Blackwell Publishing Ltd. 2004, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350Main Street, Malden, MA 02148, USA. Ratio (new series) XVII 3 September 2004 0034–0006
AbstractDispositional essentialism, a plausible view about the natures of(sparse or natural) properties, yields a satisfying explanation of thenature of laws also. The resulting necessitarian conception of lawscomes in a weaker version, which allows differences between pos-sible worlds as regards which laws hold in those worlds and astronger version that does not. The main aim of this paper is toarticulate what is involved in accepting the stronger version, mostespecially the consequence that all possible properties exist in allworlds. I also suggest that there is no particularly strong reason forpreferring the weaker to the stronger version. For example, Armstrong’s instantiation condition on universals entails thataccording to strong necessitarianism every property is instantiatedin all possible worlds. But first we do not need to accept Arm-strong’s instantiation condition, in part because his arguments forit are forceful only for a contingentist about laws and properties. Secondly, even if we do accept the condition, the consequence thatall properties are instantiated is not itself contradictory, so long as any form of necessitarianism holds. Strong necessitarianism isprima facie counter-intuitive. But for that matter so is weak neces-sitarianism. Accepting either weak or strong necessitarianismrequires denying the force of intuition in this area. And indeed wehave every reason to deny the force of intuition and its primarysource, imagination, concerning modal facts. Introduction
The received and intuitive view of laws is that they are contingent. The laws that hold in this world might not have held in someother possible worlds while in yet other possible worlds laws holdthat do not hold in this one. There have however been dissentersfrom this view. According to the dissenting view the laws of natureare necessary. This much is well known, though perhaps less wellunderstood. What is certainly less well explored is the extent ofthat necessity.
Within the necessitarian camp there are two possible views.
The weaker, more conservative view holds that laws are necessaryin a sense similar to Kripke’s necessity of identity. The claim
ᮀ(Eric Blair = George Orwell) is consistent with there being possible worlds where Eric Blair/George Orwell does not exist.1The corresponding view for the necessity of laws is this. Lawsconcern properties. Properties may or may not exist in differentpossible worlds. The conservative necessitarian will say that thelaw L(P) concerning property P is necessary, and that this requiresonly that L(P) holds in all possible worlds where P exists. Just asworlds where Eric Blair was never born are not counterexamplesto the claim ᮀ(Eric Blair = George Orwell), worlds where P doesnot exist are not counterexamples to the claim ᮀL(P). What
ᮀ(Eric Blair = George Orwell) rules out are worlds where EricBlair and George Orwell exist but are distinct individuals. What
ᮀL(P) rules out are worlds where P exists but is not governed bythe law L(P) (worlds where P exists but is governed by other lawsor by no laws at all). Thus ᮀL(P) is consistent with there beingworlds in which L(P) is not a law. It is also consistent with therebeing worlds where there exist properties Q, R etc. that do notexist in our world. In a world where Q and R exist, they may begoverned by laws L*(Q) and L†(R) that are not laws of our world. Of course ᮀL*(Q) and ᮀL†(R) will also be true in just the waythat ᮀL(P) is true. But on this weak version of the necessitarianclaim about laws, it is possible both that ᮀL*(Q) and that thereis no law L*(Q) in the actual world. On this view possible worldsare nomologically consistent with one another. No world containsany fact that is a counterexample to a law that holds in any otherworld. But worlds may be nomologically distinct in that a law thatexists in one world may not exist in another world.
The more radical necessitarian view of laws requires more than
is provided for by the analogy with identity. According to strongnecessitarianism laws of nature are necessary (in the senseexplained above) and furthermore in each possible world all laws
This is true whether one prefers the view that ‘ᮀ(Eric Blair = George Orwell)’ is true
because ‘Eric Blair’ and ‘George Orwell’ denote the same individual at all possible worldswhere Blair or Orwell exists, or the Kaplan view that the proposition ‘Eric Blair = GeorgeOrwell’ asserts of some actual individual (Blair/Orwell) that he is identical with himself,a proposition whose truth value is insensitive to changes in the possible worlds at whichit is evaluated, and so is true even at world where Blair/Orwell does not exist (because,to use Kaplan’s imagery, Blair/Orwell is loaded into the proposition, before it makes itsround-the-worlds journey). (Kaplan 1989, p. 569).
indeed hold, and so the properties they involve exist. Let us dis-tinguish between the truth of a law proposition at a world and theholding of a law at that world. This would match the distinctionbetween the truth of a proposition such as the proposition thatBlair is Orwell and the fact of Blair’s existence and identity withOrwell. Arguably the proposition can be true of a world wherethe fact does not exist.2 So the weak necessitarian view of lawsexpressed is the preceding paragraph is this:
(WN) L(P) Æ "w(it is true of w that L(P));
while strong necessitarianism is as follows:
(SN) L(P) Æ "w(L(P) holds at w)
(LH) L(P) holds (exists) at w iff it is true of w that L(P) and
Let us reserve ‘ᮀ’ for the weak necessitarian view, while
‘ᮀ+L(P)’ abbreviates the strong necessitarian view. Let L*(Q) bea possible law, i.e. ᭛L*(Q). Let then w* be a world where L*(Q). According to strong necessitarianism, in w*, ᮀ+L*(Q); and so
᭛ᮀ+L*(Q) is true of the actual world. By S5, ᮀ+L*(Q). And so wehave: ᭛L*(Q) Æ ᮀ+L*(Q). By (SN) ᮀ+L*(Q) Æ L*(Q) holds atthe actual world. And so: ᭛L*(Q) Æ L*(Q) holds. Since L*(Q)is an arbitrary possible law we can conclude that all possible lawshold in all possible worlds including the actual one. (The assump-tion of S5 is tantamount to (a) assuming that when I say that ifthe law L(P) holds in the actual world then it holds in all possi-ble worlds I mean all possible worlds, not just some proper subsetof the possible worlds that are accessible from the actual world
We may wish to distinguish between a proposition’s being true of a world and its being
true at a world. It is less plausible that a proposition can be true at a world where the cor-responding fact does not exist. That the latter is false has been argued in (Williamson2002). According to Williamson every particular exists, only that at some possible worldsthe particular is a merely possible particular, while at the remainder it is concrete. (Thisview is more attractive than it might at first appear to be – it is the conclusion of a veryplausible argument; it makes the Barcan formula true, thus permitting a very simple quan-tified modal logic) To take Williamson’s view on board this paper would have to be re-written with ‘exists’; replaced by the equivalent of ‘concrete’ for facts and properties.
Armstrong (1983, pp. 163, 166) makes a distinction between strong and weak neces-
sity along these lines. He takes ordinary necessity as applied to laws to be equivalent to mystrong necessitarian view, i.e. ᮀ+L(P), so weak necessitarianism is expressed thus: ᮀ+(theuniversal P exists Æ L(P)).
(i.e. the relevant accessibility relation is universal from the actualworld) and (b) assuming that the actual world is not privilegedas regards the accessibility relation (i.e. the relation is universalfrom all possible worlds).) According to Strong Necessitarianismthere is no difference between possible worlds as regards theirlaws; nomologically, they are identical.
In this paper I shall argue that the more radical view is entirely
consistent and deserves due consideration. The main aim of thepaper is to make clear what is involved in holding the strongnecessitarian view. While I shall not explicitly set out to defendthe strong necessitarian view let alone claim that it is superior toall competitors, I will point out some of its advantages. Dispositional essentialism and the necessity of laws
One might be a necessitarian about laws because one is a dispo-sitional essentialist about properties. According to dispositionalessentialism properties, at the very least those sparse propertiesthat appear in the fundamental laws of nature, have dispositionalessences.4 (I shall say more about what we mean by ‘property’ inthis context below.) The real essence of such a property includesa disposition to give some particular characteristic manifestationin response to a characteristic stimulus. So, for example, it mightbe the essence of negative charge to repel other negative chargesand attract positive ones.5 Let some x possess the property P. Onthe dispositional view the essence of P is some dispositional char-acter, D(S,M), the disposition to yield manifestation M in responseto stimulus S:
ᮀ(DE) ᮀ(Px Æ D(S,M)x).
(I have said that dispositional essentialism says that propertieshave dispositional essences. It is a further claim that the individ-
Cf. (Ellis and Lierse 1994) for dispositional essentialism.
Some dispositions will have stimuli that are typically or even necessarily always
present. Typically two charges will always influence one another, if only slightly, even ifthey are very far apart. However, charges can be electrically isolated from one another. Soin this case, the appropriate stimulus will be a matter of the charges not being electricallyisolated from one another.
uation of properties is to be given by their dispositional powers. If this claim is added then the conditional in (DE) and ᮀ(DE)can be replaced by a biconditional – in effect maintaining thatproperties just are their dispositional powers. That difference willnot affect what follows.)
The impact of this view on the laws of nature is that it makes
the laws of nature necessary rather than contingent. On thesimple conditional analysis of dispositional ascriptions the fol-lowing holds:
(CA) x is disposed to give manifestation M in response to
stimulus S if and only if were x to receive stimulus S, thenx would yield manifestation M.
If we symbolise the subjunctive/counterfactual conditional by‘ᮀÆ’ then (CA) can be abbreviated:
(CA) D(S,M)x ´ Sx ᮀÆ Mx.
Since (CA) is an analysis of the concept of disposition it is necessary:
ᮀ(CA) ᮀ(D(S,M)x ´ Sx ᮀÆ Mx). (DE) and (CA) give us:
(I) (Px Æ (Sx ᮀÆ Mx)).
Now consider any case where x, which is P, also acquires the stimulus S, i.e.
(IV) (Px & Sx) Æ Mx.
Since x is arbitrary we may generalise:
(V) "x((Px & Sx) Æ Mx).
We thus have a universal generalisation. The premises from
which (V) was deduced were only (CA) and (DE), which is to saythat the universal generalisation is a consequence solely of an ana-lytic proposition, (CA), and the essence of the property P, cap-tured in (DE). This means that such universal generalisations are
consequences of the dispositional essences of properties. In whichcase the argument given holds for all worlds where P exists. I.e. the conclusion, "x((Px&Sx) Æ Mx), is necessary in at leastthe weaker sense discussed above. (In brief, (CA)&(DE) ٛ"x((Px&Sx) Æ Mx) therefore ᮀ(CA) & ᮀ(DE) ٛ ᮀ"x((Px&Sx)Æ Mx).)
Since the generalisation is non-accidental it is a nomic gener-
alisation. On one view of what laws are, (V) itself states a law ofnature. This view would agree with Lewis that laws are regulari-ties but disagree with Lewis very deeply about what makes oneregularity a law and not another. On Lewis’ view it is the fact thatthe regularity is a consequence of the optimal systematisation ofall particular facts. According to the necessitarian (of this stripe)laws are those regularities whose truth is guaranteed by the dis-positional nature of one or more of the constituent properties, inthe way that the truth of (V) is guaranteed by the dispositionalnature of P. Regularities that supervene on such laws will also be laws.
A different view is that the law in this case is a relation of ‘neces-
sitation’ N among the universals P, S, and M: N(P&S,M). Thiswould be analogous to Armstrong’s view of laws. The differenceis that for Armstrong N is a contingent relation among universals,whereas the necessitarian takes N to be a relation of metaphysi-cal necessitation. On this view ‘N’ can be defined thus: N(F,G) iff
ᮀ"x(Fx Æ Gx).6 In which case N(P&S,M) is a consequence of
ᮀ(CA) & ᮀ(DE). Two of the criticisms levelled at Armstrong arethat it is unclear what his relation of contingent necessitation isand that it is unclear how it is able to necessitate anything. Clearlythese problems do not arise in this case.
The necessitarian can be reasonably relaxed as regards the
question, ‘what exactly are the laws of nature?’, for two reasons. First, whatever one takes laws to be, the derivation of ᮀ(V) andso N(P&S,M) from ᮀ(CA) & ᮀ(DE) shows that dispositionalessentialism can account for at least some of the laws of nature. An ambitious dispositional essentialist will claim that all laws ofnature may be accounted for in this way – and it is the ambitiousview that I am considering in this paper. Secondly, the disposi-tional essentialist view will regard the motor and cement of the
This is a simplification, which will allow too many entailment relations to be regarded
as nomological. It might be appropriate to restrict ‘N’ to those instances of ‘ᮀ’ that arisefrom the essence of some property in the manner just described.
universe as residing ultimately not in the laws themselves butrather in the dispositional nature of properties. The laws are, ina sense, epiphenomenal. A correction to (CA)
The flaw with the dispositional essentialist account given so far is that (CA) is false. (CA) is false thanks to finks, antidotes, andmimics. For example, a particular instance of a disposition D(S,M)is finkish if there is some mechanism in its vicinity such that thecharacteristic stimulus S has the property of causing the disposi-tion D(S,M) to go out of existence before it can produce its char-acteristic manifestation M. So, if a disposition is finkish we canhave the disposition and its stimulus but without the manifesta-tion occurring, contrary to (CA). In Charlie Martin’s example, awire is live – it is disposed to pass current to a conductor if theconductor is placed in contact with the wire (Martin 1994). Thiswire happens to be finkishly live; a mechanism exists that detectsthe presence of conductors and within a fraction of a second ofthe conductor touching the wire cuts the electricity generator,making the wire dead before it can deliver any current to the con-ductor. So at the very moment of touching the wire is live (dis-posed to deliver current to a conductor that is in contact with it)and the conductor is in contact with the wire; but no current isdelivered to the wire. We have a disposition but the subjunctiveconditional is false. In finkish cases the disposition is caused todisappear. In the cases of antidotes, the disposition remains but its normal operation is interfered with; that is, the environ-mental conditions upon which the disposition depends arealtered (Bird 1998). A poison may be deadly, disposed to kill thosewho ingest it; but if an antidote is taken the body’s physiology may be altered or protected so that the poison cannot do what itwould otherwise have done. So long as the poison is not changedby the antidote, the disposition to kill if ingested must remain(assuming dispositions to be intrinsic properties); the poison isingested; but it does not kill. Again we have a counterexample to(CA).7
Mimics (cf. Johnston 1992) are cases where the counterfactual in (CA) is true but
there is no corresponding disposition. Mimics refute the right to left implication in (CA)whereas the nomic necessitarian requires only the left to right direction.
Necessitarianism can withstand the falsity of (CA). Most com-
mentators believe either that (CA) can be modified so as still togive an analysis of dispositional concepts, or that even if no analy-sis in the traditional sense is available, there is nonetheless anintimate metaphysical relationship between dispositional ascrip-tions and conditionals – which is roughly that so long as antidotesand finks etc. are absent, then, if x is disposed to M when S, thenx will M when S.8 For example, Stephen Mumford suggests a ‘conditional conditional’ account (Mumford 1998):
(CC) D(S,M)x ´ Cx Æ (Sx ᮀÆ Mx).
This is not strictly an analysis since the conditions C (whichMumford takes to be ideal conditions) cannot be finitely and non-trivially characterised.9 (CC) may nonetheless correctly representa conceptual and metaphysical truth concerning the relationshipbetween dispositional ascriptions and counterfactuals.
It is easy then to see how the necessitarian view can be rescued.
If we replace (CA) by (CC) and ᮀ(CA) by ᮀ(CC) (the latter beingpermitted since (CC) is a statement of metaphysical fact even ifnot strictly an analysis), then we reach the conclusion: ᮀ"x(CxÆ ((Px&Sx) Æ Mx)). It is worth noting that we do indeed havelaws of this double conditional form, ceteris paribus laws, and it isplausible to suggest that the conditions encapsulated in a ceterisparibus clause are precisely the sorts of conditions (ideal condi-tions, absence of finks and antidotes etc.) that are covered by ‘C’. Strong necessitarianism and properties
Dispositional essentialism cannot itself decide between the weak(conservative) and the strong (radical) version of necessitarian-ism. While dispositional essentialism entails the weak view it is alsoconsistent with the strong view. I shall in this section explore thestrong necessitarian account in more detail in order to show thatit is indeed consistent.
One modifier is Lewis (Lewis 1997). Mellor (2000) maintains that the link holds,
Mumford in fact defends the following: D(S,M)x entails if Ci then (if Sx then Mx).
First it is clear that Mumford takes the conditional inside the parentheses to be a sub-junctive/counterfactual conditional. Secondly Mumford says things that suggest that theentailment goes both ways. But as remarked (footnote 7) the current discussion requiresonly the left to right entailment.
A world in which L(P) is true and P does not exist is a refuta-
tion of strong necessitarianism. Since necessitarianism in generaltakes L(P) to be necessarily true if true in any world, it followsthat strong necessitarianism must take P to exist at all possibleworlds if L(P) is a law at any world. Let Q be a property that existsat some possible world w. Since Q has a dispositional essencethere will be a law L*(Q) in which Q participates. Since all pos-sible laws are actual and all actual laws are necessary, it followsthat Q exists in the actual and in all other possible worlds. So anypossible property is actual. The view that all laws hold in all worlds(understood non-trivially) entails the view that all (nomic) prop-erties exist in all worlds.
What view of properties makes such a claim admissible? One
view takes properties to correspond to possible predicates; prop-erties are Fregean concepts. And there seems nothing objection-able in regarding all possible Fregean concepts as existing at allpossible worlds. But there are three reasons why this is not a viewof properties of which the necessitarian should avail him- orherself. First, this view leads to Russell’s paradox. Admittedly, wemight finesse this by somehow restricting the range of possiblepredicates that generate concepts, although no-one has yet comeup with a natural way of so doing. Secondly, Russell’s paradoxarises because there are too many possible predicates; but thereis also no guarantee that there are enough predicates to expressall the properties we might want for the laws of nature. Perhapsthere are properties that can be grasped by no possible mind andcan be expressed by no possible predicate. It is entirely plausiblethat there are laws that cannot be known or understood by anymind. So there could be properties that do not correspond to anyFregean concept. Thirdly, and most importantly, the necessitar-ian wants to account for laws as flowing from the real essence ofproperties. It is implausible to suppose that Fregean conceptshave real as opposed to merely nominal essences. But even if theydo, those concepts and their essences are metaphysically tooflimsy a foundation upon which to build an edifice of laws thatgovern the unfolding of the universe. Fregean concepts are whatLewis calls ‘abundant’ properties; we however are interested inwhat he calls ‘sparse’ properties, those which have the structureof the world to thank for their existence rather than the expres-sive possibilities of thought or language.
The most promising way to understand sparse properties is as
universals. David Armstrong, the leading proponent of sparse
properties as universals, takes it to be a necessary condition onthe existence of an universal at a world that it should be instan-tiated at that world (Armstrong 1997, pp. 38–53). On the strongnecessitarian view every possible sparse property (universal)exists. Add to this Armstrong’s instantiation condition, we havethe conclusion that every possible universal is instantiated in theactual world. This may seem an implausibly strong conclusion. Inthe next section I shall examine Armstrong’s grounds for theinstantiation condition. In the section following that I shall askwhether the view that every possible universal is instantiated isreally so implausible. The instantiation condition
Armstrong’s instantiation condition on universals is the modernequivalent of Aristotle’s in re conception of substantial forms: theforms exist through substances. Aristotle’s view conflicts with thePlatonic ante rem conception of forms as existing independentlyof any object having them. Unlike the in re conception, the anterem conception of universals would permit all properties to beactual without them all being instantiated.
Armstrong thinks of universals as ways in which things stand
towards one another. He says it is implausible that there shouldbe such ‘ways’ without there being things standing in such ways. To this Sydney Shoemaker replies that one could equally charac-terise universals as ways things can stand towards one another. And things could be thus and so without their actually being thusand so. Armstrong rejects this objection on the grounds that itmakes universals necessary beings (Armstrong 1997, p. 38). Thisof course is no objection to the strong necessitarian who happilyaccepts this proposition. This consequence of uninstantiated uni-versals is a problem only for someone adopting Armstrong’s contingentist views about laws and properties. (It might be added that one could interpret Shoemaker’s ‘can’ as a matter ofnomic rather than metaphysical possibility – a way things couldbe in this world with its laws. So for someone like Armstrong whothinks laws are contingent, universals would still be contingententities.)
Armstrong does marshal another argument against uninstanti-
ated universals, viz. that they contradict the Eleatic Principle(Armstrong 1997, p. 41):
Everything that exists makes a difference to the causal powersof something.
While the Eleatic Principle is contentious and its correct formu-lation even more so, the necessitarian I am describing is certainlyin sympathy with its motivation. Our necessitarian is a disposi-tional essentialist: properties have their causal powers essentially;indeed their causal powers may even serve to individuate prop-erties. In this sense, the necessitarian can claim to adhere to theEleatic Principle more closely that an Armstrongian or Humeancontingentist, since the latter hold that the difference a propertymakes to the causal powers of things is only a contingent differ-ence, whereas according to the necessitarian, properties maketheir differences necessarily. Armstrong’s concern is that an unin-stantiated property makes no difference to the causal powers ofanything. What Armstrong means by ‘anything’ is any actual par-ticular. To which a further pair of remarks may be made. First, ifone is a genuine, full-blooded realist about universals, one shouldbe happy to allow the ‘anything’ to quantify over universals as well. And the existence of an uninstantiated universal on the necessi-tarian view will indeed make a difference to the causal powers ofsomething, principally itself, because it is itself identical withcertain causal powers.
Secondly, the instantiation condition makes the existence of
laws and properties implausibly sensitive to contingent differ- ences (and nomically contingent difference at that) in the exis- tence or even location of rare particulars. If, as Armstrong does, one admits that particulars exist contingently, it seems plausible that one ought to be able to have two worlds that are alike in respect of their laws but differ in that one has a particular that the other lacks. But Armstrong’s position is incompatible with this. For one may imagine a law L involving a universal U where the only particular in world w1 that is affected by this law and instantiates U is the particular a. Now consider w2 which is as similar to w1 as can be except that a does not exist in w2. On Armstrong’s view U cannot exist in w2 and so neither can L. There seems no reasonable explanation as to why the removal of a con- tingently existing entity should also remove a law. The problem is made more intense when we consider that the difference between w1 and w2 need not be a difference in a’s existence but in a’s location. Let a and b be entities (and the only entities) of kind K. In w1 a is 0.9m from b; and in w1 there is a law L that
states that entities of kind K that are less than 1m apart interact in manner M, where M is a universal exclusive to this law (i.e. nothing else other than a and b are M in w1). In w2 a and b also exist and are the only K-entities. But in w2 they are 1.1m apart and so L does not apply to a and b, and so nothing in w2 is M. But if nothing is M, then according to Armstrong M does not exist and so neither does L. While we can understand that the 20cm difference in location of a between w1 and w2 may make a differ- ence to what happens to a it is implausible that it makes a differ- ence to which laws there are. Consideration of such a case lends support to Shoemaker’s contention that universals concern ways things could be not merely the way any things actually are.
The spatial version of this argument is to all intents and pur-
poses the same as Charlie Martin’s example of the non- interacting particles (Armstrong et al. 1996, p. 74) and a similar case discussed by Michael Tooley (Tooley 1977, p. 669). Arm- strongs’ reply to Martin and Tooley, adapted to this case, states that there need not be any counterfactual fact of the matter, regarding w2, as to what would have happened had a and b been 20cm closer together. Armstrong does not explain why we should believe this, except to argue that the view is not absurd, by analogy with a case of an irreducibly probabilistic event, such as a nuclear decay. In the latter case we cannot say that either ‘had the parti- cle been excited, it would have decayed’ nor can we say ‘had the particle been excited, it would not have decayed’. Even if we allow Armstrong’s position to be coherent, it gives us no reason for saying that as regards w2 it must be false that had a been 20cm closer to b, a and b would have been M, which is what Armstrong’s instantiation condition requires us to say.
Armstrong officially regards universals as part of the furniture
of the universe. He rejects nominalist views that take only partic-ulars to exist. But the instantiation condition seems to be a failureof nerve in this realism about universals. If universals really areentities in their own right, why should their existence dependupon a relationship with existing particulars? A diagnosis of thefailure of nerve is this. Armstrong’s metaphysics takes universalsto be categorical properties. That is, he denies the dispositionalessentialism espoused in this paper. Universals do not have theircausal powers essentially, but instead they have them contingently. All that can be said about the essence of a categorical property isthat it is identical with itself and distinct from other things. It maybe that Armstrong fears that such an entity is too spectre-thin to
really exist without the backing of something else more substan-tial. (Recall that Armstrong’s reading of the Eleatic Principle tookthe quantifier ‘anything’ to range over particulars but not uni-versals.) So for Armstrong, it is the instantiation of the propertyin a particular that gives it the required backbone. This is a spec-ulative diagnosis. If it is correct it is of course a motivation thatcannot apply to the dispositional essentialist view of universals. According to that view universals do have a contentful essence,their dispositional powers. Strong necessitarianism with instantiated properties
In this section I wish to explore the combination of strong neces-sitarianism with Armstrong’s instantiation condition. One reasonfor looking at this view is that it is the prima facie least plausibleversion of the strong necessitarian view. Indeed to many it maylook to be in danger of being inconsistent. The combinationrequires that every possible property is not only actual but is alsoinstantiated. The following (‘toy’) example might suggest this.10Let us assume for convenience that all laws are generalisations. One possible law says ‘all grass is green’ and another says ‘all grassis red’. Since these are both possible they are both laws. The twogeneralisations can both be true without generating any contra-diction – but only so long as there is no grass. But if every prop-erty is instantiated there will be some grass, and that grass will beboth green and red. So it looks as if the instantiation conditionleads to a contradiction in a way that strong necessitarianismwithout that condition does not.
The necessitarian answer is simple. In such a case at most only
one of ‘all grass is green’ and ‘all grass is red’ is a law. The ambi-tious dispositional essentialist holds that all laws are consequencesof the essences of properties. Let us say then that it is a law thatall grass is green. This will be because the essence of the prop-erty, being grass, involves the disposition to be green in normalconditions.11 In which case that essence will not involve a dispo-
The illustrative example is a toy example because it is unclear that the relevant prop-
erties (being grass, being green etc.) are genuinely sparse. And they certainly are not fun-damental. So ‘grass’, green’ etc. are standing in for genuinely sparse and fundamentalproperties.
We presume that the stimulus is a standing condition (see footnote 5 above), and
that this standing stimulus condition is the same stimulus for all the dispositions discussed
sition to be red. Since laws follow from the essences of properties,and since the property of being grass does not involve the dispo-sition to be red, there will be no law that all grass is red. If weregard the dispositional essences as determining the identities ofproperties, then if X and Y have distinct essences, X π Y. So if theessence of grass involves the disposition to be green, then anyproperty that involves the disposition to be red cannot be theproperty of being grass. The thought that both generalisationscould be laws arises only for those still in the grip of the contin-gentist view of laws and the categoricalist view of properties thatgoes with it. On those views any property might be involved in alaw with any other property. Clearly then the instantiation of all possible laws is impossible, for just the reason considered, that grass would be both green and red. But the dispositionalistdenies that a property can be involved in just any law with anyother property.
This response shows how to block a parallel but slightly more
subtle objection to strong necessitarianism in general (evenwithout the instantiation condition). With instantiation ‘all grassis green’ and ‘all grass is red’ lead to a contradiction. Withoutinstantiation these generalisations are consistent (they are bothvacuously true). But the laws that grass is green and that grass isred are not consistent. This is because laws entail counterfactuals. In this case we would have the inconsistent counterfactuals, ‘ifthere were grass, it would be green’ and ‘if there were grass, itwould be red’ (which are inconsistent if the existence of grass ispossible). The answer is just the same. It is not the case on thenecessitarian view that both ‘grass is green’ and ‘grass is red’ canbe laws.
A different objection to strong necessitarianism with instantia-
tion goes as follows. ‘One might imagine that the actual worldcontains more laws than we think, perhaps even all possible laws,so long as they are uninstantiated. Being uninstantiated theywould be inert and would have no impact upon us. But if theywere instantiated they would impact upon us. If all the possiblelaws were actual and instantiated, they would be doing things toexisting entities and we would expect to be able to detect them.
in this example. What is being said is that this dispositional essence of P cannot involveboth D(S,M) and D(S,ÿM). However, if complex essences are permitted, then P might well havean essence involving both D(S,M) and D(S*,ÿM), so long as S and S* are incompatible stimuli. Such essences will not lead to inconsistent laws.
But we do not detect them. The laws we do know about (even ifonly partly, at the most fundamental level) seem to be only asubset of all the possible laws. And so strong necessitarianism withinstantiation looks to be in conflict with the empirical evidence.’
The key move in this argument is the sentence ‘But if they were
instantiated they would impact upon us,’ amplified in the one fol-lowing it, ‘If all the possible laws were actual and instantiated, theywould be doing things to existing entities and we would expect tobe able to detect them.’ There is no reason to suppose that theseclaims are true. Laws form integrated, systematic sets. This muchis widely acknowledged and even forms part of the analysis of lawon the Ramsey-Lewis view. In the dispositionalist picture, inte-gration occurs because the manifestation property of one dispo-sition may be the stimulus property of another, and so on. Thismay go round in a full circle. A simple case will be when we havetwo properties. For example, gravitational mass may be regardedas the disposition to transform space-time while space-time is thedisposition to affect the motion of gravitational masses (althoughin this case it is probable that these dispositions reflect a deepersystem of integrated fundamental dispositions upon which otherdispositions supervene).12 So all the possible laws will divide upinto discrete sets (possibly singleton sets) such that each memberof the set integrates with other members of the set but not withany laws outside the set. Take any entity. The existence and natureof that entity are determined by the laws that govern it. Forexample a water molecule exists and is what it is thanks to thelaws of quantum mechanics. Those laws not only determine theexistence and nature of the molecule’s atomic and subatomicparts but also explain the existence of the molecule itself, byexplaining the ability of oxygen and hydrogen to bond together;and in turn thanks to the details of that bond, the laws of quantummechanics also explain water’s properties (such as its ability to dis-solve salt). So our entity, whatever it is, will have its existence andnature associated with some set of systematically integrated lawsbut, because they are discrete, not with any other. So entities asso-ciated with distinct sets of laws will not interact; they will becausally isolated from one another. Now consider that we asobservers are just a certain kind of complex entity governed by
This circularity may itself seem to be problematic for dispositionalism. I shall
not address that issue here – see (Holton 1999) for an explanation of why it need not beproblematic.
one of the sets of laws. From what has just been argued, we couldnot interact with and so could not detect any entity governed bya different set of laws. So it is not true that should laws from a dis-tinct set be instantiated, we might expect to detect such laws andthe entities instantiating them. Indeed we know we could notdetect them. Evaluating strong necessitarianism
In this paper I have sought primarily to show that strong necessi-tarianism is a consistent position. Are there however reasons tothink that strong necessitarianism is true – or is false? Thestrongest reason for thinking that the view is false is the force ofintuition. We have a strong intuition that the laws of nature arecontingent and a strong intuition that the existence of sparseproperties is contingent also. In the next section I shall brieflymention why I do not think that the force of intuition provides astrong objection. Note that the intuition that laws are contingentcounts just as much against weak necessitarianism as againststrong necessitarianism. So only the intuition that the existenceof sparse properties is contingent acts against strong necessitari-anism specifically. But that intuition does not seem to me to beany stronger or better founded than the former. In which case, ifone is willing to swallow one’s intuitions to the degree of goingalong with necessitarianism at all, intuition should not itself beallowed to incline one towards the weaker version rather than thestronger.
What reasons might incline one towards the stronger view?
Above we came across Shoemaker’s view that properties could beregarded as ways things can be. If we regard this ‘can’ as meta-physical rather than nomic, then the way a thing can be is some-thing that will be shared by all possible worlds, assuming S5. (Theconception of properties as ways things can be is consistent with dispositional essentialism. The latter adds that for a thing to be a particular way is for it to be disposed to behave in a certain way.)
The strong necessitarian view has an advantage over its com-
petitors in the realm of explanation. Laws and causal powers areessential components of explanations. Sometimes we are able toexplain a law or power by showing it to be a consequence of someset of deeper laws or powers. That invites an explanation of those
deeper laws and powers. A regress threatens that leads either toan infinite chain of laws or powers or, more likely, to a funda-mental set of laws or powers that cannot themselves be explained. Some have regarded this as a reason to think that the very ideaof explanation is an illusion. This is a fallacy; it is not a necessarycondition on A’s explaining B that we have an explanation for Aalso. Nonetheless one may be sympathetic with the thought thatone’s chain of explanations is less satisfactory for lacking an expla-nation of A, and even more so when told that can be no expla-nation of A, since A is fundamental. There is an explanatorylacuna if although the basic set of laws and powers could havebeen otherwise (as the contingentist holds), there is no explana-tion of their being as they are; that being fundamental we justhave to accept the brute and accidental fact that they are as theyare and not otherwise.
If one is sympathetic to that thought, then strong necessitari-
anism provides a balm. Strong necessitarianism cannot providean explanation of the same kind for the fundamental laws, sincethere are no yet deeper or more general laws to explain these. But it can provide an explanation of sorts. Being necessary, thefundamental laws could not have been otherwise. Viewed one waywe cannot ask for an explanation of the usual kind at all, for thecomparative question, ‘why do we have these laws as opposed tosome other set?’, assumes what is false, that some other set is pos-sible. Viewed another way, the strong necessitarian provides anexplanation that is the best possible. A natural way to understandwhat Hempel sought in his Deductive-Nomological model ofexplanation was that he wanted to make precise the thought thata fully satisfactory explanation shows why, given laws andantecedent conditions, the explanandum had to be. Where theexplanandum is itself a law, antecedent conditions play no role,and so to explain a law is to show how it had to be, given otherlaws, i.e. to explain L1 one shows that ᮀ(L2 Æ L1) where L2 isanother law or set of laws. By this standard, to show that a law hadto be, whatever other laws or conditions might be, is an explana-tion also, indeed as good an explanation as one could hope for. That is, if a demonstration of ᮀ(L2 Æ L1) provides an explana-tion of L1, then so does a demonstration of ᮀ(L1).13
Armstrong (1983, p. 159) records that Martin Tweedale suggests a Rationalist-
style argument based on the Principle of Sufficient Reason, in favour of necessitarianism,that is along the same lines as the proposal I put forward here. Armstrong objects thatshort of adopting a philosophy of the Absolute (the one sole reality from which all the
Imagination and possibility
The main objection to necessitarianism, whether strong or weak,is that it conflicts with our intuitions that laws are contingent. Inthis section I will very briefly indicate the line of thought I advo-cate in response to this objection. We know already that our intu-itions concerning necessity are unreliable. Frege believed, in tunewith intuition, that identity is contingent. But as Kripke has shownus this is a mistake. The fallibility of intuitions of contingency is not limited to identity. It can also be shown that some very contingent looking laws of nature are necessary – without assum-ing anything like necessitarianism (Bird 2001). Necessitarianismsimply claims that the illusion of contingency is more widespread. The source of the illusion is primarily the existence of (genuine)epistemic contingency – propositions whose truth (or falsity) isnot deducible from what we know. A related source is the associ-ation we make between possibility and imaginability. Thatimaginability is no guarantee of possibility needs no emphasizing. But a question is raised over why one ever thought there is a connection.
The following is a hypothesis about such a connection. Simple
creatures react simply to their environment as they find it. Onlywhat is actual has any effect on them. Where those creatures haveperceptual capacities, the effect of actuality is via perceptual rep-resentations of actuality. More sophisticated creatures need tothink about possibility also: the possibility that either prey or apredator is hidden in the bushes, for example. One may hypoth-esize that consideration of possibilities and their role in influ-encing behaviour might be similar to, though weaker than, themanner in which actuality influences behaviour, that is, via per-ception-like representational states. This hypothesis wouldexplain why it is natural to think of imaginability as a guide topossibility, and why it would be correct to think so to some extent. But the hypothesis gives us no reason to think that imaginabilityshould be an infallible guide. On the one hand imaginability islimited by representational capacities, so many possibilities willnot be captured by the imagination. But this failure does notharm the adaptive benefit of imagination, since the possibilities
phenomena may be deduced), one will be forced to adopt contingency in our explana-tions at some point. I am not sure why Armstrong says this. Perhaps the Strong Necessi-tarian conception of the laws makes them equivalent to the Absolute.
knowledge of which are relevant to a creature’s fitness (the pos-sibility of a predator in the bushes) are ones that a cognitivelywell-adapted creature typically does have the capacity to repre-sent. The unimaginable possibilities are often too remote to havean influence on a creature’s fitness. In the other direction, morerelevant to present concerns, the imagination will represent aspossible situations that are not possible. This failure could wellhave an influence on a creature’s fitness, since false positives(such as erroneous beliefs that there it is possible that there isprey to be found in the bushes or that a predator is near the water-ing hole) can have a deleterious impact. Even so, one should notexpect adaptive capacities to be perfect, just to be better thannearby alternatives.
An imaginative capacity that is more accurate with respect to
possibility would require an adaptation that would prevent usfrom representing as possible an identity that is in fact necessar-ily false. It does not seem that any simple improvement to humanpowers of imagination could do this, since it seems imagination(as employed in this context) is intensional rather than exten-sional. That is, how could one adapt the imagination so that it didnot present Lois Lane with a picture, for example, of Clark Kentin one place and Superman in another place? Furthermore, evenif an adaptation could occur that permits its possessor to see that alternative laws governing the same properties are not possi-ble, there is no selection pressure that would bring such an adap-tation into existence. For in so far as a creature’s fitness requiresa concern with laws (via a concern with causation), it can onlyever be concerned with the laws we actually have (and impactupon it); whatever else is or is not possible is adaptively irrelevant.14
And so if the hypothesis is correct, that the link between imag-
ination and possibility is explained by the adaptive benefits ofsuch a link, we would have no reason to suppose that such a linkshould be reliable in delivering accurate judgments when appliedto esoteric cases such as the contingency or necessity of laws. Thefunctions of intuition, imagination, and common-sense are toguide us through the near-at-hand world of middle-sized dry
Of course, I think we do have an adaptation that allows us to know the impossibility
of the relevant propositions. But this adaptation, reason, goes well beyond the power ofimagination. And the power of reason seem to be a power that extends beyond the adap-tive pressures that gave rise to it.
goods, variability among which does affect our lives. We shouldnot expect it to have much utility when faced with the science ofthe very small or very large (where it also fails us) nor with meta-physical questions. Conclusion
Dispositional essentialism, a plausible view about the natures of(sparse or natural) properties, yields a satisfying explanation ofthe nature of laws also. The resulting necessitarian conception of laws comes in a weaker version, which allows differencesbetween possible worlds as regards which laws hold in thoseworlds and a stronger version that does not. The main aim of thispaper has been to articulate what is involved in accepting thestronger version, most especially the consequence that all possi-ble properties exist in all worlds. I have also suggested that thereis no particularly persuasive reason for preferring the weaker tothe stronger version. For example, Armstrong’s instantiation con-dition on universals entails that according to strong necessitari-anism every property is instantiated in all possible worlds. But first we do not need to accept Armstrong’s instantiation condi-tion, in part because his arguments for it are forceful only for acontingentist about laws and properties. Secondly, even if we did accept the condition, the consequence that all properties areinstantiated is not itself contradictory, so long as any form ofnecessitarianism holds. Strong necessitarianism is prima faciecounter-intuitive. But for that matter so is weak necessitarianism. Accepting either weak or strong necessitarianism requiresdenying the force of intuition in this area, as indeed we have everyreason to do. Department of PhilosophyUniversity of Bristol9 Woodland RoadBristol BS8 1TBAlexander.Bird@bristol.ac.ukReferences
Armstrong, D. (1983) What is a Law of Nature? (Cambridge: Cambridge University Press). —— (1997) A World of States of Affairs (Cambridge: Cambridge University Press).
Armstrong, D., Martin, C., Place, U. and Crane, T. (ed) (1996) Dispositions: A Debate
Bird, A. J. (1998) “Dispositions and Antidotes” Philosophical Quarterly 48 pp. 227–34. ——. (2001) “Necessarily, Salt Dissolves in Water” Analysis 61 pp. 267–74. Ellis, B. and Lierse, C. (1994) “Dispositional Essentialism” Australasian Journal of Philoso-phy 72 pp. 27–45.
Holton, R. (1999) “Dispositions All the Way Round” Analysis 59 pp. 9–14. Johnston, M. (1992) “How to Speak of the Colors” Philosophical Studies 68 pp. 221–63. Kaplan, D. (1989) “Afterthoughts” in J. Almog, J. Perry, and H. Wettstein Themes fromKaplan (New York: Oxford University Press) pp. 565–614.
Lewis, D. K. (1997) “Finkish Dispositions” Philosophical Quarterly 47 pp. 143–58. Martin, C. B. (1994) “Dispositions and Conditionals” Philosophical Quarterly 44 pp. 1–8. Mellor, D. H. (2000) “The Semantics and Ontology of Dispositions” Mind 109 pp. 757–80. Mumford, S. (1998) Dispositions (Oxford: Oxford University Press). Tooley, M. (1977) “The Nature of Laws” Canadian Journal of Philosophy 7 pp. 667–98. Shoemaker, S. (1984) “Causality and Properties” in Identity, Cause, and Mind (Cambridge:
Cambridge University Press) pp. 206–33.
Williamson, T. (2002) “Necessary Existents” in A. O’Hear, ed, Logic, Thought, and Language
(Cambridge: Cambridge University Press) pp. 269–87.
PhD, NTUA, School of Chemical Engineering Thesis title: “Contribution of Gas Chromatography – Mass spectrometry in the study and identification of organic binding M.Sc, “Protection, Conservation and Restoration of Works of Art”, ARISTOTLE UNIVERSITY OF THESSALONIKI, School of B.Sc. in Chemistry, University of Ioannina, Greece European Education Programme “ERASMUS”, Dipartimento d