This is a long-planned continuation of a previous post from several months ago in which I commented on chapters 1–3 of Kirk MacGregor’s book, Molinist Philosophical and Theological Ventures (2022). MacGregor is one of the most active contemporary defenders of Molinism, a fairly popular model of divine providence that aims to reconcile a meticulous model of providence (according to which God specifically ordains everything that happens) with creaturely libertarian freedom. In my first post, I argued that MacGregor’s first two chapters, in which he explicates Molinism and develops a Biblical argument for the view, are of decent quality but also fall short in some important respects. (I would give them a “mid pass” grade.) As for his third chapter, which tackles the well-known grounding objection against Molinism, I gave it a failing grade because it never seriously engages with the objection.
In this post, I comment on MacGregor’s chapter 4, which tackles another well-known objection against Molinism, specifically that Molinism generates a vicious explanatory circle and is, therefore, internally inconsistent. I’ll call this the explanatory circularity objection. In what follows I walk through chapter 4 and explain where I think MacGregor gets it right and where he gets it wrong.
1. The Adams/Hasker version of the objection
MacGregor begins by briefly recounting the history of the explanatory circularity objection as initially presented in the 1990s by Robert Adams and William Hasker. The objection, as MacGregor presents it (p. 57), begins as follows:
- According to Molinism, the truth of all true counterfactuals of creaturely freedom (CCFs) about us is explanatorily prior to God’s decision to create us.
- God’s decision to create us is explanatorily prior to our existence.
- Our existence is explanatorily prior to all of our choices and actions.
- The relation of explanatory priority is transitive.
The objection then attempts to derive a contradiction from 1–4 by pointing out that, if we are libertarian free agents, as Molinism affirms, then our actual choices must also be explanatorily prior to the truth of CCFs about us. This closes the explanatory circle. Given transitivity, it follows that all of these terms (the truth values of CCFs, God’s creative decision, our actual existence, and our free choices) are explanatorily prior to each other. A contradiction can then be derived by noting that some of these things cannot plausibly be explanatorily prior to others. For example, our choices and actions cannot plausibly be explanatorily prior to our own existence—obviously, we can’t do anything if we don’t exist! Nor can our existence as creatures plausibly be explanatorily prior to God’s creative decision. It follows from there that some things in the circle both are and are not explanatorily prior to each other, and that’s a contradiction. Since it is Molinist assumptions together with a seemingly neutral explanatory priority relation that generates the contradiction, it appears that Molinism is the culprit and so should be rejected.
Obviously, for this objection to work, the relevant notion of explanatory priority must be univocal and transitive throughout, and this is where MacGregor rightly focuses his attention. He contends that, however the objection may be developed, the notion of explanatory priority at work will either be equivocal or non-transitive. Let’s see if he’s right about that.
2. MacGregor rebuts Adams and Hasker
Let’s now consider MacGregor’s rebuttal to the Adams and Hasker. Hasker defines explanatory priority as follows: “p is explanatorily prior to q iff p must be included in a complete explanation of why q obtains” (p. 58). According to MacGregor, this won’t do because (quoting William Lane Craig) it renders premise 1 false: “God may well have chosen to create us even if some or all true counterfactuals about us were false or even if no counterfactuals about us were true. Thus the truth of all true counterfactuals about us or even there being true counterfactuals about us need not be included in a complete explanation of God’s decision to create us” (p. 58). This line of rebuttal misses its target, however, for Craig has uncharitably switched the explanatory scenario from one in which the actual truth-values of CCFs inform God’s actual creative decision to a counterfactual one in which different CCFs are true. Adams and Hasker, however, can simply concede that God might have still created, say, Adam and Eve, even if (counterfactually) different CCFs had been true about them while still maintaining that the actual truth-values of the CCFs, which define the range of “feasible” worlds on Molinism, inform and are thus explanatorily prior to, God’s actual decision to create this world.
A better objection is that a “complete explanation” of God’s decision to create this world need not “include” information about Adam and Eve’s CCFs at all. Perhaps God chose this world for reasons entirely independent of those CCFs. But even if that’s true for Adam and Eve, it’s surely not true for every created free being in the actual world. Otherwise, middle knowledge is rendered irrelevant for divine providence concerning the actual world, something no Molinist can accept. So, even on this objection, the Adams/Hasker argument remains viable. To defend premise 1 all they need is one created individual whose CCFs are part of the complete explanation of God’s actual creative decision. They can then generate a vicious explanatory circle centered around that individual.
MacGregor’s second line of attack, again following Craig, is that Hasker’s conception of explanatory priority is subject to counterexample: “Suppose … John is going to the party [in part] because Mary is going, and Mary is going [in part] because John is going. It follows [assuming transitivity] that … John is going to the party [in part] because John is going to the party, which conclusion is obviously wrong” (p. 58). But this objection is unconvincing as well, for it’s not clear that the counterexample scenario (as described) is coherent. If John and Mary are each going only because the other is going, then that leads to the same sort of “obviously wrong” conclusion (viz., “John and Mary are going because John and Mary are going”) that MacGregor objects to. To avoid a vicious circle we must suppose that either John, Mary, or both have other reasons for going as well, which explains my parenthetical insertion of “in part.” In a coherent scenario we might suppose that John will go if and only if Mary goes and that Mary knows this. For her part, Mary is willing to go partly because John will go if she does and partly because she has non-decisive independent reasons for going. In this case, the complete explanation for John’s going must include Mary’s actually going, but the complete explanation for Mary’s going doesn’t include John’s actually going, but rather John’s going if Mary does. Applying transitivity, we get John goes because Mary goes, Mary goes [in part] because John goes if she goes, therefore, John goes [in part] because John goes if Mary goes. There’s no vicious circle here, and so I conclude that MacGregor (like Craig) has not successfully refuted Hasker.
In any case, MacGregor thinks he has refuted the Adams/Hasker version of the explanatory circularity objection and, since there are other versions of the objection to address, he shifts his focus to recent work by Philip Swenson and Andrew Law.
3. Swenson and Law on explanatory dependence
Swenson and Law (S&L) understand explanatory priority in terms of explanatory dependence, which Law defines as follows: “a fact, F1, is explanatorily dependent on another fact, F2, just in case F2 at least partly explains F1″ (p. 59). MacGregor quotes Swenson saying that explanatory dependence is “the broad or generic notion that captures what all … different cases of dependence have in common” (p. 60). The cases of dependence Swenson has in mind include, but are not limited to, logical, conceptual, metaphysical, and natural dependence. Law adds in nomic/causal explanation as well. If S&L have correctly identified a generic notion of explanatory dependence broad enough to include the various relationships in premises 1–3, then the explanatory circularity objection will be immune to charges of equivocation. If this generic notion is also transitive, then the objection succeeds. Two relevant questions, then, are how generic explanatory dependence relates to explanatory priority and whether it is in fact transitive.
MacGregor doesn’t spell out what S&L take the connection between explanatory priority and explanatory dependence to be. The most obvious idea (which may not be exactly what S&L have in mind) is that A is explanatorily prior to B just in case A “at least partly explains” B, or in other words, A is explanatorily prior to B just in case B is explanatorily dependent on A.
MacGregor objects that if explanatory dependence so understood is broad enough to include both logical and conceptual dependence, then it is “equivocal,” “incoherent,” and not transitive (p. 61). But he both overstates the problems and makes his case rather poorly. Drawing on Swenson’s examples, MacGregor says that “the mark of logical dependence is necessity” and “the mark of conceptual dependence is sufficiency” (p. 60). But this is false. Broadly understood, logical and conceptual dependence are the same thing. Both are a matter of what Hume called “relations of ideas,” and necessity and sufficiency are relevant in both cases.
Note: There is a way to differentiate between logical and conceptual dependence, but it requires narrowing “logical” dependence to contexts in which conceptual relevance doesn’t matter. In standard truth-functional systems of logic, for example, every proposition follows from, and is thus logically “dependent” on, every false proposition, even if the two proposition are conceptually irrelevant to each other. This is the ex falso quodlibet principle: from the false, everything follows. The conceptual absurdities to which this principle leads were a major motivating factor prompting the development of non-truth-functional logical systems, like modal and relevance logics.
In any case, MacGregor follows with an even more dubious claim, namely, that logical and conceptual dependence are clearly “incompatible” and “have nothing in common” (p. 60). This is grossly overstated. Across a wide range of cases, at least, logical and conceptual dependence are clearly compatible and have a lot in common. In fact, aside from explicitly truth-functional systems of logic, it’s not easy to distinguish between them.
MacGregor then makes another claim that he neither explains nor defends: logical/conceptual dependence is (in some cases) symmetric whereas no “legitimate” relation of explanatory dependence can be symmetric (p. 60). On the surface this is a plausible claim that might be used to drive a wedge between logical/conceptual dependence and explanatory dependence proper. After all, we typically think of explanatory sequences as unidirectional, like a dominoes falling in sequence, not symmetric or bidirectional. But if MacGregor means to say that no legitimate explanatory dependence relation can ever, in any instances, be symmetric, then not only does that conflict with his (and Craig’s) John and Mary example above—as presented, the example is clearly symmetric—but it verges on question-begging. For, absent some independent defense of a global anti-symmetry requirement on explanation, it amounts to stipulating a requirement that blocks any version of the explanatory circularity objection by fiat. One can also turn things around and argue that Molinism requires symmetry—explanatory relations have to run in both directions in order to accommodate both meticulous providence and creaturely freedom.
4. The fixity of the independent
The fundamental issue that the explanatory circularity objection brings to light isn’t circularity as such, but the seeming incoherence of Molinism. More specifically, there is a tension between Molinism’s commitment to both (a) meticulous providence and (b) creaturely freedom to do otherwise in the actual circumstances in which those choices are made. What closes the explanatory circle, recall, isn’t anything in premises 1–4, but rather the implications of creaturely freedom. So, on the one hand, if God’s providence is meticulous, then things like God’s middle knowledge and creative decree must be relevantly, and presumably explanatorily, prior to creaturely free actions. Whereas, on the other hand, if creatures are actually free to do otherwise, then it must in some sense be up to them whether God chose this world to create and whether their CCFs say they would freely do this rather than something else. The point of the explanatory circularity objection is ultimately to argue that, if one holds both meticulous providence and creaturely freedom true at the same time, then one lands in a vicious circularity wherein certain things are both “up to us” and not “up to us.”
We can capture this notion of “up to usness” by distinguishing between things that are fixed independently of our free choices, such that we have no “say” about them, and things that are fixed by our free choices. In this context S&L introduce the notion of the fixity of the independent. As Swenson explains, the “independent” with respect to A is that which is “not even partially explained” by A. If B is not even partly explained by A, then B is explanatorily independent of A and therefore “fixed” from the perspective of A. In other words, if B is fixed independently of A, then A has no “say” about B and it is not “up to A” whether B is or is not the case.
Put in these terms, the challenge for Molinism is that, given meticulous providence, the content of God’s middle knowledge and the content of God’s creative decree are fixed independently of all actual creaturely choices whereas, given creaturely freedom, the content of God’s middle knowledge and the content of God’s creative decree remain, in some respects, up to us and thus are not fixed independently of all actual creaturely choices. So far as I can see, Molinists have no good answer to this dilemma. If they affirm that the content of God’s middle knowledge and creative decree are (to some extent) up to us, then they undermine the Molinist commitment to meticulous providence. If, however, they affirm meticulous providence, then they must admit that we have no “say” about the CCFs concerning us or what God has decreed that we do. In short, either meticulous providence or creaturely freedom has to go. The Molinist can’t have his cake and eat it too, or so the objection runs.
5. MacGregor on counterfactual dependence
MacGregor’s strategy for rebuttal (which he doesn’t clearly articulate) is to soften the requirements on what it means to have “say” about something by asking “whether any form of counterfactual dependence counts as even partially explanatory” (p. 61). If we say, yes, then following Craig, Plantinga, and other Molinists, he’s going to maintain that we have all the “say” about the CCFs, God’s decree, etc. that we need. Because God is meticulously provident, the CCFs, God’s decree, etc. are actually prior to our choices. But because we are free and could have chosen otherwise, our choices are counterfactually prior to the CCFs, God’s decree, etc. In short, MacGregor and other Molinists rely on so-called “counterfactual dependence” to avoid the appearance of vicious circularity. In one direction (CCFs → God’s decree → our existence → our choices) we’re dealing with an actual explanatory sequence, whereas in the other direction (our choices → our existence → God’s decree → CCFs) we’re dealing with a counterfactual explanatory sequence. Because one is actual and the other counterfactual, the two sequences can’t conflict, and so there’s no vicious circularity, or so the thought goes.
It may be worth noting that this is the same strategy classical compatibilists employ in the free-will debate over against classical (leeway) libertarians. Classical libertarians affirm that genuine moral freedom requires the ability to choose otherwise and thus is incompatible with determinism. Classical compatibilists, in contrast, challenge this inference. They affirm both that moral freedom is compatible with determinism and that we could have chosen otherwise even if we were causally determined to do exactly what we actually did. They way they square this circle is precisely by going counterfactual, just like MacGregor and other Molinists do. “You could have chosen otherwise,” says the classical compatibilist, simply means that “if you had wanted to chose otherwise, then you would have chosen otherwise.” From the libertarian perspective this is what Kant famously called a “wretched subterfuge.” It amounts to changing the topic and hoping we don’t notice that a merely counterfactual ability to choose otherwise has no actual relevance.
In any case, the correct answer to MacGregor’s question about whether any form of counterfactual dependence counts as even partially explanatory is an emphatic “no,” for reasons I give in this post. Counterfactual dependence is merely a kind of necessary correlation. It is not, in any fundamental sense, explanatory, for the same reason that correlation doesn’t imply causation.
But, says MacGregor, if we answer his question negatively, then the principle of the fixity of the independent is false because “Facts are abstract objects and therefore causally effete. Facts about what A would do in a situation have no power over A to make A do accordingly in that situation. Hence A could do something incompatible with the CCFs.” (p. 62) Again, this misses the point. The issue isn’t whether facts that are fixed independently of what we do somehow constrain or causally compel us to do what we do. The issue is whether they can be “fixed facts” if they (a) they entail that we do A and (b) we can nevertheless do not-A. Keeping our focus squarely on the actual sequence, if we have the power to “do something incompatible with the CCFs,” then we have the power to nullify the facthood of those CCFs and the power to nullify the facthood of the relevant aspects of God’s meticulous creative decree. And if that’s the case, then Molinism collapses into Ockhamism. It no longer retains any providential advantage over simple foreknowledge and open theist models of divine providence.
6. Swenson’s dilemma for Molinism
As we have seen, Molinism appears to be internally inconsistent. Meticulous providence requires that the content of God’s middle knowledge and the content of God’s creative decree be fixed independently of creaturely actions, whereas creaturely freedom requires that the content of God’s middle knowledge and the content of God’s creative decree be fixed in part by creaturely actions. Swenson poses this as a dilemma by asking “Are the CCFs involving our actual choices explained by our choices or not?” (p. 63). If we answer negatively, then we say that the content of God’s middle knowledge (so far as it concerns us) is explanatorily independent of our actual choices. It’s not at all up to us which CCFs are true. But then since those CCFs specify precisely how we freely choose in every circumstance in which we find ourselves, we cannot do otherwise then how they specify, and so we aren’t free. If we could do otherwise, then it would be up to us which CCFs concerning us are true after all, contrary to the supposition. But if we answer affirmatively, then we say that the content of God’s middle knowledge (so far as it concerns us) is explanatorily dependent on our actual choices. In that case the content of those CCFs is fixed by our actual choices and so isn’t available to God independently of those choices. The result is that God can’t use that information providentially.
MacGregor doesn’t see the matter this way. He thinks a negative answer to Swenson’s question falsifies the fixity of the independent principle on the grounds that it (falsely, in his view) entails that we have no counterfactual power over our own CCFs (p. 64). I’ve already explained above why this isn’t a good rejoinder. Merely counterfactual power is no power at all. As for an affirmative answer to Swenson’s question, MacGregor thinks this renders the explanatory priority relation equivocal and non-transitive (pp. 65–68). In broad terms, his strategy is to argue that the backward explanatory relations (our choices → our existence → God’s decree → CCFs) can be understood in purely counterfactual terms, and therefore do not conflict with the forward explanatory relations (CCFs → God’s decree → our existence → our choices). Again, I’ve already explained why this isn’t a good strategy. Merely counterfactual explanations do no real explanatory work, and so they can’t capture creaturely freedom actually to do otherwise.
7. Wrapping up
We’re not even halfway yet though MacGregor’s chapter, but rather than continue with a detailed fisking (and there’s plenty left to fisk), which would probably bore most of my readers to tears, I’m going to wrap things up.
- The strongest way of posing the explanatory circularity objection against Molinism is to focus on the fixity of the independent, that is, the negative fact that there are things we have no “say” about, rather than on positive explanatory dependence. It doesn’t matter whether our CCFs, God’s creative decree, etc. positively influence or even positively explain what we do. What matters is whether they are fixed independently of what we do. If so, and if they entail that we act in such-and-such a way, then we cannot be free to do otherwise, because if we were categorically able to do otherwise, then those things would not be fixed independently of what we do.
- MacGregor’s main line of response is to insist that our own CCFs counterfactually depend on our free choices and thus that we have counterfactual power over our own CCFs and over God’s creative decree insofar as it draws upon those CCFs. I contend above (section 5), however, that there is no such thing as merely counterfactual power or merely counterfactual dependence. All real power, dependence, and explanation, is categorical, not merely conditional: A brings about B, A explains B, B depends on A, etc.
- MacGregor repeatedly makes a big deal out of the idea that all “facts” are supposedly abstract and, therefore, causally inert. For this reason, he dismisses any suggestion that the facts about our CCFs and God’s decree could possibly conflict with creaturely freedom. But the claim that all “facts” are abstract is not only false (see note below), it’s also irrelevant. The issue is not whether the facts about our CCFs and God’s decree somehow causally compel us to act as we do (they don’t), but whether our actions are entailed by facts that are fixed independently of what we do. If they are then, as Jonathan Edwards pointed out a couple centuries ago, they show that we lack freedom to do otherwise even if they do not themselves make us unfree to do otherwise.
Note: There are at least three different ways in which the word “fact” is used. (1) in modern colloquial usage, “facts” are often understood as generally accepted ideas, which therefore contrast with “opinions,” i.e., ideas that are not generally accepted. In this sense, facts are collective psychological states and therefore concrete. Most philosophers reject this usage because it’s too subjective. Before Copernicus it was a “fact” that the Earth was at the center of the solar system. Sometime afterwards it ceased to be a “fact.” That’s weird. Among philosophers, “facts” are typically understood either as (2) true propositions (“the fact that …”) or as (3) states of affairs (“the facts on the ground”). With respect to (2), while propositions are abstract, their truth or facthood often consists in their correspondence with concrete reality. With respect to (3), there are both abstract and concrete states of affairs. Triangularity entailing three-sidedness is abstract. Fluffy the cat’s being on the mat is concrete.
Overall, I give MacGregor’s chapter 4 a passing grade. It scores points for engaging with relevant recent literature (Adams, Hasker, Flint, Plantinga, Swenson, Law, Climenhaga, Rubio, etc.). I mark it down, however, because MacGregor often fails to explain himself clearly, frequently misses the point, and makes several highly dubious claims (e.g., he twice repeats John Laing’s bizarre idea that “creaturely wills … preexist in the divine mind as ideas”). Molinists looking for a carefully reasoned reply to the explanatory circularity objection are probably better off sticking with ch. 7 of Tom Flint’s now somewhat dated 1998 book, despite the fact that Flint blatantly begs the question against open futurism (see Flint, p. 165).
Pingback: The Metaphysical Muddles of Molinism – Some Thoughts on MacGregor (Part 3) – Open Future