texas Station. 83-107 issue. 'Necessity in the same volume. truth theories On modal op. see NAMS. Journal 15 Logic of Philosophical and J. My thanks go out to both institutions. Stalnaker. 44 See R.. 109-23. see Gupta. if anything has the first. Nolt also makes in 'What are Possible Worlds?'. Peacocke. pp. Mind by Freddoso. 45 This I was a fellow at the Center when for Philosophy paper was initially conceived at the University of Religion of Notre Dame the spring of 1988. 88) is stated in terms of one property's including another such that. 95 (1986). and Truth Theories'. Chris Burch's Department 510 Blocker Texas College U. p.. 41 It is not clear exactly what pp. 473-500 resp. Swoyer improved were comments in particular detailed and penetrating. extraordinarily Special thanks to Al Plantinga. pp. Al Plantinga. pp. 53. 268. and whose work continues to be my starting point.POSSIBLE WORLD SEMANTICS 389 a similar move J. 43 APW. p. 126. and Ed Zalta Kremer. 12. and C. 432-45. who first taught me about the metaphysics of modality.

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd