9
In ternational Journal for Philosophy of Religion 14:10 7-116 (1983). Martinus Ni/hoff Publishers, The Hague. Printed in the Netherlands.
HERE WE GO AGAIN: PIKE vs. PLANTINGA ON THE PROBLEM OF EVIL
KEITH J. COOPER University of Wisconsin - Madison
For some fifteen years now, Alvin Plantinga has been offering the Free Will Defense (FWD) in rebuttal to those who claim that the theist is inconsistent in affirming both the existence of God (an omnipotent, omniscient, and omnibenevolent Creator and Providence) and the existence of evil. And for some fifteen years, Nelson Pike has been publishing articles in rebuttal of Plantinga. Now that the FWD is "clothed in the complex finery of possible worlds ontology," Pike has offered a response dressed for the occasion. 1 1 will suggest, however, that he really has said nothing new. Pike's criticisms, in whatever garb, suffer in the same way now as in the past: from a misunderstanding of Plantinga's position. Plantinga's strategy is the following. To show that propositions ( 1 ) A n omnicompetent God exists and (2) There is evil are logically consistent, he need olfly find a third proposition (however complex) which is consistent with (1) and such that it and (1) jointly entail (2). Plantinga suggests for this third proposition the conjunction of (3)
It was not within (an omnipotent) God's power to create a world containing only moral good (i.e., moral good but no moral evil)
and (4)
God created a world containing moral good.
Now clearly (1), (3), and (4) entail (2). And, claims Plantinga, "these propositions are evidently consistent - i.e., their conjunction is a [logically] possible proposition" (GFE 54). Pike will contest neither the entailment nor the strategy in general, but rather the consistency of the set: he claims that (3) is inconsistent with God's omnipotence (or rather, if we understand "God" to mean an omnicompetent being, that (3) is self-contradictory). Before turning to Pike's argument, however, let us look at the general format of the critic's contention that (1) and (2) are inconsistent. The atheologian posing the problem of evil, says Plantinga, is claiming that "God could have actualized worlds containing moral good but no moral evil" (GIVE 40). I take that claim to
108 be the denial of (3), and so recast it as
(5)
It was within an omnipotent God's power to create a world containing only moral good (i.e., moral good but no moral evil).
And the argument might continue in this fashion:
(6)
An omnibenevolent God who could have created a world containing only moral good wouM, if he created at all, create such a world.
(7)
If God is omnipotent and omnibenevolent and creates a world, he creates a world containing only moral good.
(8)
This world is not a world containing only moral good.
(9)
Either God is not omnipotent, or not omnibenevolent, or he did not create this world.
The argument is valid, and (8) - which is entailed by (2) - is both an obvious truth and part of the theist's accepted beliefs. But (9) entails the denial of (1); so if (5) and (6) are necessary truths, (1) and (2) are inconsistent. 2 Pike's contention that (3) is a contradiction is, of course, equivalent to the claim that (5) is a necessary truth. And it is that contention on which I will concentrate. Notice, though, exactly what Pike will have achieved if successful. He will not have shown (1) and (2) to be inconsistent; to do that, one would also have to show that (6) was necessarily true. (And, in fact, Pike does not think (6) is true at all - that is the burden of the Augustinian "soul-making" theodicy which he favors.) But he would have dismantled Plantinga's defense of their consistency, and demonstrated the inadequacy of any theodicy which relieves God of responsibility for sin on the grounds that he could not have prevented it. And that, I suppose, is not a bad day's work. Pike reaches for "the knot at the centre of the [FWDer's] argument" (p. 458) by means of an ingenious and captivating story about a garden named Edam and possible mice ("mouse-sets"). These mouse-sets are made up of a maximally consistent set of mouse-properties, including (for some of them) the disposition to behave a certain way in Edam. God herded together all the mouse-sets with such a disposition into four cages, depending on how they would behave if instantiated in Edam; the first three cages held mouse-sets of varying moral turpitude, while cage 4 held those which, if instantiated, would always freely choose rightly in Edam. Then, in Pike's words (p. 460), Early in the afternoon on the second day of creation, God opened the door on cage 3 and thereby instantiated a mouse-set in Edam. Since the resulting mouse was left to behave in accordance with his own dispositions, he chose to [behave wrongly]. The [heavenly] hosts were astonished - indeed, bewildered .... Though most maintained a respectful silence, one of the more outspoken of the hosts (Mackie) muttered: 'For God's sake, why didn't he open cage 4?'
109 The objection made here is that God could have populated Edam by opening cage 4 but didn't; that the mouse-analogue of (5) is necessarily true. Consider what it means to be a mouse-set: (10) A mouse-set =df a maximally consistent set of mouse-properties (where a mouse-property is simply any property a mouse can rightly be said to have). Some of these properties are "dispositional choice properties," two in particular being (R) if fi'ee in Edam would ahvays behave rightly and (W) if free in Edam would sometimes behave wrongly. So when God filled the cages, he put all those mousesets containing (W) into one or another of the first three cages, while putting all those mouse-sets containing (R) into cage 4. To say that cage 4 is empty would be to say that it is impossible that a maximally consistent set of mouse-properties could contain (R), and obviously (and necessarily) false claim. Applying the analogy to possible persons, then, since (11)
A possible person =df a maximally consistent set of person-properties
and, necessarily, (12)
The property, if created fi'ee in world I41 would always choose rightly, is part of at least one maximally consistent set of person-properties
God could have created people who were "moral saints." Indeed, that he could have done so is a necessary truth, and so then is
(5)
It was within an omnipotent God's power to create a world containing only moral good.
What are we to make of this objection of Pike's? It rests, in part, on the claim that (12) is a necessary truth, which can hardly be disputed: there are coherent descriptions of possible worlds in which moral agents always choose rightly. But, Plantinga counters, we are interested in the actualization of those possible worlds (more specifically, in those which God can actualize). While the person envisioned in (12) is, necessarily, conceivable, he or she is not for that reason alone necessarily creatable. 3 The question is whether God can create a person (or mouse) with the dispositional property of always freely choosing rightly. Pike thinks the answer to that question is yes, that because there are possible persons with such a property - i.e., coherent descriptions of persons having that property - God can create some of them. Though the point is the slightest bit subtle, what is important to see here is that with respect to actual mice and creaturely essences, transworld depravity is not just an affliction, it is an infliction. If anything in either of these two ontological categories has it, the cause is ultimately to be located in the will of the creator. (p. 465).
110 But this is not as clear as Pike wishes it to be. To say that (12)
The property, if created fi'ee in world W would always choose rightly, is part of at least one maximally consistent set of person-properties
entails (13)
God can create a person who always freely chooses rightly
is to assume that, necessarily, God can make any contingent state of affairs true. If so, then (5) is a necessary truth, and so its denial (3) is a contradiction, and Plantinga's proof of the consistency of(1 ) and (2) fails. But does the above entailment hold? That it does not can quickly be seen by asking what an omnipotent being is able to do. Consider the following definition of omnipotence:
(0)
God is omnipotent =df for any proposition P, God can make P true if and only if (i) P is not a contradiction, and (ii) God makes P true is not a contradiction.
(While I won't defend this understanding of omnipotence here, I think that - with the possible addition of some temporal indexing - it is the correct one.) The first condition rules out such propositions as A married bachelor exists from being limitations on omnipotence, while the second rules out An uncreated world exists, among others; it also rules out any move from a proposition like (12) to one like (13). Still, Pike could claim that in fact (13) is not ruled out by either (i) or (ii), so that God can make it the case that (14)
A world containing only moral good exists
thereby establishing (5) as a necessary truth. The theist, of course, appears to have an out: the existence of moral good (and moral evil) depends on the freely made choices of moral agents. But God cannot make it the case that a particular free choice is made; God cannot cause a moral agent to freely make a choice. 4 So condition (ii) is satisfied, and it is no lack of omnipotence on God's part that he cannot make (14) true. There is a problem, however, with the way this move is often made. Invoking (ii) means that God makes (14) true is a contradiction, and so God cannot make (14) true. But then (5) is a necessary falsehood, and (3) a necessary truth. While this would not hurt Plantinga's consistency strategy, it is not something he would embrace - nor would any other recent theist. Moreover, if (ii) rules out (14), then it must also, for the same reasons, rule out -
(15)
A world containing moral good and moral evil exists.
And so God makes (15) true is necessarily false. But since (1) entails
111 (16)
God created the world
and since this world does contain both moral good and moral evil, (1) would be (contingently) false. The theist, then, by bringing free will into the picture, has won the battle of consistency but lost the war for truth. What the theist wants is a reading of omnipotence on which (3) and (5) are contingent propositions. (For the purposes of showing consistency it will not matter which one is true and which false.) Can such a reading be provided? I think that it can, and that Plantinga implicitly provides one. In The Nature of NecessiO, (pp. 171-173), he distinguishes between "strongly actualizing" and "weakly actualizing" states of affairs. Someone, e.g., God, can strongly actualize a state of affairs S if and only if God can cause it to be the case that S obtains; in the same way, we might say that God can strongly actualize a proposition P if and only if he can cause it to be the case that P is true. On the other hand, God can weakly actualize a state of affairs S (or a proposition P) if and only if God can create the circumstances under which someone else strongly actualizes S (or p).S Clearly, definition (0) was taking "makes P true" to mean "strongly actualizes P." But a moral world (a world containing moral good or evil, or both), depending as it does on the free choices of moral agents, would be a proper object of weakly actualizing only; God could strongly actualize neither (14) nor (15). Put another way, my contention is this:
(5)
It was within an omnipotent God's power to create a world containing only moral good
needs to be rephrased; it is equivalent to the disjunction of (5a)
An omnipotent God could have strongly actualized a world containing only moral good
and (5b)
An omnipotent God could have weakly actualized a world containing only moral good.
Since, ex hypothesi, moral good depends on the free choice of moral agents, (5a) is necessarily false. And (5b) is logically contingent on the free choices moral agents would make, if created. Therefore (5) is not a necessary truth, and the atheologian's argument (5) - (9) fails. Moreover, since (5) is contingent so is its denial (3), and Pike's first objection to Plantinga's consistency strategy fails also. What Pike needed to do, on this account, was to show that (12)
The property, if created free in world W would ahvays choose rightly, is part of at least one maximally consistent set of person-properties
112 entails either (13a)
God can strongly actualize a person who always freely chooses rightly
(13b)
God can weakly actualize a person who always freely chooses rightly.
or
But, alas, (13a) is logically false and (13b) logically contingent, and so neither can be entailed by a necessary truth (12). Whether God could create any moral saints depends on what the particular candidates for sainthood would do, if created. And, to return to Edam, that is just what Pike's story suggested. "God divided [mousesets] into subgroups in accordance with their dispositional choice properties," and each instantiated mouse "was left to behave in accordance with his own disposition" (pp. 4 5 9 , 4 6 0 ) . Why, then, think it certain that "there is a mouse-set in cage 4," or that "there are indefinitely many mouse-sets" with a given dispositional choice property (pp. 4 6 2 , 4 6 4 ) ? If freely choosing to perform some action is a dispositional choice property that one cannot be forced to instantiate, there can be no guarantee that even one - let along indefinitely many - have such a property. The occupancy level of cage 4 is, in the end, entirely up to the free choice of created mice. If it is empty, it is hard to fault God for not opening it. 6 Pike suggests the above possibility after already claiming success for his objection. Could cage 4 be empty? Could its counterpart for any other world God might have created (any Edam-variation) also be empty? Pike calls such a view the "revised" version of transworld depravity, though it is unclear that this is a revision of Plantinga and not his original position, and says he will ask two questions about it. (Let R = "Some person, if created, would always freely choose rightly.") His first question, as I understand him, is (Q1) Is R true?, but he immediately rephrases the question, asking now (QI*) Is it possible that R is true? The FWD's answer, he thinks, is no: (A1) It is not possible that R is true, which can be restated (correctly) as (40 Necessarily, R is false ("statements of the form R are contradictory"). Pike's second question is (Q2) Is it possible that R is false?, which he again rephrases, as (Q2*) Is it possible that R is a contradiction?, and answers affirmatively (as he thinks he FWDer would do) as (~) It is possible that R is a contradiction] Pike recognizes that all the FWD needs, to show the consistency of (1) and (2), is a proposition which is possibly true, and so sees ~ as the important claim. But since q~ contains a modal claim, it is (so Pike and Plantinga would agree) either necessarily true or necessarily false. That is, ~ reduces to 4). However, says Pike,
113 "the idea that it is logically impossible for there to be free persons who always choose rightly ... is patently absurd" (p. 469). Pike is accusing the FWDer of claiming that cage 4 is necessarily empty. On the contrary, says Pike, it is necessarily occupied. This is really just the same move as before, and the problem is easily seen. Clearly, (Q1) is not equivalent to (QI*), and the FWD is seeking only to answer the latter. Again, (Q2) and (Q2*) are separate questions, and the FWD is concerned with (Q2) only. Its answer to both (QI*) and (Q2) is affirmative; that answer, however, cannot be captured as either q~or 4, but as
(x)
It is possible that R is true, and possible that it is false ("statements of the form R are contingent' ).s
It is up to each instantiation of a free agent (man or mouse, at it were) to decide whether R will be true of it. While the description expressed by R is a possible one in the sense that one may imagine its being true, it is not one that is necessarily instantiable. Plantinga's premise (3) remains contingent; cage 4 may be empty, and that is all he needs. So far, the arguments I have been considering from Prof. Pike's most recent (1979) article on Plantinga and the problem of evil have met the same fate as those he raised in his 1966 exchange with Plantinga (in Journal of Philosophy). And for a very good reason: they are substantially the same arguments. In this case, anyway, it is not true that "the clothes make the man"; and just as possible-worlds garb has not changed the force of the FWD, so it has not improved the effectiveness of Pike's objections. There is, however, a third strand to Pike's argument, one which as far as I can tell receives his first attention here. 9 Unfortunately, it fares no better than the others, succumbing at the exact same point. Pike's final objection to Piantinga's demonstration of the consistency of (1)An omnicompetent God exists and (2) There is evil takes issue with Plantinga's claim that God "could have forestalled the occurrence of moral evil only by removing the possibility of moral good" (GFE 30). Grant that God could not guarantee that moral agents would always freely choose rightly. It does not follow from this that God would be unable to create a world containing free creatures who perform no wrong actions. After all, we are here working with the notion of an omnipotent and omniscient being - one who could control the actions (even the choosing actions) of creatures. Surely a being of this sort could create a world containing creatures who freely do what is right but in which no creature succeeds in performing actions (perhaps even choosing actions) that are morally pernicious. In cases where a morally wrong action could be expected to result in the suffering and sorrow of other people, it would seem that interference on God's part would, in fact, be morally required. (pp. 4 7 0 - 4 7 1 ) While the resulting action, being forced, would not add to the agent's moral character, God thereby would have prevented some (presumably undeserved) suffering to his moral credit.
114 Pike never tells us what he really thinks of this kind of world; his point is a logical rather than an ethical one. Clearly God has what Pike terms "over-power": "the power one has when one can completely determine which, if any, powers are possessed by agents other than oneself. ''1~ But then God could create a world like the one Pike describes, simply by restricting the amount of power "delegated out" to an individual whenever that person was about to do something wrong. The upshot is that
(5)
It was within an omnipotent God's power to create a world containing only moral good
is a necessary truth after all, and so its denial (3), being "openly contradictory," cannot be used to show the consistency of (1) and (2). I gladly follow Pike's lead in leaving ethics out of the discussion; for while it would be tempting to consider whether such a world as he describes preserves even a shred of moral agency, and whether it really would be possible for an omnibenevolent God to create such a world, to do so would tax both my abilities and the reader's patience. Fortunately, the logical point has already been made, and not in Pike's favor. Since (13a)
God can strongly actualize a person who always freely chooses rightly
is logically false, (5) cannot be a necessary truth. No matter how many potentially evil actions (or even choices) God prevents, he cannot by himself bring about any morally good actions (or choices) on the part of created moral agents. It is logically contingent that a created being perform any right actions at all - they might all suffer from "transworld iniquity," which is simply the infliction whereby one would perform only wrong actions, no matter what one's circumstances were. (To put it in Edamic terms, it is possible that cage 1 - reserved for those mouse-sets who would always choose wrongly - be the only one occupied.) Which this may sound so implausible as to be a mere clutching at straws, we must remember that in logic even one straw will do. Pike cannot get around the fact that (3) is logically contingent; although there is a cage 4, and God knows of its existence and location, it is possible that even he cannot bring it about that there is anyone in it. There is no inconsistency in affirming the existence of God in a morally imperfect world. One afterthought. Plantinga's consistency strategy seeks to demonstrate that (1) An omnicompetent God exists and (2) There is evil are consistent by showing that (1), (3j It was not within (an omnipotent) God's power to create a world conta&ing only moral good, and (4J God created a world containing moral good are consistent and entail (2). Yet he devotes his efforts almost exclusively to showing that (1) and (3) are consistent. What he has not shown is that (1), (3),and (4)are consistent. (Pike mentions this [pp. 4 5 4 - 4 5 5 ] , but does not pursue it.) Suppose, for example, that
115 (17)
An o m n i b e n e v o l e n t G o d would, if he created at all, create a world containing only moral good
were a necessary truth. T h e n , since it conjoined with (3) entails the denial o f (4), (3) and (4) would be inconsistent - and so (1) and (2) w o u l d n o t have been shown consistent. In fact, (17) seems clearly false. Still, to make the FWD clearly successful, one would have to establish that, as well as argue for the correctness o f a more satisfactory moral principle on which (3) and (4) were consistent. And to do so w o u l d be to m o v e from defense - defending the theist against the charge o f inconsistency to t h e o d i c y - showing that evil does n o t c o u n t at all as evidence against God's existence. Perhaps this is as it should be, for surely the theist wants a faith which is more than merely possible true. What I h o p e to have shown in this paper is that the search for such a t h e o d i c y must no proceed on the false belief that the Free Will Defense has failed. -
NOTES 1. Pike's latest article is "Plantinga on Free Will and Evil," Religious Studies, 15 (1979), 449-473, a response to Plantinga's God, Freedom, and Evil (Harper and Row, 1974) and The Nature of Necessity (Oxford, 1974). The two had exchanged views previously in Journal of Philosophy, 63 (1966), 93-108. (The quotation is from p. 450 of Pike's most recent article.) 2. Anyone holding to (1), (5), (6), and (8) = (2) would be inconsistent, so the theist who has been shown this is obligated to reject at least one of these propositions; it is understandable that (5) and/or (6) be the first to go. But if the atheologian can show that (5) and (6) are necessarily true then the choice is between (1) and (2), and the latter but not the former has the status of being an obvious truth. An entirely different question is which conjunct(s) of (the longhand version of) (1) should be the one(s) to go. 3. The phrase is that of Keith E. YandeU, from "The Greater Good Defense," Sophia, 13 (1974), 1-16. In his Journal of Philosophy response to Pike, Plantinga distinguished between coherent descriptions and possible actualizations. 4. Though this point has been hotly debated among philosophers, both Pike and Plantinga take it to be true (a virtue, in my opinion). Clearly the FWD requires that it be so. 5. These are my formulations, not Plantinga's. 6. Plantinga, of course, argues for the possibility of the truth of (3) in terms of "transworld depravity." To say that a free moral agent has transworld depravity is to say that if created no matter in what circumstances (which possible worlds) - that person would perform at least one morally wrong action. To put it somewhat differently, every possible instantiation of that person would go wrong at least once. Plantinga claims that (TD) Every possible person suffers from transworld depravity is a logically contingent proposition, and were it true (3) would also be true. But if (TD) entails (3), and (TD) is logically contingent, (3) cannot be logically false. Then (3)'s denial, (5), is (at best) logically contingent, and the atheologian's argument fails. 7. For those who are symbolically inclined, the exchange might be put in this fashion: -
(Q1) R?, (Q1 e) OR?, (Al) ~ O R, (dp)~ R ;
(Q2) O~R?, (Q2*) O, ~ R L
(These are my reconstructions of Pike's questions, as well as of ~band ~.)
8. ( X ) 9
(@ ~, t ~ R .
116 9. In Section V. Pike treats this also in his masterful "Over-Power and God's Responsibility for Sin," presented at the Notre Dame Conference on the Philosophy of Religion in April 1981. The point was made also by J.L. Mackie in "Evil and Omnipotence," Mind, -64 (1955), 200-212. 10. "Over-Power and God's Responsibility for Sin," p. 11 of the mimeographed (and revised) paper I was graciously provided with by the University of Notre Dame philosophy department.