Figuring out Zagzebski's take on Molinism

In this blog, I discuss the first couple sections of Zagzebski's chapter on Molinism in her book The Dilemma of Freedom and Foreknowledge (1991). 

Zagzebski admits that middle knowledge is a powerful theory (I've never understood objections from the likes of John Martin Fischer that middle knowledge doesn't do anything to explain the foreknowledge/free will issue). And let's be honest. Theological squabbles aside, the theory, the model, taken on its own, looked at conceptually, is amazingly elegant. Even though I realize there's a degree of annoyance critics of a view feel when they're constantly being told that they've misunderstood the view (a charge Calvinists often level against their critics), I believe there's something to this. It's very hard to find a solid criticism of Molinism, on a purely philosophical level, that hasn't (to some degree) misunderstood some element of Molinism itself. (And it goes without saying that I would never take the view that merely understanding a view is sufficient to make that view true - the closest I get to saying something like that is in considering the concept of God in the context of ontological arguments). I attribute it to the idea that the metaphysics of modality is really hard to do and think about oftentimes. 

(Also, as an aside, I'll try to do a blog on Zagzebski's take on Ockhamism - that God's past beliefs aren't accidentally necessary because they're soft facts about the past. There were some parts of the chapter I found a little murky, but it would be worth the effort to try and write out what I think is the takeaway. I will say that there was not a little surprise in how Zagzebski burst my bubble a little. The glib way with which I understood Ockhamism is now something I better understand. Each presentation of an analysis of 'soft fact' was something I was nodding my head in agreement only to have Zagzebski clearly and decisively show me that there was something fundamentally wrong with it. So, I'm grateful for that.)

Zagzebski argues that middle knowledge aims to demonstrate the logical consistency, and the coplausibility, of foreknowledge (with all of Zagzebski's qualifications: Divine, Provident, Essential Infallibility and Omniscience) and libertarian free will, the compatibility of the doctrine with Aquinas's theory of God's mode of knowing, and how God can know what I freely do by knowing His own essence (that last part is the part that often get neglected: there's always some parenthetical remarks about God knowing these things conceptually, as opposed to perceptually, and we're left to our own devices to pin down what that means analytically on our own). Molina explicitly mentions the "knowing His own essence" qualification and Zagzebski is sure to provide that quotation. 

Mention is made of Plantinga's now famous, accidental excavation of the doctrine when crafting his response to Mackie's version of the logical problem of evil (which I find puzzling a little: Zagzebski contrasts Ockhamism and Molinism - and they do seem to be conceptually distinct - but I find Plantinga using middle knowledge with the problem of evil and Ockhamism with the problem of foreknowledge and free will: looking forward to reading Christopher J. Kosciuk dissertation blending the two to see some sort of reconciliation here). 

There is new, exciting breakthrough explanations of the possibility/feasibility distinction usually only talked about in the way William Lane Craig has presented the distinction. Craig will often tell his listeners/readers to go to Thomas Flint's book (Divine Providence: The Molinist Account) to get a better, fuller investigation of the distinction, and I wonder how many people actually take Craig's advice. Flint's book is a wonderful philosophical exploration of the doctrine. You'll find very interesting meditations on the concept of feasibility, but also on a concept that Flint calls a "galaxy". I had thought that was a locution that Flint coined, but his book was published in 1998. I'll have to do some further investigation to see if Flint was using this earlier, but my point is that Zagzebski's book was published in 1991 (!) and she begins a discussion of the concept of a "galaxy" alongside what she calls "world-germs" seven years before Flint's book: and this is all in the context of fleshing out the concept of "feasibility" (she doesn't use the word "feasibility": she uses the word "realizable", which harkens back to Robert Adams). (Once again, the point is, if you're willing to dig beneath the surface a bit, you'll see an entire conceptual genealogy, with a family history, and that these concepts don't sprout out of some ahistorical void.)

To understand what a galaxy is we have to first understand what a world germ is. A world germ 

probably contains certain substances and laws and the results of any direct action by God on these substances. We need not settle just how far God's direct action goes in the creation, since the point is that whatever God does, he does not bring about a complete world. His creative activity is compatible with a great number of possible worlds, probably an infinite number of them. What makes the actual world this particular world rather than some other one compatible with what God has created is determined by both God's direct action in the world subsequent to the creation and by the free action of the creatures God has made (p. 129). 

The "results of any direct action by God" on the substances are called counterfactuals of divine freedom (there's concordance with how Plantinga uses the term "counterfactuals of divine freedom" in his Methodological Naturalism essay (1997): "Perhaps the demand for law cannot be met. Perhaps there are regularities, but no laws; perhaps there is nothing like the necessity allegedly attaching to laws. Perhaps the best way to think of these alleged laws is as universally or nearly universally quantified counterfactuals of divine freedom." Then, in footnote 11: "That is, propositions that state how God (freely) treats the things he has made, and how he would have treated them had things been relevantly different. Nearly universally quantified: if we think of them this way, we can think of miracles as going contrary to law without thinking of them (inconsistently) as exceptions to some universal and necessary proposition."). 

There "is a set of possible worlds compatible with that world-germ. Let us call each such set of worlds a galaxy" (129). She then represents this "diagrammatically":

1. world-germ 1: world w1.1, w1.2, w1.3, w1. . . . GALAXY 1
2. world-germ 2: world w2.1, w2.2, w2.3, w2. . . . GALAXY 2
3. world-germ n: world wn.1, wn.2, wn.3, wn. . . . GALAXY n

Zagzebski broaches the question: Why doesn't God actualize a world where everyone chooses the good and no one chooses evil? 

Now suppose that there are possible worlds that contain free creatures who always choose good. Why doesn't God just create those worlds? The short answer is that it is possible that there are true propositions such as the following:

(i) If God created world-germ 1, world w1.2 would be actual, (ii) If God created world-germ 2, world w2.4 would be actual, (iii) If God created world-germ 3, world w3.1 would be actual,

where in each case the world that would have resulted given God's creative activity would have been one containing evil (perhaps very great evil).

So even if, say, worlds w1.1, w2.3, and w3.2 have no evil in them, they would not result even if God did his part in bringing them about. It is not God's fault, then, that there is evil, even if there are possible worlds with free creatures and no evil. Such worlds are unrealizable (p. 130).

This is very interesting and I think critics are on the wrong methodological foot when they leap immediately into a critique here. Let's try to understand the modal point being made. We are to believe in the possibility of these counterfactuals of world actualization. Presumably, individual counterfactuals of creaturely freedom constitute (in some way) these counterfactuals of world actualization. We can incorporate the strong/weak actualization distinction in this context as well. It looks as though God strongly actualizes world-germs, along with that "determined by both God's direct action in the world subsequent to the creation", and that God softly actualizes "the free action of the creatures God has made" (p. 129) - the creature strongly actualizes their own free actions 'once they are made': the 'being made' an additional instance of strong actualization, along with their being continually conserved in existence. 

In section 1.2 (The Middle Knowledge Solution), Zagzebski admits that if Middle Knowledge is true, then it provides a way out of the Foreknowledge/Freedom dilemma, because the accidental necessity of God's infallible beliefs (distinct from logical and causal necessity) doesn't transfer to our choices (which involves denying Transfer of Necessity Principles), and so our choices remain contingent even if implied by, entailed by, or strictly equivalent to, God's infallible, past beliefs. If such necessity doesn't transfer, then saying we have power over whether or not we do something doesn't say anything about whether we have power over the truth of counterfactuals or the truth of God's past beliefs. 

In section 2 (Conditional Excluded Middle and the Asymmetry of Time), Zagzebski argues (against the theory of Middle Knowledge) that it's rational to deny that Conditional Excluded Middle applies to counterfactuals of creaturely freedom (CEM-CCF), but that since Middle Knowledge requires CEM-CCF, Middle Knowledge doesn't explain how God knows the truth of CCFs prevolitionally. If one affirms CEM-CCF, then one denies the asymmetry of past and future time. It's in this section that I completely disagree with Zagzebski here (or, at least, I should say that, based on what I've read, what she says here doesn't do anything to move me). 

Let me explain. First, Zagzebski seems to be making the same move that Gregory Boyd makes in interpreting the Counterfactual Square of Opposition (hereafter, just 'Square'). Zagzebski and Boyd seem to be saying that we should construct the Square such that might and might-not counterfactuals are not contraries (contraries are such that they can't both be true, even if they can both be false), and the reason they seem to be saying this is that they're semantically equating 'could' with 'might'. Based on this equation, since it's perfectly sensible to say that something could and could not be the case, they see conjoined might/might-not counterfactuals as perfectly legitimate. But this isn't how the counterfactual square of opposition is understood at all (1. 'might' and 'might-not' counterfactuals are contraries. 2. 'would' and 'would-not' counterfactuals are contraries. 3. 'would' is logically prior to 'might' and contradictory to 'might-not'. 4. 'would-not' is logically prior to 'might-not' and contradictory to 'might'.), and semantically equating 'coulds' and 'mights' just obfuscates things worse. 

Let's take a quick look at William Lane Craig's Question of the Week (#89), Gregory Boyd’s Neo-Molinism., and also some points made in response to Boyd in the book, Four Views on Divine Providence (2011), specifically Craig's rebuttal to Boyd's chapter, God Limits His Control. (I'll only list what I think are the relevant points, not every point)

1. Question of the Week: First, "there is an important difference between what a person can do and what he might do in any given set of circumstances." Second, "on the Molinist view would-counterfactuals logically imply might-counterfactuals, so that both are true and known to God." Third, "on the traditional semantics for counterfactual conditionals, might-counterfactuals are simply defined—contrary to their usage in ordinary language—to be the negations of would not-counterfactuals." 
2. Response to God Limits His Control: First, "Boyd proposes a reform of the English language" (this one is the most important: for the sake of brevity here, I only include this point as a thesis sentence - I'll unpack this more in response to Zagzebski). Second, "one can embrace a tensed theory of time and causal indeterminism without sacrificing the bivalence of future contingent statements". 

By my lights, these five points completely undercut some of Zagzebski's key points in her case. For example, she says: 

the openness of the future means that the totality of states of affairs in the past and present is not sufficient to determine that the future be what it is going to be. So The totality of present and past > actual future is false. That is, it is false that if the past and present were just as it is, the actual future would follow. Alternative futures might follow instead.

Here Zagzebski unjustifiably equates what might follow with what could follow and uses that equation to negate the truth of what will or would follow. That equation is supposed to justify the move that since it's perfectly sensible to say that something 'could and could not' happen, it's perfectly sensible to say that something 'might and might not' happen; and if something 'might and might not' happen, it's false that something 'would or would not' happen (since she - and Boyd, it seems - are making the 'mights' logically prior to the 'woulds', and then making 'might not' contradictory to 'would', and 'might' contradictory to 'would not'. All of this is semantically catastrophic because, while the 'could and could-not' propositions are perfectly compatible (there's nothing wrong with alternative possibilities), 'might and might-not' propositions are contraries! Therefore, they can't be doing what Zagzebski is trying to say they're doing. 

What's baffling to me is that it seems like Zagzebski is aware of all this in discussing her examples of foward-looking counterfactuals (one whose antecedent is about the past/present and the consequent is about the future). She cites Lewis/Pollock as supporting what I've been saying but then disagrees with them on the basis of her semantic equation of 'mights' and 'coulds'. Let me quote her at length here (emphasis in bold is mine): 

To take a particular example, consider the following subjunctive conditional, where the antecedent is about the actual present and past and the consequent is about the future:

(4) If I lived my life the way I have up to now (May 1988), I would be living in Chicago in 1999.

It seems to me that (4) is false. Defining the might-counterfactual in terms of the would- counterfactual, as Lewis and Pollock do, (4) is equivalent to the following.

(5) ~ (If I lived my life the way I have up to now (May 1988), I might not be living in Chicago in 1999).

But (5) is surely false, and it seems false because it seems true instead that the following:

(6) If I lived my life the way I have up to now (May 1988), I might be living in Chicago in 1999.

(7) If I lived my life the way I have up to now (May 1988), I might not be living in Chicago in 1999).

I have not proposed an account of counterfactuals in this book, but it is clear to me that there is a large class of paradigm cases of the might-counterfactual in which the antecedent is about the actual past or present and the consequent about the future. In these cases we think that both A > might B and A > might not B are true. This means that the principle of Conditional Excluded Middle (CEM), (A > B) or (A >-B), is false when A is about the actual past and present and B is about either the actual or possible future. CEM fails in these cases because one mark of the difference between the future and the past is just that the future is such that there is more than one alternative that might be the future, whereas there is only one alternative that (now) might be the past. The ordinary notion of time explained in Chapter 1, then, supports the rejection of CEM for propositions in this category.

End quote. Notice that Lewis and Pollock (and Stalnaker, by the way) support my view of the Square: the 'would' contradictory to the 'might not', and therefore, if it's true that I would do something, it's not true that I might not do something (since 'woulds' logically imply 'mights', not 'might-nots'). But Zagzebski is forced to disagree with this because it 'seems true' that some conjoined might/might-not counterfactual is true. But the only reason (that I can see) why she would think this is because she's semantically equating 'mights' with 'coulds' and 'might-nots' with 'could-nots', an equation that is semantically forbidden. This equation leads her into thinking that there is this 'large class of paradigm cases of might-counterfactuals' that just don't exist! Also, notice why she thinks this in the last bold emphasis: there is more than one alternative that might be the future. This is not true. If something S will/would be, then it's not true that S 'might not' be, even though it will continue to be true that S could and could not be. 

On the other hand, Zagzebski thinks that Molinist are correct to say that some CCFs are true. But she believes that citing examples of true CCFs isn't sufficient for demonstrating the truth/plausibility of the theory of Middle Knowledge. If "many free choices are not counterfactually implied by even the totality of the past up to the time of the choice", then "God cannot know these choices by knowing counterfactuals and his own will" (pg. 141). But the problem is that Zagzebski's entire case for thinking that many free choices are not counterfactually implied by the past up to the time of the choice is based on inferences drawn from semantic conflations. If the semantics, properly understood, forbid those conflations, then there's no such obstacle to the truth/plausibility of the theory of Middle Knowledge. Therefore, if you accept Middle Knowledge, and Middle Knowledge implies CEM, that's perfectly okay, since CEM doesn't have the consequences that Zagzebski seems to think it has.