ON BEHALF OF MAVERICK MOLINISM

Jonathan L. Kvanvig

In clarifying and defending Molinism, Thomas Flint argues against a position he terms Maverick Molinism. This version of Molinism maintains that, though counterfactuals of freedom have their truth-value logically prior to God’s acts of will, God could have so acted that these counterfactuals would have had a different truth value from that which they actually have. Flint believes this position is flawed, and presents an argument for rejecting it. I argue that Flint’s argument against Maverick Molinism is flawed, and suggest that the Maverick has a position with advantages over more traditional versions of Molinism.

Molinism is constructed to preserve a strong view of God’s sovereignty as well as a libertarian conception of human freedom. It does so by maintaining that human beings are free in the libertarian sense, but that there are counterfactuals of freedom that describe what a person would do (freely) in given circumstances. Other views that deny that human beings are free in the libertarian sense may also affirm that there exist counterfactuals of freedom, but they will not hold that the kind of freedom in question is the kind libertarians cherish. So we might describe the Molinist position as accounting for God’s sovereign control over the course of affairs in virtue of his knowledge of which counterfactuals of libertarian freedom are true and which are false.

Beyond this minimal commitment of any Molinist are various embellishments. Some embellishments are dictated by logic. For example, this characterization leaves open the question of whether counterfactuals of libertarian freedom are true in virtue of God’s choices, i.e., whether their truth value is divinely postvolitional. There is a straightforward argument against the view that counterfactuals of freedom have their truth value in virtue of God’s acts of will. For if God makes it true that Joe would kick his dog in certain circumstances, then Joe does not have the power to do other than to kick his dog in those circumstances. And if he does not have this power, then the counterfactual in question is not a counterfactual of libertarian freedom. If this argument is sound, as I will grant here, it is a further commitment of Molinism to hold that counterfactuals of libertarian freedom are divinely prevolitional–that is, their truth value does not obtain in virtue of any act of God’s will.

Though I am granting the soundness of the above argument, I want to emphasize its significance a bit. Note that the concept of prevolitional truth is not understood here in terms of what God has control over. To say that a truth is prevolitional is only to deny that its truth value is the direct result of God’s choices. To ask whether God has power or control over these truth-values is to ask a question that must be answered by some inference from the concept of prevolitional truth and other theistic commitments about the nature of God.

This last point leads to the heart of what I want to argue, for many standard Molinists affirm not only that counterfactuals of libertarian freedom are prevolitional, but that they are also beyond God’s control. They hold, that is, that there is nothing God could do which would have resulted in a counterfactual of libertarian freedom having a different truth value from what it actually has. Consider, for example, Thomas P. Flint’s account:

For, though God’s middle knowledge will as a matter of fact remain the same no matter what he does, this isn’t a necessary truth. To put the same point more precisely: any element of God’s middle knowledge will be true no matter what God does, but it is entirely possible for that element to be false and yet for God to choose to act.^[i]

Flint claims that though it is possible for such counterfactuals to be false and for God to choose to act (this conjunction is possible in part because counterfactuals of libertarian freedom are contingent), the truth-value of such counterfactuals is thoroughly insulated from God’s activity. Flint explicitly puts his point in future-tense terms: any counterfactual of freedom will be true no matter what God does. For this claim to be true, all that needs to be true is that there is nothing God could do to change the truth value of any counterfactual of freedom. As we shall see below, however, I think Flint intends a different point, for the position he wishes to argue against doesn’t claim that God can change the truth value of counterfactuals of freedom, but rather that God has counterfactual power over them. So what Flint is committed to is the claim that, for any proposition x which is a true counterfactual of libertarian freedom, there is no choice or action within God’s power to perform such that, were he to perform it, x would be false.^[ii]

Not only do typical Molinists endorse this claim that there is nothing God could do to that would have made such counterfactuals false, they identify it with the earlier claim that counterfactuals of libertarian freedom are prevolitional. Flint’s language is especially instructive in this regard. The context in which Flint discusses this issue regards a certain kind of counterfactual, the particular kind being unimportant, regarding which

some who are otherwise favorably disposed toward Molinism might be tempted to challenge the alleged prevolitional status of such truths. That is, they may want to say that a Molinist can hold that, whichever truth-value [such] a counterfactual . . . has, there nevertheless were things God had the power to do such that, had he done them, that counterfactual would have had the opposite truth-value.^[iii]

Flint labels the position in question “Maverick Molinism,” and argues that it is untenable.^[iv] He thereby commits himself not only to the view cited above that there is nothing God can do to change the truth-value of counterfactuals of freedom, but that he doesn’t have counterfactual power over them either. Maverick Molinists hold no brief about the first point, but disagree with him about the second.

I don’t think Flint’s argument is successful. I will show that the argument he gives is flawed, and that his position imposes a severe epistemic burden on Molinism. This burden is damaging, because Molinism enjoys a privileged status in that it resolves the apparent conflict between divine sovereignty and human freedom by appealing to counterfactual claims that are endorsed regularly by common folk. That is, it relies on the kind of judgments that people ordinarily make or presuppose in everyday life about what free individuals would do in certain circumstances.^[v] If it incurred no epistemic burdens along the way, the Molinist could be content with rebutting criticisms of his view, relying on the intuitive advantages of the position to provide sufficient warrant for belief. Flint’s position loses this intuitive advantage because it endorses problematic claims along the way, so his position leaves the Molinist in a position of needing to do more than rebut criticisms in order to establish the tenability of the view. The lesson will be that the most attractive versions of Molinism will refuse to endorse Flint’s restrictions on Molinist family membership.

I begin with Flint’s argument against Maverick Molinism, which proceeds as follows. First, he argues that even if a counterfactual such as A ® B is postvolitionally true, so that there are propositions C and D within God’s power to bring about such that C ® (A ® B) and D ® ~(A ® B), these latter two truths must be prevolitional on pain of an infinite regress of things that God would have to do to make A ® B true. Call this step the vicious regress premise.

Next, Flint claims that the Law of Conditional Excluded Middle (LCEM) should be endorsed by all Molinists, including the Maverick. The context of the discussion leaves open that Flint only wishes to endorse a limited version of that law, applied to counterfactuals of world actualization, where, using the formulas of the previous paragraph, B is a complete possible world, and A, C, and D are states of affairs maximal with respect to what God can strongly actualize. I will initially evaluate the argument applied to counterfactuals of freedom generally, and then to the more restricted class of world actualization counterfactuals, so I will begin by attributing to Flint an unrestricted endorsement of LCEM.^[vi] Call this step the LCEM hypothesis.

Finally, Flint presents the reductio step. For counterfactual of world actualization A ® B, consider the expanded counterfactuals C ® (A ® B) and C ® ~(A ® B). By the LCEM hypothesis, one of these two must be true, regardless of the value for C. So suppose C = A. Then either A ® (A ® B) or A ® ~(A ® B). Since the first is logically equivalent to A ® B, the two could not differ regarding their status as prevolitional or postvolitional. But the vicious regress premise requires that they do so differ. Hence, the only option is that A ® ~(A ® B). But this claim can’t be true either. For ~(A ® B) entails (A ® ~B) (by LCEM), and so A ® ~(A ® B) is logically equivalent to A ® (A ® ~B), which is logically equivalent to A ® ~B. This latter claim must therefore be prevolitional, and since it is equivalent to ~(A ® B), this latter claim must be prevolitional as well. So, both A ® B and ~(A ® B) have to be prevolitional, but that is contrary to the Maverick’s hypothesis. Hence Maverick Molinism is not a legitimate form of Molinism at all.

There are two major flaws in this argument, as I see it. First, the LCEM step is suspect. Flint never argues for the claim that Molinists ought to endorse such a principle, and I can’t see why the reasons for thinking that LCEM is false don’t apply to the counterfactuals the Molinist endorses. I must admit that I am unpersuaded by the purported counterexamples to LCEM, such as the relationship between the two claims Reagan and Gorbachev are compatriots and Reagan was born in Russia. I would not endorse either the claim that the first counterfactually implies the second, or that it implies its negation, but that refusal might arise from epistemological uncertainty rather than the failure of LCEM.^[vii]

It may be that Molinists don’t need to accept LCEM, but only a restricted version of it applying to counterfactuals of creaturely freedom. Yet, if Molinists use such examples as the one above as part of the explanation of the failure of LCEM, they should also find counterexamples of the same sort regarding counterfactuals of creaturely freedom. We have as much reason to refuse to endorse either of

If I were offered both Blue Bell ice cream and Godiva chocolates, but could only choose one of the two, I freely would choose the ice cream

and

If I were offered both Blue Bell ice cream and Godiva chocolates, but could only choose one of the two, I freely would not choose the ice cream

as we do with the two claims regarding Gorbachev and Reagan.

Note also that there is no rescue squad encountered by insisting that the antecedent of such a counterfactual of creaturely freedom be maximal, fully specific regarding the entirety of the circumstances in which I am situated. For even if the antecedent contains all feature of the history of a possible world up to the moment of free choice in question, there is no reason to be found for thinking that the world in which I take the ice cream is closer to the actual world than the world in which I do not take the ice cream.

It is not hard to see why a Molinist would find (a suitable refinement of) LCEM attractive, however. For the Molinist wishes to retain a strong view of sovereignty and if there are violations of LCEM regarding counterfactuals of freedom, it looks as if there would be a kind of uncertainty in the mind of God regarding what would happen were he to actualize the antecedent of a false counterfactual of freedom. For, according to the Molinist, it is in virtue of God’s knowledge of counterfactuals of freedom that he knows which world would be actual in virtue of his creative acts of will, and his sovereign control over the course of affairs depends on this middle knowledge. If, however, God’s middle knowledge is “gappy,” if it leaves undetermined which of a variety of alternatives would occur in certain circumstances, then God’s sovereign control over the course of history is threatened. So it is not surprising to find a Molinist such as Flint endorsing LCEM.

An explanation is not a justification, however. What is needed is an argument why the law is true in spite of the weight of the evidence against it. Perhaps such an argument can be found by highlighting the relationship between philosophical work on vagueness and LCEM. A standard view of vagueness is metaphysical–that vagueness is a part of the world, and not merely a feature of our language or a limitation on our epistemic capacities. The metaphysical view of vagueness has an intuitive advantage over the two alternatives just mentioned, but careful investigation of the issues may require endorsing an epistemic view of vagueness,^[viii] with a concomitant endorsement of the counterintuitive idea that there really is a precise number of grains of sand needed to form a heap. Along the same lines, it may be counterintuitive to have to claim that one of the two claims

If I were offered both Blue Bell ice cream and Godiva chocolates, but could only choose one of the two, I would choose the ice cream

and

If I were offered both Blue Bell ice cream and Godiva chocolates, but could only choose one of the two, I would not choose the ice cream

is true, but such counterintuitive results may be forced on us. Perhaps, that is, the similarity relation between worlds really is precise and is such that no two distinct worlds can be equally similar to a third. I won’t press this argument here, but I’m inclined to think that the only way to provide an adequate motivation for the idea that two distinct worlds can be equally similar to a third is by treating similarity as a vague (and perhaps contextual) relation. If I’m right, then if the epistemic view of vagueness is the only defensible position, LCEM may be true after all.

Suppose, however, that no such argument is successful; how should we think about Molinism in such a case? Perhaps the conclusion to draw is that the plausibility of Molinism is tied to LCEM, as suggested above, but there may be another way out. Perhaps God would only create compatible with LCEM, so that even though there are possible violations of LCEM, there are no actual violations of it, and hence God’s sovereignty is preserved. Such a view requires endorsing Einstein’s conception of God as one who doesn’t throw dice with his creation (though it is interesting to note that it doesn’t require holding that there are no irreducibly probabilistic events–these can be tolerated as long as there are counterfactuals of indeterminacy analogous to counterfactuals of freedom). On the negative side, however, such a view still faces the modal problem of allowing the possibility that God’s sovereignty is compromised by human freedom.

The point of this discussion is to highlight the centrality and controversiality of the LCEM step of Flint’s argument. It is controversial in part because Flint’s discussion seems to indicate an acceptance of the law driven by Molinist requirements instead of some decent argument for it, and in part because the best arguments that I know of for endorsing the law have precisely the same counterintuitive aspects involved in the epistemic view of vagueness. Since I find the epistemic view of vagueness to be superior to known alternatives, I am not as bothered by the analogy as will be those less enamored with that view. Equally important, the step is central to Flint’s argument, for without it, the reductio step can’t be developed, and no decisive reason will have been given for rejecting Maverick Molinism.

Even if the reductio step fails, however, there still may be a residual worry regarding Maverick Molinism. For even without the reductio, it is hard to see where the regress should be stopped, and without some account of why it should be stopped at some particular point, perhaps the best conclusion to draw is that it should never have been started in the first place. This conclusion is just a denial of the Maverick’s position, so even if we politely decline the invitation to endorse LCEM, there may be still be a worry that the Maverick cannot give a non-arbitrary account of where to draw the line.

The worry depends on the regress in question being vicious, and it is not obvious that it is. All Flint shows is that, if embraced, the regress requires that God do an infinite number of things in order to exercise control over the truth value of any counterfactual of freedom. But why think that God can’t perform such an infinite task? It is far from clear that infinite tasks are themselves impossible,^[ix] and if they are not, then we have a prima facie case that God, an omnipotent being, could perform one.

Yet, I don’t wish to rescue the Maverick by appealing to the possibility of infinite tasks (in part because of disanalogies between the two cases), so I will grant Flint the point that the regress is vicious. Still there is a second major problem with the argument, arising from the distinction between what is in God’s control and what is postvolitional for God.

As I granted above, it is a logical requirement on Molinism that counterfactuals of freedom be prevolitional. Hence if we define a version of Molinism which counts some counterfactuals of freedom as postvolitional, the resulting position will be inconsistent. So, pretty clearly, Maverick Molinism better not affirm that counterfactuals of freedom are postvolitional. Yet, it looks as if that is precisely how Flint interprets the view. In fairness to Flint, things are not quite that simple, since the context of discussion concerns only those counterfactuals of world actualization which have a specification of a complete possible world as their consequent and a largest state of affairs that God can strongly actualize as antecedent. But it’s not clear why someone attracted to the Maverick position need view such counterfactuals as different in kind from ordinary counterfactuals of freedom. And if we do not treat them differently, then we must say of this special kind of counterfactual that it too is divinely prevolitional. What the Maverick denies is not the prevolitional status of such truths, but only the claim that their truth value is outside God’s control. The crucial point maintained by the Maverick should be that a counterfactual’s being divinely prevolitional is not the same as a counterfactual’s being outside God’s control.

Flint conflates these two claims, as the passage quoted earlier shows. In the first sentence (“[S]ome who are otherwise favorably disposed toward Molinism might be tempted to challenge the alleged prevolitional status of such truths”), Flint makes clear that the position he is discussing challenges the prevolitional status of certain counterfactuals, while the second sentence (“That is, they may want to say that a Molinist can hold that, whichever truth-value [such] a counterfactual . . . has, there nevertheless were things God had the power to do such that, had he done them, that counterfactual would have had the opposite truth-value”) identifies this claim with the claim that there are things God could have done that would have resulted in those counterfactuals having a different truth value. This latter claim is clearly a claim about which truth are within God’s control, not a claim about which truths are prevolitional.

A Maverick need not conflate these two claims. It can be granted that the first claim entails the second, that is, that if a truth is divinely postvolitional, then it’s truth value is within God’s control. The second does not entail the first, however. That is, it could be true that certain claims are true independently of God’s acts of will and yet there be acts or choices that God could have performed which would have resulted in those claims being false.

The argument that the power claim does not entail the volition claim is simple. There are lots of truths that are true independently of our wills, and yet we have or had the power to act in such a way that those truths would not have been true. For example, snow is found at certain precise spatiotemporal points north of where I live, but no act of will on my part caused the snow to be there. Yet, there are things I could do or could have done which would have prevented the presence of snow at some such particular spot. (I’m not claiming I can control the weather, but I can control whether snow is on a particular piece of ground, still owning the shovel so necessary to endure winters in northern Indiana.) So the power claim simply does not entail the volitional claim.

Of course, there could be something about God’s will that makes claim about what he has power over entail claims about his volitions. But it is instructive that Flint nowhere gives such an argument. Molinism’s opponents do have such an argument, for they can maintain that all contingent truths depend directly on the will of God. Molinists cannot give such an argument, however, for it is constitutive of their view that some contingent truths are true independently of the will of God. Since the Molinist grants this point already, and since the general entailment claim about the relationship between power and volition is false, I see no reason to think a Molinist is forced to embrace the entailment claim when it is restricted to divine scope.

Moreover, it would be paradoxical indeed for there to be such an argument. For take any contingently false counterfactual A ® B. Even if A ® B is false, it is still possible that there is some contingent proposition C which is such that if it were true, A ® B would be true.^[x] For example, it is contingently true that if I were offered a choice between asparagus and beans for dinner, I’d not choose beans. Things could have happened to alter my preferences, however. The proliferation of special pests that blunted the growth of beans but allowed asparagus to flourish would have led to much more asparagus consumption in my youth, and to a preference for the unusual taste of beans over that of asparagus (I’m assuming here that the core of the explanation of my preference for asparagus over green beans is the rarity of the former over the latter in my diet). So, suppose that this is an example in which C ® (A ® B) is true, even though the counterfactual embedded in the consequent is false. For that embedded counterfactual to be within God’s control, we need only assume that God could make C true (could strongly actualize that state of affairs), and it would be perplexing indeed to find someone denying that claim. So, if God could actualize C, then God has power over the truth value of A ® B, even if we assume with Molinism that the truth value of A ® B is prevolitional.

One might think that the example above is convincing only because the antecedent of the conditional is not maximal, i.e., that its antecedent does not include the entire history of a possible world up to the preference for asparagus. Maximal specificity is no more help here, however, than it was earlier. What is at issue regarding whether a particular counterfactual is true is how close, or similar, possible worlds are to each other. If there is a world God could actualize such that the “A ® B”-worlds are farther away from the world thus actualized than some “~(A ® B)”-worlds, then there is something God could do that would have made that counterfactual false. Maximal specificity of antecedents is simply a red herring here; the argument turns only on the critical fact that the similarity relation between worlds allows four worlds to be such that the first is closer than the second to the third, while the second is closer than the first to the fourth.

So this version of (Maverick) Molinism has an important advantage over Flint-style Molinism. For Flint-style Molinism is saddled with having to impose additional restrictions on God’s omnipotence in the face of the above, if its defenders insist on maintaining that God has no control over the truth value of counterfactuals of libertarian freedom. Flint-style Molinists will have to hold in cases like that above that God cannot in fact actualize C, even though C is contingent and not obviously of any of the usual sorts of contingent truths that defenders of omnipotence allow to be irrelevant to the claim of omnipotence.^[xi] If we talk about entire worlds rather than sets of circumstances, Flint-style Molinism will have to infer the unactualizability of world C from the fact that the truth value of A ® B is prevolitional, an inference threatened by a confusion between prevolitionality and lack of control.

So there is an alternative position to the one Flint attacks that doesn’t construe counterfactuals as postvolitional, but still claims that they are within God’s control. What happens when we try to apply Flint’s argument to this position? In a word, it collapses. In particular, the vicious regress step must be abandoned. That step, recall, claimed that if A ® B were postvolitional, then C ® (A ® B) had better be prevolitional, for otherwise a regress is launched that would require God to do an infinite number of things in order to fix the truth value of A ® B. But if A ® B is conceived of as prevolitional, as well as any proposition embedding A ® B as the consequent of a longer counterfactual, then the truth-value of such propositions doesn’t require any particular divine activity in order to obtain. There are, of course, things God could do that would make their truth-value other than what it is, since the truth-value of such claims are within the reach of God’s power, but no regress of actions is launched since no action by God is required to fix the truth-value in the first place.

One might still worry that there is a regress problem in the neighborhood, in the following way. Suppose again that A ® B is false and that C ® (A ® B) is true. Isn’t there the regress worry that if God chooses to exercise his power over A ® B, that there will be an infinite series of things he’ll have to do to exercise such power? The idea here, I take it is that there might be a further claim D ® (C ® ~(A ® B)), which is also true, and so if God choose to make A ® B true, he’ll not only have to actualize C, but also actualize ~D. The last point contains the mistake: he’ll only have to fail to actualize D, and that requires no special action on his part, so no infinite regress is launched.

Throughout all of this there remains the residual worry that a lot of contingent truths are turning out to be true independently of God’s volition, and that is, to my mind, the most serious difficulty with this entire discussion. I want to reiterate, however, that this aspect is a central tenet of any Molinist position, and the point of this paper is not to take on objections to Molinism per se, but rather to investigate a particular kind of family squabble within the Molinist community. It is part of official Molinist doctrine that many contingent truths are true independently of God’s choices, so in the context of the present dispute, one cannot argue against Maverick Molinism on grounds that it implies that there is a large number of such truths.

Flint began his discussion of Maverick Molinism by suggesting that not even charity should allow us to recognize this position as a legitimate version of Molinism. Instead, Flint claims, it should be viewed as something more akin to heresy. Perhaps the correct picture is that it is only here that Molinism and orthodoxy form a philosophically happy union.

Texas A&M University

Endnotes

---

[i]. Thomas P. Flint, Divine Providence: The Molinist Account, (Ithaca, NY: Cornell University Press, 1998), p. 175.

[ii]. ibid.

[iii]. ibid. p. 65.

[iv]. ibid, p. 65ff.

[v]. See Flint, Chapter 1, for a more complete discussion of the intuitive advantages for Molinism.

[vi]. Flint suggests such an unrestricted endorsement in Chapter two, where he queries whether there is any good reason for a Molinist to prefer definitions that do not involve commitment to LCEM to those that do. See, especially, pp. 49-50.

[vii]. The standard route to giving up LCEM appeals not just to purported counterexamples, but to the standard semantics for such. One reason for being dissatisfied with that defense is that it confuses testing certain counterfactuals with making them true, i.e., if my son proposes that I would eat a particular piece of chocolate were it put in front of me, we should be able to test his claim by putting that piece of chocolate in front of me. But, alas, we might fail, for the counterfactual could be false had we not run the experiment, but made true because of my actions when the chocolate is put in front of me. A good theory of counterfactuals should not confuse testing and making true in this way.

[viii]. For an argument to this effect, see Timothy Williamson, Vagueness, (Oxford: Oxford University Press, 1996).

[ix]. See Mark Sainsbury, Paradoxes, (Cambridge: Cambridge University Press, 1995), especially chapter 1 for a discussion of the issues surrounding the possibility of infinite tasks.

[x]. There is a further generalization on this point that I will ignore in the text: even if there should be no value for C of the sort required, there still might be some finite set of ordered propositions p₁...p_n relative to the false proposition A ® B such that for l£m<n, where A ® B = p_n, some claim of the form p₁®[...p_l...®(p_m®...®p_n)] is true. That is, there might be a finite string of propositions connected by counterfactual arrows to A ® B such that the parenthetical groupings begin with the rightmost connectives and work back to the leftmost, so the first counterfactual arrow connects p₁ with everything after it, and similarly for each p_i, where the resulting proposition is true even though A ® B is false.

[xi]. E.g., truths about the past, limiting truths about the power to sin or the capacity to make something too big to move, egocentric truths about things that only a particular person has the power to do, etc. The classic paper, to my mind, on the required restrictions for an adequate understanding of omnipotence is Flint and Freddoso, “Maximal Power,” in The Existence and Nature of God, ed. Alfred Freddoso (Notre Dame: Notre Dame University Press, 1983), pp. 81-113.