DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
Philosophers of religion have debated for a long time about whether divine foreknowledge is inconsistent with human freedom. The latest, but surely not the last, words on the subject are a substantial contribution to the contemporary literature in philosophy of religion.1 Recently, however, a novel element has been injected into this perennial debate. Richard R. La Croix has argued that divine foreknowledge, interpreted in a plausible way and taken together with certain other doctrines about the nature of God, is inconsistent with divine freedom? As one might expect, some of his arguments are strikingly similar to those which concern divine foreknowledge and human freedom. But others, which have to do with God's foreknowledge of his own decisions and actions, introduce the theme of divine self-knowledge into the debate in an interesting fashion. So, for this reason if for no other, the arguments La Croix sets forth are worth attending to. The strategy La Croix adopts in arguing is complex. He begins with an argument to show that anyone who holds that God has foreknowledge and is omniscient, eternal and immutable is committed to the view that God is everlastingly omniprescient. He then tries to demonstrate that this doctrine of divine omniprescience entails two principles about the relations between God's knowledge and his actions. Next he argues in favor of three general principles concerning the nature of decision-making. Using these five principles, he constructs two schematic arguments, each of which has as its conclusion the claim that God can at no time make any decisions about what he will do at any later time. From this he infers that the doc-
t See, for instance, Nelson Pike, "Divine Omniscience and Voluntary Action," The Philosophical Review, January 1965; Alvin Plantinga, God, Freedom and Evil, (New York, 1974), Part I, Section B; and Nelson Pike, "Divine Foreknowledge, Human Freedom and Possible Worlds," The Philosophical Review,
April 1977. Richard R. La Croix, "Omniprescience and Divine Determinism," Religious Studies, September 1976.
220
INTERNATIONAL IOURNAL FOR PHILOSOPHY OF RELIGION
trine of divine omniprescience has as a consequence the unwelcome thesis that there is no divine freedom. In this paper I propose to poke holes in this tightly knit fabric of argumentation at several different places. I shall begin with a brief reconstruction of the reasoning which leads to the doctrine that God is everlastingly omniprescient. Next I shall prove that one of the principles concerning the relations between God's knowledge and his actions, which La Croix supposes is entailed by that doctrine, is not as a matter of fact among its consequences. This will suffice to show that one of his arguments, constructed to show that divine omniprescience implies the absence of divine freedom, fails. Then I shall present counterexamples to two of the principles about decisionmaking advocated by La Croix. This will serve to show that La Croix has not provided good reasons for thinking that his other argument about the relation between omniprescience and determinsm is, as stated, sound. And, finally, I shall argue that, even if it should be true that God at no time decides what he will do at later times, this does not entail that God is not free or that he can only do what he does.
Two features of the arguments La Croix constructs are worth mentioning explicitly at the outset. First, La Croix assumes that future events are, while future, possible objects of knowledge. For the sake of argument, I shall follow him in making this assumption. Second, he makes no mention of the familiar difficulties associated with the opacity of propositional attitude contexts, the puzzles connected with quantifying into modal contexts, or the distinction between modalities de re and modalities de dicto. I too shall ignore these matters. To deal with them adequately would make both his arguments and my criticisms complicated to the point of obscurity, and I believe that nothing crucial, at the points where I wish to raise objections, hangs on such technicalities. Let us begin, then, with an explication of the doctrine of divine omniprescience. As La Croix first presents it, this is the claim that at some time God knows everything that will happen after that time. This doctrine can, I believe, be made quite precise with the aid of the following two definitions:
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
221
1. God is omniprescient at time t = df For every time t' and every event e, if time t is earlier that time t' and even e occurs at time t', then God knows at time t that event e will occur at time t'. 2. God is sometimes omniprescient = df For some time t, God is omniprescient at time t. La Croix observes that if God is ever omniscient and endowed with foreknowledge he is then omniprescient in this sense. However, this doctrine is compatible, as he notes, with the claim "that God is n o w omniprescient while he was not omniprescient, say, prior to creation and will not be omniprescient, say, at the day of judgment." '~ To obtain a stronger doctrine of divine omniprescience La Croix proceeds to invoke the traditional doctrines of divine immutability and divine eternality. He is well aware of the problems involved in trying to define the notion of divine immutability. Thus, for example, God may have the property of being prayed to by Moses at one time and lack that property at another time without himself having undergone any change or mutation. So, instead of trying to formulate a definition of immutability, La Croix offers us what can plausibly be interpreted as the following pinciple concerning God's immutability: . If God is immutable, then, if for some time t God is omniprescient at time t, then, for all times t', if God exists at time t', then God is omniprescient at time t'. Omniprescience is, according to La Croix, not the sort of property an immutable God could possess at some times and not at others. As he puts it: "On the assumption that God is immutable, and on an intuitive understanding of divine immutability and divine omniprescience it would appear to follow that if God is ever omniprescient, then he is a l w a y s omniprescient." ~ Or, at least, it would appear to follow that if God is ever omniprescient, he is omniprescient whenever he exists. But La Croix interprets the claim that God is eternal as asserting that he exists at every time instead of as stating that he exists, somehow, outside of time. On this interpretaion, a suitable definition of divine eternality may be framed as follows:
3 I b i d . , p. 367. 4 I b i d . , p. 368.
222
INTERNATIONAL JOURNAL FOR PHILOSOPHY OF RELIGION
4. God is eternal = df For every time t, God exists at time t. And, if we now assume that God is immutable, sometimes omniprescient and eternal, we can then deduce from our assumptions, together with ( 1 ) - ( 4 ) , the following principle: . For all times t and t' and for all events e, if time t is earlier than time t' and event e occurs at time t', then God knows at time t that event e will occur at time t'. The doctrine expressed by (5) might be thought of as the doctrine that God is everlastingly omniprescient. And it is clear that La Croix proposes to work with this doctrine of omniprescience, for he explicitly states that the doctrine of divine omniprescience is to be understood as the doctrine "that there is no time at all, ever, at which God does not have knowledge of all future events." 5 It seems to me that many theists are, explicitly or implicitly, committed to this doctrine. Indeed, I suspect that some reflective theists would maintain that God is not merely, as a matter of fact, everlastingly omniprescient but that he is necessarily everlastingly omniprescient. And so, we must now ask whether there are any nasty surprises for theists among the consequences of the doctrine of divine everlasting omniprescience.
II
In what follows some of the numbered sentences are quoted directly from the paper I am discussing. I have taken the liberty of renumbering these sentences so that they fit in with my numbering system, but I have marked each quoted sentence with an asterisk. ~ In some of the quoted sentences La Croix uses the letters 'TI', 'Te' and 'T3' schematically to stand for times. Though he is not explicit about this matter, I believe that the context makes it clear that he intends that certain restrictions be placed on the times picked out by such schematic Ibid., p. 369. 6 Thus, my sentence (6)* is the first sentence (5) and sentence (12) on p. 373;
(7)" is the second sentence (5) on that page; (8)* is (7) on that page; (10) ~ is (P2) on p.370; (31)*-(34)* are the first sentences (1)-(4) on p. 373; (35)* is (Q1) on p. 371; and (36)* is (Q2) on p. 372.
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
223
letters. In the first place, he assumes, I think, that time is linearly ordered. Secondly, he always supposes that T1 is earlier than T:, and that T~ is later than T1 and earlier than Ta. The reader will find it helpful to keep these assumptions in mind in interpreting the arguments which follow. La Croix presents in schematic form two arguments for the following schematic conclusion: 6.* It is not possible for God to decide at % to do a at T~I. The second of these arguments has among its premisses these two schemata: 7.* If God knows at T1 that he will do a at T:~, then it is not possible at Te for God to refrain from doing a at Ta. 8.* If God knows at T1 that he will not do a at T:,, then it is n o t possible at T2 for God to do a at T3. La Croix believes that the schemata expressed by these sentences follow from a schematic principle which is entailed by the doctrine of divine everlasting omniprescience. That schema may be adequately expressed as follows: 9. If God knows at T1 that he will do (will not do) a at T~, then it is not possible at T2 for God to refrain from doing (to do) a at T~. As a point of pedantic accuracy, the schema La Croix actually exhibits is slightly different. It has this form: 10.* If God knows at Ti that he will do (will not do) a at T3, then at T2 or T~ it is not possible for God to refrain from doing (to do) a at T3. There are several reasons for my decision to replace (10) * with (9). First, the disjunction at the beginning of the consequent of (10) *, T2 or T~, surely represents a typographical error of some sort. Second, because we are ignoring problems connected with quantifying into modal contexts, it makes no difference whether the first temporal qualifier in the consequent comes before or after the modal operator.
224
INTERNATIONAL IOURNAL FOR PHILOSOPHY OF RELIGION
Third, (9) has that temporal qualifier in the same place (7)* and (8)* do. And, fourth, the argument La Croix gives to show that what (10)* expresses is entailed by the doctrine of divine everlasting omniprescience depends in no way upon the features which distinguish it from what (9) expresses. Since the argument that the doctrine expressed by (5) entails the schema expressed by (9) will be the focus of my first major critical attack, it merits being quoted at length: The doctrine of divine omniprescience entails (9) because if (9) is false then it is possible for God to do each one of two things, namely, know at TI that he will do a at T~ and also re]rain from doing a at T3. But by the doctrine of divine omniprescience, if God refrains from doing a at T3, then God knows at T1 that he will refrain from doing a at Ts. So, if (9) is false, then it is possible for God to do each one of two things, namely, know at T1 that he will do a at T3 and also know at T1 that he will not do a at T~. Furthermore, by the doctrine of divine omniprescience, if God knows at T1 that he will do a at T3, then God does do a at Tz (otherwise he cannot know at T1 that he will). So, if (9) is false, then it is possible for God to do each one of two things, namely, do a at T3 and also reJrain from doing a at T3. 7 And, since, according to La Croix, the doctrine of divine everlasting omniprescience and the negation of (9) jointly entail contradictions, the doctrine of divine everlasting omniprescience has what (9) expresses among its logical consequences. Though this argument has the form of a rather complex reductio, it rests, I believe, upon a simple logical fallacy. In order to see that this is the case, let us look in detail at what follows from the assumption that (9) expresses something false. To simplify matters, I shall follow La Croix in omitting consideration of the case expressed by the parenthetical phrases in (9). It will become obvious that arguments exactly parallel to those I give can be constructed to deal with that case. If what (9) expressed were false, what is expressed by its antecedent as well as what is expressed by the negation of its consequent would be true: 7 La Croix, op. cit., p. 570. I have substituted '(9)' for '(P2)' throughout the quotation.
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
225
11. God knows at T1 that he will do a at Ta. 12. It is possible at Ts for God to refrain from doing a at T:~. To these two assumptions we may add as premisses three necessary truths. The first of them might be thought of as a consequence of the doctrine of divine everlasting omniprescience, as La Croix suggests, but it might equally well be considered a consequence of any adequate philosophical analysis of knowledge. The second represents a case of that implication from actuality to possibility about which even such a skeptic as Bayle said that it is "as clear as that two plus two equal four" and is "one of the clearest and most incontestable axioms of metaphysics." 8 They are the following: 13. If God knows at T1 that he will do a at Ta, then God does a at T,~. 14. If God does a at Ta, then it is possible that God does a at T~v 15. If it is possible at T2 for God to refrain from doing a at T:~, then it is possible that God does not do a at T:~. From (11), (13) and (14), we may conclude: 16. It is possible that God does a at T3. And from (12) and (15), we may infer: 17. It is possible that God does not do a at T3. But (16) and (17) do not contradict one another. Nor do they entail the necessary falsehood: 18. It is possible that God does a at T:~ and God does not do a at T~. To suppose that (16) and (17) entail (18) would be to commit the same fallacy involved in inferring from the premisses that it is possible that there are unicorns and that it is possible that there are no unicorns the conclusion that it is possible that there are unicorns and there are no unicorns. But it is obviously not the case that all things 8 Pierre Bayle, Historical and Critical Dictionary: Selections, tr. R.H. Popkin, (Indianapolis, 1965), pp. 152, 168.
226
INTERNATIONAL JOURNAL FOR PHILOSOPHY OF RELIGION
possible are also compossible. So, while it is true that, if (9) is false, it is possible for God to do a at T3 and it is possible for God to refrain from doing a at T~, it does not follow that, if (9) is false, it is possible for God both to do a at T8 and to refrain from doing a at T3. And, in fact, if (9) is false, God does a at Ta and does not refrain from so doing, though his refraining is a possibility. Taking the doctrine of divine everlasting omniprescience to be an additional explicit premiss in the argument does not serve to strengthen it. As La Croix observes in the passage I have quoted at length, that doctrine has in the present case this consequence: 19. If God refrains from doing a at Tz, then God knows at T1 that he will refrain from doing a at T:~. But, suppose we grant that the following analogue of (15) also a necessary truth:
is
20. If it is possible at T~ for God to refrain from doing a at T3, then it is possible that God refains from doing a at T 3. Still, (12) and (20) together only allow us to deduce this conclusion: 21. It is possible that God refrains from doing a at Ts. And, because (21) is prefixed with a modal operator, while the antecedent of (19) is not, we cannot validly infer from (19) and (21) the conclusion: 22. God knows at T1 that he will refrain from doing a at T3. But perhaps we should grant to La Croix, for the sake of argument, a modalized version of (19). Thus, the following principle about divine omniprescience seems plausible enough, even though it is not obviously a consequence of (19) and La Croix has not explicitly argued for it: 23.
If it is possible that God refrains from doing a at T3, then it
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
227
is possible that God knows at Tj that he will not do a at T3.
Then from (21) and (23) we may deduce: I
24. It is possible that God knows at T1 that he will not do a at T:~. Moreover, surely we may add to our premisses the following necessary truth, another instance of the implication from actuality to possibility: 25. If God knows at T~ that he will do a at T3, then it is possible that God knows at T~ that he will do a at T> And now we may deduce from (11) and (25) this conclusion: 26. It is possible that God knows at T1 that he will do a at T3. But, of course, (24) and (26) do not contradict one another. Nor do they together entail the necessary falsehood: 27. It is possible that God knows at T1 that he will do a at T3 and that God knows at T1 that he will not do a at T,~. As before, the fallacy here consists in inferring from the possibility of each of two conjuncts, taken, individually, the possibility of their conjunction. So, while it is true that, if (9) is false, it is possible for God to know at T1 that he will do a at T3 and it is possible for God to know at TI that he will not do a at T:~, it does not follow that, if (9) is (alse, it is possible for God to know at T~ both that he will do a at T:~ and that he will not do a at T.~. And, in fact, if ~9) is false, God knows at T1 that he will do a at T~ and does not know at T1 that he will not do a at T3, though such knowledge is a possibility. I confess that I can find no better arguments in the passage by La Croix quoted at length above than the two which contain the fallacy I have been at pains to expose. For this reason, it seems clear to me that he has not succeeded in showing that (5) en-
228
INTERNATIONAL JOURNAL FOR PHILOSOPHY OF RELIGION
tails (9) or (10)*. Thus, the attempt to show that (6)* is a consequence of, among other things, (5) by means of an argument which includes (7)* and (8)* among its premisses breaks down. In short, this argument does not establish that there is a relation of entailmen,t between the doctrine of divine everlasting omniprescience and the claim that God can never decide at any time what he will do at a later time. Moreover, it is evident that there is no reason for a theist, even one who accepts the doctrine of divine everlasting omniprescience, or for anyone else, to accept (7)*, (8)*, (9) or (10)*. It is easy to see why this is so. Suppose someone accepts what is schematically expressed by .(11). From (11) and (13) such a person can infer: 28. God will do a at Tz. But such a person would err grievously in inferring from (28) the following: 29. It is necessary that God will do a at Tz. This is the elementary logical blunder most often committed in arguments from foreknowledge to fatalism. Yet (29), or something equally strong, is needed in order to support the inference to this conclusion: 30. It is not possible that God will refrain from doing a at Ts. Yet it is from this conclusion that the negation of (12) does fop low. So (9) certainly does not express a necessary truth; its antecedent does not entail or strictly imply its consequent. Hence, for all that has been said so far, a theist can consistently maintain both that God has foreknowledge of every action he will, in fact, perform and that some or all of the actions which God will, in fact, perform are such that he might refrain from performing them, though, to be sure, he will not so refrain. And, since one can hold such a view while complying with all the proprieties of logic, it is open to a theist to claim that (7)*, (8)*, (9) and (10)* are in fact false and that, for this reason, the argument
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
229
La Croix bases upon them is unsound. Nothing La Croix says in support of (9) or (10)* serves to show that such a view would be mistaken or even to make it appear the least bit implausible.
Ill
But La Croix has another arrow in his quiver, a second argument whose conclusion is (6)*. Since it raises some fascinating questions about decision-making, it will turn out to be worthwhile to examine it with care. Its premisses, schematically represented, are these: 31.*
If it is possible for God to decide at T.~ to do a at Ta, then God has not yet decided at TI with respect to doing a at Ta. 52.* If God has not yet decided at T1 with respect to doing a at T3, then God does not know at T1 whether or not he will do a at T:3. 33.* If it is possible for God to decide at T.~ to do a at T~, then God does not know at T1 whether or not he will do a at T~. 34.* It is false that God does not know at T1 whether or not he will do a at T:~. It is evident that (6)* does follow from this group of premisses. Moreover, it becomes clear that (34)* is a consequence of the doctrine of divine everlasting omniprescience once we recall that TI is understood to be earlier than T:~. And, obviously, (33)* follows from (31)* and (52)*. But what is there to be said in justification of (51)* and (32)*? According to La Croix, they are, respectively, consequences of these two schematic principles about the concept of making a decision: 35.* 'x decides at T., to do a at T~' entails 'there is a time T1, prior to T._,, at which x has not yet decided with respect to doing a at T3 ~.
36.* 'x has not yet decided at T1 with respect to doing a at T~' entails 'x does not know at T1 whether or not he will do a at T~'.
230
INTERNATIONAL JOURNAL FOR PHILOSOPHY OF RELIGION
What we need to ask ourselves is whether (35)* and (56)* express truths so that we may find out whether they justify (51)* and (52)* Unfortunately this task is made somewhat difficult by the fact that La Croix nowhere explicitly states what he takes entailment to be. However, because philosophers whose opinions differ on what sufficient conditions for entailment are agree that strict implication is a necessary condition for entailment ", (35) * and (36) * can legitimately be taken to imply, respectively: 37. Necessarily, for all agents x, actions a and times T2 and Ts, if Te is is earlier than T3 and x decides at Te to do a at Tz, then there is a time T1, prior to Te, at which x has not yet decided with respect to doing a at T~. 38. Necessarily, for all agents x, actions a and times T~ and T3, if T~ is earlier than T3 and x has not yet decided at T1 with respect to doing a at T% then x does not know at T1 whether or not he will do a at T~. However, each one of this pair of principles seems open to grave, perhaps even fatal, objections. Consider first the proposition expressed by (37). It is, I suppose, possible that time itself has a unique first moment. As long as we stick with the assumption that time is linearly ordered, the notion of the first moment of time can be understood, in the Russellian manner, as some time such that every moment of time diverse from it is later than it. Indeed, there are some reasons for thinking that time actually does have such a first moment. For the metaphysical thesis that there is no time without change has an air of plausibility about it, and some cosmologists tell us that change began when the 'big bang' set the universe to expanding. But, then, it is possible that at the first moment of time God made all his decisions about what he was going to do at every subsequent time, in which case there would be no time prior to any divine decision at which God had not yet made that decision. In other words, we are entitled, 1 believe, to affirm at least this much: 39. It is possible both that God decides at the first moment of time to do a certain thing at some later time and that there is no time, 9 G . E . Hughes and M. I. Cresswell, A n Introduction to Modal Logic, (London, 1968), p. 26.
DIVINE FQREKNOWLEDGE AND DIVINE FREEDOM
231
prior to the first moment of time, at which God has not yet decided to do that thing at that time. And, of course, because what (39) expresses is inconsistent with what (37) expresses, we are consequently entitled to reject (37). I suspect that the reason why La Croix has not seen this possibility is that he takes the doctrine of God's everlasting existence to imply that time itself has neither first nor last moments and is of infinite duration. He says that "the eternity of God is explained as unending duration." '~ But this is simply a mistake. The doctrine that God is everlasting asserts no more than that he exists at every time, as the definition expressed by (4) makes abundantly clear. This doctrine by itself implies nothing at all about whether or not time has a first or a last moment or about whether its duration is finite or infinite. Nevertheless, though it is possible that there is a first moment of time, it must be acknowledged that it is also possible that there is not. So it seems to be possible that for every time at which God makes a decision there is a time prior to that decision. In short, what (31)* expresses does not appear to be necessarily false, and so it seems that La Croix may assume it if he wishes. The point is that, because what (57) expresses is false, his justification for this assumption fails; it is, in the context of his argument, a gratuitous assumption. In other words, La Croix has provided us with no reason for supposing that what (31)* expresses is true. For this reason, a theist who accepts the doctrine of divine everlasting omniprescience would be on firm ground in challenging the epistemic credentials of the argument which depends upon (31) * However, since La Croix clearly does accept the doctrine that God can only be everlastingly omniprescient if time has no first moment, perhaps we should concede to him for the sake of argument the assumption that there is no first moment of time. Once this concession has been made, then La Croix would, I believe, have a sound argument for a premiss which resembles (31)*. For suppose that at some time T,, God does decide to perform a certain action at a later time T::. Then, because time has no first moment, there is a time T1, prior to T.,, at which God has not yet made that particular decision. 10 La Croix, op. cir., p. 368.
232
INTERNATIONAL IOURNAL FOR PHILOSOPHY OF RELIGION
Hence, given this assumption, we are entitled to affirm something much like (31)*, namely, this schema: 40. If God decides at Te to do a at T3, then God has not yet decided at T1 with respect to doing a at T~. Then, if there is nothing wrong with the rest of the argument we are engaged in examining, we could use it to establish, not (6)*, but the weaker conclusion: 41. God does not decide at T2 to do a at Tz. Notice that in formulating (40) and (41) I have left out the modal operator which prefixes the antecedent of (31) * and whose negatio,1 prefixes (6)*. This is because it is not even clear that (31)* follows from (55)*. Certainly (31)* does not seem to follow from (57). Here the formal fallacy is the same as that involved in inferring from the premiss that, necessarily, if there are unicorns, there are horned horses, the conclusion that, if it is possible that there are unicorns, then there are horned horses. If such inferences were legitimate, it would be all too easy to prove the existence of any sort of possible but imaginary beast. Similarly, suppose that it is possible for God to decide at Te to do a at T~. This means that in some possible world God does decide at T 2 to do a at T3. But suppose in addition that in the actual world God intends at every moment of time prior to T2 to do a at T~. Then it is not the case that there is some T1 such that God has not yet decided at T1 with respect to doing a at T~. What this example shows is that (31) * itself might well be false. Since a parallel objectio~ cannot he directed against (40), it would seem reasonable to take (40) instead of (31)* as a premiss of the argument to be considered and to regard (41) instead of (6)* as its conclusion. But is the remainder of the argument flawless? Well, the only premiss we have not yet scrutinized is (32)*. Once we look carefully at (32) *, however, some serious problems can be raised about whether what it expresses is true and about (38), which is its justification. The reason for this is not hard to explain; perhaps it is most clearly stated in terms of a counterargument. Suppose that at a certain time T.~ God decides that he will perform a certain action at a later time T~. This divine decision is itself an event, and so, according to the
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
233
doctrine of divine everlasting omniprescience, at every time T1 earlier than T~ God knows that at T2 he will decide to perform that action at T3. Hence, given that God does make such a decision, someone who holds the doctrine of divine everlasting omniprescience ought to affirm: 42. God knows at T~ that he will decide at T 2 that he will do a at T~. Moreover, God surely is neither so foolish as to decide to do things he cannot do nor so inconstant as to change his mind. And so such a person ought also to assert: 43. God knows at TI that, if he will decide at T,2 that he will do a at T3, then he will do a at T:~. But, since God knows elementary logic, we may infer from (42) and (43): 44. God knows at T1 that he will do a at T:~. Therefore, if we assume that God does decide at Te that he will do a at T:~, then we ought to assert: 45. God has not yet decided at Ta with respect to doing a at T~ and God knows at T1 that he will do a at T:v In short, God knows what he will do as a result of his decisions in advance of making those decisions just because he knows what his decisions will be before he makes them. But, since what (45) expresses is inconsistent with what is expressed by (32)* and (38), if we accept (45) we must reject both (32)* and (38). Now (45) is certainly not explicitly contradictory. And who knows enough about the divine nature to be justified in claiming that it expresses something logically impossible? But it sounds odd. What are we to say about this? Normally, we humans do not know what we wil! do as a result of our decisions in advance of making those decisions, but then we are evidently not omniprescient. So arguing from what is normal in cases of human decision-making hardly seems de-
234
INTERNATIONAL JOURNAL FOR PHILOSOPHY OF RELIGION
cisive when what is at issue is what an omniprescient being could know. Yet the abnormality of God's case may suffice to account for the odd sound of (45). Moreover, it does appear that sometimes the outcome of a decision can be known in advance. Suppose Jones knows that his friend, Brown, plans to decide tomorrow whether to go to London next week. Knowing very well Brown's tastes and circumstances, Jones knows that Brown will decide to go to London next week. Knowing a great deal about Brown's abilities and opportunities, Jones also know that Brown will go to London next week. So, it would seem that in certain cases one of us can know about another what he will do as a result of a certain decision before that decision is made. Cases of self-knowledge of a similar sort are a bit harder to imagine, but they seem to me logically possible. Suppose Smith knows that, if White invites him to the evening concert tomorrow morning, he will then decide to go that evening, and that, if he then decides to go, he will go. And now suppose that Smith learns that White will invite him to the concert tomorrow morning. Smith, knowing a bit of logic, infers that he will then decide to go to the concert and that he will go to the concert that evening. Thus, it would appear, Smith knows what he will decide to do before he makes his decision and knows what he will do as a result of that decision before he makes it. There seems to be no logical impossibility in this story, and so it appears that we have some reason to reject (38). Even less do similar stories about God have the air of logical impossibility. After all, what (42) expresses entails only that God will decide at T2 that he will do a at Tz. It does not entail that it is necessary that God will so decide. And it certainly does not entail that God will not so decide or that his decision will be in any way constrained or bogus. Similarly, what (44) asserts entails only that God will do a at T:~. It does not entail that he must do so or that he cannot do otherwise. So, if there is something wrong with the argument from (42) to (45), it must lie with the initial supposition that God does decide at T2 that he will do a at T~. But this too seems to be logically possible. How, then, are we to deal with this puzzle? Perhaps it would be helpful to state the issue in a slightly different way. There can be no doubt that the following three propositions form an inconsistent triad:
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
235
46. There is a time T~ at which God decides that he will perform an action a at some subsequent time T3. 47. If there is a time T2 at which God decides that he will perform an action a at some subsequent time T.~, then there is a time T~, prior to T2, such that God knows at Ya that he will decide at T., to perform a at T~. 48. If there is a time T~ at which God decides that he will perform an action a at some subsequent time T:~, then there is no time T1, prior to T:~, such that God knows at T1 that he will decide at T2 to perform a at T:~. Now (47) is a consequence of the doctrine of divine everlasting omniprescience together with the assumption, granted here for the sake of argument, that time has no first moment. Hence, someone who holds that doctrine must, on pain of inconsistency, reject either (46) or (48). La Croix would have us reject (46). But it seems that (46) expresses something logically possible and that (48) expresses something which is not logically necessary. It would appear, then, that we could accept (46) and reject (48). What La Croix needs is a proof that (48) expresses a truth. But, if my analysis so far has been correct, no such proof has been provided. Still, it may be that (48), (41) or even (6)* does express a truth despite the fact that La Croix has not successfully demonstrated that any one of them does.
IV
So, La Croix might, at this point, simply dig in his heels and insist that (41) or even (6)* just does express a truth. I myself can think of no convincing argument to support such claims, but perhaps others would be prepared to say a word in behalf of this kind of dogma. Maybe some people will think it self-evident and not in need of justification. Thus, it might be said that it is just obvious that there can be no foreknowledge of genuine decisions. Or perhaps the appeal will be to our ordinary concept of making a decision. For instance, it might be thought to be conceptually necessary that, if anyone ever really makes a decision at a time, then no one has foreknowledge of that decision and whatever action may result from it. All this sounds quite implausible to me, for reasons I have already stated, but perhaps
236
INTERNATIONAL
JOURNAL FOR PHILOSOPHY OF RELIGION
we ought to allow it as an hypothesis in order to see if anything interesting can be deduced from it about what an omniprescient being can or cannot do. In particular, we would like to know whether it is the case, as La Croix believes, that it is "a consequence of the doctrine of divine omniprescience, that there is no d i v i n e lree will." 11 For the sake of argument, then, let us suppose that, on the assumption that God is everlastingly omniprescient, we may assert this very strong modal thesis: 49. It is not possible that there are a t, a t' and an a such that t and t' are times, t' is later than t, a is an action and God decides at t that he will do a at t'. What (49) tells us, in effect, is that God cannot make a decision at any time. God, of course, may still have intentions and purposes, but they do not change with time. He, as it were, never makes up his mind to act; his mind is always made up. Thus, whatever intentions and purposes God has he always has. God's intentions, we might say, are fixed, or, more picturesquely, we might say that from eternity God intends to do whatever it is that he is going to do. Hence, God's intentions never change, and he is possessed of an admirable constancy of purpose. And so we may, on his hypothesis, affirm about God the following principle: 50. Necessarily, for all times t and t' such that t' is earlier than t and for all actions a, God intentionally does a at t if and only if God intends at t t to d o a a t t and knows at t' that he is going to do aatt.
Needless to say, principles similar to the one expressed by (50) could be stated using the language of 'willing' or 'purposing' instead of the language of 'intending'. Exactly what unwelcome consequence is supposed to follow from this principle? Let us assume that for some particular action A and some particular time T it is asserted: 51. God intentionally does A at T. 11 I b i d . , p. 374.
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
237
From (50) and (51).we may infer: 52. For all t' earlier than T, God intends at t' to do A at T and knows at t' that he is going to do A at T. But we cannot deduce from (51) the fatalistic or deterministic conclusion: 53. It is necessary (logically or nomologically) that God intentionally does A at T. And so we cannot deduce from (50) and (51) together the fatalistic or deterministic conclusion: 54. It is necessary (logically or nomologically) that, for all t' earlier than T, God intends at t' to do A at T and knows at t' that he is going to do A at T. For (54) follows from (50) only if we also assume (53), and (53) does not follow from (50) and (51). Similarly, if we assume (52), we can deduce (51) from it together with (50). But (54) does not follow from (52). So, although (53) does follow from (54) and (50), it does not follow from (52) and (50). God, of course, does exactly those things he always antecedently intends to do and knows he will do, and he intends to do and knows he will do exactly those things he subsequently does. But it is consistent with this doctrine to hold that it is possible, logically and nomologically, for God to do things other than those he actually does, in which case he would antecedently have intended to do those other things and known that he would do them. It is also consistent with this doctrine to hold that it is possible, logically and nomologically, for God to intend to do and know that he will do things which are different from the things he actually does. From the assumption that God knows from eternity every act he will perform, we cannot infer that it is impossible for God to perform any act he does not actually perform or that it is impossible for God not to perform any act he does actually perform. Even given the assumption that God makes no decisions to act at particular times, this inference rests on nothing more substantial than the modal fallacy found in the traditional arguments from foreknowledge to fatalism. Since,
238
INTERNATIONAL IOURNAL FOR PHILOSOPHY OF RELIGION
at bottom, an inference of this sort is, as I have shown, what La Croix must rely on in his argument that the doctrine of divine everlasting omniprescience entails that there is no divine free will, that argument is ultimately no better than the traditional argument for fatalism. Even though God has, according to this doctrine in its strongest form, his intentions to act and his foreknowledge of his actions from all eternity, his intentions, actions and foreknowledge could all have been other than they are. Therefore, God's intentions, knowledge and actions are not constrained by laws of logic or nature to be exactly what they are in fact, even if he is everlastingly omniprescient. The most that we can legitimately infer from the doctrine of divine everlasting omniprescience, even when coupled with the dogma that God can never make decisions, is the conclusion that God is necessarily such that he never changes his mind about what he intends to do and, hence, what he does.
V
By way of conclusion, a brief summary of my argument may prove to be helpful. In the first section, I reconstructed the doctrine of divine everlasting omniprescience which follows from the claim that God has foreknowledge and is omniscient, immutable and eternal. In the second, I examined one of the arguments La Croix offers in support of the claim that this doctrine of omniprescience entails that God never makes any decisions. I showed that this argument depends upon a premiss which neither follows from the doctrine of divine everlasting omniprescience, despite the argument La Croix sets forth to prove that it does, nor is in its own right a necessary truth. I then pointed out that La Croix has given us no reason at all for accepting this principle. In the third section, I examined the other argument La Croix presents for the conclusion that the doctrine of divine everlasting omniprescience entails that God never makes any decisions. It rests on two general principles about decisions. I showed that the first of these is, as stated, false, but I conceded that a consequence similar to the one La Croix wishes to draw from it might nonetheless be true. So I allowed that he might assume this consequence provided he were willing, as he clearly is, to take for granted the doctrine about time which makes it plausible. But then I argued that there are counter-
DIVINE FOREKNOWLEDGE AND DIVINE FREEDOM
239
examples to the second of these two principles, and I claimed that La Croix has, therefore, given us no reason to accept the particular consequence of that principle he needs for his argument. However, I acknowledged that the conclusion of this argument might be true despite the fact that La Croix has not succeeded in establishing its truth. And so, in the fourth section, I examined the consequences of the dogma that God can never make any decisions. I showed that the claim that those actions which are such that God performs them and, hence, intends and foreknows from eternity that he will perform them are also such that God necessarily performs them arises from the very same fallacious form of modal inference which so often plagues discussions of divine foreknowledge and human freedom. All that follows from the assumption that God can never make any decisions is the unexciting conclusion that God is necessarily such that he never changes his mind about what he will do. In each possible world where he exists, God does exactly those things he has always intended to do. But, since God's intentions may vary with possible worlds, so too may his actions, and so we may not conclude that God does what he does necessarily. Perhaps it would be appropriate to end by giving a sketchy but positive account of how the claim that God exercises free choice might be reconciled with the claim that he at no time makes any decisions. We imagine God confronted with a choice among possible worlds he can actualize. Such a choice cannot take place at a time because decisions at times take place within possible worlds. To choose to actualize a possible world is to determine, among other things, whether time itself will have a first moment and whether its duration will be infinite. So let us further imagine that God chooses to actualize a possible world whose time has no first moment. In so doing, God, despite the fact that he makes no decisions at any time, determines the actions he will perform at each moment of time and the intentions to act he has at every moment prior to acting. This, in turn, determines the foreknowledge he has of how he will act at every moment of time. But God chooses among possible worlds freely in the sense that he could have chosen to actualize a possible world other than the one he does choose. And, if he had so chosen, he would have had different intentions than those he does have and would have foreknown and performed other actions than those he does perform. So it is up to God, and to no one else, to determine, without logical or
240
INTERNATIONAL IOURNAL FOR PHILOSOPHY OF RELIGION
nomological constraint, what actions he performs at any time and what intentions to act he has for all times before he acts. The only thing God cannot do, on this conception, is to alter his choice among possible worlds once it is made. He cannot, in other words, change his intentions once he has freely formed them or his actions once he has determined what they are to be. If he is unable to make decisions at times, God is necessarily such that his intentions remain constant through time. Yet he remains free precisely because he himself determines what constant intentions he has and what he is going to do at each moment of time. PHILIP L. QUINN
Brown Universtiy