Philosophical Review

Freedom and Foreknowledge Author(s): John Martin Fischer Source: The Philosophical Review, Vol. 92, No. 1 (Jan., 1983), pp. 67-79 Published by: Duke University Press on behalf of Philosophical Review Stable URL: http://www.jstor.org/stable/2184522 . Accessed: 24/02/2011 13:15
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=duke. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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.

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend access to The Philosophical Review.

http://www.jstor.org

The Philosophical Review,XCII, No. 1 (January 1983)

FREEDOM AND FOREKNOWLEDGE

John Martin Fischer

incompatible with human freedom.' If God is eternal and omniscient,then it mightseem that my freedom now to do other than what I'm doing mustbe the freedomso to act thata factabout the past (God's priorbeliefabout mypresentactivity) wouldn'tbe a factabout the past. But since the past is "fixed,"itseems thatifGod exists,then I am now not free to do other than what I'm doing. Many philosophers have been attracted to an Ockhamist response to this argument.2 Both the Ockhamist and the incompatibilist can distinguish between "hard" and "soft"factsabout the
'Nelson Pike, "Divine Omniscience and VoluntaryAction,"Philosophical Review 74 (January 1965), pp. 27-46; "Of God and Freedom: A Rejoinder," Philosophical Review 75 (July 1966), pp. 369-379; and "Divine Foreknowledge,Human Freedom and Possible Worlds,"Philosophical Review 86 (April 1977), pp. 209-216. Pike also discusses the same basic argument in the fourthchapter of his book, God and Timelessness, (New York: Schocken Books, Inc., 1970), pp. 52-86. 2Some examples are: MarilynAdams, "Is the Existenceof God a 'Hard' Fact?"Philosophical Review76 (October 1967), pp. 492-503; and WilliamL. Rowe, Philosophy ofReligion,(Encino, Dickenson, 1978), pp. 154-169. The approach sketchedbelow is called "Ockhamist"because Williamof Ockham distinguishedbetweenpropositionsabout the past whichare necessaryand those whichare not and argued thatamong those propositions about the past whichare not now necessaryare certainpropositionsabout God. (WilliamOckham, Predestination, God'sForeknowledge, and FutureContingents. MarilynMcCord Adams and Norman Kretzmann (trans.),(New York: Appleton-Century-Crofts, 1969), pp. 46-47; 92.) Roughly,Ockham claims that those propositionsabout the past which are true by virtueof contingent future events are not now necessary. Such propositions,it mightbe said, express "softfacts"about the past. A usefuldiscussionof the Ockhamist approach can be found in: Arthur Prior, Past, Present, and Future,(Oxford: Clarendon Press, 1967), pp. 121-127. John Turk Saunders agrees with the Ockhamist that certain propositions about God express soft facts:John Turk Saunders, "Of God and Freedom," Philosophical Review 74 (April 1966), pp. 219-225. Saunders holds a positionwhichis even strongerthan Ockhamism,since he believes that neithersoftfactsnor hard factsneed be fixed:John Turk Saunders, "The Temptations of Powerlessness,"American Philosophical 5 Quarterly (April 1965), pp. 104- 107. 67

A powerfulargumentcan be made thatGod's omniscience is

JOHN MARTIN FISHER

past; the hard factsare fixedwhile the softfactsneed not be fixed. But the Ockhamistclaims thatGod's priorbeliefabout my present activity is a soft fact about the past and hence not fixed; my freedom is thus preserved. Some Ockhamistseven claim thatthe very existence of God is also a softfactabout the past. I shall argue thata veryattractive presentation of the Ockhamist approach, one explicitlyformulatedby MarilynAdams, is inadequate.3 There are significantproblems with Adams' attemptto characterizethe hard fact/soft factdistinction. Further,I shall presenta general challenge to anysortof Ockhamistattempt to explain this distinction.
I. PIKE's ARGUMENT

Nelson Pike claims to exhibitthe incompatibility of human freedom and divine foreknowledge,relative to certain plausible assumptionsabout God's nature.4These assumptionsreflect central featuresof the standardJudeo-Christian conception of God. Pike explicitlyadopts the assumption that if God exists, then God is essentially omniscientand God is eternal.On Pike's account,God is omniscientif and only if God believes all and only true propositions, and we might say that God is essentiallyomniscientif and onlyif God is omniscientin all possible worldsin whichGod exists. Pike says that God is eternal if and only if God has alwaysexisted and always will.5 FollowingPike's presentationin a different article,I assume that the term "God" is a descriptiveexpression used to mark a certain role,rather than a proper name.6 Whoever occupies the role of God is omniscient,omnipotent,eternal,etc. In contrast,the term "Yahweh" is a proper name; it refersto the person who actually occupies the role of God (if God exists). It is not necessarilytrue
3Adams, op. cit. 4Pike, "Divine Omniscience," pp. 26-31. 5Thus Pike conceives of God's eternality as sempiternality-existence at all times.This conception is shared by the Ockhamist;it can be contrasted withthe atemporal conceptionof eternality held by Boethius and Aquinas. 6Pike makes thisassumptionexplicitin "Omnipotence and God's Ability to Sin," American Philosophical Quarterly 6 (1969), pp. 208-216, esp. pp. 208-209. 68

FREEDOM AND FOREKNOWLEDGE

that Yahweh is omniscient,omnipotent,eternal,etc.; it is logically possible that some other person has been God.7 Since "God" is being used here as a nonrigiddesignator,thereis some ambiguity in the assumptionsabout God's attributes. "God is essentially omniscient"does not mean thatthe person who is in fact God is essentially but rather,thatnecessarily, omniscient, whoever is God is omniscient.In termsof possible worlds,God is essentially omniscient just in case for any possible world in which there is a person who is God, thatperson is omniscient. (One can assume that if God is eternal in a particularworld, then it followsthatthere is one and the same person who is God at all timesin thatworld. Pike need not accept thisparticularassumption,as it is not crucial to his argument.) Though thisis the approach to the term"God" thatPike appears to adopt, it mightseem to be an unusual and unappealing position. I shall follow Pike in adopting this interpretation, but it is important to note-that Pike could just as easily embrace the stronger interpretation according to which the person who is in factGod is God. essentially Nothing in Pike's proof, or in my criticismof Adams' Ockhamism,restson adopting the weaker ratherthan the strongerinterpretation of God's attributes. In effect,Pike also appears to adopt what mightbe called the "fixed past" constrainton power attributions: (FPC) It is neverin anyperson's powerat a timeT so to actthatthe
past (relative to T) would have been differentfrom what it actuallywas.

Pike's view about the fixity of the past implies not only that one cannot causally influencethe past; it implies thatno person is free to do somethingwhichis such that,were he to do it,the past would have been different fromwhat it actuallywas. Pike's argumentis essentiallyas follows.Suppose Jones did X at time T2 and God exists. Since God exists, it follows from God's that He existed at T1 (a time prior to T2). Let us call the eternality person who was God at T1, "Y." Since Jones did X at Ti2, it follows
7C. B. Martin argues for this approach in the fourthchapter of his Religious Belief,(Ithaca: Cornell Press, 1964). 69

JOHN MARTIN FISHER

fromGod's omnisciencethatHe believed at T1 thatJones would do X at T2. Now if it was withinJones' power at T2 to refrainfrom doing X, then (1) it was in Jones' power at T2 to act in such a way that Y would have been God and would have held a false beliefat TI, or (2) it was in Jones' power at T2 to act in such a way that Y would have been God but wouldn't have held the beliefHe held at TI or (3) it was in Jones' power at T2 to act in such a way that Y wouldn't have been God at T1. But (1) is ruled out by God's essential omniscience,and (2) and (3) are ruled out by (FPC). Hence itwas not inJones' power at T2 to refrainfrom doing X. If the argument is sound, it can easily be and essentialomniscience generalized to show thatGod's eternality human being free at any time. with agent's any are incompatible about divine foreIt should be pointed out thatincompatibilism knowledge and human freedom needn't entail incompatibilism and human freedom.The problemis about human foreknowledge foreknowledgebecause of God's essentialomdeeper with-divine niscience; perhaps it was in Jones' power at T2 so to act thatSmith (who actually held only correct beliefs) would have held a false betweendivineand beliefat T1. Pike wantsto insiston an asymmetry human foreknowledge.8
II. HARD AND SOFT FACTS

It is sometimesin one's power so to act thatfactsabout the past be facts.John Turk Saunders discusses such a fact: wouldn't thispaperin from writing it is truethatif I had refrained Although it thanitis in that wouldhavebeenother 1965,Caesar'sassassination
8Pike says in his original paper: "The importantthing to be learned fromthe studyof Smith'sforeknowledgeof Jones' action is thatthe problem of divine foreknowledgehas as one of its pillarsthe claim thattruthis arises No problemof determinism connected withGod's beliefs. analytically when dealing with human foreknowledgeof future actions. This is beconnected withhuman beliefeven when (as cause truthis not analytically conjoined to belief." in the case of human knowledge) truthis contingently Pike, "Divine Omniscience," p. 43. Thus it is clear thatPike as well as the between hard and softfacts. Ockhamistneedsthe distinction

70

FREEDOM AND FOREKNOWLEDGE

thispaper,it would not have precededby 2009 yearsmywriting did nothaveitin mypower wouldbe absurdtoarguethatI therefore from thispaper in 1965.9 writing to refrain It is obvious thatthe mere factthatifSaunders had refrainedfrom writinghis paper, then Caesar's assassination wouldn't have preceded Saunders' writinghis paper by 2009 years did not render Saunders incapable of refraining;relative to 1965, "Caesar died 2009 years prior to Saunders' writinghis paper" expresses a soft factabout the past. Of course, it was not in Saunders' power so to act that Caesar would not have died on the steps of the Senate. the factthatCaesar died on the steps Relativeto Saunders' lifetime, of the Senate is a hard factabout the past. Pike agrees withthe Ockhamistthatthereare both hard and soft factsabout the past.'0 It is not easy to provide a precise characterPike himeselfprovides factdistinction. ization of the hard fact/soft no such account, though he claimswe can recognizeclear examples 1I1The disagreementbetween Pike and the Ockhamist of each sort. is about where to draw the line. Pike's position is that if the ordinarynotionsof beliefand existenceare applied to God, then God's beliefat T1 and God's existenceat T1 (includingthe factthatY was God at T1) are hard factsabout the past relativeto T2. And if they were softfactsabout the past relativeto T2, thiswould show thatwe were ascribingbeliefs and existence to God in a special, nonstandard way. fact distinction,the appropriate inGiven the hard fact/soft of the past should be of Pike's claim about the fixity terpretation made explicit: any powerat a timeT so to actthat (FPC*) It is neverin anyperson's
Saunders' arithmetic is wrong 9Saunders, op. cit.,p. 224. Unfortunately, since there is no 0 B.C. or 0 A.D. Hence, Caesar's death preceded Saunsake, however,I shall his paper by 2008 years! For simplicity's ders' writing ignore thisand proceed withSaunders, Pike, and Adams in adding a year to history. 10Pike,"Of God and Freedom," pp. 369-370; Rowe makes a similar betweenfactswhichare "simplyabout the past" and factswhich distinction are not, in William Rowe, op. cit.,pp. 162-165. 1 "Pike, op. cit.

71

JOHN MARTIN FISHER

to T) wouldhavebeendifferaboutthepast(relative hardfact was.12 whatit actually entfrom whichshe MarilynAdams presentsan account of the distinction against Pike's attack.It willbe usebelieves supportscompatibilism ful to consider MarilynAdams' attemptat givingan account of the
distinction:

P is at leastin (B) "Statement partabouttimeT'

df.

"The happeningor not happening, actualityor nonacat T is a of something tuality necessarycondition of the of P." truth

is: l2Put in termsof possible worlds,the fixed past constraint (FPC*) A possible world W* (in whichan agent does other than what he does in W at T) can establish that the agent had it in his power at T in W to do otherwiseonly if W and W* have the same hard factsabout the past relativeto T. In "Pike on Possible Worlds, Divine Foreknowledge and Human FreeReview88 (July1979), pp. 433-442, Joshua Hoffman dom," Philosophical of the past. Hoffmanconstrues of the fixity Pike's interpretation criticizes the occurrence Pike as claimingthat the possession of everypower entails to or nonoccurrence of past circumstances.That is, Hoffman attributes ascribinga particularpower to Pike the claim thatthe truthof a statement an agent at a time in a world W entailsthat the past be as it is in W (Hoffman, pp. 441-442). Pike himself puts the constraintin a misleading way, saying: "If we assume thatwhatis withinmypower at a givenmomentdeterminesa setof possible worlds, all of the members of that set will have to be worlds in which what has happened in the past relative to the given moment is preciselywhat has happened in the past relative to that moment in the actual world." Pike, "Divine Foreknowledge,"p. 215. But nothingin Pike's position requires acceptance of the radical doctrine attributedto him by commitshim to the claim thatif an Hoffman. Pike's fixed past constraint agent performsan act in world W, then any possible world W* in whichhe the act musthave the same past as W, if W* is to refrainsfromperforming establish that the agent can in W refrain from performingthe act. But theremaybe possible worlds(includingW) in whichthe agentcan perform (relativeto the the act (and doesperformthe act) in whichthe past histories fromone another; hence, the truthof a time of the act) are all different of need not entailthe past history.Hoffman's criticism power-ascription unscathed. Pike misses the mark and leaves the fixed past constraint

72

FREEDOM AND FOREKNOWLEDGE

Saunderswrote before "Caesardied 2009 years Thus thestatement, hispaper"is at leastin partabout44 B.C., sinceCaesar'sdeathat that ofthetruth ofthat statement. It is alsoat condition time is a necessary hispaperin 1965 leastin partabout1965A.D. sinceSaunders' writing ofthat Given condition ofthetruth statement. A.D. is also a necessary as follows. of a "hard"factmaybe explained (B) thenotion P expresses a (C) "Statement T" 'hard'fact abouta time
=

df.

"P is notat leastin partabout any time futurerelativeto

T. ts13

Adams uses this account to present an Ockhamist response to Pike's argument. On her account, God's belief at T, and the fact that Y was God at T1 are deemed softfactsabout T,. Adams claims that her account shows why "Caesar died 2009 years before Saunders wrote his paper" does not express a hard factabout 44 B.C. But her account does notexplain thisunless it is interpreted to implythatno sentenceexpresses a hard fact.Adams saysthat"Caesar died 2009 yearsbeforeSaunders wrotehis paper" is at least in part about 1965, since Saunders' writing his paper in 1965 is a necessaryconditionof the truth of thatstatement. But this seems plainly false; the statemententails that Caesar's death and Saunders' writing his paper be separated by 2009 years,but it does not entail any two particulardates for the two events. The statemententailsthatthe twoeventsstand in a certaintemporalrelation, but it does not entail thattheyoccur on any specificdates. Hence, Saunders' writinghis paper in 1965 is nota necessaryconditionof Caesar's death being 2009 years prior to Saunders' writing his paper, if we interpret"Q is a necessaryconditionforP" as "P entails One might reply that since it is true that Saunders wrote his paper in 1965, "Saunders wrote his paper in 1965" is materially implied by "Caesar died 2009 years prior to Saunders' writinghis paper." So if we interpret"Q is a necessaryconditionforP" as "P materiallyimplies Q," Saunders' writinghis paper in 1965 is a
13Adams,op. cit.,pp. 493-494.
Q"

73

JOHN MARTIN FISHER

necessaryconditionof the truthof "Caesar died 2009 yearspriorto Saunders' writinghis paper." But it is obvious thatif thissense of "necessarycondition" is adopted, then no sentence will express a dishard factabout 44 B.C. So Adams' account of Pike's intuitive of no explication tinctionis inadequate as it stands. Adams gives the notion of a necessarycondition by referenceto which she can say that "Caesar died 2009 years prior to Saunders' writinghis paper" does not express a hard factabout 44 B.C. Consider also the statement, "John F. Kennedy was assassithis statementexnated." Given the entailment interpretation, presses a hard fact about 1961, since it does not entailthe occurrence of anything subsequent to 1961. Of course, there are logically possible worlds in which Kennedy was assassinated in 1961. But we wantto say thatin 1962 (and in 1963, untilNovember 22nd), it was withinOswald's power so to have acted thatKennedy wouldn't have been assassinated. And again, it is obvious that the material implication interpretationof "necessary condition" is inadequate. the inadequacy of the enillustrate further Complex statements tailmentaccount of "necessarycondition."If Jones did not believe at T1 thathe would do X at T2, then "Either Smithknew at T1 that Jones would do X at T2 or Jones believed at T1 thatJones would do X at T2" should notexpress a hard fact about T1; the Ockhamist would say thatJones mighthave been able so to act at T2 thatthis disjunctivestatementwould be false. Yet on Adams' account, the statementexpresses a hard fact about T1, since its truthdoes not entail that anythinghappens afterT1; the truthof the disjunction happens (or failsto happen, etc.) after does not entailthatanything In defense of Adams' approach, one mightofferthe following account of a necessarycondition:Q is a necessaryconditionforP if and onlyifP would not be true (or have been true) ifQ weren'ttrue the "counter(or hadn't been true). Let us call this interpretation to say that It is plausible condition. of necessary factual"account a have wouldn't it his paper in 1965, then if Saunders hadn't written been the case that Caesar died 2009 years prior to Saunders' writing his paper. Thus, Adams could say, on the counterfactualaccount, that "Saunders wrotehis paper in 1965" is a necessarycon74
T1.

FREEDOM AND FOREKNOWLEDGE

dition of "Caesar died 2009 years prior to Saunders' writinghis paper." Also, it is perhaps reasonable to say (though I'm not sure) thatif Oswald had not shot Kennedy in 1963, then Kennedy would not have been assassinated. If thisis so, then Adams could say that "Oswald shot Kennedy in 1963" is a necessaryconditionof "John ifJones hadn't done X at F. Kennedy was assassinated." Similarly, either Smithknewat T1 that false that would have been then it T2, Jones would do X at T2 or Jones believed at T1 thatJones would do X at T2. Thus, Adams could say that "Jones did X at T2" is a necessarycondition of the disjunction. both plausible But thereis another sortof problemwhichafflicts accounts-both the counterfactualand entailmentinterpretations of "necessarycondition."Suppose "Smithexistedat T1" is true.It is a necessarycondition of the truthof this statement(on both the and entailmentaccounts) thatit is not the case that counterfactual Smith existed for the firsttime at T2. It is obvious that Smith's existingat T1-entails that he doesn't exist for the firsttime at T2* timeat T2, then he wouldn't And if Smithhad existed for the first have existed at T1, so the counterfactualaccount fares no better than the entailmentaccount. Thus, by (B), the statement"Smith existed at T1" is at least in part about T2; by (C) the statementfails to express a hard factabout T1. But since Smithneed not be eternal (or essentiallyomniscient),this is a disastrous result for Adams' account. The same sort of argument shows that Adams must say that "Jones believed at T1 thatJones would do X at T2" does not express a hard factabout T1. This is because "It is not the case that timeat T2 thathe would do X at T2" is a Jones believed forthe first necessaryconditionof 'Jones believed at T, thathe would do X at of the Also, it is a necessarycondition (on both interpretations) truthof the statement,"Piece of salt S dissolved at T1j," thatS did expresses a not dissolve at T2. One wants to say thatthisstatement hard factabout T , but Adams' account does not capture thisintuition (since the statementis at least in part about T2). It is not easy to see how Adams could provide an account of 'necessary condition" which would avoid all the problems raised above. Withoutsuch an account, she hasn't presentedan adequate between hard and softfacts. explanation of the distinction 75
T2 "

JOHN MARTIN FISHER

III. THE INCOMPATIBILIST'S CONSTRAINT

Various contemporaryOckhamistshave argued thaton any acceptable account of the distinctionbetween hard and soft facts, God's prior beliefwillbe a softfactabout the past. I shall not here accounts of the distinction; discuss particularcompatibilist further on the account of the distinction rather,I shall sketcha constraint charmightuse to defeatanycompatibilist whichan incompatibilist That is, I shall develop an explanaacterizationof the distinction. tionof the claim thatGod's priorbeliefis a hard factabout the past; thisexplanation will notimplythat humanforeknowledgeis also a hard fact about the past. This mightprovide a way in which Pike claim and the asymmetry could defend both his incompatibility thesis-the thesis that God's foreknowledgeundermines human freedom in a way in which human foreknowledgedoes not. Consider the factthatCaesar died 2009 years priorto Saunders' his paper. What lies behind our view thatthisfactis not a writing hard factabout 44 B.C.? We mightsay thatit is a softfactabout 44 B.C. because one and the same physicalprocess would have counthis paed as Caesar's dying 2009 years prior to Saunders' writing per, if Saunders wrote his paper in 1965, and would not have his counted as Caesar's dying2009 years priorto Saunders' writing his paper in 1965. This captures paper, if Saunders hadn't written of the "futuredependence" of softfacts;a softfactis a factin virtue events which occur in the future. suppose that Smithknew at T1 thatJones would do X Similarly, at T2. Smith'sknowledgeis a softfactabout T1 because one and the same state of Smith's mind (at T1) would count as knowledge if Jones did X at T2, and would not count as knowledgeifJonesdidn't do X at T2. Exactly the same sort of futuredependence explains why both facts-the fact about Caesar's death and the fact about Smith's knowledge-are soft facts. might insist on the followingsort of Thus, an incompatibilist the factdistinction: constrainton an account of the hard fact/soft only way in which God's belief at T1 about Jones at T2 could be a softfactabout the past relativeto T2 would be ifone and the same stateof the mind of the person who was God at T1 would count as belief(or not a belief one beliefifJones did X at T2,-buta different at all) ifJones did not do X at T2. But it is implausibleto suppose thatone and the same stateof the mind of the person who was God 76

FREEDOM AND FOREKNOWLEDGE

at T, would count as different beliefs given different behavior by Jones at T2. Suppose again thatJones did X at T2. Y (being God) believed at T, thatJones would do X at T2. Let's say thatY's mind was in states at TI; this constitutedHis believing thatJones would do X at T2. Now if Y's mind were in state s and Jones did notdo X, Ys mind being ins would stillcount as a beliefthatJoneswould do X. (In this case, Y wouldn'tbe God, since he would have a falsebelief.)Hence, Ys mind being in s at T, would notcount as one beliefifJones did X at T2 and another belief (or not a beliefat all) ifJones did not do X at T2. Someone mightagree that the incompatibilist's constraint is appropriatebut disagree withwhat I have said about its application. That is, one mightargue thatifJones hadn't done X at T2, thenthe state of God's mind that actually constitutedHis believing that Jones would do X would not have constitutedthat belief. This position might be supported by extending Putnam's point that meaningsand beliefsain't in the head.'4 Accordingto Putnam,my belief that water is wet-the state of my mind that constitutes in fact,my believing that-would have been a different belief-the beliefthatXYZ is wet-if lakes and oceans on earth had been filled withXYZ rather than water. On this approach, the state of God's mind at T, that counts as His belief that Jones will do X at T2 counts as that belief partlyin virtueof the fact thatJones does in factdo X at T2' But thispictureof God's omniscienceis highly implausible.God's omniscience would be seriously attenuated if the same state of God's mind at T, would constitutedifferent beliefs about Jones, depending on Jones' behavior at T2. The followingis a more appealing picture of God's omniscience. An Ockhamist mightdeny the appropriateness of the constraint, claiming that while it's not true that one and the same state of God's mind at T, would constitutedifferent beliefs,depending on Jones' behavior at T2, it is truethatGod's mind would have been in a different stateat T, (from the one it was actuallyin), ifJones had not done X at T2. Whereas
'4Hilary Putnam, "The Meaning of 'Meaning"' reprintedin HilaryPutnam, Mind, Language, and Reality, (London: Cambridge University Press, 1975), pp. 215-271, esp. pp. 223-227. Robert Stalnaker suggestedto me the idea for the incompatibilist's constraint and pointed out the relevance of Putnam's point to it.

77

JOHN MARTIN FISHER

Y's mind was actuallyin states at T1, itwouldn'thave been in s, had Jones not done X at T2. If the Ockhamist makes this move, however, he weakens his argument to the conclusion that God's belief at T1 is a soft fact between soft facts such as about T1. There is now an asymmetry his paper and Caesar's dying2009 years prior to Saunders' writing one hand, on the X at T2, do will Jones that T1 at knowing Smith's and God's beliefat T1 thatJones willdo X at T2, on the other.But it was the assimilationof these sortsof factswhichwas the ground for claimingthat God's belief at T1 is a softfactabout T1. can agree withthe Ockhamistthat the facts The incompatibilist discussed above about Caesar's death and Smith's knowledge are "spurious" factsabout the relevanttimes.They are temporalanalogues of factsinvolving"mere Cambridge" spatial properties,such as the propertyof being ten miles south of a burningbarn. But if is rejected,then it is open to him to constraint the incompatibilist's argue that God's prior belief is a genuinefactabout the past. The constraintI have proposed captures the incompatibilist's is acceptable,then of the past. If thisconstraint notion of the fixity and the asymmeclaim his incompatibility both defend could Pike trythesis. There is, however,one formof Ockhamismwhichisn'tdefeated Jones' Consider again, "If itwas within by the proposed constraint. power in Jones' it was (3) then X, from doing to refrain at T2 power T1." at God been at T2 to act in such a way that Y wouldn't have There are two ways in which it mightbe true that it was in Jones' power at T2 so to act that Y wouldn't have been God at T1. First, Jones could have had it in his power at T2 so to act thatY wouldn't have existed at T1. Second, Jones could have been free at T2 to act in such a way thatY (though existing)wouldn'thave filledthe role of God at T1. The Ockhamistmightagree withPike thatthe existence of a particular person is a hard fact about a time, but he mightinsistthat the fact that the person is God is nota hard fact about a time. Adams) thatthe fact Thus, the Ockhamistmightclaim (following that Y had the propertyof being God at T1 is a softfactabout T1. This is because the fact that Y was God at T1 depends upon the truthof Y's beliefs about futurecontingentevents; indeed, since God is eternal, the fact that Y was God at T1 depends on the fact that Y existed at T2. 78

FREEDOM AND FOREKNOWLEDGE

should pointout thatfromthe claim that But the incompatibilist Y's occupyingthe role of God at T, is a softfactabout T, itdoes not follow thatJones could have at T2 so acted that Y wouldn't have been God at T1. There are softfactsabout the past whichare such thatone cannot now so act thattheywouldn't have been facts.For instance, on Tuesday, it was a soft fact about the past that on Monday it was the case that the sun would rise on Wednesday morning.'5 But on Tuesday, one could not have acted in such a way thatit wouldn't have been the case thaton Monday it was true that the sun would rise on Wednesday. Thus, even if the factthatY was God at T, is a softfactabout TI, to establishthatJones could have so acted at T2 thisdoesn't suffice that Y wouldn't have been God at T,. Further,it is theologically implausible to suppose that any human agent is free so to act that the person who is actuallyGod wouldn'tbe God. This would make of God dependent on human actions in an unacceptthe identity able way; such a God would hardly be worthyof worship. So, whereas the factthatY was God at T, mightbe a softfactabout T,, an Ockhamistwho claims thatone could have at T2 so acted thatY wouldn't have been God at T, would posit an unacceptable viewof can be defended even if Pike's claim thatthe God. Incompatibilism factthat Y was God at T, is a hard factabout T, were false.
IV. CONCLUSION

Adams' formulationof Ockhamism is inadequate. I have not fact distinction here argued that no account of the hard fact/soft can be given which captures the Ockhamist intuition.Rather, I have posed a challenge to Adams' Ockhamismand have presented motivationfor thinkingthat any Ockhamist the incompatibilist's challenge account willbe unacceptable. I have thusissued a twofold factdistincto formulatethe hard fact/soft to the Ockhamist: first, tion in a way whichyieldsOckhamism,and second, to explain why constraintis inappropriate.'6 the incompatibilist's Yale University
borrow this sort of example fromRowe, op. cit.,p. 165. have benefitedfromcommentsby Carl Ginet,Norman Kretzmann, T. H. Irwin, and Judith Jarvis Thomson. I am especially indebted to Robert Stalnaker, many of whose suggestionshave been incorporatedin this paper.
151 161

79