Articles O M N I S C I E N C E A N D F R E E D O M FOR EVIL
JOSEPH
RUNZO
Chapman College
T h e a p p a r e n t i n c o m p a t i b i l i t y o f d i v i n e o m n i s c i e n c e with h u m a n free will, a n d the free will d e f e n s e as an a t t e m p t e d t h e o d i c y , are recurrent p h i l o s o p h i c issues. T h e s e t w o issues are o f t e n assessed s e p a r a t e l y . But they a r e logically i n t e r d e p e n d e n t . A n d the s o l u t i o n to the first is essential to the c l a r i f i c a t i o n o f the second. By casting the first o f these p r o b l e m s in terms o f G o d ' s f o r e k n o w l e d g e a n d the free, evil a c t i o n s o f h u m a n s , A u g u s t i n e p e r s p i c u o u s l y exposes the p o i n t o f intersection o f these t w o issues: Yet I still cannot see how G o d ' s foreknowledge of our sins can be reconciled with our free choice in sinning. God must, we admit, be just and have foreknowledge. But I would like to know by what justice God punishes sins which must be; or how it is that they do not have to be, when l t e foreknows that they will be; or why anything which is necessarily done in l t i s creation is not to be attributed to the Creator.'
H o w c o u l d G o d be o m n i p o t e n t , o m n i s c i e n t , a n d perfectly g o o d , a n d yet allow evil to exist? A u g u s t i n e a n d m a n y o t h e r s have a t t e m p t e d to solve the p r o b l e m o f evil with the free will defense: the r e s p o n s i b i l i t y for evil, o r at least m o r a l evil, is to be a s c r i b e d to the free a n d i n d e p e n d e n t a c t i o n s o f h u m a n s , n o t to G o d . But as A u g u s t i n e saw, the success o f the free will defense is logically d e p e n d e n t on the s o l u t i o n to a n o t h e r difficulty: viz., if an o m n i s c i e n t G o d f o r e k n o w s w h a t I shall d o - a n d surely, it seems, H e m u s t - then I c a n n o t act o t h e r t h a n I d o a n d , c o n s e q u e n t l y , I d o not act freely. H e n c e , the insistence that G o d is o m n i s c i e n t seems logically i n c o m p a t i b l e with the free will defense a g a i n s t the p r o b l e m o f evil. I will a t t e m p t to s h o w t h a t divine o m n i s c i e n c e is c o m p a t i b l e with h u m a n free will. In o r d e r to d o so 1 will focus c e n t r a l l y on an analysis o f the logic o f " o m n i s c i e n c e . " T h a t analysis will in t u r n p r o v i d e an i n d i c a t i o n o f h o w the free will defense can f u n c t i o n as a t h e o d i c y . It will be a s s u m e d here t h a t G o d , if H e exists, is b o t h o m n i s c i e n t a n d a t e m p o r a l being. A n d it will be a s s u m e d t h a t h u m a n s d o have free will.
l n t J P h il Rel 12:131-147 (1981) 0 0 2 0 - 7 0 4 7 / 8 1 / 0 1 2 3 - 0 1 3 1 $02.55. f~) 1981 Martinus N i j h o f f Publishers, The Hague. Printed in the Netherlamg,.
132
It will be helpful to begin with a preliminary analysis of "omniscient." "Omniscient" means "knows everything." But of course God neither knows how to ride a bicycle, 2 nor knows what it feels like t o y o u when you are in pain, nor knows who the first man's human ancestors were. There are three senses of " k n o w " : (i) have a competence - i.e., know h o w to, (ii) be acquainted with, and (iii) know a truth - i.e. know that. To say that God is omniscient is to say that God has all knowledge of the third sort. And of course God, like any common knower, cannot k n o w falsehoods though He can know the truth that a certain proposition, P, is false. As a first approximation, then, we can formulate the following definition of "omniscient": D1
A being, x, is omniscient if and only if, for any proposition, p, if p, then x knows p.
However, omniscience is a stronger notion than that expressed by D1. Cognizers can possess contradictory beliefs - we often do. And regardless of the multitude of truths which some actual being, X, knew, if X concurrently believed that the first man had human ancestors, that printing was invented in 1850, and that Beethoven only wrote eight symphonies, X could not be omniscient. Thus, "omniscient" can be redefined more accurately as:
D~ A being, x, is omniscient if and only if (i) for any proposition, p, if p then x knows that p, and (ii) for any proposition, p, if not-p then x does not believe p. God, qua omniscient, knows everything knowable - that is, He knows everything which a cognizer logically could know. Additionally, I will assume that the omniscient being God would know with certainty everything knowable. It would be senseless to suggest that God in fact knows that P is the case, yet that He only thinks that He knows that P and is not sure that He knows that P. Clearly, an omniscient being would know with certainty the truth of all propositions in the following categories: 1. Metalogical and logical truths 2. Metalinguistic and analytic truths 3. Propositions denoting actually occurrent past states of affairs 4. Propositions denoting actual present states of affairs 5. Propositions which can be logically deduced from propositions in (1)-(4) 3.
133 Moreover, an omniscient being would not only know with certainty all the truths in (1)-(5), but would infallibly know those truths. (Knowing all logical possibilities together with all actually occurrent past and present states of affairs, an omniscient being would have absolutely incontrovertible grounds as justification for his true beliefs regarding specific past and present states of affairs.) A harder question, though, is whether or not an omniscient being can know the truth of the following sorts of propositions: . Propositions denoting future contingent states of affairs which either (a) involve the future acts of creatures, or (b) do not involve any future free acts of creatures. Some philosophers - Augustine for example - have held that God can know future contingents; others have held that He cannot.' However, the answer to this issue depends on the answer to the logically prior question, " I f an omniscient God can know the future, how can He know it?" I will begin, then, by addressing this latter question, particularly as it relates to propositions about future free will acts of creatures (category [6a]).
11 In discussing the possibility of God's foreknowledge of the free acts o f human agents, Anthony Kenny observes that knowledge is justified true belief; and the justification of a past belief would have to be past grounds for the belief; and nothing in the past could be adequate grounds for a belief about my current action unless it necessitated that action, s
On the same topic, A. N. Prior has made a similar point in noting that "nothing can be said to be truly 'going-to-happen' (futurum) until it is so 'present in its causes' as to be beyond stopping; . . . . -6 Would an omniscient God, then, foreknow free human acts in virtue of the fact that He foreknows the causes of those acts? An action could be "necessitated" or "present in its causes" in either of two senses: it might be causally necessitated, or logically necessitated. The latter possibility is ruled out here because, ex hypothesi, the action in question is a contingent future state of affairs. 7 Therefore God could not foreknow human free acts on the grounds that their occurrence is selfevident. But for the theist employing the free will defense, the former possibility, that the free acts of human agents are causally necessitated, is also eliminable insofar as it is incompatible with the notion of human free will presupposed by the free will defense. To say that an agent acts freely is to say, in part, that if the agent had
134 chosen to do otherwise, then, ceteris paribus, s the agent would have done otherwise. But notice that for the theist employing the free will defense, the fact that if one had chosen otherwise, one would have done otherwise, is irrelevant if one could not (causally) have chosen to do other than one did. For if human " f r e e acts" are causally determinable in this fashion, then whenever humans commit evil acts, God couM (causally) have prevented those actions from occurring by causing the agent to choose otherwise, yet have retained the agent's " f r e e will" - since in that event, the agent " w o u l d have chosen otherwise and would have done otherwise." But then it is hard to see how God would not be blatantly culpable for the evil committed by the human agent. Hence, if God exists and is not morally culpable for human moral evil because humans act out of their own free will, then to say that an act is free must mean both that if the agent had chosen otherwise, he would, ceteris paribus, have done otherwise, and that it was causally possible for the agent to choose otherwise. 9 There can be no causally sufficient antecedent conditions for a free act if the moral evil of human agents is, as Augustine says, " n o t to be attributed to the C r e a t o r . " Therefore, if an omniscient God could foreknow the future contingent free actions of creatures, He could not foreknow this on the basis of the causal antecedents of those actions. Now, even though he himself employs the free will theodicy, Augustine holds that God can foreknow the future free actions of His creatures. Although God cannot foreknow these acts on the basis o f a knowledge of causal antecedents, Augustine offers a different explanation o f God's foreknowledge: If you foreknewthat someone was going to sin.... [this] does not of itself necessitatethe sin. ... As you, by your foreknowledge,know what someoneelse is going to do of his own will, so God forces no one to sin; yet He foreknows those who will sin by their own will.~~ Our knowledge of what our intimate friends will do does not make their actions any the less free; analogously, Augustine reasons, God's foreknowledge does not reduce human freedom. Since we do not strictly infer from causal antecedents what our friends will freely do, this sort of knowledge of future contingents would seem to provide a foreknowledge which is compatible with free will. However, Augustine's analysis fails. Imagine that you have a friend, Robbins, who has an insatiable appetite for ice cream. You go to a restaurant together, expecting that he rill order ice cream for dessert. He does, and you exclaim, " I knew that you would order ice c r e a m . " While common "foreknowledge" like this obviously leaves the free will of our friends intact, it is not sufficiently analogous to explain God's foreknowledge. What you foreknew - if anything - was the truth of the proposition "Robbins will order ice cream for dessert." On a
135 view like Augustine's, to be omniscient, a being must additionally know the truth of a rather different proposition - namely, "At Tn (some specific time) Robbins will order ice cream." At best, we know the truth of general propositions about the future; an omniscient being must, it would seem, know the truth of propositions whose referents are singular events, or states of affairs, together with their precise temporal location. More importantly, we achieve our quasi-foreknowledge of the future free acts of other humans by knowing something of their dispositions, inclinations, desires, etc. But as long as their acts remain free in the sense of having no causally sufficient causal antecedents, this sort of knowledge could never suffice for a certain knowledge of the truth of definite future propositions of the form, "At Tn, A will do X . " If God can foreknow with certainty the future free actions of creatures, His knowledge cannot be based on a like knowledge of human dispositions, etc.
I!1 The failure of Augustine's attempted analogy suggests, however, that perhaps God foreknows the future free actions of creatures apart from any evidence or grounds. In this vein, Anthony Kenny briefly attempts to solve the problem of how God could foreknow future free actions by suggesting that even in non-theological contexts there seem to be cases where true belief, without grounds, constitutes knowledge. One such case is our knowledge of our own actions. Commonly, we know what we are doing with our hands, and we do not know this on the basis of any evidence or grounds. ~1
It is true that if asked " H o w do you know that your hand is behind your back?", it is quite proper and sufficient to reply, "I just know it." Yet although one does not infer this truth and one might be initially unable to give good reasons for one's belief, there are, necessarily, good grounds for the belief. There is a causal connection between my hand being behind my back and my belief that my hand is behind my back. If there were no such causal connection, I could not know that my hand is behind my back. More formally, D. M. Armstrong has proposed the following helpful definition for such cases of non-inferential knowledge: A knows p non-inferentially if, and only if, A has no good reason for p but: i) A believes p; ii) p is true; iii) The truth of p is empirically necessary for A ' s belief that p.,2
136 As we have seen, for the free will defense to succeed, there can be no causally sufficient antecedents for a future free will act. Thus, applying Armstrong's analysis, there would be no causal conditions which could make a truth about a future free creaturely act empirically necessary for an omniscient being's belief in that truth. Hence, Kenny's suggested parallel between our knowledge of the location of the parts of our bodies and God's knowledge o f the free actions o f creatures breaks down. Additionally, even if no causal connection between bodily states of affairs and knowledge about those states of affairs were necessary for our own body location knowledge, Kenny's analogy is self-defeating. For the stronger one claims this suggested analogy is between our knowledge of our actions and God's knowledge of our actions, the more it is implied that the actions of creatures are in some sense extensions of God's actions. But this is inconsistent with the supposition on the free will defense that free creaturely acts are independent of God's active will. In God and Timelessness, Nelson Pike suggests a different and rather interesting analogy to explain how God could foreknow apart from any evidence or grounds. Pike suggests that God might know apart from any inferring or calculating much as a "crystal-ball gazer ... 'sees' future events in his crystal." ~3 Pike asks us to imagine a crystal-ball gazer who never makes a mistake but who also does not know how well what he says will come about does correspond to what in fact occurs. The central point of Pike's analysis is that if this seer is right every time, l think we would have to admit that he knows what is going to happen in the future - though we might also have to admit that we do not understand how his knowingdevice works, i,
Either there is a causal connection - although neither we nor the seer knows it - between the operation of the crystal-ball and the occurrence of future events, or there is no such causal connection. Suppose first that there is no causal connection between the functioning of the crystal-ball and the occurrence of some future event, E. In that case, the seer could not truly know that E will occur. The only helpful sense here in which we would like to know, as Pike says, " h o w his knowing-device works" is to discover some causal connection between E and the effects displayed by the ball. Otherwise, the crystal-ball simply is not a knowing device. At most, the seer - and God by analogy- would just correctly believe that E will occur. But why should we call this knowledge? Not knowing how the crystal-ball works, or how God might " f o r e k n o w " E, we are only justified in concluding that the seer, and God, just happen to be right - even if consistently right - in their beliefs. As Plato points out in the Meno, knowledge and mere true belief are distinct states of cognition despite the fact that as long as one " h a s true opinion about that which the other
137 knows, he will be just as good a guide if he thinks the truth, as he who knows the t r u t h . . . " 1~ Knowledge claims m a k e no sense unless we can, in principle, enunciate h o w one knows, and thereby why one does not possess mere true belief. We c o m m o n l y make knowledge claims apart f r o m any attention to this question only because we m a k e such claims in the context of a presumption that the subject does have proper grounds for his putative knowledge. God cannot, then, " j u s t k n o w " the future without grounds. 1~ In order to meet this objection to Pike's analysis, we might try to amend Pike's story. Suppose the hypothetical crystal-ball gazer does become aware of his frequency o f success. And, to give the strongest case, suppose now that there is some causal connection between the crystal-ball and future events. Could the seer's knowledge of his success provide good reasons to justify his beliefs, enabling him to have knowledge about the future? I do not think so. For although the seer will now have grounds for his beliefs, they are not good grounds and do not justify those beliefs. There are two possibilities to consider. Either the generalization describing the relation of crystal-ball effects to the occurrence o f future events involves a causal law(s) or it does not. Suppose first that the generalization does involve a causal law(s). Even so, the effects of the crystal-ball could not be causally sufficient for the occurrence o f future h u m a n free will acts, or there would be no free will in the sense requisite for the free will defense. But then the fact o f a future free will act cannot be strictly inferred from the effects o f the crystal-ball even if one knew all the relevant operative causal laws. Therefore the seer could not use the effects of the crystal-ball as a justification of knowledge claims about future free will acts. And, a f o r t i o r i , the suggested analogy with G o d ' s knowledge fails since G o d ' s knowledge is certain knowledge and the crystal-ball could not possibly provide certain knowledge. Turning to the second alternative, that no actual causal laws are operative here, presumably the functioning of the crystal-ball is in fact neither a necessary nor a sufficient condition of the occurrence of future free will acts, and the causal connection between the two is merely incidental. The seer's beliefs would, then, be like believing that it will rain on any day when one gets out of bed left foot first, because one has noticed that, in fact, every time one gets out of bed left foot first it has rained that same day. Assuredly, the fact that it rains and a left foot egress f r o m bed are only coincidentally related. And if the effects displayed by the crystalball and the occurrence of certain future events are likewise only coincidentally related, the seer would not k n o w the future any more than the foot-conscious rain predictor knows that it will rain. Therefore, even assuming that the seer knows the frequency o f his success, the crystal-ball gazer will not serve as an analogy for how God might foreknow the future free will actions of creatures.
138 Finally, if God cannot "just foreknow" without grounds, and if it is assumed (as, e.g., Pike assumes) that God does not have foreknowledge on the basis of evidence about the causes o f the future contingent free acts which He foreknows, one last alternative comes to mind. Could God have justified true beliefs about the future simply because He knows that He is God and omniscient? This is not merely to suggest that God just knows that He foreknows. That is a vacuous move - for .why suppose God does know that He foreknows given that we have as yet no grounds for acknowledging that He does foreknow? Rather, on this account God would know that since He is omniscient, every belief which He has must, by definition, be true. Thus, if God knows that He believes that future event E will occur, then He also has good grounds for believing that E will occur since He knows that He is God and omniscient. Even granting that we do not first need an answer to the question, " H o w does God know that He is God and omniscient?", this account does not explain how God could have justified true beliefs about future states of affairs. Let us say that God believes that Nero will fiddle tomorrow while Rome burns. What God's knowledge that He is omniscient would justify on this account is God's second order belief that His first order belief that " N e r o will fiddle tomorrow while Rome burns" is true, given that He does hold this first order belief. However, we have not as yet seen that God would have any good grounds for such first order beliefs, in and of themselves. So why, in the first place, would an omniscient God possess such a belief as " N e r o will fiddle tomorrow while Rome b u r n s " ? One reasonable possibility is that an omniscient God - just like any common cognizer - simply correctly guesses, and so believes, the truth of propositions about the occurrence of future states of affairs involving free will acts of creatures. 1, This possibility is worth assessing in some detail. But unless God would hold first order beliefs like " N e r o will fiddle t o m o r r o w , " His knowledge that He is omniscient is irrelevant to whether or not He possesses justified beliefs about future contingents.
IV Consider the assumption, then, that God could correctly guess, say a t / ' 1 , that " X will do A at T3." The conditional, " I f God correctly guesses at 7"1 that ' X will do A at T3,' then X does A at T3," is surely a true conditional. And if an omniscient God could antecedently correctly guess the occurrence of all actually occurring, future creaturely acts, it would seem that for any human agent, X, the consequent - " X does A at T3" - will be true at all moments subsequent to TI. But then at the moment at which the human agent does act, the agent could not have done otherwise. Conse-
139 quently, no human actions wodld be free. 18 Hence, on the assumption that humans do exercise free will, either God is not omniscient or He does not exist. There are two objections to this line of reasoning. First, as Gilbert Ryle has observed, " c o r r e c t " functions " m o r e like a verdict than a description. '''9 O f course one's future tensed guess cannot be correct unless the predicted state of affairs obtains. One's guess, however, is not correct until the predicted state o f affairs does obtain. Unlike past or present tensed guesses, future tensed guesses are initially neither correct nor incorrect. Future tensed guesses do not even turn out (at first) to have been correct to incorrect. At the moment when they become correct or incorrect, future tensed guesses only turn out to be correct or incorrect (unless they are nonsensical). Hence, no omniscient being, any more than you or I, can correctly guess at 7"1 what will later occur contingently at T2. In reply, it might be suggested that although God cannot guess correctly at T,, He can at least guess at T1 that " X will do A at T3" - a guess which turns out at T3 to be correct. Thus we might say that God held a true belief at 7", .20 And, this reply continues, it seems true that " i f God believes truly at TI that ' X will do A at T3', then X will do A at 7'3." But then at any moment after TI, the antecedent of this true conditional would already denote a past state of affai.rs, and the consequent would then be true. Universalizing this point for all human agents and all purportedly free human acts, there would be no human free will acts. The second objection is directed against this reply. The second, and more important, objection to the notion that God might foreguess or have true future tensed beliefs about states o f affairs which have not yet obtained is that the truth of non-logically necessitated propositions about the existential status o f future contingent states of affairs is indeterminate. The truth of such propositions about future contingents is indeterminate because the existential status of all future contingent states of affairs is indefinite. Obviously there are innumerable states of affairs which obtain at some moment which is temporally posterior to some (in fact every) previous moment. If Nero fiddles tomorrow while Rome burns, then he fiddles after all the events which occurred today. And if Nero fiddles tomorrow while Rome burns, then it is true tomorrow that Nero fiddles while Rome burns. But it is not true today. Although we can refer today to the state of affairs, " N e r o fiddles at T t o m o r r o w , " our statement is neither true nor false today. For it is still an open question whether the statement, " N e r o fiddles at T t o m o r r o w , " refers to the state of affairs which actually obtains. At the least, any such statement about possible future states o f affairs should be understood as implicitly governed by the sentential operator, " I f Nero continues to exist in the future . . . . " And possible states of affairs, even
140 probable states of affairs, are not actual states of affairs; likewise a possible truth, even a probable truth, is not a truth. The only kind of future states of affairs - in an admittedly odd sense of " f u t u r e " - about which we can make antecedently true existential statements is logically necessitated states of affairs. Thus a citizen of Nero's Rome might state the truism that " A t 12:00 noon tomorrow Nero cannot be both dead and alive." This kind of antecedent truism can be stated only because among future states of affairs, logically necessitated states of affairs alone are unavertable. In contrast, Nero's future fiddling (or lack of fiddling) is a contingent, and hence still avertable, state of affairs. In this context, it is helpful to consider Charles Peirce's analysis of the concept of " t h e future": future facts are the only facts that we can, in a measure, control; and whatever there may be in the Future that is not amenable to control are the things that we s h a l l be able to infer, or s h o u l d be able to infer under favorable circumstances. 21
If we follow out this line of thought, we can formulate the following sufficient condition 22 for a future contingent state of affairs: /)3
A future state o f affairs, S, is contingent if S is, at least in principle, controllable in some measure by an antecedent free will act.
It follows f r o m / ) 3 that propositions about future contingents are neither true nor false prior to the purported moment of occurrence of the specified state of affairs, x3 For future contingent states of affairs can be averted, and any putative state of affairs which is still avertable has an indefinite existential status. In general, while it may be that there always will be some specific, contingent future, it is never the case that there is, now, an identifiable contingent future. Thus for example, there can be no list of future free acts. To rephrase a remark of Ryle's, the list of free acts and the list of committed free acts is identical. ~4 And with no future free acts - good or evil - to be foreknown by any knower, it would appear that divine omniscience is compatible with the continued occurrence of human free will actions. However, Nelson Pike argues in God and Timelessness that the question of the prior truth of propositions of the form, " ( h u m a n agent) X does A at T , " is irrelevant to the issue of the compatibility of divine foreknowledge and human free will. Suppose Jones does A at T~. Pike reasons, if " J o n e s does A at T 2 " is true, then it must be known (infallibly believed) at T1 by an omniscient being. This is not to say that it is t r u e a t T I - it m e a n s only that it is b e l i e v e d (or known) a t T I . 2s
141 First, note that unless " J o n e s does A at T~" is true at 7',, God cannot know at 7"1 that " J o n e s does A at T2." And, I have argued, since it is neither true nor false at TI, this proposition is not knowable at 7",. Second, if as Pike himself implies, " J o n e s does A at T2" is not true at T,, what reason is there for supposing that God even believes at TI that " J o n e s does A at T , " ? As definition D2 indicates, in ascribing omniscience to God, we only imply that God believes what is knowable. Hence, although " J o n e s does A at T2" turns out to be true at T, (and thereafter), God would remain omniscient even if He does not believe at T~ that " J o n e s will do A at T2." And why should God commit Himself on the issue o f what Jones might do? In view o f the fact that God cannot know what Jones will do, it would seem that a suspension of judgment would be the wise course for God to follow. 26 To return to our earlier example, I take it then that God would not even believe at 7", that " N e r o will fiddle at T3." And clearly enough, the knowledge which God logically could have, and would have, about Nero's future actions - e.g., that "either Nero will fiddle tomorrow, or Nero will not fiddle t o m o r r o w " - leaves Nero free the next day to fiddle, or to refrain from fiddling. Universalizing this for all human actions, human free will is compatible with a temporal God's omniscience.
V Nevertheless, perhaps the significance for the free will defense of the unknowability of future free will acts is even more powerfully exhibited by an additional consideration beyond this assessment of human free will acts. After posing the foreknowledge-free will problem in On Free Choice of the Will Augustine sees that one issue which naturally arises is, If everything of which God has foreknowledge happens, not by will, but by necessity, shouldn't you be careful lest you say that God does what He is going to do by necessity too, and not by will? ~7
Thus, if God could foreknow that " X will do A at T," it seems unavoidable that X does A at T even if the agent, X is God Himself. However, God could not foreknow, for instance, that " G o d freely opens the waters o f the Red Sea at T." For, if God acts freely and his actions are contingent, 28 then before T there can be nothing true or false for God to foreknow here - otherwise God's own actions would be unavoidable. Now, in order to see the full consequences of this, consider a case where God does not actively perform a miracle. Could God foreknow that "Vesuvius will erupt at T " ? I suppose that what God would be said to foreknow in a case like this is that (a) given the operative causal laws, the
142 current state o f the universe is causally sufficient to bring about an eruption, and (b) God does not intend to suspend the operative causal laws, or p e r f o r m other miraculous feats which would disrupt the sequence of causes leading to an eruption. Item (b), though, involves G o d ' s own free will choice not to disrupt normal causal processes. Although G o d could infallibly know at TI what H e intends - at T, - to do at some later m o m e n t , T2, H e could not know at 7"1 what H e will in fact freely do at T2. (Similarly, we can know today what we intend - today - to do tomorrow, even though we cannot know what we will in fact freely do tomorrow.) Consequently, if God exists, sentences expressing propositions about future contingent states of affairs should be understood as governed by a sentential operator o f the form, " I f God does not change His mind, then . . . . " Thus, the inability of God to foreknow His own future free will acts provides another reason why God cannot foreknow the occurrence of (any) future contingent states of affairs. For, all future contingent states o f affairs would either be, or would result directly from, a free will act of G o d ' s , or they would involve His free will act to allow, or not to allow, causal process to continue. Keeping this point in mind, we can now address the sort of forceful argument which Nelson Pike offers in G o d a n d Timelessness for the incompatibility of h u m a n free will with the infallible omniscience of a temporal God. If God is essentially omniscient and a temporal being, then, says Pike, " I f Jones does A at T2, then [God] believes at T I that Jones does A at T 2 . " He then reasons that if Jones has free will, it is within Jones's power at T, to refrain f r o m doing A. In that case, if G o d exists, Jones has the power so to act that either (a) what G o d believed at T~ was false or (b) G o d did not hold the belief which H e did hold at 7",. Pike argues that since both (a) and (b) are impossible, infallible divine foreknowledge is incompatible with h u m a n free w i l l ) 9 In God, F r e e d o m , a n d Evil, Alvin Plantinga attempts to defend the compatibility of omniscience and h u m a n free will by reasoning against this argument of Pike's that We are told.., both that in the actual world God believesthat Jones does X at T2 and also that it is within Jones' power to refrain from doing Xat T2. Now consider any world Win which Jones does refrain from doing X. In that world, a belief that God holds in the actual world in Kronos - is false.... But it does not follow that in WGod holds a false belief. For it doesn't follow that if W had been actual, God would have believed that Jones would do X at T2. Indeed, if God is essentially omniscient (omniscient in every world in which He exists) what follows is that in W God did not believeat 7"1that Jones will do X at 7"2;He believed instead that Jones will refrain from X. S o . . . it was [not] within Jones' power to bring it about that God held a false belief at 7",.3~ In his recent article, "Divine Foreknowledge, H u m a n Freedom and Possible Worlds," Pike effectively rebuts this objection of Plantinga's.
143 As Pike points out, The question is not whether there is just some possible world or other in which Jones refrains from doing X at 7"2. What must be asked is whether there is a possible world, having a history prior to 72 that is indistinguishable from that of the actual world, in which Jones refrains from doing X at 72. The answer is that there is not. All such worlds contain an essentially omniscient being who believes at TI that Jones does X at 72. There is no possible world o f this description in which Jones refrains from doing X at 72. ~' "
If God does in fact do so in the actual world, once God comes (infallibly) to believe before 7"2that "Jones will do A at T~," Jones would not then be free to refrain from doing A at I"2 in the actual world. However, despite the effectiveness of his reply to Plantinga, Pike fails to show that infallible divine omniscience and human free will are incompatible. For both Pike's and Plantinga's analyses here depend on the false assumption that some cognizer, such as God, might foreknow human free will acts. On the one hand, God cannot know the truth of propositions denoting future contingent states of affairs involving the future free acts of creatures (category [6a]). And on the other hand, as a result of His inability to foreknow His own free will acts, God cannot know the truth of propositions denoting future contingent states of affairs which do not involve any future free acts of creatures (category [6b]).32 Consequently, in terms of possible worlds, we can derive the following principle, P, regarding God's omniscience. P
In any possible world, IV, in which God exists, and for any temporally successive moments 7"1 and T2, and for any agent X (including God) and any action A in IV, God logically cannot at 7"1 know the truth of any proposition of the form " X will do A at T2."
In sum, God could not foreknow the truth of any propositions referring to future contingent states of affairs. And even the knowledge that He is God and omniscient, would not provide God with a justification for beliefs about the future - for God would also know that He did not know the future. Yet this does not impugn God's omniscience. For there is nothing here for any cognizer to know - whether an omniscient or a finite knower. While not knowing the truth of propositions in category (6), then, what God would know is the truth of propositions in categories (1)-(5). And since God would know these truths infallibly (and with certainty), whatever God knows, He knows infallibly. 33
144 u We are now in a position to assess briefly how the free will defense can function as a theodicy. Setting aside the question of God's possible culpabilities for physical evils, I will restrict my comments to the function of the free will defense vis-a-vis moral evil. The free will defense against the problem of moral evil is often presented in terms such as these: although God foreknows the evil actions which humans will perform, since humans so act out of their own free will, humans, not God, are responsible for moral evil. But in view of the logical limitations on omniscience which are encapsulated in principle P, the free will defense should be stated instead in terms of two salient conditions. (1) Humans are directly morally responsible for the moral evils which they commit or allow, and (2) God is not even indirectly responsible for those moral evils since for any specific human agent, X, and any specific evil human act, A, God cannot (logically)foreknow the truth of any proposition of the form " X performs A at T." Now, principle P not only entails that free human actions are compatible with divine omniscience; it also helps explain why God is not culpable for human moral evil. Principle P tells us more than just that for any particular evil human action, God cannot foreknow the occurrence of that evil action. More strongly, principle P entails that God could not have foreknown either whether or not there would be any moral evil in the actual world, or, if there were moral evil, how much moral evil there would be. Imagine the actual moment, To, at which God decides to instantiate a created world. Which world would He create? Presumably a world containing as little moral evil as possible. But which world is that? For any morally responsible agent which God creates, He could not foreknow how much evil - or good - that actual moral agent would produce. Moreover, given human free will acts of reproduction, 34God could not even foreknow how many human moral agents there would be - each agent acting for evil and/or good. And finally, God could not even have a sure foreknowledge that if He creates certain conditions, then some human agents or other will commit evil acts, even though the identity of the actual malefactors is dependent on the free, interpersonal relations among the human agents. For if every particular moral act of every human agent is indeed a freelycommitted act, then it must be causally possible - however improbable that no human agent ever in fact commits an immoral act. Therefore, at To, for any possible world W containing free moral agents, which God instantiates, He could not foreknow the eventual incidence of morally evil versus morally good actions in W. God would only be morally reprehensible for actual human moral evil if (a) the probability35that there would be a high preponderance of moral evil
145 over good in the actual world renders morally unjustified the risk taken in producing the (non-moral) good of existent, free moral agents," and if (b) God could have foreknown (a). It is not at all clear that God could have foreknown (a). For if God logically cannot know how He will Himself act in the future, how many human agents will come into existence, and what free will act any creature will commit in any specific instance, it is difficult to see what basis there would be for assigning a probability to any specific, possible future incidence of evil. Finally, principle P also indicates something about what God would have instantiated at the moment of creation, To. On principle P, God could not have instantiated that created world which is now the actual created world. God could not knowingly instantiate any specific possible world and this is all the more true for any possible world containing created agents with free will. What God would instantiate at To is not a completed possible world, but a set of entities - e.g., causal laws, material states of affairs, dispositional properties, etc. - such that after To only a restricted subset of the set of all logically possible worlds becomes the set of all logically possible worlds which could be the actual world. That restricted subset would be those possible world which included at To just those causal laws, etc., actually instantiated at To. I take it that what God would know at the moment of creation is which logically possible worlds are no longer candidates for being the actual world. What God would not know at the moment of creation is what subset of the then potential actual worlds will remain potential actual worlds at any later time. While the number of logically possible worlds which could turn out to be actual continuously decreases as time passes, at any specific moment, T,, there is a set of such possible worlds which possess precisely the same description as the actual world has up until Tn. And it is these many possible futures for the actual world which make possible human freedom for good, or for evil. 37
NOTES
1.
2.
3.
4.
St, Augustine, On Free Choice o f the Will, trans. A n n a S. Benjamin and L. H. Hackstaff (Indianapolis: Bobbs-Merrill, Library of Liberal Arts, 1964), Bk. 3, Sec. 4, p. 94. Although God would presumably know how bicycles are ridden by others, to know how to "ride a bicycle" means, in part, to be able oneself to sit on a bicycle, pedal it, and either ride " n o h a n d s " or hold onto the handle bars. E.g., if God knows that Beethoven was a bachelor and that bachelors are unmarried men, then He knows that Beethoven was not a married man. This particular consequent would be subsumed under category (3), but category (5) is not superfluous since G o d ' s knowledge of what is logically possible but not actual would come under category (5). In a sense, T h o m a s Aquinas, for example, holds a position which is between these two
146
5.
6.
7. 8. 9.
extremes. Aquinas argues that God knows future contingents, but not qua future. (See Summa Theologica, Q. 14, Art. 13 and De Veritate, 2, 12.) Anthony Kenny, "Divine Foreknowledge and Human Freedom," in Aquinas: A Collection of Critical Essays, ed. Anthony Kenny (Garden City, N.Y.: Doubleday & Co., Anchor Books, Modern Studies in Philosophy, 1969), p. 268. A. N. Prior, "The Formalities of Omniscience," from Philosophy (1962), repr. in Readings in the Philosophy of Religion, ed. Baruch A. Brody (Englewood Cliffs, N.J.: Prentice-Hall, 1974), p. 422. See Jonathan Edwards, Freedom of the Will, Sec. 12, I1. E.g., someone else does not physically restrain the agent from carrying out his choice. Indeterminism need not be endorsed here. An agential view of human action would allow that each free will act is caused, but that not every event is caused by another event free acts are caused, at least in part, by the agent. (See, for example, Roderick Chisholm, " H u m a n Freedom and the Self", in William K. Frankena and John T. Granrose, Introductory Readings in Ethics [Englewood Cliffs, N.J.: Prentice-Hall, 1974], pp. 289-94). Augustine, On Free Choice of the Will, Bk. 3, Sec. 4, pp. 94-95. Friedrich Schleiermacher offers the same analysis of God's foreknowledge (see The Christian Faith, par. 55). Kenny, "Divine Foreknowledge and Human Freedom," p. 268. D . M . Armstrong, A Materialist Theory of the Mind (London: Routledge and Kegan Paul, 1968), p. 189. Nelson Pike, God and Timelessness (London: Routledge and Kegan Paul, 1970), p. 55. Ibid. Plato, Meno, 97b. This passage is translated by Benjamin Jowett. (I mean to agree with this distinction in itself, but not necessarily with the criteria which Plato uses to distinguish knowledge from true opinion.) I think that, like Pike, Paul Helm mistakenly assumes, in "Divine Foreknowledge and Facts," that God need not have evidence or grounds for His knowledge (Paul Helm, "Divine Foreknowledge and Facts," Canadian Journal of Philosophy, Dec. 1974, p. 306). In "The Formalities of Omniscience" (p. 420), A. N. Prior concludes that future contingents can at most be correctly guessed. Thomas Aquinas takes up a similar issue concerning God's foreknowledge rather than, as here, divine fore-guessing, in the Summa Theologica, Q. 14, Art. 13. For an assessment of Aquinas' reply, see Anthony Kenny, "Divine Foreknowledge and Human Freedom," pp. 260f. Gilbert Ryle, " i t Was to Be," in Dilemmas (Cambridge: Cambridge University Press, 1966), p. 20. Implicitly in God and Timelessness, and explicitly in "Divine Foreknowledge, Human Freedom and Possible Worlds," (Philosophical Review, 1977, pp. 209-16), Nelson Pike analyzes "omniscience" as "believes all true propositions and believes no false propositions." However, it seems to me that we would not regard a being as omniscient who claimed to possess innumerable true beliefs - but who did not claim to know anything. Thus (as definition D2 indicates), I think that it is inadequale to define "omniscience" in terms of mere belief. Hence, the shift at this point to "mere belief" would apply, at most, only to God's future tensed beliefs. The past and present tensed beliefs of God must correspond to instances of His knowledge. Charles S. Peirce, Collected Papers of Charles S. Peirce (Cambridge, Mass.: Belknap Press of Harvard Univ. Press, 1960), v. 5, p. 312. Since it is at least logically possible that there be a purely random occurrence, R, such that for a possible state of affairs, S, the current state of the universe together with R is -
10.
11. 12. 13. 14. 15.
16.
17. 18.
19. 20.
21. 22.
147
23. 24. 25. 26.
27. 28. 29. 30. 31. 32. 33. 34. 35. 36.
37.
causally sufficient for not-S, D~ is not additionally a necessary condition for future contingents. Peirce himself draws this conclusion, a position which Prior also endorses in "The Formalities of Omniscience" (p. 424). See Ryle, "It Was to Be," p. 26. Pike, God and Timelessness, p. 71. Pike develops his argument specifically against A. N. Prior's position on this issue. A different objection than that noted here to Pike's analysis of Prior's treatment of foreknowledge and freewill is offered by Paul Helm in "Pike on Prior on Action," Philosophical Studies, 1974, pp. 141-143. Augustine, On Free Choice of the Will, Bk. 3, Sec. 3, p. 91. Assuming that God is not - as is often held - immutable, then God might decide to perform some action, and then at the last moment freely change His mind. Pike, God and Timelessness, pp. 58f. Alvin Plantinga, God, Freedom, andEvil, (New York: Harper and Row, Harper Torchbooks, 1974), pp. 70-71. Nelson Pike, "Divine Foreknowledge, Human Freedom and Possible Worlds," Philosophical Review, 1977, p. 216. At the least, God would will to keep from interfering in normal causal processes. Thus, although "omniscience" and "infallibility" are logically distinct notions, the extension of these terms is identical with respect to God's knowledge. And possibly the ever present possibility of God changing His own mind about how many humans He will allow to exist. Notice that the issue here is not what the balance of moral good over evil turns out in fact to be in the actual world. It might be objected that the risk involved in creating creatures with free will who might very well commit the horrendous moral evils which do occur would surely outweigh any plausible reason for taking a chance on sucha world. This raises the issue of theplausibility of God's existence given the fact of evil, rather than the question of the logical compatibility of God's existence with the fact of evil. And it seems to me that this issue is not susceptible to argument so much as to persuasion and valuation. The answer here depends on how much emphasis one places on the non-moral good of free will, the emphasis one places on actual moral evils which are committed, and that placed on the moral, as well as the non-moral, goods produced by actual human agents. I am indebted to Nelson Pike for conversations which helped me develop several of the ideas in this paper, and to Gregory Kavka and Jack Meiland for their valuable comments on an earlier version of this paper.