August 7, 2018

A Theological 'Slippery Slope' Argument for Compatibilism

When I first began studying philosophy, I was a convinced libertarian about free will. My reasons included supposed direct introspection together with what I now take to be two distinct but related intuitions, which I will now call the consequence argument intuition and the buck-stopping intuition. (I wouldn't have explained them like this back then, of course: I'm trying to do some autobiographical rational reconstruction.) The consequence argument intuition is the notion that if an event is necessitated—whether logically, metaphysically, or causally/nomologically—by factors outside my control, then that event is itself outside my control, and an event outside my control is clearly not a free action of mine. The buck-stopping intuition is the notion that if some other agent (e.g., God) bears some kind of ultimate responsibility for my choice, then the choice is not free.

The more time I've spent thinking about this matters (and it's been over 15 years now), the more difficult I find it to make any sense out of these intuitions or the notion of freedom they seem to support. I certainly still feel their pull to some extent, but I just don't know what to do with them. Furthermore, I think my previous belief in direct introspection of libertarian free will was just a confusion stemming from a failure to consider what it would be like to possess compatibilist free will.

In recent philosophy of religion publications (e.g., here, pp. 10-11 and here, pp. 10-12), I've been saying: "You might think that my views here entail omnidetermination [i.e., the view that God determines every detail of the course of history]. I'm not convinced of that, but if you did convince me, I'd just shrug and endorse omnidetermination. It's not so bad." Today, I have a strong opinion about what is the best form of libertarianism (some kind of broadly Reidian agent-causation theory) and what is the best form of compatiblism (a 'Frankfurt-style' theory, along the lines of Gary Watson), but I don't exactly endorse either. On the one hand, my old libertarian intuitions (however hard they are to make philosophical sense of) are still strong enough to make me somewhat uncomfortable with even the best version of compatibilism. On the other hand, even if the libertarian theory could overcome the formidable empirical and metaphysical problems it faces, it's not clear that it succeeds on its own terms. The reason agent-causal libertarianism is (to my mind) a better theory than event-causal libertarianism is that it inserts the agent qua agent into the explanation of the action, and this is very important for overcoming the randomness objection: using a random number generator to determine the action of an AI would not make the AI more free, even if the random number generator was genuinely undetermined, and it's unclear how the event-causal libertarian can do any better than this. (There have been many attempts, and I haven't read nearly all of them, but I've read quite a few and I'm pretty convinced that this strategy just won't work.) Even on the agent-causal theory, however, there is some question about whether we have really succeeded in inserting the agent qua agent. An agent is influenced by motives, character, etc. So the really pressing thing if the agent-causal theory is to succeed on its own terms is for it to make sense of some kind of non-causal influence of motives, character, etc., on action in a way that makes sense of agency and makes agency free in a way that compatibilism can't. (In a historical paper I've argued that Reid does recognize this problem. However, he irritatingly ceases to philosophize and appeals to common sense just at the crucial moment! Because, you know, he's Reid...)

As can perhaps be gathered from the above, the core idea that has led me to take a skeptical attitude to my libertarian intuitions is that when it comes to examining philosophical theories of free will, I just can't see how the theories that respect the libertarian intuitions make me freer than those that don't. After all, if actions were random that would respect the intuitions, but it wouldn't make them free. The intuitions are negative; we need some positive characterization of freedom to fill them in, and the failure of these positive characterizations gives me some suspicion that the intuitions may just be mistaken.

This brings me to the theological 'slippery slope' argument promised in the title of the post. At present, I consider the following argument to be among my strongest reasons for leaning toward compatibilism:

  1. If God exercises providential control, then God knows counterfactuals of creaturely freedom.

  2. If God knows counterfactuals of creaturely freedom, then either Molinism is true, or acausal Thomism is true, or determinism is true.

  3. If Molinism is true, creatures do not have any form of freedom stronger than that available on acausal Thomism.
  4. If acausal Thomism is true, creatures do not have any form of freedom stronger than that available on determinism.

  5. Therefore, if God exercises providential control, then creatures do not have any form of freedom stronger than that available on determinism.

  6. God does exercise providential control.

  7. Therefore, creatures do not have any form of freedom stronger than that available on determinism.

The reason I call this a 'slippery slope' argument is that proponents of Molinism often claim that they allow a stronger form of freedom than is available on acausal Thomism, and acausal Thomists often claim that they allow a stronger form of freedom than is available on determinism, but I think that this is not the case, so that once one admits divine knowledge of counterfactuals of creaturely freedom one slides right down into the compatibility of free will and determinism. Note that this doesn't mean determinism is actually true (it's probably not), just that the kind of freedom available is a kind that wouldn't necessarily be eliminated by determinism.

It might be asked why I take this to be an argument for compatibilism rather than fatalism (the denial of free will). The answer is, I think it's pretty clear that when an action follows from an agent's character and motives in the right sort of way without outside interference the action deserves to be called 'free' in some sense, and the agent bears responsibility for it in some sense. Further, I think it's clear that there are such things as character and motives and actions sometimes follow from them. The questions that remain include: is this thing 'character', which I am confident exists, something that could exist in a fully deterministic world? Is there a coherent and plausible stronger sense of 'freedom' (and perhaps a correspondingly stronger sense of 'responsibility') than this character-and-motives sense? Might indeterminism somehow make possible these stronger sense of freedom and/or responsibility? The argument, as I understand it, suggests negative answers: character should be construed compatibilistically, and following from character and motives is the strongest sense of freedom/responsibility we can reasonably attribute to creatures.

Let me unpack the argument a bit and explain why it looks like a formidable argument to me (though as I said above I'm not fully committed to endorsing it).

First, counterfactuals of creaturely freedom are claims about what creatures would do if they were placed in certain specified circumstances, e.g., (in Tom Flint's quirky example) if Cuthbert went to the pet shop, he would buy an iguana. If God doesn't have knowledge of this form, then God doesn't know what would have happened if God had created in slightly different ways, and this seriously limits God's freedom and providential control. Think of it this way: sitting unimpeded in an airplane cockpit with all of those buttons and dials and so forth in front of you doesn't give you control over the plane unless you know what happens if you push this button and also what happens if you don't.

Now, there are three basic views about how God could know what free creatures would do in counterfactual scenarios. First, the Molinist view is that God simply knows all these things from eternity, prior to God's decision to create. These facts are contingent, but beyond God's control. This, I admit, does allow God to exercise very significant providential control. However, consider the conditional claim, if a student offered me €1,000,000 to give an undeserved First, I would refuse. (A 'First' in the British/Irish marking scale is like an A in the American system.) Now this proposition and its truth or falsity would ordinarily be thought to be connected in some way to my moral character, and I might even be thought to have some responsibility for it. However, on the Molinist view, this proposition was either true or false from eternity and (assuming that I am never actually faced with this choice) nothing I can do makes any difference to its truth or falsity. I have no more control over it than God does. In a way, insofar as the conditional is made to float completely free of my character, it even seems to me that this might make my actions less free (because more random) than compatibilist views on which free actions are actions following from my character. Molinists have, of course, tried to finesse this point in various ways, but the bottom line is, I'm not convinced that having such things determined by nothing at all makes me any more free than if they were determined by God or prior causes or whatever.

The second view, acausal Thomism, holds that although my actions are undetermined with respect to the sequence of ordinary natural causes ('secondary' causes, in the traditional terminology) nevertheless God can choose how these actions will go. For comparison, imagine a genuinely indeterministic coin flip, that is, one in which the past and the laws of nature (or the sequence of ordinary natural causes) does not determine which way the coin will land. The acausal Thomist would hold that God, without disturbing the natural causal order, nevertheless as the creator and sustainer of that causal order is able to choose the outcome of such indeterministic events. Since God can determine these outcomes, God knows what would result in any possible scenario. (It is controversial whether Thomas Aquinas actually held this kind of view, but it has been called 'Thomism' since at least the 17th century.)

Now, I actually lean toward this kind of view (see here), but not because I think it makes us any more free than causal determinism does. I lean toward it because I think that causal determinism is empirically disconfirmed and because I think it provides a metaphysically and theologically satisfying account of God's relationship to creation. Obviously, this view requires just straightforwardly giving up what I called the 'buck-stopping intuition'. Once we've given that up, what does causal indeterminism buy us? Maybe not much. Most acausal Thomists would say that it buys us primitive agent-causation within the natural/created order. As I said above, I take this possibility seriously, and maybe it would somehow make us more free, but more work is needed to explain in what, positively, this freedom would consist. Otherwise we're back at randomness and, again, I think a random action is less free than an action determined by one's own character.

Finally, determinism holds that prior causes determine or necessitate all of our actions. Clearly, on such a view God, in virtue of God's understanding of the causal order, can know what we would do in any hypothetical scenario. Further, the causes may work through our character and motives and deliberation and so forth. Clearly a kind of freedom is possible here. However, as I said above, strict determinism is empirically disconfirmed (though my understanding is that neurology is as good as deterministic for practical purposes). But the question isn't whether determinism is true, but rather whether it is compatible with free will, and the argument aims to show that the kind of freedom we actually enjoy is no stronger than the kind we could enjoy if determinism were true.

If someone rejects (or weakens) the notion of divine providential control, then that person will perhaps want to endorse some form of libertarianism inconsistent with divine knowledge of counterfactuals of creaturely freedom. Even views of this sort, however, often focus merely on negative notions of freedom and don't actually explain how they are making us more free. This, indeed, is for me the core problem: libertarian theories often take on board highly questionable metaphysical, theological, and sometimes even scientific claims in order to satisfy some extremely vague intuitions about what freedom is not, without every getting around to a positive account of what freedom is and how this libertarian notion is better than compatibilist notions. My question is always: does this notion of freedom, with all of the baggage it carries, actually explain how I can be more free than I would be if my actions were determined by my character and motives? If not, why bother?

Posted by Kenny at August 7, 2018 1:15 PM
TrackBack URL for this entry:


Ken, that's as very interesting post. Tell me how the acausal Thomist explains how God controls the outcomes of indeterministic events without causing them. I have actually defended this view without knowing that it's called acausal Thomism. On my account what God does is *predict* the outcome of his initial creation. Since God's predictions are perfect, then if God predicts that a fair coin indeterministically lands tails, then a fair coinindeterministically lands tails (the counterfactual, 'were God to predict p, then it would be the case that p' is true in every world, so necessarily God can bring about p without causing p. Predictions are not causes.). Same for the free actions of libertarian free agents. But how does the thomistic God do this? Thanks.

Posted by: Mike at August 7, 2018 2:41 PM

Hi Mike,

Thanks for stopping by! In fact, I decided to write this post while reading your paper in the recent Hugh McCann volume this morning. (I might have more to say about that later.)

The Thomistic view relies on a distinction between what was traditionally called 'primary causation' and 'secondary causation'. In my "Foundational Grounding" paper ( I suggested that this is a misnomer: primary causation is not really a kind of causation at all (as the term 'causation' is used in contemporary philosophical English). It's really a type of grounding or ontological dependence. So the basic idea is that when we say that billiard ball A causes billiard ball B to move, the relation between A and B is not a relation that God ever bears to anything (except perhaps in miracles and/or the Incarnation). Ordinarily, God bears the so-called 'primary causation' relation (I call it 'foundational grounding', since it's not a kind of causation), whereby God upholds or sustains the causal order in being. The nature of this relation is such as to allow God non-causal control over all of the events in the sequence, but God does so from outside and so is not 'a cause among causes'.

One common way of explaining this (that I also employ in my paper) is to compare it to an author's control over a fiction. There is a chain of causes and effects within the fiction, and some of them might be deterministic and some might be chancy. The author, however, stands outside the fiction and constructs the causal order in whatever way she or he sees fit, and the author is perfectly capable of decreeing that on a particular occasion within the fiction a fair coin will be flipped and land heads. There's no contradiction between the author decreeing that the coin be fair and also decreeing a particular outcome because the author is not part of the causal chain that exists inside the fiction.

If I remember right, I think I got the phrase 'acausal Thomism' (as opposed to just plain 'Thomism') from Tom Flint's chapter in the Oxford Handbook of Philosophical Theology. Of course Flint doesn't endorse this view. Matthews Grant defends a version of it, though he doesn't seem to object to saying that primary causation is a species of causation (something I find problematic). It is not controversial that Aquinas employs a distinction between primary and secondary causation, but Eleanor Stump argues that Aquinas (unlike the later Thomistic tradition) does not endorse omnidetermination and is an incompatibilist about human freedom and divine omnidetermination.

Incidentally (and perhaps not surprisingly) this Thomistic theory was found congenial by Calvinists and makes an appearance in chapter 3 of the Westminster Confession (

Posted by: Kenny Pearce at August 7, 2018 2:58 PM

One weakness of the argument is that what you are calling acausal Thomism is (historically, at least) actually an account of how God can exert providential control while assuming that some counterfactuals of freedom are undefined (that's in fact one of the big points of controversy between historical Molinists and Banezian Thomists) -- there is nothing in the account that requires that all questions about what a free agent would do have answers; the free agent only grounds a range of possibilities, and the divine contribution only restricts it to one if God specifically decides that it needs to be this way in his plan (e.g., by giving you the grace of a firmly good character). Whatever one thinks of such a view, it seems to complicate the slippery slope aspect of the argument: God's providential control only requires knowledge of counterfactuals of freedom directly relevant to what he is actually intending for the actual world; this doesn't seem to require that all counterfactuals of freedom have a definite truth value.

Incidentally, although it doesn't affect your argument at large as far as I can see, I don't think you've quite characterized Molinism right. Traditional Molinists would not say that the counterfactual of freedom floated completely free of one's character -- on their view God has 'supercomprehension' of your character, knowing it so perfectly that he knows exactly what you would do in any possible order of nature, and then he just actualizes the possible order of nature in which you do such-and-such. As I say, this doesn't seem to affect how your overall argument works, but character is actually quite important to historical Molinism. (As it is for Leibniz, whose views on these subject are modified Molinism; Adam's character, as being a bundle of his properties, is already included in any counterfactual about him.)

Also incidentally, since it's just a minor example: While people often say that actions being random wouldn't make them free, if you look at the history of this claim, the arguments for this are remarkably sketchy and sometimes outright dubious. If we're talking about human actions already, in what way would their being random not make them free actions? When people try to explain this idea, they most commonly fall back on morality -- that randomness would not be adequate for an account of moral freedom. But this is a confusion -- it is an argument that, if we assumed that random human actions were free, it wouldn't be enough to give us morality; that seems plausible, but it doesn't give us any reason to think the claim itself is false -- you could accept that argument and also hold that random human action is one kind of free action.

Posted by: Brandon Watson at August 7, 2018 10:15 PM

Hi Brandon,

I have to admit I'm not too familiar with the late Medieval stuff: I know what Leibniz (and to a lesser extent Bayle) says about this, and then the recent stuff by Tom Flint, et al. The Thomistic view is standardly presented as not requiring middle knowledge, i.e., as saying that all divine knowledge is either natural knowledge or free knowledge, where natural knowledge is necessary and prevolitional and free knowledge is contingent and postvolitional. I guess the view you describe does fit that mold, it just requires some propositions with no truth values. The version I myself like is closer to what I take to be Leibniz's view (but I'm not sure whether we agree about Leibniz interpretation). It seems to me that the 'big picture' of the Thomistic system of philosophical theology supports this kind of view (on which everything that isn't determined by the necessary facts about essences is determined by the divine will), but of course there are lots of possible moves that can be made to avoid the result, and I'm not a Medievalist so I don't want to make any very specific claims about what was actually said in the period.

Related to the above, when I was characterizing 'Molinism', I mainly meant Flint. (These are among the perils of switching back and forth between my historian hat and my analytic philosopher of religion hat!) However, the view you are calling 'traditional Molinism' does sound more like Leibniz than like Flint, which is very interesting!

On the last paragraph, I did write sloppily there. I should have made two claims: (1) no purely random event can be an action, and (2) injecting randomness into the decision making process cannot make the decision more free.

Posted by: Kenny Pearce at August 7, 2018 10:43 PM

Just a superficial thought in passing (nice piece, by the way); you say:

Sitting unimpeded in an airplane cockpit with all of those buttons and dials and so forth in front of you doesn't give you control over the plane unless you know what happens if you push this button and also what happens if you don't.

It seems to me that if I know what will very probably happen if I do or don't (the usual state of affairs), and if I also have three wishes, then I am much better off than if I know what will definitely happen but have no such wishes. So the main thing is actually the wishes (and what we see in the Bible is the occasional use of a miracle).

Posted by: Mr Martin Cooke at August 8, 2018 3:33 AM

Hi Martin,

Yes, I thought about discussing probabilistic knowledge, but I thought that would complicate matters unnecessarily. A lot depends here on (1) what kind or degree of providential control one wants to preserve, and (2) what view of divine interventions in the natural order one has. There are a lot of possible alternative views here.

Posted by: Kenny Pearce at August 8, 2018 7:36 AM

"God upholds or sustains the causal order in being. The nature of this relation is such as to allow God non-causal control over all of the events in the sequence, but God does so from outside and so is not 'a cause among causes".

Tell me if I'm misunderstanding this. I like the analogy with creating fiction, but isn't there this disanalogy? The fiction is created from the author's imagination. So the author brings it into existence. God's work of sustaining the causal order is not (is it?) creating the causal order (in the way that the author is creating the fiction). The causation, for instance, when A causes B to move, is something that God sustains, but there must be something there (viz., the causal connection) that gets sustained. God is not creating the causal connection, just sustaining it.

Posted by: Mike at August 8, 2018 11:06 AM

God both creates and sustains the causal order. I myself (in "Foundational Grounding") don't draw any distinction between creating and sustaining: on my model, God's creative act constitutes History (the total causal sequence) so that same act is needed to hold History in being. Another comparison I use is to a dance that exists only as long as the person continues dancing. In this kind of case, the same act creates the dance and holds it in being.

I do think some acausal Thomists have tried to draw a distinction between creating and sustaining, but note that if someone is a serious Thomist (i.e., if someone accepts some of Aquinas's other core commitments) this gets to be pretty tricky, since God is atemporal and perfectly simple, so strictly speaking God performs a single timeless act and that act is identical with God. (Personally, I'm not sure I can make sense of the identity claim.) So I'm actually not quite sure how someone in this general camp who wanted a distinction between creating and sustaining would go about drawing it.

Posted by: Kenny Pearce at August 8, 2018 11:14 AM

Post a comment

Return to