Home » Library » Modern Library » Richard Gale Freedom

Richard Gale Freedom

Freedom and the Free Will Defense (1990)


The following article was originally published in SOCIAL THEORY AND PRACTICE, Vol. 16, No. 3, Fall 1990.


It is my purpose to explore some of the problems concerning the relation between divine creation and creaturely freedom by criticizing various versions of the Free Will Defense (FWD hereafter).1 The FWD attempts to show how it is possible for God and moral evil to co-exist by describing a possible world in which God is morally justified or exonerated for creating persons who freely go wrong. Each version of the FWD has its own story to tell of how it is possible that God be frustrated in his endeavor to create a universe containing moral good sans moral evil. The value of free will is supposed to be so great that God is morally exonerated under such circumstances for creating the Mr. Rogers type persons you know, the very same people who are good sometimes are bad sometimes. If it is objected that God could not be unlucky in this manner, that it necessarily is within his power to create goody-goody persons, either by supernaturally willing in his own inimitable manner that it be so, which is the theological compatibilist objection, or by a judicious selection of the initial state of the universe and operant causal laws which together entail that every free action be morally right, which is the causal compatibilist objection, the response is that it is logically incompatible that a creaturely free action be determined by God or by anything external to the agent, such as causes outside of the agent.

We will begin with Plantinga’s version, according to which God foreknows how possible free persons will act if actualized but faces certain contingent limitations that might preclude his being able to actualize any goody-goody type persons, and then go on to consider versions that deny God such foreknowledge, thereby finding his exoneration for permitting moral evil in excusable ignorance or lack of knowledge.

Plantinga’s Version

Plantinga’s version is developed in a dialectical manner as a response to the objection of the sophisticated theological compatibilist who reasons as follows.

L. It is necessary that God can actualize any possible world w such that it is logically consistent that God actualizes w;

1. It is logically consistent that God actualize a possible world in which all free persons always freely go right;


M. It is a necessary that God can actualize a possible world in which all free persons always freely go right.

Plantinga is first going to argue for the contradictory of M:

-M. It is possible that God cannot actualize a possible world in which all free persons always freely go right.

which can aptly be called the “depravity premise,” because it claims that every possible free person could be such that were God to attempt to actualize it the resulting concrete person would freely go wrong at least once. As a step along the way to proving -M, Plantinga will prove the counter-Leibnizian proposition:

CL. It is necessary that there is some possible free person that God cannot actualize.

Once -M is established, Plantinga can complete his FWD by adding that in the possible world in which what -M claims to be possible is realized God is morally exonerated for his creation of the Mr. Rogers type persons.

Before we can understand Plantinga’s arguments for -M and CL, we must see what he means by “possible person” and “possible world,” as well as their instantiation and actualization respectively. A possible person is a higher order platonic entity, being a set of abstract properties that may or may not be instantiated, while a possible world is a set of abstract states of affairs (or propositions) that may or may not obtain (or be true). More precisely, a possible person is a maximal compossible set of properties each of which could be possessed by a single person. The set is compossible in that it admits of the logical possibility of coinstantiation by a single concrete individual, and it is maximal because for every property that could be possessed by a person either this property or its complement is included in the set. A possible world, on the other hand, is a maximal compossible set of states of affairs or properties. I will use “universe” to refer to a maximal spatiotemporal aggregate-the totality of what exists and happens in space and/or time-and “cosmos” to refer to the maximal aggregate of actually existent entices. Of special interest for our purpose is a possible significantly free person, that is a possible person containing the property of being free in respect to some action having moral significance. For such a person there is at least one morally significant action A and set of circumstances C such that this person includes the disjunctive property of either-freely-doing-A-or-freely-refraining-from-doing-A-in-C. Since a possible person is maximal, it also includes the property of doing A or the property of not doing A. For every possible free person containing the property of freely-doing-A-in-C there is a numerically distinct possible person that includes all of the same properties save for its including freely-refraining-from-doing-A-in-C instead. Let us call such a pair of possible free persons an “incompatible pair.” Whenever you freely perform an action you instantiate one member of such a pair to the exclusion of the other. In what follows we shall consider only significantly free possible persons and, for short, will call them “possible persons.”

With these terminological and ontological points out of the way, we can begin to approach Plantinga’s argument for CL, the counter-Leibniz proposition. The argument attempts to show that for any incompatible pair God will be contingently unable to actualize one person in the pair. Let our specimen incompatible pair be P and P’, who include all of the same properties save for P’s including freely-doing-A-in-C and P”s instead including freely-refraining-from-doing-A-in-C. Our sophisticated Leibnizian will hold that each is such that God can actualize or instantiate it, though he cannot, of course, coinstantiate them. Plantinga, the Libertarian, disagrees, because he thinks it is logically inconsistent that God actualizes or instantiates either of them. The reason is that God’s actualizing or instantiating of P. for example, consists in his causing there to exist a person having all of the properties included in P. and thereby God causes or determines this person freely to do A in C; but the incompatibilist premise tells us this is logically inconsistent. For the Libertarian it must be the agent itself that is the agent-cause of a free act, not some condition external to the agent, though such conditions might limit the range of possibilities from which the agent can choose and “incline” it in a certain direction, requiring greater effort of will for it to pursue an opposite course of action.

If God cannot actualize a possible person simply by supernaturally willing that it will be actualized, how does he do it? It is here that Plantinga has an incredibly ingenious and controversial story to tell. Again we must begin with some terminology. What God does to actualize what I will call a “diminished possible person” and then leave it up to the created person what it will freely do. Each possible person contains a diminished possible person that is its largest proper sub-set of properties that is such that for any action A it neither includes or entails freely doing A nor includes or entails freely refraining from doing A, in which a property F includes or entails another properly G just in case it is logically impossible that F be instantiated and G not be. We will also refer to such a sub-set as a “freedom-neutral” set of properties. Each property included in a set of properties could be freedom-neutral and yet the set as a whole not be, for the set could contain either-freely-doing-A-or-freely-refraining-from-doing-A and doing A. Any incompatible pair will contain as proper sub-sets the same diminished person or set of freedom-neutral properties. Thus, P’s diminished person, DP, is numerically one and the same as P”s. God performs the same creative act when he endeavors to actualize P as he does when he endeavors to actualize P. namely, he supernaturally wills that the diminished person DP be instantiated or actualized. Intuitively, we can think of this as God’s creatively determining every feature of the universe up until the time at which the created person, the instantiator of DP, freely does A or freely refrains from doing A.

The question is what would result if God were to instantiate DP. Would the instantiator of this diminished person or set of freedom-neutral properties freely do A or freely refrain? Plainly, it must do one or the other, since it has the disjunctive property of either-freely-doing-A-or-freely-refraining-from-doing-A. Thus, it is either true that

F. If DP were instantiated, the instantiator would freely do A.

Or it is true that

F’. If DP were instantiated, the instantiator would freely refrain from doing A.

Let us call a subjunctive conditional whose antecedent reports the instantiation of a diminished possible person and consequent the performance of a free action a “free will subjunctive conditional,” for short an “F-conditional.” If F is true, then were God to instantiate DP, it would result in P’s being actualized; whereas, if F’ is true, were God to actualize DP, it would result in P”s being actualized. Since F and F’ are logically incompatible, it follows that if F is true God is unable to actualize P’, and if F’ is true God is unable to actualize P. But necessarily one of them is true and therefore necessarily true that God cannot actualize P or cannot actualize P’, which proves CL.

This proof assumed that the law of conditional excluded middle holds for F-conditionals. Herein the necessarily true disjunction is formed not from the disjunction of an F-conditional with its negation, as is the case when the weaker law of excluded middle is applied, but from the disjunction of an F-conditional with an F-conditional containing the same antecedent and the denial of the former’s consequent, as is the case above with the disjunction of F and F’. Plantinga has another proof for CL that applies only the law of excluded middle to F-conditionals. It begins with what Plantinga calls “Lewis’s lemma,” which, when translated into my terminology, says that God can actualize a possible person P containing the property of freely doing A only if it is true that if God were to actualize its diminished person DP, the instantiator would freely do A. It next is claimed by appeal to the law of excluded middle that it is either true or false that F. If it is false, then, given Lewis’s lemma, God cannot actualize P. and, if it is true, then he cannot actualize P’.

From Plantinga’s argument for CL it is only a few short steps to the establishment of the depravity proposition, -M. At the outset let us confine ourselves to possible persons that include the property of being free with respect to only one action, such as persons P and P’ above. What we establish then can be generalized to more complex possible persons. Any incompatible pair of such simplified persons is a Dr. Jekyll and Mr. Hyde pair, the former being the one that contains the property of freely doing A (which is the morally right thing to do), the latter the property of freely refraining from doing A (which is the morally wrong thing to do). It has already been shown in the argument for CL that God might not be able to actualize P. the Dr. Jekyll member of the pair, since F could be false. But what could be true for this particular Dr. Jekyll and Mr. Hyde pair could be true for all of them. Every incompatible pair of this sort could be such that it is true that if God were to instantiate the diminished possible person common to both, the instantiator would freely do the morally wrong alternative. Under such unfortunate circumstances, God can actualize only Mr. Hyde’s, and therefore will not attempt to instantiate any of these simple possible persons, assuming that his brand of benevolence requires that there be a favorable balance of moral good over moral evil.

The result can be generalized so as to apply to more rich possible persons that contain the property of being free in respect to more than one action. It could still be the case for every such person that it is true that if God were to actualize its diminished person, the instantiator would freely go wrong with respect to at least one of these actions. And this suffices to establish -M – that it is possible that God cannot actualize a possible world in which all free persons always freely go right.

At this point Plantinga can complete his FWD by claiming that in the possible world in which the truth-values of the F-conditionals preclude God from actualizing any Dr. Jekylls or, more generally, possible persons containing the property of always freely doing what is right, he is excused for creating the Mr. Rogers sort of persons, provided for the most part they freely go right, which I am sure is true for the Mr. Rogers types. In such a world God can plead that he did the best he could but was stymied by F-conditionals. This completes our rough sketch of Plantinga’s FWD account of the possible world in which God is unlucky and thereby morally exonerated for allowing moral evil.


The account given so far of Plantinga’s Story of Creation fails to deal with some key issues concerning F-conditionals: their modal status; God’s knowledge of them; and whence they derive their truth-values, assuming that they have them at all. Different ways of handling these issues produce different versions of the FWD.

A careful examination of the text of Plantinga’s several presentations of his FWD would show that he is committed to the following three theses.

I. Every F-conditional has a contingent truth-value, that is, is contingency true or contingently false.

II. God knows the truth-value of all F-conditionals prior to his creative decision.

III. God does not determine the truth-values of F-conditionals.

Theses I and II together comprise the doctrine of God’s “middle knowledge.” Another way of formulating the doctrine of God’s middle knowledge is that God foreknows for every diminished possible persons what free actions would be performed were that person to be instantiated.

We will hold off giving Plantinga’s reasons for I until we consider FWD’s based on its rejection. Thesis II logically follows from thesis I, on the assumption that God’s omniscience requires that he knows and believes all and only true propositions. According to the Libertarian-type incompatibilist premise in his FWD, it would be inconsistent for God both to determine the truth-value of an F-conditional and actualize its antecedent, which would be analogous to God predetermining a free action by determining both the initial state of the universe and the operant causal laws.

If God does not determine the truth-values of the F-conditionals, who or what does? There is an answer to this that is implicit in the platonic ontology employed in Plantinga’s FWD. Since possible persons, including diminished possible persons, are sets of abstract properties, they exist in every possible world, as do all abstracts. Abstract entities have both essential and accidental properties. The number two has the property of being even in every possible world but has the property of being Igor’s favorite object in only some. Our old friend, diminished possible person DP, being a set of properties, has the property of containing the same properties in every possible world, such as the property of being free with respect to A. However, it also has some accidental properties, among which is the following: being-such-that-if-it-were-instantiated-its-instantiator-would-freely-do-A. In some worlds it has it and in others not. In virtue of this, the F-conditional, that if DP were instantiated, its instantiator would freely do A, is true in some worlds but not others. It is all right to call this funny property of DP a “dispositional property” provided we are clear that it is not a disposition of DP to freely perform A if instantiated (abstract entities, with the possible exception of God, cannot perform actions) but a disposition to have its instantiator freely do A.

But what, it will be asked, determines whether a diminished person has one of these funny dispositions? As they used to say in the Bronx, “Don’t ask!” Here’s where the regress of explanations hits the brick wall of brute, unexplainable contingency. There are no further elephants or tortoises upon whose back this contingency rests. Let us now consider some objections to Plantinga’s FWD.

According to its Story of Creation the F-conditionals limit God’s power in a way similar to that in which fate limits the power of the Greek gods. In both cases there is a force or power above and beyond the control of the individual that limits its powers to do what it wants. The idea that God must be lucky, that he must be dealt a favorable poker hand of F-conditional facts, if he is to be able to create a universe containing moral good sans moral evil, strikes some as blasphemous, as a radical distortion of the orthodox concept of God’s omnipotence. While Plantinga’s account of omnipotence is not every theist’s cup of tea, certainly not that of the great medieval theists, it might be the cup of tea that will prove most digestible and healthy for theism in its effort to construct an adequate defense for God’s permitting moral evil.

There are a number of objections of the “God-can-do-more” variety. One version is based on a theological compatibilist view of God as a mere foreknowing enabler. Another appeals to the power of God’s grace as enabling him to determine that all persons always freely go right. Yet another objection is based on God having the power to step in just in the nick of time when he foresees, on the basis of his middle knowledge, that someone will freely go wrong, by either preventing this wrong choice or causally quarantining the culprit from the surrounding world so that no innocent persons are harmed. I believe that all of these God-can-do-more-than-Plantinga-allows objections can be met, but cannot explore that now.2

Instead I want to press an opposite objection-that God cannot consistently do as much as is required by Plantinga’s FWD. It will be argued at some length that God, in virtue of having middle knowledge, has a freedom-canceling control over created persons. And because these created middlemen aren’t free, the buck of moral blame for seeming moral evils cannot stop with them but must reach through to God, which destroys the FWD’s attempt to show how God can escape blame, although not responsibility, for these evils. I will begin by marking the distinction between blame and responsibility.

In general, one is responsible for an occurrence that she was fully able to prevent, that is, had the power, opportunity, and necessary knowledge to prevent. God, for example, is responsible for moral evil, since he could have prevented it by electing not to create any free persons. An especially pertinent case is that in which one person delegates some of his power to another but retains the power to revoke the delegated power.3 In a dual control student driver car the instructor can throw a switch that gives the student control over the car but still retains the power to regain control over the car by flipping the switch the other way. If the car should be involved in some foreseeable untoward incident while the student is in control, the instructor, along with the student, is responsible, but it could be that only the student is blameworthy. Whether the instructor shares blame will depend on whether she has a good reason for not having retaken control of the car–for example, the resulting harm was minor and the student can best learn by being free to make mistakes.

The relation of God to created free persons is similar. By creating free persons God delegates some of his power to them, but he still retains the power, called “overpower” by Nelson Pike, to rescind their power, either in part or wholly. Because God can withdraw his gift of free will-flip the big switch-he is responsible along with created free persons for the moral evil they cause. But, like the driving instructor, he might have a good excuse that frees him from sharing the blame with those to whom he has delegated some of his power. The FWD supplies such an excuse. God can be responsible but not blameworthy for the evils caused by created beings only if they are free. But, I will now argue, according to the Plantinga Story of Creation, they are not. He never succeeded in flipping the switch that gave them the power to freely control their own lives.

The first stage of my argument establishes that God causes the actions of created free persons according to the FWD. I’ll begin with a fallacious argument for this that can then be fixed up to work. According to the FWD, when God instantiates a diminished possible person, say DP, there is a true F-conditional known to God to the effect that if DP were instantiated, its instantiator would freely do A. This shows that DP’s being instantiated is sufficient for the instantiator’s freely doing A. But God’s willing that DP be instantiated is causally sufficient for DP’s being instantiated. Since causation is transitive, it follows that God’s willing that SP be instantiated is causally sufficient for the instantiator’s freely doing A.

The fallacy in this argument for God being the cause of a creaturely free action jumps out at us. DP’s being instantiated is only subjunctive-conditionally sufficient for the instantiator’s freely doing A, not causally sufficient for it, at least according to the Libertarian account given by the FWD. Since one of the links in the sequence going from God to A is non-causal, the principle of the transitivity of causation cannot be applied. By interposing the indeterministic process reported by the F-conditional between God’s initial act of will and the eventual doing of A, the FWD supposedly cuts the link of causal sufficiency between them. God’s act of will is only causally necessary at best for A, since if he were not to have willed as he did, A would not have occurred.

The first thing that must be done in fixing up this fallacious argument is to show that under certain circumstances a sufficient cause can reach through the interposition of a relation of subjunctive-conditional sufficiency of an indeterministic sort. Then it will be shown that this very circumstance obtains in the FWD. Consider this stochastic machine. When its button is pressed, a stochastic process, such as they decay of a radioactive element or the spinning of a wheel of fortune, is triggered, the outcome of which determines whether a poisonous gas will be released into a crowded stadium that will result in the deaths of 50,000 innocent people. It might be necessary to add for you sports fans that they would die before seeing the end of the game, this qualification being necessary so that you are convinced that this is a truly evil outcome. When the button is pressed, either this outcome will ensue or it won’t. Therefore, either it is true that if the button were to be pressed, this horrendous outcome would ensue or it is true that if the button were pressed, this outcome would not ensue. Let us assume, furthermore, that we mortals cannot discover by any discursive methods which of these subjunctive conditionals is true, any more than we can for similarly matched F-conditionals.

Imagine the case in which I chance on the scene and inadvertently press the button, resulting in the horrendous outcome. Given that I did not have “middle knowledge” of what would result from pressing the button and did not intend to bring about or even risk bringing about this outcome, I am blameless for the resulting evils. Furthermore, I do not even cause these evils. Were we to infer by appeal to the transitivity of causation that I cause them, we would be guilty of the above fallacy. Let us change the circumstances so that I now have middle knowledge via some ESP faculty and press the button so as to bring about the deaths. In this case my action is a sufficient cause of the deaths, and is so in spite of the interposition of a stochastic process. Furthermore, I am blameworthy for the deaths, unless I have got a mighty good excuse, for example, that they were Mets fans.

While there is no doubt that this is what people on the street would say, it might be objected that their concept of causation is confused; for the only difference between the two cases is my psychological state, what I know and intend, and how can this determine whether or not I cause the deaths? If what was at issue was the physicist’s concept of causation, this would be a powerful objection. But this is not the concept of causation in question. Rather, it is the forensic one that concerns moral and legal responsibility and blame, which is the very concept that figures in the FWD, since it is concerned with the assignment of responsibility and blame to God and man.

It might be urged by Plantinga that while both God and created free persons were fully able to have prevented moral evil, only the latter are to blame for it, since only God could have a morally exonerating excuse for not doing so. God is the creator of the universe, not they, and thus he alone could say that his allowing such evil was the price that had to be paid for there existing any free persons at all, which is something like my “They are Mets fans!” excuse for intentionally pushing the button. In both cases, the evil in question was necessary for the realization of an outweighing good.

Notice that the response that has been made on Plantinga’s behalf does not claim that God does not cause moral evil, only that he is not blameworthy for it since he has an excuse that cannot be available to created free middlemen. This excuse collapses if these middlemen are not free, since then the buck of blame could not stop with them. And this is just what I will now argue.

Since God creates free persons with middle knowledge of what will ensue, he sufficiently causes the free choices and actions of these persons. This alone does not negate the freedom with which these acts are done, for one person can cause another to act without thereby rendering the act unfree. As a rule, the more the external event only triggers a deep-seated character trait or natural disposition of the agent the less difficulty there is in treating it as not abrogating the free will of the affected agent. When I induce a person of amorous nature to call Alice for a date by telling him that she is desirous of going out with him, I cause him to act but do not usurp his free will since prominent among the causes of his action are his own deep-seated character traits, which traits were not imposed on him by me. I didn’t have to “work on him”-to drug, hypnotize, or brainwash him to get him to call Alice. Unfortunately, God’s way of causing created persons to act is not of this innocent sort. It is freedom-canceling.

My argument for this is anthropomorphic in that it applies the same freedom-canceling principles that apply to man-man cases to the God-man case. Whether it is permissible to reason in this anthropomorphic manner will be considered subsequently. Obviously, any analogy between man and God will be an imperfect one, since there are such striking disanalogies between the two. For this reason I do not see my argument as in any way conclusive. At best, it might take the smirk off the face of a Free Will Defender and replace it with a worried grin. I will try to derive these freedom-canceling principles by examining paradigm cases in which one man or finite person has a freedom-canceling control over another.

1. The Case of the Sinister Cyberneticist. Imagine a Stepford Wives type situation in which a cyberneticist operates on his wife’s brain or replaces it with a pre-programmed computer-analogue, so that he can inculcate in his wife the desired psychological make-up comprised of various desires, wants, dispositions, etc. As a consequence, she is always amorous, anxious to cook and clean, and so on. To an uninformed observer her actions will appear free and voluntary, since they emanate from and are explainable by her own psychological make-up. But her cyberneticist husband has imposed this make-up on her. Her lack of freedom of the will is not due to the fact that this make-up has been determined by factors external to herself (no man is either an island or a causa sui) but rather to the manner in which it has been determined, namely through the machinations of another person for the purpose of controlling her response to stimuli. The case of the Insidious Hypnotist and Manchurean Candidate type Barbaric Brainwasher who have gained an habitual ascendary over the will of another by inculcating in them a certain psychological make-up are similar.

Our intuitions about these cases suggest the following freedom-canceling sufficient condition for man-man cases:

C1. If M1‘s action and choices result from psychological conditions that are intentionally determined by another man M2, then these actions and choices are not free.

Under these circumstances, M2 has a freedom-canceling control over M1, not in virtue of determining Ml‘s actions and choices, but rather causing M1 not to have a mind or will of his own. It isn’t so much Ml‘s actions and choices that are not free but M1 himself; and, in virtue of Ml‘s lack of global freedom, his specific actions and choices are not free.

2. The Case of the Evil Puppeteer. Stromboli has poor Pinocchio wired up in such a way that he controls his every movement. An observer who fails to notice the wires might falsely believe that Pinocchio’s behavior was fully free and voluntary. Stromboli controls Pinocchio, not via having imposed on him an inner network of dispositions, motivations, or intentions, but by exerting a compulsive force over him that renders such inner factors irrelevant. There need not be actual wires connecting the controller with the “puppet.” It could be a wireless radio hookup such as exists between a controller and a remote control toy airplane or between the Horrible Dr. Input and a brain in a vat that in turn has a radio control hookup with a shell body in the manner described in Daniel Dennett’s “Where Am I?”

By a coincidence that rivals that of the preestablished harmony it could be the case that every time the external controller causes the “puppet” to perform some movement the “puppet” endeavors on its own to perform this movement. This is a case of causal over-determination in which there is more than one sufficient cause of a given occurrence. While the puppet’s action is unavoidable in that it would have made this movement even if it had not endeavored to, there are those, like Locke, who would still call it free. I do not share Locke’s intuitions in this matter.

What is it about these cases that makes us say the controller, be it the Evil Stromboli or the Horrible Dr. Input, has a freedom-canceling control? It is that most of the behavior is caused by and subject to the whim of the controller. This suggests that

C2. M2 has a freedom-canceling control over M1 if M2 causes most of Ml‘s behavior.

Is God’s relation to created persons in the FWD such that it satisfies C1 and/or C2? If it satisfies either, no less both, the FWD is in trouble. I submit that it satisfies both, and thus it is time for the nervous smile to replace the smirk.

It is clear that it satisfies C1, since according to the FWD God intentionally causes a created free person to have all of its freedom-neutral properties, which include its psychological make-up. The Free Will Defender will make the Libertarian claim that these inner traits only “incline” but do not causally determine the person to perform various actions or act in a certain regular manner, but this does not make the God-man case significantly disanalogous to the type 1 man-man cases. For even if we imagine that our intentional psychological-trait inducers could render it only probable according to various statistical laws that their victims would behave in certain characteristic ways, they still would exercise a global freedom-canceling control in which the person is rendered non-free due to her not having a mind of her own.

The God-man relation in the FWD also satisfies C2, for when God instantiates diminished possible persons or sets of freedom-neutral properties he does have middle knowledge of what choices and actions will result, and thereby sufficiently causes them. And he does so quite independently of whether or not he is blameless for the untoward ones among them.

During the discussion of my paper, David Blumenthal offered an interesting objection to C2. He begins by asking us to imagine that M1 is “massively ignorant” in that for every action M1 performs there is an unknown fact such that if M1 were to have known it M1 would have refrained from doing it. There is another person M2 who knows all of these that are unknown to M1. So far there is no problem about M2 having a freedom-canceling control over M1. Although M2 has the power to cause all of Ml‘s actions by feeding M1 the relevant facts at the right time, M2 does not actually exercise this power. We are now to imagine that M2 exercises this power by informing M1 on every occasion of the relevant unknown fact, thereby causing all of M1‘s actions. We have already seen that I can cause someone to call Alice for a date by informing him of some relevant fact without usurping his free will in doing so. Why can’t I similarly cause all of his actions without thereby negating his free will in doing them?

This slide from some to all is suspect. People have different intuitions about the all case. Michael Slote thought that M1 would feel dependent upon M2‘s ubiquitous timely advice, so much so as to feel bereft of the sort of independence that is necessary for being free. Just think of the cold sweat that would engulf M1 if she were called on to act before M2 called her. She could say to M2 what many a mother has said to her offspring, “What’s the matter, you never call.” M1 would have the same sort of radical doubts about her own agency as she would if M2 were correctly to predict her every action, in spite of her efforts to falsify them.

Be that as it may, Plantinga’s creator God’s way of causing our actions is less benign than is M2‘s. For God’s instantiating a diminished person alone sufficiently causes all of the instantiator’s actions. (The truth of the relevant F-conditional is not among the causes of these acts, since a proposition cannot cause anything.)

Plantinga would agree that if God’s relation to created persons satisfies C2, he has a freedom-canceling control, for he has said that “If God causes them always to do only what is right, then they don’t do what is right freely.”4 But Plantinga might be conceding too much in accepting C2. It might be objected that for M2 to have freedom-canceling control over M1 it is not enough that M2 cause most of M1‘s behavior: M2 also must have counterfactual control over M1 in virtue of which M2 can cause M1 to behave in ways ocher than chose in which M1 in fact behaves. Whereas Stromboli and Dr. Input have this additional counterfactual control over their victims, God does not have it over created persons. For while God causes the instantiator of DP to do A, he does not have the power to cause this instantiator to do other than A, given that it is true that if DP were instantiated, its instantiator would freely do A. God could have prevented the instantiator from doing A by not instantiating DP, but this is not causing the instantiator to do ocher than A-non-existent persons do not act.

Granted that there is this disanalogy between God and our finite controllers in that only the latter have this sort of counterfactual control, what follows? Not that God does not have freedom-canceling control over created persons in virtue of satisfying C2 (as well as C1), but that there is a stronger sufficient condition for having freedom-canceling control that he does not satisfy, viz.

C3. M2 has a freedom-canceling control over M1 if M2 causes most of M1‘s behavior and also has the counterfactual power to cause M1 to act differently from the way in which M1 in fact acts.

In general, to satisfy one sufficient condition for being X does not require satisfying every sufficient condition for being X.

The objector might retort that having counterfactual control is necessary for having freedom-canceling control: the “if” in C3, accordingly, is to be replaced by “only if.” This is not particularly plausible for two reasons. First, if C3 is turned into a necessary condition, it follows that C1 is unacceptable and that therefore the Insidious Cyberneticist and the others do not have the freedom-canceling control, which is not what we want to say. Second, God, although lacking counterfactual control, has an additional power over created persons that Stromboli and Dr. Input do not have-God both creates and determines the psychological make-up of his “victims.” This additional power of God’s should at least counterbalance his lack of counterfactual power and thereby make him at least as good a candidate as our finite controllers for having freedom-canceling control.

Furthermore, it should be obvious by now that the FWD’s gambit of having something other than God determine the truth-values of the F-conditionals does not succeed in showing that God does not cause the free acts of created persons. Stromboli and Dr. Input were not excused from being the cause of their victim’s behavior because they did only “half the job”-determined the causally relevant instantial conditions but not which causal laws hold. Analogously, God is not excused from being the cause of the free acts of created persons because he did only half the job-determined which diminished possible persons get instantiated but not the truth-values of the relevant F-conditionals. If this does not convince you, try these counterfactual thought experiments. Our finite controllers do only half the job by determining which causal laws hold after they come upon their victim in some instantial state, and God does only half the job by determining the truth-values of the F-conditionals after he comes upon concrete instantiations of various diminished possible persons. Certainly, we want to say of both God and the finite controllers in these thought-experiments that they cause their “victim’s” behavior and have a freedom-canceling control in virtue of C2 alone.

So far, it appears that God’s relation to created persons satisfies both C1 and C2 (but not C3) and that he thereby has a freedom-canceling control over them. But there skill remain some disanalogies between the God-man and the man-man cases that have not been explored. One of them concerns the fact that the finite controllers in our type 1 and 2 cases were a sinister bunch who meant no good for their victims, whereas God is benevolent and intends the best for his created beings. This makes no difference in regard to having freedom-canceling control but only in how the movie is titled. One is titled The Horrible (Sinister, Insidious, Barbaric) Dr. Input (Cyberneticist, Hypnotist, Brainwasher) while the other is tided The Incredible (Fabulous, Stupendous, Marvelous) Supernatural Predeterminer. One is a horror movie and the other is not; but neither involves free persons.

Another tack is to argue that God’s relation to man is so disanalogous to man’s relation to man as to render the freedom-canceling principles, such as C1 and C2, that holds for the latter inapplicable to the former. It is not just that God is quite different from humans in the lingo of the streets “He is something else! “-but that he is different in just those respects that make these principles inapplicable to his relation to created persons. He literally is out of it, not a part of the universe. No insult intended, but he is as unnatural as you can get; in fact, he is supernatural. He does not cramp our elbow room in the way in which finite people do. In this respect he is crucially unlike our bevy of sinister finite controllers. These people ride herd on their fellow humans. God does not do so. This is not an epistemological point concerning our being unaware of God’s causal efficacy in bringing about things the world, but an ontological one having to do with the radical difference in the way his causal efficacy works from that in which a finite controller’s works.

It is just such anti-anthromorphic considerations that are at the foundation of theological compatibilism. And I am very sympathetic to them. Were I to be a theologian in my next reincarnation as a result of my sinful life in this one, this is the line I would run. Unfortunately, Plantinga cannot avail himself of this strategy for averting the objection that God assumes a freedom-canceling control over created persons in his FWD. The reason is that his FWD must take the anthropomorphic route in its rejection of theological compatibilism, for it claims that God cannot determine the free acts of persons without negating their freedom. And the only basis for this claim is that if one person were to do this to another it would be freedom-canceling. In other words, God cannot get away with determining the free actions of persons, because this would violate C1 and C2 or both-the very principles that operate in man-man cases.

We cannot allow Plantinga to be a good-time anthropomorphist: to reason anthropomorphically when warding off the objection of the theological compatibilist, and then refuse to do so for the purpose of rebutting the charge that God has assumed a freedom-canceling control over created persons. Thus, Plantinga is caught on the horns of a dilemma. If he reasons anthropomorphically, his FWD collapses because it imputes to God a freedom-canceling control over created persons. And if he does not reason anthropomorphically, again his FWD collapses, this time because it has no reply to the objection of the theological compatibilist. But either he reasons anthropomorphically or he does not. Therefore his FWD collapses.

I conclude that my God-cannot-do-as-much objection has considerable force against the Plantinga version of the FWD, since it imputes middle knowledge to God. Without it, God’s instantiating the antecedent of a true F-conditional no longer would count as a case of his causing the instantiator to do what the consequent reports. Created persons, then, could serve as suitable scapegoats for moral evil. This naturally gives rise to the question whether a viable version of a FWD can be constructed that denies middle knowledge to God.

Versions of the FWD sans Middle Knowledge

Since God’s middle knowledge is composed of the conjunction of

I. Every F-conditional has a contingent truth-value.


II. God knows the truth-value of all F-conditionals prior to his creative decision.

there will be two ways of constructing a FWD sans middle knowledge. The first version, which is ably championed by Adams, denies thesis I and thereby thesis II, since not even an omniscient being can know what isn’t true. This renders God blameless for permitting moral evil, since he could not have known in advance the moral evils that would result from his creation of free persons. The second version accepts thesis I but denies thesis II, again rendering God blameless in virtue of an excusable lack of knowledge. Herein there was something to be known, unlike the first version, but there was no way in which God could have divinely known it. God winds up watching the history of the universe unfold in just the way parents watch their child play in a hockey game. In both cases, there are a lot of grimaces and groans as they observe unforeseen errors and transgressions.

Because both versions have God instantiate the antecedent of F-conditionals without foreknowledge of what the created persons will freely do, they face the objection that God is acting in a recklessly immoral way by shooting crap at our expense. No red-blooded theist would accept the wimpy moral intuition underlying the Reckless-objection, and would give God’s creation of free persons in both versions as a counterexample. The objection also faces an ad hominem type rebuttal in that no existent person, except for a few gripers, are apt to make it; for if God hadn’t elected to roll the dice, they wouldn’t even exist, and supposedly, they are glad that they do.

Although both versions make the same lack of knowledge excuse available to God, they differ significantly in their epistemological and metaphysical underpinnings, and thus require separate consideration. The outcome of our discussion will be that both versions face formidable objections: the first version, because of its denial of I, renders it anomalous how God is able to create any free persons; and the second version waters down God’s omniscience in an unacceptably anthropomorphic manner.

First Version: Adams develops this version in the course of defending the attack on middle knowledge by certain late sixteenth-century Dominicans against their Jesuit opponents, Molina and Suarez. Adams’s denial of thesis I is built upon the Libertarian account of F-conditionals, according to which the act reported by the consequent is neither causally determined nor determined by anything other than the instantiator of the antecedent. Given this account, he does not see how an F-conditional could possibly be true. Thus, it appears to be logically or conceptually impossible for it to be true and therefore it necessarily lacks a truth-value. Unlike the “neuter” or “indeterminate” propositions of Lukasiewicz, it does not become true or false with the passage of time. Adams says that he doubts if they “ever were, or ever will be, true.”5 This means that even if the antecedent should be instantiated and the instantiator subsequently perform the action reported by its consequent, the F-conditional does not become true. And a fortiori, this sequence of events does not show that the F-conditional was true all along, for there is no present truth to cast a backward shadow. Adams gives us a choice between F-conditionals being necessarily false and being necessarily neither-true-nor-false. The common denominator of these options is that F-conditionals necessarily are not true.

Adams’s argument for the denial of thesis I is not made fully explicit. One who denies thesis I typically is a warranted-assertibility theorist who holds a proposition to be true only if it is in principle epistemically supportable. But this isn’t Adams’s line, for he says that a proposition reporting a future contingent (for example, the sea fight tomorrow, which we finally know didn’t happen since Kruschev decided not to challenge Kennedy’s blockade of Cuba), although in principle not warrantedly assertible, “can be true by correspondence to the actual occurrence of the event they predict.”6 This suggests that for Adams a necessary condition for a proposition being true is that it have an external correspondent. The “external” qualification precludes the correspondent of the proposition that p being the fact that p; for, given that a fact is a true proposition, this would make that p the correspondent of that p.

In the above discussion of thesis III it was suggested that for Plantinga the external correspondent of the true F-conditional, that if DP were instantiated its instantiator would freely do A, is the abstract diminished possible person DP having the peculiar dispositional property of being-such-that-if-it-were-instantiated-its-instantiator-would-freely-do-A. According to Adams’s commentary, Suarez also held this view. While not objecting to this account’s platonism, Adams objects that he does not


have any conception, primitive or otherwise, of the sort of habitudo or property that Suarez ascribes to possible agents with respect to their acts under possible considerations.7

This autobiographical fact hardly constitutes a refutation of the Suarez-Plantinga position. The rejection of a position by saying that “I do not quite understand this” mercifully has gone the way of rejection by appeal to desert landscapes. Not being able to understand is not always an achievement. I suspect that the reason that Adams does not understand what it is for a diminished possible person to have one of the “dispositional” properties is that he does not see how an F-conditional can be true, and his reason for this will emerge in our discussion of whether thesis I is true, to which we now turn.

For Plantinga, thesis I is a consequence of the law of the excluded middle.


Our question is really whether there is something Curley would have done (had he been offered the bribe)….The answer is obvious and affirmative. There is something Curley would have done, had that state of affairs obtained.8

If DP were instantiated, given that its instantiator has the disjunctive property of either-freely-doing-A-or-freely-refraining-from-doing-A, its instantiator either freely does A or freely refrains. Thus, it is either true that

F. If DP were instantiated, its instantiator would freely do A.

or it is true that

F’. If DP were instantiated, its instantiator would freely refrain from doing A.

So far I cannot find anything to object to. According to Lewis’s lemma, if neither F not F’ is true, it follows that God is unable to instantiate either P or P’, in which P and P’ form an incompatible pair having DP in common and differing only in that P contains freely-doing-A while P’ contains freely-refraining-from-doing-A. For if it is not true that if God were to instantiate DP, its instantiator would freely do A, God is unable to instantiate P. and likewise mutatis mutandis for P’. And this can be generalized, so that if no F-conditional is true, God cannot instantiate any possible person, which would be the shipwreck of the FWD, since any possible world in which God is morally exonerated for allowing moral evil is one in which He succeeds in creating free persons and thus is able to do so.

Adams, surprisingly for a Free Will Defender, says something that seems to imply that God is unable to instantiate any possible person: “In other words, I deny that God could have made free creatures who would always have freely done right.”9 This seems to say that God couldn’t have instantiated any possible person containing the property of always freely doing right by instantiating its diminished possible person, because the relevant F-conditionals are not true. I assume that this is why Adams put “would” in italics. But this reason for God’s being unable to instantiate any goody-goody possible person can be generalized to all possible persons, since no F-conditional is true. Thus, God is unable to instantiate a Mr. Rogers type possible person. Adams is not rejecting Lewis’s lemma-quite the contrary. It is exactly because he accepts it and denies claim I that he reaches the anomalous conclusion that God cannot instantiate any possible person.

Maybe my interpretation of Adams is unduly uncharitable, for certainly as a Free Will Defender he does not want to wind up with this anomalous conclusion. Another way of understanding him is as denying both claim I and Lewis’s lemma, rather than denying claim I and accepting Lewis’s lemma, which resulted in anomaly. On this new interpretation, that God instantiates DP and this is followed by the instantiator freely doing A entails that God was able to instantiate P but not that F was true. But this new position is not without the appearance of anomaly, for it seems obvious that F would be rendered true by this sequence of events. Adams better have a very good argument for why our intuitions deceive us in this matter, however, I can’t find any such argument in his essay. I get the feeling that he is making impossible demands on an F-conditional first, neither its antecedent nor anything else can nomically necessitate its consequent; and, second, for it to be true there must be this very relation of nomic necessitation between them. That the antecedent probabilize the consequent in virtue of various dispositions and inclinations of the instantiator of DP isn’t enough.

At this point, Adams no doubt will hit the ball back into Plantinga’s court and ask him to supply an explanation of how F possibly could be true. Plantinga’s above claim that there is something Curley would have done if offered the bribe suggests a “minimalist” account of F-conditionals. While F-conditionals usually involve a probabilizing relation between their antecedent and consequent due to a set of freedom-neutral properties often containing dispositional properties, they need not. For an F-conditional to be true it suffices (though isn’t necessary) that the instantiation of its antecedent both precede and be necessary for the event reported by its consequent. This minimalist view could aptly be called the “Stage-setting” account. F is true if the instantiation of DP is in fact followed by the instantiator’s freely doing A, since the former is necessary for the ensual of the latter. While the law of conditional excluded middle holds for stage-setting subjunctive conditionals, it does not apply to all subjunctives, such as “If I were now to snap my fingers, Caesar would have freely crossed the Rubicon,” in which there isn’t the right sort of conditional-ensual relation, in this because there is neither the right sort of temporal order between what is reported by antecedent and consequent nor the required stage-setting relation. Because a stage-setting F-conditional need not involve anything more than the instantiation of its antecedent both preceding and being necessary for the event reported by its consequent, it will resist a possible worlds analysis in terms of the closeness relation between different worlds.10 For a world in which the instantiator of DP freely does A and one in which it freely refrains from doing A need not involve any other difference between them, such as a difference in their deterministic or statistical laws.

I think that this stage-setting analysis of F-conditionals in terms of their truth-conditions captures the ordinary way in which they are used, and explains why we are perplexed by the claim that DP’s instantiation could be followed by its instantiator freely doing A without it being true that if DP were instantiated, its instantiator would freely do A. This is just as anomalous as saying That God’s instantiation of DP could be followed by the instantiator freely doing A without God having been able to instantiate P.

Second Version. This version accepts both thesis I and Lewis’s lemma, understood in terms of minimalist stage-setting F-conditionals, and thereby escapes both types of anomalies. Its special gimmick is to deny thesis II, thereby enabling God to reply to questions about his foreknowledge by saying that although there was a truth of the matter at the time he deliberated about creating free persons, there was no way he could have known then that these moral evils would result from his decision to instantiate certain diminished possible persons. This second version, because it accepts thesis I and thesis III, also makes available to God, in addition to the excuse of ignorance, the I-was-stymied-by-F-conditionals excuse of Plantinga’s above FWD. The latter excuse, of course, could be offered by God only after the fact, not at the time of his creative decision, as it can be in the Plantinga version.

It is plain that in accepting thesis I and rejecting thesis II, the second version rejects the traditional definition of God’s omniscience according to which God knows every true proposition and is replacing it with the weaker definition that

O. For any proposition p, if it is true that p and it is logically possible that someone know that p, then God knows that p.

The problem is why is it logically impossible for anyone to know an F-conditional in advance? The answer is that knowledge requires a justificatory explanation and none could be given in advance for an F-conditional, since neither its antecedent nor anything else determines its consequent. This certainly holds true for finite persons, since their knowledge is discursive and thus in need of a justificatory explanation. But God is supposed to know things in his own inimitable supernatural way that is denied to us finite beings, just as he is supposed to be able to do or bring about things that we cannot, such as create things ex nihilo. Moreover, we cannot even conceive of how God knows and does these things, though we can, pace Berkeley, conceive of there being things that we cannot conceive of. Maybe God’s justification for believing an F-conditional that has not yet had its antecedent instantiated is that he knows that he is God and thereby omniscient and thus whatever he believes is true. To restrict God’s omni-powers to what we can conceive of is a radical anthropomorphizing of God. For this reason a traditional theist cannot adopt this second version.

But this is not the end of the matter, since the “traditional” account of God’s omniscience is a theoretical reconstruction of the Biblical notion of an all-knowing personal Deity that, in spite of its essential omnitemporality, is through and through temporalistic. Because God is time-bound, he does not know everything about the future, such as what future decisions he himself will make and what responses will be made by created persons to his overtures. He periodically returns, incensed at the turn of events because of his disappointment at these responses. Contemporary process theologians give a theoretical reconstruction of God’s omniscience that is closer to the temporalistic God of the Bible. Maybe it is the sort of religiously available God that is needed by the working theist?


  1. This work appears in a slightly different version as a chapter of my On the Nature and Existence of God, forthcoming from Cambridge University Press in 1991.
  2. For a full treatment of these objections, see Chapter 4 of my forthcoming On the Nature and Existence of God.
  3. Nelson Pike gives an excellent account of such cases in his “Over-Power and God’s Responsibility for Evil,” in The Nature and Existence of God, ed. A. Freddoso (Notre Dame: Notre Dame University Press, 1983).
  4. Plantinga, in his “Self-Profile” in Alvin Plantinga Profiles Volume, ed. James Tomberlin and Peter van Inwagen (Dordrecht: Reidel, 1985), p. 45.
  5. Adams, “Middle Knowledge and the Problem of Evil,” American Philosophical Quarterly 14 (1977): 110.
  6. Adams, p. 110, my italics.
  7. Adams, p. 12.
  8. Plantinga, The Nature of Necessity (Oxford: Oxford University Press, 1975), p. 181.
  9. Adams, p. 117.
  10. Adams gives a penetrating critique of such analysis in his essay.

Richard M. Gale

Department of Philosophy

University of Pittsburgh

all rights reserved