The Fallible God of Molinism

I recently listened to the exchange on Molinism and Calvinism between William Lane Craig and Paul Helm on Justin Brierley’s Unbelievable? radio program. It was more of a conversation than a debate, but it’s still worth a listen. In this post I want to expand on a point Helm raised but didn’t himself develop. I’ll first summarize the main tenets of Molinism before discussing what I regard as a serious objection to it. (Be patient — the first half of this post is just set-up.)

Molinism is a philosophical theory designed to reconcile a strong view of divine providence (according to which God foreordains all things) with a libertarian view of free will and a synergistic view of salvation (according to which God doesn’t cause anyone to repent and believe; instead sinners freely cooperate with God’s resistible grace in order to be saved). According to Molinism, God is able to providentially direct events by means of his middle knowledge, that is, his knowledge of what any libertarian-free creature would choose in any specific circumstances. For example, God knew prior to his decision to create this world whether I would freely choose a Boston Kreme if I were to go to Dunkin’ Donuts at noon on February 19, 2014, in such-and-such exact circumstances. God is therefore able to plan events down to the very last detail by prearranging the precise circumstances in which his creatures will find themselves and make their free choices. God doesn’t cause those choices, but he does guarantee them in some strong sense by orchestrating circumstances in light of his middle knowledge.

Molinists are therefore committed to three key claims. First, God foreordains all things, including the free choices of his creatures. Craig was quite emphatic about this during the exchange with Helm. He stated that God “preordains” everything and quipped that he could affirm almost everything the Westminster Confession of Faith says about God’s eternal decree and God’s providence — apart from the confession’s deliberate rejection of Molinism! (Note the final clause in WCF 3.2.)

The reason Molinists such as Craig want to affirm this is because they recognize that the Bible has a very strong view of providence (e.g., Gen. 50:20; Matt. 10:29-30; Acts 4:27-28; Rom. 8:28; Eph. 1:11). And they should be commended for recognizing this (although, as I’ll explain shortly, the biblical view is even stronger than they think).

Secondly, by affirming a libertarian view of free will, Molinists are committed to the idea that if a person S freely chooses A in specific circumstances C, then it must have been possible for S not to have chosen A. It may be more likely (perhaps even overwhelmingly so) that S will choose A over not-A, but it must be possible for S to have chosen otherwise. (Here I’m glossing over the distinction some libertarians make between derivatively and non-derivatively free choices. In what follows, for simplicity’s sake, I’m going to ignore derivatively free choices; I don’t think the distinction affects my argument.)

Molinists are therefore committed to the following:

(1) If S freely chooses A in C, then it is possible for S not to choose A in C.

Or to put the point in terms of possible worlds:

(1′) If S freely chooses A in C, then there is at least one possible world in which S does not choose A in C.

Note that in these statements C here should be understood as, in effect, the entire history of the universe up to the time of S’s choice. The point is that S’s choice isn’t determined (or at any rate causally determined) by any prior events or states of the universes (including S’s character, past experiences, memories, beliefs, desires, etc.).

Thirdly, the Molinist affirms that God has middle knowledge, that is, knowledge of the (inaptly named) counterfactuals of freedom. God therefore knows, prior to his decision to create a particular world, subjunctive conditional truths like the following:

(2) If S were in C, S would freely choose A.

It is this knowledge, the Molinist claims, that allows God to foreordain events. Since God knows what every possible creature would freely choose in every possible circumstance, he can plan accordingly. He can know in advance how things would go if he were to create particular people and arrange for them to make their free choices in particular circumstances.

This means that God cannot actualize just any possible world. Molinists such as Craig will distinguish between possible worlds and feasible worlds. All feasible worlds are possible worlds, but not all possible worlds are feasible worlds. A feasible world is a possible world that God can actualize on the basis of his middle knowledge: his knowledge of truths like (2). He can “weakly actualize” any one of these feasible worlds by “strongly actualizing” those elements of the world under his causal control, viz., which creatures exist and which circumstances they find themselves in. By his middle knowledge, God knows which possible worlds are feasible worlds, and he decides (on the basis of certain criteria) which of those feasible worlds to (weakly) actualize. And that is the Molinist view of divine foreordination.

It should be noted that Molinists aren’t always clear on what counts as circumstances in truths like (2). Do the circumstances include God’s decree or God’s foreknowledge, for example? Fortunately, we don’t need to settle that question here, so long as we recognize that whatever is included under C in (2) must also be included under C in (1) and (1′).

[Addendum: Greg Welty has pointed out that I slipped up here. See below for his comments and my responses. In short, it does matter how the circumstances are defined, but I think the Molinist faces serious problems either way.]

Well, so much for the background. The question I want to pose is simply this:

Is God infallible on the Molinist view?

I suspect most Molinists will be inclined to say yes. On the very face of it, we would expect God to be infallible. “Fallible God” hardly sounds right! More significantly, the Bible teaches explicitly that God’s plans cannot fail. God’s purposes are always accomplished:

Then Job answered the Lord and said: “I know that you can do all things, and that no purpose of yours can be thwarted.” (Job 42:1-2)

Many are the plans in the mind of a man, but it is the purpose of the Lord that will stand. (Prov. 19:21)

The Lord of hosts has sworn: “As I have planned, so shall it be, and as I have purposed, so shall it stand… For the Lord of hosts has purposed, and who will annul it? His hand is stretched out, and who will turn it back?” (Isa. 14:24, 27)

“Remember this and stand firm, recall it to mind, you transgressors, remember the former things of old; for I am God, and there is no other; I am God, and there is none like me, declaring the end from the beginning and from ancient times things not yet done, saying, ‘My counsel shall stand, and I will accomplish all my purpose,’ calling a bird of prey from the east, the man of my counsel from a far country. I have spoken, and I will bring it to pass; I have purposed, and I will do it.” (Isa. 46:8-11)

“For as the rain and the snow come down from heaven and do not return there but water the earth, making it bring forth and sprout, giving seed to the sower and bread to the eater, so shall my word be that goes out from my mouth; it shall not return to me empty, but it shall accomplish that which I purpose, and shall succeed in the thing for which I sent it.” (Isa. 55:10-11)

Of course, the Molinist is familiar with these texts. The strong view of divine providence expressed in Scripture is one of the reasons he favors Molinism over other positions that share a commitment to libertarian free will.

However, it should be clear from the above that according to the Molinist view there are possible worlds in which God’s plans fail. For the Molinist is committed to the claim that although God knows that S would choose A in C, and he actualizes C because he plans for S to choose A, it is nonetheless possible for S not to choose A in C. (Craig clearly affirm this point a couple of times in his exchange with Helm.) In other words, there are possible worlds in which God actualizes C so that S will choose A, but S doesn’t choose A. There are possible worlds in which God’s eternal decree doesn’t come to pass, because libertarian-free agents do otherwise than he had planned.

The upshot is this: on the Molinist view there are some possible worlds in which God is fallible. Indeed there are many, many such worlds. Any world in which God’s plans fail is a world in which God is fallible. It seems to me that this conclusion is built in to the Molinist system.

So what could a Molinist say in response? One reply would be to concede the point at to say that God isn’t necessarily infallible: he isn’t infallible in every possible world, but he’s infallible in this world (at least). God is contingently infallible.

But I think there are several serious problems with this response. In the first place, the very notion of “contingent infallibility” is logically suspect. On the face of it, infallibility is a modal concept. It’s concerned with possible failure, not merely actual failure. Infallible means not fallible, and fallible means capable of failure. If a person is fallible, it’s not because he has failed or because he will fail but because he could fail. After all, there’s nothing logically inconsistent in saying that S is fallible but S didn’t (or won’t) actually fail. What would be flatly inconsistent would be to say that S is fallible but S cannot fail.

Hence we wouldn’t describe someone as infallible simply because he actually succeeds in every case, even though he might have failed at any point. Infallibility surely entails that one cannot fail, even in principle. So it’s not clear that the notion of “contingent infallibility” is even coherent.

If I’m right about this then Molinists shouldn’t say that God is infallible. For if God isn’t infallible in every possible world then he isn’t infallible in any possible world, including the actual world. (I note for the record that this argument presupposes some widely-held modal principles which I’m not going to defend here, precisely because they’re widely held!)

Secondly, these divine-failure worlds present a problem for Molinists who are committed to perfect being theology (which is most of them, I would guess). If God is only contingently infallible, it follows that God doesn’t possess maximal greatness: for a being who is necessarily infallible is greater than a being who is contingently infallible. A being who is infallible in all possible worlds is greater than a being who is infallible in only some possible worlds. (It’s noteworthy that Alvin Plantinga, who introduced the notion of maximal greatness in his defense of the ontological argument, is one of the most prominent advocates of Molinism.)

The only way out of this for the Molinist, so far as I can see, is to take the route I suggested above: to abandon the claim that God is infallible (in this world or in any other possible world).

It seems to me that all of this exposes an unresolved tension at the heart of Molinism. When we ask the question, “Can humans frustrate God’s plans?” the Molinist is pulled in two directions. On the one hand, he’ll want to answer no. After all, God foreordains all things! God has an eternal decree. It’s inconceivable to the Molinist that God’s decree should fail.

Yet at the same time, the Molinist must also want to answer yes, because of his commitment to libertarian free will. He wants to affirm (as Craig does explicitly) that although S would choose A in C (and will in fact choose A if God has decreed it) it’s nonetheless possible for S not to choose A in C — and thus it really is possible for S to act contrary to God’s plans.

In short, the Molinist wants to have his cake and eat it too.

Another way to look at the problem is to ask whether the non-feasible worlds really are possible. Suppose that God opts for world X out of all the feasible worlds he considers. It hardly seems coherent to imagine God thinking, “I’m going to weakly actualize X — that’s my decree — but I know there’s a real possibility that I’ll end up with some other world, because of the real possibility that libertarian-free agents will act otherwise than I’ve planned!” But if that isn’t coherent, in what sense are these other worlds really possible worlds?

As I see it, this tension at the center of Molinism arises because it aspires to be a deterministic indeterminism. ‘Indeterminism’ because of its commitment to libertarian free will. ‘Deterministic’ because God’s decree somehow (we know not how) determines in advance that his creatures will make certain choices. It may not be a causal determinism, but it’s deterministic nonetheless (as many non-Molinist Arminians, such as Roger Olson, can clearly see). If preordaining that S chooses A doesn’t mean predetermining that S chooses A then what on earth does it mean?

The Molinist Cup-and-Ball Trick

Molinism is certainly an impressive theory. But it’s only impressive in the way that a slick cup-and-ball trick is impressive. It’s philosophical sleight-of-hand par excellence. Molinists have to be skilled in the art of misdirection. Now you see the determinism… and now you don’t!

40 Responses to The Fallible God of Molinism

  1. I think your question is a death blow for Molinism. I once held the Molinist view but have since come to more fully understand the Reformed position. Here’s a good quote that summarizes Reformed thought: “The Reformed agree that God knows what would happen under all conditions, but they reject the notion that this knowledge is ever ultimately based on man’s autonomous decisions. Human decisions, they argue, are themselves the effects of God’s eternal decrees (see Acts 2:23, Rom. 9:10-18, Eph. 1:11, Phil. 2:12-13).”—John M. Frame “Scientia Media” from Evangelical Dictionary of Theology

  2. David Reimer

    Everything I know about Molinism (more or less), I know from reading this blog post. So…!

    Two thoughts came to mind: (1) concerns this claim:

    In other words, there are possible worlds in which God actualizes C so that S will choose A, but S doesn’t choose A. There are possible worlds in which God’s eternal decree doesn’t come to pass, because libertarian-free agents do otherwise than he had planned.

    Why? Is there a non sequitur here? Why should not-A thwart the decrees of God? From what I’ve read here (the sum total of my Molinist learning!) there is an assumption, isn’t there?, that God’s purposes can only be realized by S choosing A. Presumably a Molinist would say that God can accomplish his purposes equally certainly should S choose not-A. (I imagine I’ve fouled up something in (1) and (1′) and (2) here, but … it’s late.)

    Which brings me to my second thought. (2) The Job 42:2 text suggests a doctrine of divine unthwartability. There is (what I guess would be) a Molinist like text in Exodus 13:17 (pick your translation! here’s ASV, for convenience): “And it came to pass, when Pharaoh had let the people go, that God led them not by the way of the land of the Philistines, although that was near; for God said, Lest peradventure the people repent when they see war, and they return to Egypt…”.

    Isn’t this God, looking down the road and seeing trouble ahead, and so choosing route B rather than route A? I suppose it’s also explicable as God “actualiz[ing] C because he plans for S to choose A”. (And, of course, Ex 13:18 has God leading them in circles.) God took them round about to avoid Philistine aggression … presumably a Molinist would say that had the Israelites made a bee-line for the Red Sea, God would still have ensured that they arrived in the end in the Promised Land … because, as Job confessed, “no purpose of yours can be thwarted”.

    Sorry for blethering on!

    • Hi David,

      “Is there a non sequitur here? Why should not-A thwart the decrees of God?”

      Not-A would thwart the decree of God because, ex hypothesi, God has decreed that S choose A. The issue isn’t whether God can accomplish his purposes by means of A or not-A. The issue is whether, God having already decreed that S choose A, it is possible for S to choose not-A. You’re considering the pre-decree situation; I’m discussing the post-decree situation.

      As for Exodus 13:17, Calvinists can agree with Molinists that there are true counterfactuals of freedom, that is, truths about what people would do (or would have done) in alternative circumstance. The issues are (1) whether these truths are brute contingent facts independent of God and (2) whether God’s decree is dependent on such truths. The mere fact that there are counterfactuals expressed in Scripture does not as such select between Calvinism and Molinism.

  3. Pingback: James Anderson on Molinism’s Slight of Hand | Remington's

  4. David,

    1) Suppose God’s purpose is for S to choose A.

    2) Molinism is more than counter-factual knowledge. So Exodus 13:17 isn’t a Molinist text. Craig admits that Scripture doesn’t support Molinism per se.

  5. “Note that in these statements C here should be understood as, in effect, the entire history of the universe up to the time of S’s choice.”

    Doesn’t that give them a solution? Just make sure that God’s decree/foreknowledge is not included in the C which is specified in the relevant counterfactual. Thus, in no possible world is there a mismatch between God’s foreknowledge and what creatures choose.

    After all, on the Molinist view God’s middle-knowledge isn’t part of the ‘history’ of the world he actualizes. Why should we think his foreknowledge is? The history of the world includes those things causally relevant to the choice, but why think divine foreknowledge is causally efficacious, influential, or something similar?

    You ask, “Do the circumstances include God’s decree or God’s foreknowledge, for example?” This is a good question, but you answer that it doesn’t matter, and I’m not sure why. Molinists seem to think it does matter. While Molinists say agents can do otherwise in the same circumstances, no Molinist would say that it’s possible for an agent to act contrary to God’s actual foreknowledge. For that isn’t part of the ‘circumstances’ in which the agent makes his choice. The circumstances include all causally relevant aspects of the agent’s environment, and divine foreknowledge isn’t causally relevant.

    “For the Molinist is committed to the claim that although God knows that S would choose A in C, and he actualizes C because he plans for S to choose A, it is nonetheless possible for S not to choose A in C.”

    Yes, but if S were not to choose A in C, then God would have foreknown something different. So again there is never a mismatch between God’s foreknowledge and creaturely action. If such a set of circumstances were to have been actualized, it would come with a different content to God’s foreknowledge. God is not so stupid as to have his decree with respect to C be out of accord with the middle-knowledge he has. He’s not going to decree C as a way of getting S to choose A, if the relevant counterfactual says that S wouldn’t A in C.

    Yes, given the contingency of the counterfactual, it’s possible that S doesn’t A in C. But that doesn’t mean that it’s possible for the agent to act contrary to God’s foreknowledge, so that there’s a mismatch. For if S were to refrain from A in C, then the relevant counterfactual would have been false. And in that case, God wouldn’t have actualized C as a way of getting S to do A, because he would know that would get him nowhere. Since no possible world would be actualized apart from God actualizing circumstances per Molinist providence, no possible world would involve a mismatch between divine foreknowledge and what happens.

    “He wants to affirm (as Craig does explicitly) that although S would choose A in C (and will in fact choose A if God has decreed it) it’s nonetheless possible for S not to choose A in C — and thus it really is possible for S to act contrary to God’s plans.”

    Here you seem to be conflating “not to choose A in C” and “act contrary to God’s plans.” To be sure, it’s possible for S to refrain from A in C, but if he were to do so, that would be a world that God actualized in the usual Molinist sense, according to his plan. So once again in no possible world does S choose contrary to God’s plan.

    “Another way to look at the problem is to ask whether the non-feasible worlds really are possible.”

    I think what you state here is a problem for Molinists. But I’m not sure it’s ‘another way to look at the problem.’ I think it’s a different problem.

    If it is the same problem, then it might admit of the same solution (that I’ve suggested above). It’s not that God actualizes C – knowing that in C, S would choose A – but then thinks that given his actualization of C “there’s a real possibility that I’ll end up with some other world.” Given the truth of the relevant counterfactual and his choice to actualize the antecedent, things could not turn out differently. Nevertheless, the truth of the relevant counterfactual is contingent, and his will is contingent, and as long as there is a possibility for those things to be otherwise there’s a possibility for a different outcome. But the different outcome would come with different foreknowledge, and so all is well.

    • Hi Greg,

      You’re above my pay-grade but I’m interested in this topic so I hope you don’t mind if I throw out some thoughts I had while reading your response.

      >>Just make sure that God’s decree/foreknowledge is not included in the C which is specified in the relevant counterfactual.

      If these are possible worlds in which God is present and active (say, in concurrence or something like that), how could God’s decree and foreknowledge not be a part of C?

      >>After all, on the Molinist view God’s middle-knowledge isn’t part of the ‘history’ of the world he actualizes. Why should we think his foreknowledge is?

      Because God’s foreknowledge is part of his free-knowledge?

      >>While Molinists say agents can do otherwise in the same circumstances, no Molinist would say that it’s possible for an agent to act contrary to God’s actual foreknowledge.

      Dr. Craig seems to say exactly this in his discussion with Helm. Maybe he wasn’t being careful. But it’s not clear to me how a molinist can maintain that I or anyone else “can do otherwise” given God’s foreknowledge. Later in your comment you seem to rely on what Fischer calls the “back-tracking conditional”… but as Fischer points out, thats no solution (only the presumption that there is one). Unless one doubts the fixity of the past.

      >>Given the truth of the relevant counterfactual and his choice to actualize the antecedent, things could not turn out differently.

      So, the world as it now stands is accidentally necessary.

      >>Nevertheless, the truth of the relevant counterfactual is contingent, and his will is contingent,

      But now accidentally necessary…

      >>and as long as there is a possibility for those things to be otherwise there’s a possibility for a different outcome.

      In what significant sense is there “a possibility for those things to be otherwise” given that they are now accidentally necessary?

      • Remington,

        “If these are possible worlds in which God is present and active (say, in concurrence or something like that), how could God’s decree and foreknowledge not be a part of C?”

        Because God’s foreknowledge exerts no causal influence whatsoever upon the agent’s choice. So why again is it relevant to the specification of ‘circumstances’? The libertarian point is that I can do otherwise despite any and all causally relevant factors – both near and remote – being exactly the same. As seen below, Molinists typically cash out ‘circumstances’ in terms of history, laws of nature, and divine action. Divine foreknowledge doesn’t qualify.

        By the way, if you want to make God’s foreknowledge comprise part of the circumstances in which libertarian agents act, then the argument in this blog post has very little to do with Molinism. If it is a general condition on libertarian agency that the agent must be able to do otherwise even given the same exact foreknowledge by God, then clearly God’s foreknowledge is fallible if human agency is libertarian. No appeal to Molinism is needed to generate this conclusion.

        What’s missing in this discussion is what Molinists have actually written on the topic. It’s not as if they haven’t thought about this. So consider Thomas Flint, “Two Accounts of Providence,” in Thomas Morris (ed.), _Divine and Human Action: Essays in the Metaphysics of Theism_ (Cornell Univ. Press, 1988). Here Flint affirms, as “virtually definitive of the libertarian position,” principle (C):

        “(C) Necessarily, for any human agent S, action A, and time t, if S performs A freely at t, then the history of the world prior to t, the laws of nature, and the actions of any other agent (including God) prior to and at t are jointly compatible with S’s refraining from performing A freely.”

        He states (C) explicitly in a context in which he is considering how we ought to specify the circumstances of the agent’s choice in contrast to other views of how to do this (such as the Thomist view). God’s foreknowledge isn’t part of the history of the world, it’s not a law of nature, and it’s not an action on the part of God. So why should it be in the antecedent?

        Likewise, on p. 40 of _Divine Providence: The Molinist Account_, Flint says that the ‘circumstances’ in a counterfactual “specifies the complete set of nondetermining circumstances in which the creature is placed.” It seems awkward at best to regard divine foreknowledge as a ‘circumstance’ in which a creature is ‘placed.’

        Does God really need middle knowledge in order to answer the question: “Hmm, I wonder what would happen in a world in which I infallibly foreknow that the creature does A? What would happen in *those* circumstances?” God doesn’t need middle-knowledge to figure that out. Rather, he figures out what creatures will in fact do by way of combining his middle-knowledge with his decree to create, and then (logically) he gets his free knowledge.

        Imagine if we were to include divine foreknowledge in the antecedent of counterfactuals:

        If S were in C (and thus, foreknown by God to do A), then S would freely do A.

        If God’s foreknowledge is infallible, then this makes the conditional necessary (wide scope), and thus a necessary truth. I think it’s up to the critic of Molinism to make the case that divine foreknowledge belongs in the specification of ‘circumstances’. Why should the Molinist agree here? (Notice that James implied it didn’t matter what we do here, whereas your view seems to be that it’s crucial to include it.)

        “Because God’s foreknowledge is part of his free-knowledge?”

        I don’t think that’s sufficient for being part of ‘the history of the world,’ as Molinists appeal to that notion in their system of providence. But let’s go with this anyway. Even granting this assumption, there’s little reason to think this impacts the infallibility of foreknowledge. So consider two possible worlds, and two counterfactuals that describe those worlds:

        CF1 If Cuthbert were in C, Cuthbert would freely buy the iguana.
        CF2 If Cuthbert were in C, Cuthbert would freely refrain from buying the iguana.

        Let’s say that CF1 is true, CF2 is false, and God wants a world in which Cuthbert freely buys the iguana. God decides to actualize C, and thus foreknows that Cuthbert buys.

        Could Cuthbert refrain from buying? Sure, says the Molinist. And what if he *were* to refrain? How should we talk about that? Well, some things would be different. First, CF1 would be false and CF2 would be true. After all, these counterfactuals are contingent and so their truth changes from world to world. Second, no world will ever be actualized apart from God’s creative decree. So if Cuthbert were to refrain, that would be a world in which CF2 is true, and God would have exploited his middle knowledge of CF2 to get Cuthbert’s refraining. And, obviously enough, given that middle knowledge he would have had the free knowledge of Cuthbert’s refraining.

        Generalizing from this example, it follows that in no possible world is there a mismatch between Cuthbert’s actions and God’s foreknowledge. So no challenge to infallibility.

        Now let’s strengthen CF1 and CF2 as you prefer:

        CF1’ If Cuthbert were in C (which includes being foreknown by God to freely buy the iguana), Cuthbert would freely buy the iguana.
        CF2’ If Cuthbert were in C (which includes being foreknown by God to freely refrain from buying the iguana), Cuthbert would freely refrain from buying the iguana.

        If C *includes* foreknowledge, then these aren’t the same circumstances, and the common nomenclature of ‘C’ above is misleading. So you’re not giving us a case in which *libertarian* freedom trumps the infallibility of divine foreknowledge. For, *on your own assumptions*, you’re not giving us a case in which the agent does otherwise *in the same exact circumstances*. What you need is a Molinist scenario in which an agent *acting libertarianly* is able to defy foreknowledge. But you don’t have one.

        “Later in your comment you seem to rely on what Fischer calls the ‘back-tracking conditional’… but as Fischer points out, thats no solution (only the presumption that there is one). Unless one doubts the fixity of the past.”

        But the burden is on you to characterize God’s foreknowledge as partaking of “the fixity of the past” or “accidental necessity”. Why should the Molinist think this? In addition, even if (1) God’s foreknowledge deserves to be called ‘past,’ and (2) the past is temporally fixed, (3) Craig and Freddoso – and Molina – deny that temporal necessity is closed under entailment. Think about all of Craig’s counterexamples in *The Only Wise God*: Newcomb’s Paradox, relativity theory, time travel, precognition, etc. Again, it’s not as if Molinists haven’t given this some thought.

        Finally, when you say, “that’s no solution,” solution for *what*, exactly? What problem does this blog post have in mind, and is Fischer really talking about that problem?

        • “What’s missing in this discussion is what Molinists have actually written on the topic.”

          I’m not sure to what extent that’s directed at me, but let me say that my post was prompted by specific comments made by Craig in his exchange with Helm.

          Here’s a relevant snippet, which came toward the end of the discussion:

          Helm: “Why can’t they [i.e., human libertarian-free agents], as it were, go off in the other direction [i.e., than the direction God decreed]?”

          Brierley [echoing Helm's question]: “Why couldn’t they go off-script?”

          Craig: “Well, they could! That’s the whole idea!”

          Now, maybe Craig misspoke here. Perhaps on reflection he would have said something different in response, along the lines you described. But that’s what he actually said.

      • Greg,

        Thanks for indulging me.

        Let me start at the bottom of your reply and then jump back up:

        >>Finally, when you say, “that’s no solution,” solution for *what*, exactly? What problem does this blog post have in mind, and is Fischer really talking about that problem?

        James Anderson and John Martin Fischer aren’t addressing the same thing, and I don’t think I implied that they were. But the issues are related at the level of whether foreknowledge is compatible with PAP and, it seemed to me, especially in light of issues you raised.

        >>Because God’s foreknowledge exerts no causal influence whatsoever upon the agent’s choice. So why again is it relevant to the specification of ‘circumstances’

        It certainly seems to entail something about the causal power of the agent (unless you deny the fixity of the past, as I mentioned before, or that accidental necessity is closed under entailment) , whether that’s a causal influence or some other type of influence seems irrelevant. As Fischer says, “this concession does nothing to vitiate the force of the commonsense point that the past is fixed” and “then it seems highly dubious to distinguish causally relevant from causally irrelevant features of the past (in regard to their fixity)” (On Molinism, Oxford Studies in Phil of Rel: Vol 1, p. 33).

        You go on to note that Molinists like Craig and Freddoso deny that accidental necessity is closed under entailment. Fischer addresses this in the paper I was referring to.

        >>It seems awkward at best to regard divine foreknowledge as a ‘circumstance’ in which a creature is ‘placed.’ Does God really need middle knowledge in order to answer the question: “Hmm, I wonder what would happen in a world in which I infallibly foreknow that the creature does A? What would happen in *those* circumstances?” God doesn’t need middle-knowledge to figure that out.

        Isn’t divine foreknowledge part of our circumstance? God may not “need” middle-knowledge to “figure out” the scenario you present, but it seems to me perfectly fine for God to see something like “In a world in which I have foreknowledge of the creature’s doing A…” And since God exists (omnisciently) in every possible world then I think something like that is what God does see.

        >>I think it’s up to the critic of Molinism to make the case that divine foreknowledge belongs in the specification of ‘circumstances’. Why should the Molinist agree here? (Notice that James implied it didn’t matter what we do here, whereas your view seems to be that it’s crucial to include it.)

        I haven’t suggested it’s crucial (or at least I didn’t intend to suggest that). But at any rate it seems obvious to me that foreknowledge belongs to the specification of our circumstance in the actual world. It seems to me that insofar as God exists in every possible world then God’s foreknowledge comprises a part of that world. You don’t think it’s causally relevant, but I guess I’m with Fischer in not seeing it (or maybe I should go re-read Fischer).

        What you say about Cuthbert and the iguana seems to get back to the issue James Anderson is pointing to and Paul Helm was pointing to in the dialogue with Craig. If Cuthbert can refrain from choosing to buy the iguana in CF1′, then there’s a possible world in which he does.

    • Thanks, Greg. Excellent comments, as usual. See, this is why you need to write that book!

      I have to deal with some other stuff today (you know, the stuff we get paid for) but I will post some remarks in response later on.

    • Hi Greg, your criticism is exactly the one I had on mind (and I look forward to James response). However, when you say, “The history of the world includes those things causally relevant to the choice,” did you leave out an *only*, i.e., “it includes [only] those things causally relevant to the choice.” It seems so, because if not, how is it that FK and the decrees removed from the history? I don’t think this is right, though. It should be all the intrinsic or non-relational facts. If we only hold fixed those causally relevant to the choice, we get to remove most of the hard facts about the past, and I don’t think that’s right. For there’s more ways LFW can be ruled out other than just the causal antecedents. We might say, from a LFW perspective, that those antecedents *control* (in some L-freedom undermining sense) our actions. But, another way to rule out LFW is not by showing the relevant agent’s actions are *controlled* but, rather, *out of* the agent’s control (again, in some L-freedom undermining sense).

      • Paul,

        Yes, I meant to say “only”. Thanks for the catch.

        “If we only hold fixed those causally relevant to the choice, we get to remove most of the hard facts about the past, and I don’t think that’s right. For there’s more ways LFW can be ruled out other than just the causal antecedents.”

        I’d like to hear more about these “hard facts about the past” that are not causal antecedents of my choice and yet undermine my LFW. I’d particularly be interested in any such facts that would be relevantly analogous to the mere fact of divine foreknowledge. It’s this latter assumption that’s key, I think.

    • Greg,

      If I can boil it down, your objection comes to this. Whether or not C in (1) and (2) includes God’s decree or foreknowledge is relevant, contrary to what I claimed. If C does include them, then God must be fallible if S can refrain from choosing A in C (contrary to God’s decree). But Molinists will deny that C includes God’s decree or foreknowledge, because they aren’t causally relevant. So in the possible world in which S refrains from choosing A in C, God’s decree is different (because the counterfactuals are different). In that world, God has decreed that S will refrain from choosing A in C. So God’s decree does indeed come to pass in that world. No problem!

      Is that a fair summary?

      Several things to say in response. First, you’re correct that I slipped up at that point. It does make a difference whether or not C includes God’s decree or foreknowledge. So I thank you for the correction.

      However, I don’t think that gets the Molinist out of trouble. In what follows, I’m going to focus on God’s eternal decree, as I did in my original post. You focused on the foreknowledge issue in your reply, but I want to focus on the Molinist concept of God’s eternal decree, because that decree is active and intentional on God’s part. It is God’s decision to strongly actualize circumstances so as to weakly actualize a particular world. I’m not concerned (in this post) with the traditional foreknowledge/freedom dilemma, which isn’t distinctive to Molinism. I’m concerned here with the distinctive Molinist understanding of the divine decree.

      First of all, is God’s decree causally irrelevant? Not obviously so. Because of his decree, God (strongly, causally) actualizes C, and we agree that C is causally relevant. So it seems to me that God’s decree is indeed causally relevant, even if we concede that his foreknowledge is not.

      Secondly, preserving divine infallibility by excluding the divine decree from C has the bizarre and, I suggest, thoroughly unbiblical consequence of giving human agents counterfactual control over God’s decree. If, when S enters C, S has the power to choose A or not-A, S has at that moment the power to determine whether God has decreed an A-plan or a not-A-plan. To put it bluntly, on this view God’s will is bent to the wills of his creatures. Saying that it is only bent in a counterfactual sense hardly lessens the offense! This is quite at odds with the biblical view of divine sovereignty.

      Indeed, it’s a strange model of divine infallibility that hangs on the divine decree counterfactually tracking the libertarian-free choices of human agents. You might as well say that a train infallibly decrees its destination because it necessarily follows its tracks. If a junction gets switched, no problem, because that only means the train decreed a different station!

      Another implausibility of this view is that it requires us to detach the circumstances from the causal explanation of the circumstances. Suppose S finds himself in C and is about to choose. According to the Molinist, C is fixed but the causal explanation for C is not fixed. For S is free at this moment to choose A or not-A. If S chooses A, that in effect makes it that God eternally decreed A for some reason R1. But if S chooses not-A, that makes it that God eternally decreed not-A for some other reason R2. God, not S, determined C, and yet we’re to believe that S now has the power to determine (at least in part) why God determined C!

      Indeed, isn’t it odd to think that at the moment S makes his choice in C, the occurrence of C is determinate, but the reason for which C occurs is indeterminate (because it hangs on S’s indeterminate choice)?

      Perhaps you’ll say, “Yes, these are problems, but they’re different problems than the one you originally raised.” Maybe so. But as I see it, if the proposed solution to a problem (the need to preserve divine infallibility) immediately gives rise to further problems, one might say that the original problem hasn’t really been alleviated.

      Or to state the problem another way: depending on how the circumstances for human choices are defined, the Molinist must either (1) deny divine infallibility or (2) adopt a very artificial and unbiblical understanding of divine infallibility (rather like the open theist’s reconstruction of divine omniscience).

      There are a few other things I could say about the possibility of non-feasible worlds, but this will have to do for now.

      • James,

        “Is that a fair summary?”

        Sure.

        “First of all, is God’s decree causally irrelevant? Not obviously so. Because of his decree, God (strongly, causally) actualizes C, and we agree that C is causally relevant. So it seems to me that God’s decree is indeed causally relevant, even if we concede that his foreknowledge is not.”

        OK, so we’re talking about ‘fallibility’ in terms of efficacy of decree rather than modal accuracy of knowledge. For the record, I think the term ‘fallibility’ is typically used in the epistemological context of divine foreknowledge, whereas ‘irresistibility’ or ‘absolute efficacy’ are the preferred terms if we’re talking about the metaphysical context of divine decree. It seems strange to apply the epistemological concept of fallibility to divine decree. (Perhaps that’s why Remington immediately brought up a Fischer article that is devoted to Molinism and the freewill/foreknowledge dilemma.)

        But if we’re going to emphasize that we’re talking about divine decree rather than foreknowledge, and we want to characterize the decree as ‘causal’ (and we surely do), then we should be clear about *what* God is causing (i.e., strongly actualizing) by way of his decree. He is causing C. He is not, on the Molinist view, causing S’s A-ing. That is something S causes.

        So sure, let’s put ‘God decrees C’ (read: God causes C) into the antecedent of the counterfactuals:

        CF1 If God were to cause C and put S in C, then S would freely choose A.
        CF2 If God were to cause C and put S in C, then S would freely refrain from choosing A.

        Say CF1 is true and CF2 is false, and knowing this, God strongly actualizes C and weakly actualizes S’s A-ing (by way of S causing his own A-ing, of course). Could S have done otherwise? Yes. Could S have done otherwise *in the same C*? Yes! Does this threaten the efficacy of the divine decree? I don’t see how.

        Perhaps you think the Molinist must now envision a possible world in which God causes C but S doesn’t choose A. There goes S, thwarting the divine decree! But here we must step back and consider *why* God is deciding to cause C. And the answer is that he’s consulted the relevant counterfactuals and proceeded accordingly. In the possible world in which S freely refrains from A-ing in C, CF2 would be true and CF1 false. God would know this. Thus in *that* possible world, God would cause C knowing that S would refrain from A-ing. His decree would take that into account. He wouldn’t be frustrated because he caused C and then things ‘didn’t go his way’. He had a reason, in that world, for causing C, and there’s little reason to think that his reason there is the same as in all the other possible worlds.

        If Molinism is a global theory of providence, then no world can be actualized without being actualized by way of God’s causing C after reflecting upon the truth-value of the relevant counterfactuals. It’s hard to see, then, how it *could* be the case that the results don’t match up with God’s purposes. No possible world can be actualized without its having a C that is caused by God for the purpose of realizing the consequents of the counterfactuals that are true in that world. So there isn’t even the possibility of a mismatch.

        So I entirely concede the causal relevance of God’s decree (as opposed to his foreknowledge). But I’m not convinced that gets you the result you want.

        “Secondly, preserving divine infallibility by excluding the divine decree from C has the bizarre and, I suggest, thoroughly unbiblical consequence of giving human agents counterfactual control over God’s decree. If, when S enters C, S has the power to choose A or not-A, S has at that moment the power to determine whether God has decreed an A-plan or a not-A-plan.”

        It all comes down to what you mean by ‘power to determine’. If you mean that their choice literally causes or determines the past, including divine choice, then yes, that’s problematic. But I think you intend something that falls short of literal causation, because you speak of humans having ‘counterfactual control over God’s decree’. What kind of ‘control’ is that? Does it coerce God to create them? Does it coerce God to put them in specific circumstances? Does it change the past? Is God, when he considers the truth-value of the counterfactuals, now forced to create anyone at all? I don’t think any of that follows. God could consult the counterfactuals and decide not to create anyone. That doesn’t sound like his decree is being ‘controlled’ by agents in any sense. Indeed, there wouldn’t be any agents unless he decided to create them.

        “If, when S enters C, S has the power to choose A or not-A, S has at that moment the power to determine whether God has decreed an A-plan or a not-A-plan.”

        Again, what is meant by ‘determine’? Literally determine? Cause? That doesn’t seem implied. Rather, S has the power to do something, such that if he were to do it, God would have decreed something different all along. That doesn’t speak at all to *why* God decrees such-and-such. Does the creature *make* God decree something different? No. In fact, it was entirely open to God to decree that there be no agents choosing A or not-A at all. So God knows what S would do in C. Whatever decree God settles on is *informed* by this knowledge, but the actual decree is not ‘determined’ or ‘controlled’ by the creature. He could consult the counterfactuals, and then just shrug and move on, not creating anything.

        “Indeed, it’s a strange model of divine infallibility that hangs on the divine decree counterfactually tracking the libertarian-free choices of human agents. You might as well say that a train infallibly decrees its destination because it necessarily follows its tracks. If a junction gets switched, no problem, because that only means the train decreed a different station!”

        This would be a good analogy, if the agents were actual the way the tracks and junction are actual. As you know, nothing creaturely is actual when God makes his decree. God isn’t ‘tracking’ any *actual* creaturely choices. If you want to talk about a Molinist train, the fact that there are any tracks or junction switches at all is due to the choice of the train at the outset. The train doesn’t ‘necessarily follow’ anything. As you can see, such a Molinist train isn’t much of an analogy to a real train.

        “For S is free at this moment to choose A or not-A. If S chooses A, that in effect makes it that God eternally decreed A for some reason R1. But if S chooses not-A, that makes it that God eternally decreed not-A for some other reason R2.”

        I can’t tell if by “in effect *makes it*” you are speaking of causation, or not. As you know, Molinists will eschew the causal interpretation. You are clearly implying at the least some sort of (metaphysical?) asymmetry that is supposed to reflect poorly on God in the Molinist system. But if S chooses not-A, all that is revealed is that it was true from eternity that he would as a matter of fact choose not-A. But then God knew that too from eternity, and made his decree accordingly. He certainly wasn’t caused by that truth to do anything. God’s actual intentions are settled from eternity. It is also true that he could have intended something else from eternity. I’m not sure how, on this story, anything ‘gets settled’ *at a future point in time*. The counterfactuals and their actual truth-value are available to God from eternity.

        • Thanks for the further comments, Greg.

          As Paul surmised, I’ve configured the blog so that comments can only be nested to the third level, otherwise the text columns start to get too narrow.

          Regarding your complaint about my vocabulary: I didn’t write the post for philosophy grads or profs, so I don’t feel obliged to stick to what that audience would consider “typical usage”. The notion of an infallible decree is perfectly clear and understandable.

          I reject the reduction of “God decrees” to “God causes”. God doesn’t merely decree C (where C is the intramundane circumstances, as you’re taking C to denote). He decrees everything. He decrees that S will choose A, and therefore he decrees C in order that S will choose A. All this to say, God decrees S’s choosing A in C (and because of C). I don’t believe C can be abstracted from God’s reasons for actualizing C. And that’s why I think God’s decree should be included in the antecedents of (1) and (2).

          “Again, what is meant by ‘determine’? Literally determine? Cause? That doesn’t seem implied. Rather, S has the power to do something, such that if he were to do it, God would have decreed something different all along. That doesn’t speak at all to *why* God decrees such-and-such. Does the creature *make* God decree something different? No.”

          I’m not sure why you’re missing the point here. According to the Molinist, it is ultimately up to S whether or not S chooses A. But then it is ultimately up to S whether or not God has decreed an S-chooses-A world.

          I realize that on the Molinist view you’re defending, if S were to choose otherwise, God would have decreed otherwise, because the relevant counterfactuals would be different. So it appears that the counterfactuals are ultimately up to S, and God’s decree hangs on the counterfactuals. God’s decree (at least in part) is ultimately up to S. As I said before, on this view God’s will is bent to the will of his creatures.

          You criticize my train analogy because the agents aren’t actual when God decrees. Well, sure, but I don’t see how that affects my point. I acknowledged that the divine decree counterfactually (not actually) tracks the choices of humans. So rewrite the analogy if you want: “If a junction were to have been switched, no problem, because that only means the train would have decreed a different station!” How does that look better for God?

          “God isn’t ‘tracking’ any *actual* creaturely choices.”

          He’s counterfactually tracking creaturely choices.

          If the counterfactuals are up to the creatures, then God’s decree is (at least in part) up to the creatures. And if they aren’t up to the creatures, then all the talk of “power to choose otherwise” is vacuous.

          The whole issue here comes to this: Does God’s decree guarantee that S chooses A? If so, then Molinism is really just a form of determinism: non-causal determinism, but determinism nonetheless. If not, then God’s decree must be fallible.

  6. It seems to me that you’re mixing up the logical moments of God’s knowledge. If S were in C, S could choose A B C or D. But, it is impossible for S to choose all options simultaneously. Therefore, S must freely choose one option. To put it in modal terms, there is a possible world where S chooses A, another where S chooses B, etc. God knows all of the possibilities here. However, once in C, only one of these worlds could be actualized by S’s free decision and God knows which world that would be. For example, let’s suppose S chooses A. You’re right that there is a possible world where S chooses B, but, there is not a world where S would choose A and chooses B. God’s middle knowledge logically follows his knowledge of possibilities and it seems to me that your argument tries to conflate the two. I may be misunderstanding, though. If that is the case, then please correct me.

  7. Pingback: Credo Magazine » Credo’s Cache

  8. Pingback: Late January 2014 Presuppositional apologetics Links | The Domain for Truth

  9. Pingback: Do Not Be Surprised… This ‘n’ That (31 January 2014) | Truth2Freedom's Blog

  10. [I'm not sure why there's not a 'reply' link to every comment. But this is a reply to one of James's shorter comments. I've already given a lengthy reply to his lengthy comment, above.]

    James,

    “I’m not sure to what extent that’s directed at me, but let me say that my post was prompted by specific comments made by Craig in his exchange with Helm.”

    It was directed at Remington. If a person has espoused a view in peer-reviewed publication, and also talked about the same view in a podcast, I tend to give priority to how the view has been worked out and defended in the more ‘serious’ context. It’s a lot easier to leave something out in a real-time interview (such as various qualifications or precisifications) then in a manuscript you’ve worked on over time.

    “Helm: “Why can’t they [i.e., human libertarian-free agents], as it were, go off in the other direction [i.e., than the direction God decreed]?””

    They can go off in the other direction, other than what God has decreed. But if they were to do that, it would only be because God had decreed something else. Both the human choice and the divine decree could be otherwise. They can go off the *actual* script, and if they were to do so, then another divine script would have been actual. In no situation do they go off the actual divine script.

  11. [Again, I'm not sure why there's not a 'reply' link to every comment. But this is a reply to Remington's reply above.]

    Remington,

    “James Anderson and John Martin Fischer aren’t addressing the same thing, and I don’t think I implied that they were. But the issues are related at the level of whether foreknowledge is compatible with PAP and, it seemed to me, especially in light of issues you raised.”

    You said that reference to a ‘backtracking conditional’ wasn’t a ‘solution’ to some kind of problem. Here you specify what the problem is: “whether foreknowledge is compatible with PAP.” But it seems to me that Fischer is making precisely my point in his *Oxford Studies in Phil of Rel* chapter on “Molinism”. Fischer there concludes that Molinism “is not relevant to providing an answer to the problem of reconciling God’s foreknowledge with human freedom” (42). As Stump makes clear, and as Flint concedes, denying that accidental necessity is closed under entailment is not a *distinctive* thesis of Molinism. One could deny it and not be a Molinist. Or one could affirm it and be a Molinist (preferring, instead, to deploy Boethianism or Ockhamism to free divine foreknowledge from the fixity of the past).

    Even as Molinism “is not relevant to providing an answer to the problem of reconciling God’s foreknowledge with human freedom,” so Molinism is not relevant to generating the threat to divine infallibility that is the subject of James’s post. Any libertarian theist who believes in infallible divine foreknowledge is going to be saddled with the problem that James brings up. It’s not *Molinist* commitments per se that lead to this difficulty. It’s *any* view that supplies us with the premises of the freewill/foreknowledge dilemma: human libertarian freedom and infallible divine foreknowledge. His post is misleadingly titled “The Fallible God of Molinism”. It should instead be titled “The Fallible God of Anyone Who Hasn’t Yet Solved the Freewill/Foreknowledge Dilemma” :-) In short, James’s post doesn’t bring up any problems that are peculiar to Molinism.

    As I’ve tried to point out, granting the Molinist point of view, there is no problem. But that’s because, as Fischer points out, granting the Molinist point of view presupposes a solution to the fw/f dilemma. I’m not sure this is news.

    “It certainly seems to entail something about the causal power of the agent (unless you deny the fixity of the past, as I mentioned before, or that accidental necessity is closed under entailment)…”

    That’s a pretty huge ‘unless’! Molinists have plenty of options here, as your parenthetical comment makes clear. Notice that nowhere in “Molinism” does Fischer critique Ockhamism (although he does elsewhere, just as many others defend it). That’s a live option for many Molinists when it comes to the fw/f dilemma. Fischer’s point in “Molinism” is to argue that *Molinism* doesn’t provide a solution to the problem, and so it needs to be patched up by reference to another strategy (Boethian or Ockhamistic). Let’s say a Molinist does exactly this. Surely it would then be unreasonable to say they are committed to a “fallible God”.

    “You go on to note that Molinists like Craig and Freddoso deny that accidental necessity is closed under entailment. Fischer addresses this in the paper I was referring to.”

    Sure. And Fischer, following Perszyk, regards Molinism as an “unstable and highly implausible” (36) position on this account, since it is at odds with their acceptance of the Consequence Argument. But how does that show that Molinists are committed to God’s being fallible? They expressly *deny* this. Rather, their denial of closure is implausible given other things they believe (the Consequence Argument). So what?They are not ‘committed’ to the fallibility of God. They are committed to a move that *preserves* the infallibility of God, a move that at worst is dialectically infelicitous when it comes to defending their libertarianism.

    “God may not “need” middle-knowledge to “figure out” the scenario you present, but it seems to me perfectly fine for God to see something like “In a world in which I have foreknowledge of the creature’s doing A…” And since God exists (omnisciently) in every possible world then I think something like that is what God does see.”

    I don’t disagree. As I said, such knowledge would be natural knowledge on the part of God: necessarily, in a world in which God foreknow S’s freely A-ing, S freely As. No problem with that. The Molinist would agree. I’m just bringing out that surely the insertion of foreknowledge into the ‘circumstances’ specified in the antecedent trivializes the whole point of Molinism, rendering it providentially otiose. One would need a very good argument for thus including it in counterfactuals that are the basis of a system of providence.

    “But at any rate it seems obvious to me that foreknowledge belongs to the specification of our circumstance in the actual world. It seems to me that insofar as God exists in every possible world then God’s foreknowledge comprises a part of that world. You don’t think it’s causally relevant, but I guess I’m with Fischer in not seeing it (or maybe I should go re-read Fischer).”

    Well, what you find ‘obvious’ I guess I do not. You might want to read the point made by Neal Tognazzini as Fischer summarizes it on p. 29 fn. 21. Not to flatter myself, but I had already made the point by way of written commentary on the article before I read that footnote. I think it’s the most sensible thing the Molinist can say here. Fischer’s critique presupposes view (1), that *in some possible world* God ‘knows in advance’ what will take place. Notice that on this view, God’s foreknowledge is ‘in’ possible worlds. But the Molinist can easily advert to view (2) and deny the required temporal framework. Rather, “God’s decreeing that a particular world be actual brings into being the temporal framework—a framework that did not exist antecedently… On this picture, God atemporally decrees that a particular possible world come into being, along with its associated spatio-temporal framework.” Thus on this picture, it is very hard to make God’s foreknowledge part of the ‘history’ of the actual world, much less a ‘circumstance’ in which we choose. Fischer admits that “it may be open to a Molinist to pursue this approach.”

    “If Cuthbert can refrain from choosing to buy the iguana in CF1′, then there’s a possible world in which he does.”

    Speaking for myself, I almost never use as the standard of my analysis of one’s published position what they say in some oral conversation for a podcast. There’s just too many corners cut for the sake of a quick presentation. As soon as heard the claim attributed to Craig, it just didn’t matter to me that he said it. Passages where he clarifies the issue in print immediately leapt to mind, and I think we should give such commentary priority when we are critiquing a philosopher’s oral dialogues.

  12. Hi Greg, this is in reply to your request for elaboration. In answer to your question about there being no reply feature for some comments, that’s because I assume James has set some limits on how deep a nested set of comments can go, for if there were no limits, it can get pretty unwieldy.

    Okay, since I know you desire brevity (and clarity), let me try to at least meet the former:

    As I mentioned, we might think, at least on libertarian assumptions, of antecedent causal factors that thwart our having (libertarian) free will as *controlling* our actions. These undermine “ultimate control.” But, another way an agent’s actions can fail to be free is for them to be *out of* his control. I’ll assume we can agree with this:

    • It is not necessary that, if an agent’s actions are *out of* his control, then they are caused (esp. in a way that undermines free will).

    One obvious example, just to make the point at the moment, is this: the problem of luck purportedly shows that our actions are *out of* our controlled, and this doesn’t entail that they are controlled.

    So, I’m thinking of hard facts that could render our future actions *out of* our control. Here’s two examples, that I believe are relevant to the foreknowledge case:

    Now, we agree with this, I think: If S believed that P at t, then this is a hard fact about the past at all ts > t. So, we suppose Reagan believed in 1983 that Grenada was a threat to American interests. Thus, in 2014, this is a hard fact about the past.

    Now let me cite Fischer’s construal of the fixity of the past, FP

    (FP) = For any action Y, agent S, and time t, S can perform Y at t only if there is a possible world with the same “hard” past up to t as the actual world in
    which S does Y at t.

    Let’s suppose that, in part, our freedom is the power to add to the given past. If we have LFW, then our power is to make the world-as-it-has-gone-up-until-now go one way or another. Now consider the Reagan world. My LFW power to stand-at-t-in-2014 is the power to make the world-where-Reagan-believes-that-Grenada-is-a-threat-to-American-interests either include or exclude my standing-at-t-in-2014.

    If we make a minor adjustment and change Reagan’s belief in 1983 to a belief about the future, namely, whether you will respond at t to my comment at t or not (I’m not sure why he’d have such a belief!), then this appears just as hard as the above. But on our add-to-the-past model, you have the power to make it the case that reagan’s belief was *true* or not. And Reagan’s belief isn’t causally relevant to your choice to reply to me at t (where t > 1983) or not.

    But make one more adjustment, keeping the rest of the story the same otherwise. Having been blasted by a cosmic ray from Alpha Centauri, it is now metaphysically impossible for Reagan to ever hold a false belief. On the above story, Reagan’s belief is a hard fact, and so by FP, we hold it fixed. His belief is still *causally* irrelevant to what you do, i.e., his belief doesn’t cause you to do what you do. However, this does show that you don’t have the power to prevent yourself from responding to me.

    My basic point here is that this is a hard fact about the past that isn’t causally relevant to what you do, and yet it nevertheless can undermine (L) free will by putting your doing otherwise action *out of* your control.

    Here’s another way to make the point: Suppose you sit at t3. This is immoral and angers God. He decides to punish you at t1 for sitting at t3. Say, he puts you in jail from t1 to t2. You’re locked up, undergo mental anguish, are deprived of certain goods, you miss work, etc. Being in jail, and the attendant consequences, are hard facts about the past. So, by FP, we hold them fixed when asking if you can refrain from sitting at t3. But, it doesn’t seem that you can so refrain. One reason is that, assuming you do, you make it the case that God punishes you unjustly, and we assume that that’s metaphysically impossible for God to do. However, your being locked up doesn’t *cause* you to sit at t; in fact, on the contrary, your sitting at t3 plausibly causes *God* to have the belief he does and lock you up.

    These are just some examples of hard facts that don’t cause you to do what you do, yet they nevertheless undermine LFW, at least where “doing otherwise” is considered a necessary component.

    • Paul,

      Thanks so much for your concise and illuminating reply. It’s clear to me you’re referring to Fischer’s paper, “The Truth about Freedom: A Reply to Merricks.” That paper is available here:

      http://www.andrewmbailey.com/jmf/Truth_Freedom.pdf

      Merricks wrote a reply to that paper. It is available here:

      https://pages.shanti.virginia.edu/merricks/files/2011/09/Merricks-F-and-F-PR.pdf

      Of course, you already know this.

      “Now, we agree with this, I think: If S believed that P at t, then this is a hard fact about the past at all ts > t. So, we suppose Reagan believed in 1983 that Grenada was a threat to American interests. Thus, in 2014, this is a hard fact about the past.”

      For sure, *we* agree with this, and what you say after this seems to follow handily for me and my house. But Merricks’s explanation of his not being “sure what exactly a ‘hard fact’ is supposed to be” (578) is sophisticated enough that it gives me pause. He seems to give a non-Ockhamistic account of how God’s beliefs can depend on the future, an account that doesn’t avail itself of backwards causation (578-79). I find it really interesting, and I’m not sure why a Molinist couldn’t avail himself of it. Once the dependence relation is explained in the manner Merricks suggests, it’s harder to characterize God’s belief as the kind of ‘hard fact’ that undermines LFW.

      • Hi Greg,

        Actually, I was mostly referring to Fischer & Todd’s response to Merricks’ response. :)

        Two things: (1) I was offering examples of facts that could undermine LFW actions without undermining them by deviantly causing them (and I suspect Merricks could agree with the controlled/out of control distinction).

        I think the latest Fischer/Todd rejoinder is very good (I believe I sent it to you), and the implementation of the pre-punishment case completely avoids Merricks’ “sophisticated reply.” For, as they argue, if his argument works to show that we can do otherwise given God’s foreknowledge, it works to show we can do otherwise if pre-punished, and the latter is very implausible.

      • Paul,

        Could you provide the title of that response for me? I was actually searching for it earlier and couldn’t find it.

  13. I admit that I haven’t read through the comments, but I wanted to mention something that I think really undermines the argument here: namely, Molina himself had a means to thwart this argument.

    You wrote, “In other words, there are possible worlds in which God actualizes C so that S will choose A, but S doesn’t choose A. There are possible worlds in which God’s eternal decree doesn’t come to pass, because libertarian-free agents do otherwise than he had planned.”

    Molina, on the other hand, explicitly held that if we were to choose otherwise, God would have known otherwise. That is, on Molina’s view (and on the view of William Lane Craig), if we were to choose ~A instead of A, then God would have known ~A instead. Molina, in “On Divine Foreknowledge” makes this extremely clear.

    Of course, you could argue that the means by which Molina argues that this is possible is itself mistaken, but the argument is complicated and I’m not going to reproduce it here. My main point is the very core of the argument you’ve given simply misrepresents the position of molinists. Molina (and Craig) held that if a creature were to choose otherwise, then God would know it. Period.

    Thus, on Molina (and Craig’s) view, it would still be impossible for God’s knowledge to fail. It is merely that the content of God’s knowledge would be different.

    My point here is not to defend that position. Rather, all I am saying is that the argument presented here fails to actually interact with the view of Molina himself and prominent molinists like Craig.

    • Thanks, J.W. Maybe you should have read through the comments first. :)

      My argument focuses on God’s decree, which is distinct from his foreknowledge. See my interaction with Greg Welty in the comments.

  14. Remington,

    Fischer and Todd, “The Truth About Foreknowledge and Freedom,” *Faith and Philosophy*, Vol. 30 No. 3 July 2013, 286–301.

  15. James,

    My comment about your terminology was just a suggestion, based on my experience. You didn’t write the post for philosophy grads or profs, but you sure wrote it *in response to* a couple of philosophy grads and profs. But maybe Craig and Helm are less accomplished than I thought! :-)

    You say that “The notion of an infallible decree is perfectly clear and understandable,” but as I’ll try to argue, I actually think it’s the ambiguity of the term that is contributing to our impasse.

    It’s great that you “reject the reduction of ‘God decrees’ to ‘God causes’.” But that’s the Molinist position as well. On the Molinist view God decrees everything, including that S will choose A. (He doesn’t decree, of course, that S *would* choose A in C. Rather, taking that subjunctive knowledge into account, he decrees that the choice will in fact happen. And if he refrains from that decree, the choice will not in fact happen.)

    You emphasize that God “decrees C *in order that* S will choose A.” The Molinist holds to this as well (as long as you don’t mean by ‘in order that’ something like ‘as the cause of’). In fact, I noted this detailed, contentful understanding of the decree when I said: “No possible world can be actualized without its having a C *that is caused by God for the purpose of realizing the consequents of the counterfactuals that are true in that world*. So there isn’t even the possibility of a mismatch.”

    You go on to say, “God decrees S’s choosing A in C (and *because of* C).” Here the Molinist would demur. It isn’t a part of the Molinist view that agents choose what they choose because of their circumstances, as if their circumstances caused them to choose a particular thing. But nevertheless, Molinists will share your skepticism that “C can be abstracted from God’s reasons for actualizing C.” They will think the very reason that God actualizes C is so that “S choosing A” will come to pass.

    It’s still not clear to me how any of this harms the infallibility of God’s decree at all, on the Molinist system. Let’s say that God decrees S’s choosing A in C. In decreeing this, God decrees his own causal activity, which is God’s causing C. (As you put it, “Because of his decree, God (strongly, causally) actualizes C, and we agree that C is causally relevant.”)

    Because God’s decree is posterior to his middle-knowledge, it takes account of middle-knowledge. A necessary condition for God’s decreeing S’s choosing A in C is the truth that S would choose A in C. If that wasn’t true, then God wouldn’t know it, and not knowing it, he wouldn’t decree S’s choosing A in C.

    It follows that in no possible world does God decree S’s choosing A in C, and yet S refrains from choosing A in C. In no possible world is there a mismatch between God’s decree, and what S ends up doing. (Yes, there are possible worlds in which S does otherwise in C, but those wouldn’t be worlds with the same decree, in which God causes C as a means of getting S to choose A. So they aren’t worlds that frustrate the same decree, or any decree, for that matter.) So there is no case for fallibility here. It is exactly parallel to the infallibility of God’s foreknowledge: in no possible world does God foreknow S’s choosing A in C, and S refrains from choosing A in C. It’s pretty clear why foreknowledge would have infallibility if the decree does, since the foreknowledge is posterior to the decree, and grounded in it.

    Perhaps if you can clearly describe, on Molinist premises, a possible world in which God decrees S’s choosing A in C, and yet S refrains from choosing A in C, you would have a case for the fallibility of the decree. I haven’t yet seen you show why such a world would arise on the Molinist scheme.

    And now we’re in a position to evaluate your closing dilemma:

    “The whole issue here comes to this: Does God’s decree *guarantee* that S chooses A? If so, then Molinism is really just a form of determinism: non-causal determinism, but determinism nonetheless. If not, then God’s decree must be fallible.”

    You’re clear here that you’re not talking about *causal* determinism. After all, it is no part of the Molinist scheme that God causes S’s choosing of A. Rather, God just causes C, and C doesn’t causally determine S to choose. And I assume you’re not talking about *logical* determinism either. So what is the nature of this ‘determinism,’ a ‘determinism’ that doesn’t cause agents to do anything? As far as you’ve argued, this alleged feature of Molinism doesn’t impact human freedom at all, and invoking such Molinist ‘determinism’ is like a man who keeps a toothless old circus lion around the house in order to experience the thrills of the jungle. (Apologies to Allan Bloom.) It’s not much of a threat.

    Now, if you were to start talking about gunslingers then it might be easier for me to see that Molinism is indistinguishable from efficient causation. :-)

    I personally think that talk of an ‘infallible decree’ is unhelpful because it splices together epistemological and metaphysical categories. What is an ‘infallible decree’? Is it a decree that occurs in all possible worlds? No, the decree is contingent. Is it a decree which is such that, given there is the decree, events couldn’t fail to match it? Sure. But that’s compatible with the contingency of the decree. The decree doesn’t exist in all possible worlds, but in no possible world in which there is the decree, do events fail to match it. Is it a decree that causally determines circumstances? Molinists are fine with that. Is it a decree that causally determines human choices? Why should Molinists accept this?

    I’ll admit the terminology does have precedence in Reformed thought. Calvin apparently uses it in his commentary on Isaiah 2:3 – “*the law will go forth out of Zion, and the word of the Lord from Jerusalem*; for this is an infallible decree of God, which no diversity or change of events will make void.” But Molinists will agree: no change of events will make void God’s decree.

    “I realize that on the Molinist view you’re defending, if S were to choose otherwise, God would have decreed otherwise, because the relevant counterfactuals would be different. So it appears that the counterfactuals are ultimately up to S, and God’s decree hangs on the counterfactuals. God’s decree (at least in part) is ultimately up to S. As I said before, on this view God’s will is bent to the will of his creatures.”

    I really think this continues to mischaracterize the Molinist view. Yes, “the counterfactuals are ultimately up to S.” But God’s decree ‘hangs on’ the counterfactuals only in the sense that God must decree in a way that is *consistent with* the counterfactuals. They are a necessary but not sufficient condition for any particular decree. It doesn’t follow from this that the counterfactuals tell God what to do. God is free to *not* decree that S chooses A in C. He is free to *not* place S in C. He is free to *not* create S at all. The counterfactuals don’t require him to do any of these things. The counterfactuals are available to him as a resource but they don’t bend his will.

    “I’m not sure why you’re missing the point here. According to the Molinist, it is ultimately up to S whether or not S chooses A. But then it is ultimately up to S whether or not God has decreed an S-chooses-A world.”

    I can only ‘miss the point’ if there’s a point to be missed, so touche! ;-) But again, it just seems to me that this is very misleading. It is flat-out false to say that, on Molinism, “it is ultimately up to S whether or not God has decreed an S-chooses-A world.” To be sure, it is up to S whether “if S were in C, S would freely choose A” is true. But granting its truth, it is *not* up to S whether or not God creates S, or creates S in C. So how can it be possibly ‘up to S’ whether or not “God has decreed an S-chooses-A world”? The counterfactual could be true, and God decrees such a world, or doesn’t decree such a world. It’s a strange sense of ‘up to’ that is compatible with that.

    • Greg,

      Thank you for your patience with me. And I’m sorry I haven’t been able to get back to this until today. I can’t continue this indefinitely, so I’m happy to let you have the last word. I’m just going to make two points in response and leave it at that.

      You go on to say, “God decrees S’s choosing A in C (and *because of* C).” Here the Molinist would demur. It isn’t a part of the Molinist view that agents choose what they choose because of their circumstances, as if their circumstances caused them to choose a particular thing.

      See, this is part of my objection to Molinism. It’s why I say Molinists want to have their cake and eat it: a deterministic indeterminism. Sure, I understand that on the Molinist view the circumstances don’t cause the agent to choose what he chooses. But the circumstances have to be somehow relevant to why S chooses A, otherwise the circumstances are entirely redundant in the counterfactuals. One might as well just throw Pythagoras’s theorem into the antecedents of the counterfactuals as well!

      So how do the circumstances relate to S’s choice? Do they make it probable that S will choose A — some very high degree of probability short of determination? But clearly that’s not sufficient to underwrite an infallible divine decree. So what guarantees that S will do A in C when the time comes? It’s not good enough to say, “The truths of the counterfactuals,” because that begs the very question at hand. Either S has the power to refrain from A in C given that God has already decreed that S do A in C or he hasn’t. If he hasn’t, how is this not a species of determinism?

      Do you see what I mean when I say that Molinists want determination without determinism? If you don’t, I’m not sure how we can move forward on this.

      The second point pertains to the last paragraph of your reply. I didn’t claim that it is entirely up to S whether or not God has decreed an S-chooses-A world. Of course, it is up to God whether to create S, etc. My point is that if the counterfactuals about S are up to S, and God’s decree is conditioned on the available counterfactuals (which is part and parcel of Molinism), then what God is in a position to decree is partly up to S. What God can decree is constrained by S (and by every other libertarian-free possible agent). It is in that sense that the will of God has to be ‘bent’ to the wills of his creatures.

      BTW, I have to disagree that talk of an “infallible decree” is unhelpful. It has more than precedence in Reformed thought and Calvin. The language (albeit not the exact phrase) is right there in the WCF and LBCF: “Although in relation to the foreknowledge and decree of God, the first cause, all things come to pass immutably and infallibly…”

  16. Hi James,

    Thanks everyone for the great discussion! This has been quite interesting to read.

    I pretty much agree with what you said in response to Greg a while back, that after Greg’s comments, we are left with two choices:

    “(1) deny divine infallibility or (2) adopt a very artificial and unbiblical understanding of divine infallibility”

    It doesn’t seem clear to me why creatures effecting God’s decrees counter factually entails an unbiblical understanding of divine infallibility, perhaps you could clarify.

    This choice also seems to be a pretty big concession with respect to your blog post. Your original conclusion that Molinists must reject infallibility carries a very large intellectual cost since Molinists would be uncomfortable rejecting infallibility. However, I doubt many Molinists would be shocked that their doctrine entails that creatures effect God’s divine decree counterfactually.

    Looking forward to your responses if you find the time!

    • A pretty big concession? I’m not so sure. :)

      I still maintain that Molinists can only save divine infallibility by artificially excluding God’s decree from the circumstances in which agents make their libertarian-free choices. They have to maintain that the intramundane circumstances don’t determine that S will choose A, but God’s decreeing those circumstances (in light of the counterfactuals) somehow does determine that S will choose A (on pain of divine fallibility).

  17. BTW, I just want to say something that might clear the air when it comes to this whole dialogue about James’s blog post on “The Fallible God of Molinism”. I think I’m segregating things in my mind that I think *should* be segregated, but it may be that there’s a difference of opinion among us about that. Specifically, Molinists hold that:

    (ID): In no possible world does God decree that S A’s in C, and S doesn’t A in C.

    Someone may wonder how Molinists can affirm Infallible Decree like this. Can’t the agent do otherwise? Here the Molinist adverts to another principle:

    (CT): If S were to not-A in C, then another counterfactual would have been true, and God would have incorporated that counterfactual into his decree, such that God would have decreed that S not-A’s in C.

    Molinists might support this Counterfactual Truth by saying that the agent has the power to do something, such that if he had done it, then God would have done something different, different things would have been true, etc.

    Now, it might be that although CT and its support strikes the Molinist as very plausible, as a matter of fact they’re all messed up here, metaphysically speaking. There are clever arguments from pre-punishment that provide counterexamples to CT and/or its support.

    What would follow from that difficulty is *not* that “the God of Molinism is fallible” or that “God is fallible on the Molinist view” or that “according to the Molinist view there are possible worlds in which God’s plans fail”. No, they are clearly committed to ID. That’s their view! Rather, at worst, they’ve given an inadequate metaphysical defense of how ID could be true.

    When an Arminian comes to us Calvinists and says, “According to the Calvinist view, people are robots and God is a moral monster,” don’t you think that’s a misrepresentation? Isn’t it better for them to say, “Calvinists say that humans have free agency and that God is holy, but I don’t see how they give a consistent defense of those commitments”?

    Here’s a quicker analogy: “Molinists *say* that God knows these counterfactuals, but we all know they haven’t solved the grounding objection. Without grounds, there aren’t any truths to know. Therefore, Molinists believe in an ignorant God. The God of Molinism is ignorant.” Surely something has gone awry here.

    • Greg,

      You make a good point here. If I had the time over again, I would make a different argument than the one I made.

  18. Adam Omelianchuk

    I think Greg Welty said everything I wanted to say, only much better. Molinists do not assume that God’s decree is unconditioned; rather it is conditioned by his middle knowledge, and the outcome of that degree in conjunction with his middle knowledge will most certainly come to pass. I will give you some props though as you seem to draw attention to a distinction between a weak and a strong sense of infallibility. On Molinism if God decrees x, then x will come to pass. On Calvinism, if God decrees x, then x must come to pass. Thus, there is the infallibility of certainty, and the infallibility of necessity that makes all the difference, and the former is contained the latter, but not vice versa.

    • Thanks for the comment, Adam.

      My problem with this is that “infallibility of certainty” is an epistemological category rather than a metaphysical one. You might as well say that the Simple Foreknowledge view can affirm “infallibility of certainty” because it affirms exhaustive divine foreknowledge, in which case how does Molinism have any stronger a view of divine infallibility? The issue I’m focusing on here is a metaphysical one rather than an epistemological one: it’s about whether God’s decreeing X determines that X comes to pass.

      • Adam Omelianchuk

        I see what you are saying, James. I think the point I am trying to make is that the decree without middle knowledge is fallible (which is a problem for simple foreknowledge), but in conjunction with middle knowledge, it is not. The divine decree is therefore conditioned by God’s middle knowledge, and whatever God plans will come to pass. We might even say that decree-failing worlds are possible, but in light of God’s middle knowledge, they are not feasible for God to create given that God is wise and rational.

        In any event, this is an interesting challenge to Molinism, and I wonder if you worked it out with the proper qualifications, you might have reductio argument in your toolbox.

  19. Pingback: Is Molina’s God Fallible? A response to James N. Anderson | J.W. Wartick -"Always Have a Reason"

  20. Is there a difference between a possible world where God makes himself fail and a possible world where he makes a rock too heavy for him to lift? It seems similar somehow. The modality of the (im)possible is confusing here, and is your point that Molinism makes God in effect makes a stone he cannot lift in a possible world? Does that matter as long as it is never real?