Replies by John Martin Fischer

John Martin Fischer

University of California, Riverside

Editor | Gregg D. Caruso, Corning Community College, SUNY (USA).

Correspondence | John Martin Fischer, University of California, Riverside; Email: [email protected], [email protected]

Citation | Fischer, J.M. (2017). Precis, Our Fate: Essays on God and Free Will (New York: Oxford University) Press, 2016). Science, Religion and Culture, 4(2): 79-101.

DOI | https://dx.doi.org/10.17582/journal.src/2017/4.2.79.101

Reply to Griffith

I am very grateful for the extremely thoughtful and generous essay by Meghan Griffith. She raises various important issues, which are worth thinking seriously about, but I want to focus on what I take to be her main point. Griffith agrees with me that it is plausible that God’s foreknowledge is incompatible with human freedom to do otherwise. She also agrees with me that this still leaves the question of whether God’s foreknowledge rules out acting freely (and moral responsibility). I defend semicompatibilism about God’s foreknowledge and moral responsibility; thus I hold that acting freely and moral responsibility are compatible with God’s foreknowledge, even if God’s foreknowledge rules out freedom to do otherwise.

Griffith is more circumspect here. She suggests that God’s foreknowledge would not in itself eliminate acting freely and moral responsibility, but she argues that adding causal determinism to God’s foreknowledge would. This is because causal determinism implies that there is an “independent” source of the action that is external to the agent, and God’s foreknowledge constitutes something akin to manipulation. She emphasizes the distinction between “mere” (as I would put it) causal determination and determination that starts from the specific intentions of an agent (human or divine). She contends that we are in general more troubled when there is an agent at the beginning of the sequence leading to the action under consideration that when there is no such agent (and the sequence starts naturally).

In explaining her position here, Griffith writes:

It can be hard sometimes, when thinking about divine power, knowledge, and so on, to get away from the feeling that we are being set up. If we are lucky, we are set up to succeed. If not, we may be set up to fail. Determinism on its own doesn’t have these overtones.

… The divine case feels different, presumably because there is a knowing agent involved. Interestingly, these intuitions seem like the reverse of the dialectical situation Fischer presents. As Fischer tells it, one is probably more willing to grant compatibility between the divine and responsibility than between determinism and responsibility. But there is a sense in which it is easier to dismiss the threat of determinism (provided that God is out of the picture). Without an omniscient and omnipotent being, there is less of the feeling of a set up. There is less of a feeling that someone else is controlling us in some responsibility-undermining way. There is less of a feeling that we are being manipulated. (Griffith)

So, on Griffith’s view, both causal determinism and God’s foreknowledge expunge human freedom to do otherwise. God’s foreknowledge (or, perhaps, causal determinism) in itself does not eliminate freedom to do otherwise. But adding causal determinism to God’s foreknowledge provides a package of views that imply the lack of acting freely and moral responsibility; here there is an independent deterministic source of our actions external to us brought about by an intuitively “manipulating” agent. The problem is not just external sourcehood, and it is not just manipulation. The combination is a toxic cocktail.

I argue in the introductory essay in Our Fate that God can have foreknowledge of free human actions in the absence of causal determinism. If this is correct, then even on Griffith’s views about God’s foreknowledge and manipulation, (or control—or being “set up”) by another agent, it is possible that we could have God’s foreknowledge and human free action (in the sense of acting freely) and moral responsibility. I think Griffith would agree with this conditional, and she might even agree with the antecedent (that God could have foreknowledge of free human actions absent causal determination of those actions).

I do not agree with Griffith that causal determinism entails that there are problematically independent external sources of our actions. But put this aside. I’d like to consider more carefully whether God’s foreknowledge would constitute a set-up—one that is responsibility-undermining. Unlike human agents, God is necessarily morally perfect. So if we are set up by God, at least we are being set up by a morally perfect being. This might change one’s intuitive views about the “manipulation” in question. Our intuitions might be influenced by the ordinary sort of manipulation by morally imperfect human beings. Further, the term “manipulation” already has a negative connotation, which may also be influencing us, and I would suggest, distorting our intuitions. I would at least suggest that we use “initial design,” rather than “manipulation.” This seems more accurate, and the term may have fewer negative connotations (at least for some).

In any case, I just don’t think it should matter for moral responsibility whether or not there is an agent at the inception of the relevant sequence (issuing in the action in question). Griffith points out that elsewhere I have given some examples that I take to show that this difference (with respect to the existence of an agent at the beginning of the sequence) does not make a difference (as regards attributions of moral responsibility). (Fischer 2011; for further development of my views about the Zygote Argument and related manipulation—or initial design—arguments, such as Pereboom’s famous Four-Case Argument, see Fischer 2017.)

In this argument for the “no difference” thesis, I start with an ordinary case of sexual intercourse between John (not me, as I’m happily married to Tina!) and Mary in a deterministic world. We here assume that this innocent enough act (it is consensual) leads to Ernie’s conception and then his A-ing thirty years later, where Ernie meets the compatibilist conditions for acting freely. Surely there is no special bar to holding Ernie morally responsible for his action—no bar, that is, except determinism, and I start here with the assumption that semicompatibilism is true. (After all, this is not an argument for semicompatibilism, but for the no-difference thesis.)

Now we add in that John and Mary intend to have sexual intercourse in order to conceive an individual who will A thirty years down the road. That is, we add in a specific intention about Ernie’s subsequent behavior. Suppose they conceive Ernie, and he does the desired act thirty years later. Again, he meets the compatibilist conditions for acting freely; we can assume that he acts “in the same way” as in the first case. Let’s say that in both cases Ernie acts from his own, reasons-responsive mechanism (or he “identifies” with the desire that motivates him, or that his values line up with this desire, and so forth). I think that Ernie is just as responsible in this version of the case as in the first. He is morally responsible in both cases, because the relevant parts of the history of the actions are the same. The relevant parts of the history are proximal, not radically distal. The addition of John and Mary’s intention makes no difference, in my view.

At this point it is helpful to recall a passage from Griffith:

Even though John and Mary believe their action will lead to Ernie’s action 30 years later, and even though they intend that their action do so, it is difficult to see them as having manipulated anything—unless, of course, we were to fill in the scenario to make them just like the divine [agent]. Why do John and Mary believe what they do? Do they have any evidence? If not, this may explain why it is difficult to see them as having manipulated the result. …The divine agent has full justification for the belief that Ernie will do what he does in 30 years and knows that what Ernie does will be the inevitable outcome of at least one of the divine agent’s actions. So my question is whether we might think that Ernie is still responsible in the John-and-Mary case because we do not really have any indication that their intention can control the eventual action. … In the divine case, we have a being with full knowledge and considerable power, thus we have reason to think the intention of this agent can control the eventual action. (Griffith)

But imagine now that John and Mary know that the world is causally deterministic, and they actually know the complete state of the universe (or the relevant parts) and the laws of nature, so they can tell exactly when to have intercourse in order to produce Ernie. They have intercourse at that exact time, intending to produce Ernie, an agent who will A thirty years later. They specifically intend that their intercourse will lead to Ernie’s A-ing thirty years later. We can assume that Ernie’s (or any agent’s) A-ing thirty years later is counterfactually dependent on John and Mary’s sexual intercourse at exactly the time in question. So it seems that they can “control” the results, just as God could. But my intuition is that adding this information about their evidence and their capacity to control the relevant outcome in no way changes the fact that Ernie is fully morally responsible for what he does thirty years later. Again, Ernie acts from his own, reasons-responsive mechanism, and John and Mary’s intentions, together with their sufficient evidence about the future, do not seem to be relevant to Ernie’s moral responsibility. They are radically distal, and do not affect what seems to be intuitively relevant about the history of Ernie’s action.

There is a sense in which John and Mary “control” Ernie in this version of the case, and thus they are at least in part morally responsible for Ernie’s A-ing thirty years later. But this does not diminish Ernie’s full moral responsibility for his action. As Harry Frankfurt has pointed out, one can be fully, although not solely, morally responsible for an outcome. In this case Ernie is fully, if not solely, responsible for A-ing.

Ernie is fully morally responsible because he acts from his own, reasons-responsive mechanism, just as he does in the first two versions of the case. And there is no responsibility-relevant difference between the first two versions and the third. In my view, moral responsibility is a matter of the way the relevant action is produced, not whether the agent has alternatives available to her. (This is an “actual sequence” approach to moral responsibility.) The moral of the story (or stories) above, for me at least, is that moral responsibility is a matter of the proximal, rather than distal, way the action is produced. Moral responsibility is a matter of the history of the action—how the actual sequence unfolds—where the relevant part of the history is proximal, rather than distal (and, especially, radically or remotely distal). Specifically, the act of Ernie’s A-ing has the same proximal history in all three cases. That is, the action takes place in the same way, in responsibility-relevant respects, in all three cases. Given that there is no special bar to concluding that Ernie is responsible in the first case, it follows that there is none in the third.

Since this version of the John-and-Mary case is just like a similar case in which God is at the inception of the sequence, I conclude that God’s foreknowledge would not rule out moral responsibility. His foreknowledge, together with His providence, provides no special or additional problem, when considered in conjunction with causal determinism. Incidentally, the same clearly holds for the Goddess Diana, in Mele’s famous Zygote Argument scenario. (Mele 2006) John and Mary are just like Diana, although perhaps they have more fun in starting the sequence in motion!

In my view, causal determinism per se does not crowd out acting freely and moral responsibility. I thus do not believe that an independent deterministic source external to the agent threatens her acting freely (and moral responsibility). (I seek to justify this view, at least in part, in my response to Hunt below.) Finally, I do not think that adding in God’s foreknowledge makes a difference. I accept the no-difference thesis, according to which there is no difference, as regards moral responsibility, between sequences begun by agents with specific intentions and blind nature, so to speak. This kind of “no-difference” thesis is widely shared, and it is a crucial premise in the famous (or infamous) Manipulation Argument for incompatibilism about causal determinism and acting freely. It is interesting that Griffith denies the no-difference thesis in aid of arguing for the incompatibility of God’s foreknowledge and acting freely (moral responsibility) in the context of causal determinism.

Compatibilists in general start with the intuition that it is at least plausible that we are morally responsible in this world. Most compatibilists are willing to acknowledge that, for all we know, our world was created by an essentially omniscient God. Upon reflection, it does not seem to compatibilists that how the world was created thousands of years ago should affect the basic intuition that we can be morally responsible. A compatibilist can certainly entertain the possibility that God created the world, intending that it be the best of all possible worlds, and having specific intentions about all of its features. The compatibilist will bite the bullet and argue that this is not as unpalatable as one might have thought. As Mark Twain said about Wagner’s music, “It’s not as bad as it sounds.”

But I do not think that it is obvious and uncontentious that we should side with the compatibilist here. I do side with the compatibilist, but I recognize that reasonable people can disagree. One’s intuitions about the role of distal creators seem to divide those antecedently inclined toward compatibilism (“natural compatibilists,” like me) and those inclined toward incompatibilism (“natural incompatibilists,” like Griffith). I am grateful to Griffith for bringing out and exhibiting this theoretical divide in a clear way.

Reference List

• • Fischer, John Martin. 2011. The Zygote Argument Remixed. Analysis 71: 267-72.
• • Fischer, John Martin. 2017. Responsibility, Autonomy, and the Zygote Argument. The Journal of Ethics 21:223-38.
• • Mele, Al. 2006. Free Will and Luck. New York: Oxford University Press.

 

 

Reply to Hunt

Hunt’s comments are (predictably) thorough, insightful, and sympathetic. I start with his suggestion that God can play an equivalent role to that of Black in the Frankfurt-style Cases (FSCs). After all, God would be a “factor” that ensures that the agent (Jones) cannot choose or do otherwise, and yet He does not play any role in the unfolding sequence leading to the action. When a factor has these characteristics in a given scenario, it seems irrelevant to the agent’s moral responsibility, although its presence renders the agent unable to do otherwise. Hunt suggests that the context of divine foreknowledge provides a better counterexample to (PAP) than any of the (other) Frankfurt-style cases. Of course, God is not a preemptive over-determiner (a failsafe device), as in the FSCs, but arguably He plays the same role as regards moral responsibility and alternative possibilities. (For this suggestion and helpful critical discussion, see Widerker 2000.)

I agree with Hunt’s suggestion, and I’m very grateful for it. It seems to me to be exactly right. I think prudence requires that we PAP-deniers have a plausible defense of the FSCs, as they are developed by Frankfurt and his followers—with their signature structure involving preemptive overdetermination—in the context of causal determinism. Elsewhere I have sought to do precisely this. (Fischer 2010) But I also think we should welcome Hunt’s suggestion about the context of divine foreknowledge.

Given the Pike-style argument in its various regimentations, it is very plausible that divine foreknowledge rules out human freedom to do otherwise. When people have considered the possibility that God functions in relevant respects like Black in an FSC, they have worried that God cannot know the future actions of human beings without relying on causal determinism to provide the required evidence of the belief in question. But they have thought that it would be question-begging or otherwise dialectically inappropriate simply to assert that Jones is morally responsible in an FSC. After all, Jones’s behavior would (on the current assumption) be causally determined, and many would resist the idea that a causally determined agent could act freely and be morally responsible. But in the introductory essay to Our Fate (summarized in the Precis), I have sought to address this worry; I have offered an account of God’s foreknowledge of future human actions in an indeterministic world—the bootstrapping view. Let’s assume, here, that the account “works”, and that God can have such knowledge.

Given the assumption of God’s existence (conceptualized as above) and the soundness of the argument for theological incompatibilism, it would follow that Jones cannot choose or do otherwise. Indeed, Jones has no alternative possibilities at all—no flickers of freedom. This is a tremendous advantage of the context of divine foreknowledge as a generator of counterexamples to PAP. In the standard dialectical context (in which we evaluate the Frankfurt-style counterexamples), where we make no assumption about causal determinism or God’s foreknowledge, we have to address the fact that we cannot eliminate via an FSC all alternative possibilities. Those pesky flickers of freedom keep popping up. There is a huge (and daunting) literature on this subject—the flickers keep getting fanned.

On the bootstrapping view conjoined with the Pike-style argument, the flickers are entirely extinguished, and it would also seem that Jones acts freely and is morally responsible. (To reiterate: crucial to my argument here is the separation of freedom to do otherwise and acting freely—the freedom component of moral responsibility.) As Hunt puts the view of the PAP-denying defender of theological compatibilism, “Look, divine foreknowledge isn’t in any way interfering with, or even explaining, the person’s action. Despite the individual’s lack of alternatives, this is one free agent!”

So I agree with Hunt that the context of divine foreknowledge provides the resources for an argument against PAP. This is deeply similar to the FSC-based argument against PAP, but better. Further, this context provides a strong plausibility argument for the compatibility of God’s foreknowledge and moral responsibility. David, Thank you for your service--to semicompatibilism. In previous work, Hunt has called this sort of view (about the relationship between God’s foreknowledge and human moral responsibility, “the Augustinian View.” (Hunt 1996) I embrace this, and in my opinion, it is appealing to invoke Augustinianism in support of semicompatibilism about causal determinism and moral responsibility.

But of course more argumentation would be needed to establish semicompatibilism about causal determinism and moral responsibility. Most obviously, we need to address the “source ncompatibilist,” who takes it that if causal determinism were true, the source of our behavior would not be “in us,” in the sense required for moral responsibility. (A related, but slightly different, worry comes up for God’s foreknowledge.) I wrote in the introduction to Our Fate that the context of divine foreknowledge could point us toward a response to the source incompatibilist, who believes that causal determination would imply that the source of the action would be external to the agent (and out of her control). I observed that in the context of divine foreknowledge, God’s belief is a sufficient condition for the action in question, and yet it is external to the agent (and presumably out of her control). If we think that, nevertheless, the agent can be morally responsible, then the simple view of sourcehood has to be sharpened, in order to get to the notion of sourcehood implicated in the incompatibilist’s argument (about God’s foreknowledge and moral responsibility). We can then evaluate it within the debate about the relationship between causal determinism and moral responsibility.

I suggested that the source incompatibilist will at least need to advert to “explanatory factors” external to the agent and out of her control. That is, the source incompatibilist should require that there be no factor external to the agent and out of her control that helps to explain the action. God’s belief does not explain the action; it is the other way around. In contrast, a causal factor along the deterministic path to the choice and action may well be explanatory. Hunt helpfully reminds us that I need to say more about sourcehood, and, in particular, about the suggested notion.

Let me try. The first thing to note is that explanation is a pragmatic notion; the adequacy of an explanation is (in my view and many others’) always relative to a particular context of inquiry involving a particular set of interests. There is no such thing as “the explanation” of an action tout court. But we can put this aside, implicitly supposing that we are thinking of the appropriate context (of moral responsibility attribution).

Let’s say causal determinism is true. And let’s further suppose that you suggest to me that we go to dinner and a movie tonight. If I agree, it might well be that your suggestion explains our (and a fortiori my) act of going to dinner and the movies. Your making this suggestion is external to me and, we can assume, out of my control. And yet I think that nothing has been invoked so far to suggest that I didn’t act freely or that I couldn’t be held morally responsible. Or, better, nothing that is uncontroversial and uncontested within the dialectical context of an evaluation of the compatibility of causal determinism and moral responsibility has been thus far invoked. In my view, explanatory factors in themselves, just in virtue of their being explanatory, do not threaten moral responsibility; and this doesn’t change if we assume that the context is causally deterministic.

Perhaps when we are thinking that such a factor exculpates, we are thinking that it is a factor that explains the behavior in question in part by adverting to a responsibility-undermining factor. These are special kinds of explanations. I think this is right, and I think that both compatibilists and incompatibliists should agree. A factor’s being explanatory does not eo ipso threaten moral responsibility, even in a deterministic world and the factor is external and out of the agent’s control. It is only when the factor explains the behavior via positing an external responsibility-undermining factor (such as brainwashing, irresistible urges, phobias, manipulation, and so forth) that it threatens moral responsibility.

Of course, it is contested in the envisaged dialectical context whether an explanatory factor’s being deterministically related to the behavior is responsibility-undermining. My point in the book was that God’s prior belief is an external factor out of the agent’s control that is sufficient for the behavior, and thus the source incompatibilist (about causal determinism and moral responsibility) cannot point to this sort of factor as ruling out moral responsibility, insofar as we take it that God’s foreknowledge is compatible with human moral responsibility. That is, we start with the assumption (at this point in the dialectic) that God’s foreknowledge is compatible with human moral responsibility. Given that God’s foreknowledge constitutes an external factor out of an agent’s control, this cannot be what rules out moral responsibility in the case of causal determinism.

Further, I pointed out that the source incompatibilist might refine his notion of sourcehood to include explanatory factors. Here I have sought to show that a factor’s being a deterministic explanatory factor external to and outside the control of the agent does not necessarily threaten the agent’s moral responsibility. The source incompatibilist needs to contend that it is only when a deterministic factor that is external to the agent and out of his control explains an action via positing a responsibility-undermining factor that the agent is not morally responsible. Perhaps this can be simplified to: when a responsibility-undermining factor is on the route to the action, and helps to explain the action, then the agent is not morally responsible for the action. (This is source incompatibilism: it leaves it open that factors that occur (or obtain) and rule out access to alternative possibilities, but are not explanatory—are not on the route to the action—do not threaten moral responsibility.)

Now of course we have to figure out whether causal determination in itself and apart from ruling out alternative possibilities is responsibility-undermining. It might seem that we haven’t made any progress. I agree that we haven’t made much progress, but even a little is warmly welcomed in thinking about these ancient issues. I suggest that recognizing that God’s prior beliefs are external, out of the agent’s control, and sufficient for his behavior, at least nudges one toward rejecting source incompatibilism. At the least, it should nudge us toward sharpening the notion of sourcehood that plays such a crucial role in debates about the relationship between causal determinism and moral responsibility. In the end, we need to figure out whether causal determination in itself is a responsibility-undermining factor. (I noted this point in the last sentence of my very first article on moral responsibility [in Ancient Times]: Fischer 1982. There I wrote, “Both the compatibilist and the incompatibilist alike can unite in conceding that enough information is encoded in the actual sequence to ground our responsibility attributions; as philosophers we need to decode this information and see whether it is consistent with deterministic causation.” Much of my work since then has sought to show that it is: examples are Fischer 1994 and Fischer and Ravizza 1998,)

Finally, Hunt highlights a mystery (that had not gone unnoticed by me) in the bootstrapping account of God’s foreknowledge of free human actions in an indeterministic world. Hunt expresses the worry well:

Suppose that God is in a KCS [knowledge-conferring situation] with respect to Jones’s mowing [his lawn] tomorrow, and he’s also in a KCS with respect to Jenkins’ mowing tomorrow—a KCS at least equal in its knowledge-conferring power to God’s KCS with respect to Jones’s mowing tomorrow. Suppose further that Jones will mow tomorrow, but Jenkins won’t. God must believe (and therefore know, with certainty) that Jones will mow, on pain of knowing less than human beings (e.g., Jones’s neighbor) might know; but God mustn’t believe that Jenkins will now, on pain of believing falsely. How does God know that he should believe in the first case and not in the second? (Hunt)

A mystery indeed. My view is that God just finds Himself believing that Jones will mow, but not that Jenkins will mow. There is no evidence-based, or other internally accessible, difference for God between the two situations. In a sense, the difference in God’s beliefs is a bare difference. How does God believe that Jones, but not Jenkins, will mow? How does He do it? It is a mystery.

In Chapter Four of Our Fate, I distinguish between two ways in which one might answer the “how” question: a “philosophical explanation” and a “nuts-and-bolts” explanation. Taking a bit of liberty with the characterization I previously offered, we can understand a “philosophical” explanation to explain how (say) the bootstrapping view of God’s foreknowledge in an indeterministic scenario is not incoherent, how it fits well with other doctrines, and perhaps helps to solve important philosophical problems. In contrast, a “nuts-and-bolts” explanation is the kind one might find in a “how-to” manual or cookbook: a step-by-step “recipe”. I have offered a philosophical explanation, but not a nuts-and-bolts explanation, of God’s foreknowledge of future actions of human beings in an indeterministic world. More specifically, we do not have a nuts-and-bolts explanation of how God is able to believe that Jones, but not Jenkins, will mow his lawn tomorrow. What we do have is a general account of God’s foreknowledge of future human actions—its philosophical requirements and implications.

I pause to note here that the bootstrapping view might be deemed a nuts-and-bolts type of explanation, rather than a philosophical explanation. So someone might conclude that we do in fact have a nuts-and-bolts explanation, but not a philosophical explanation. That is, the worry would be that the explanatory situation is exactly the opposite of what I have suggested. I would however resist this view. We can interpret the bootstrapping view “objectively,” rather than “subjectively.” I explain this distinction in my reply to Rhoda below, but here it suffices to note that the bootstrapping view can be put in terms of what is (logically or analytically) required for God’s knowledge of future human actions, rather than how God subjectively gets to this knowledge. That is, it can be formulated as an account of, or explanation of, how God’s beliefs about future human actions can count as knowledge, rather than how God’s mind “works” in order to generate the belief in question.

We clearly lack a nuts-and-bolts explanation of God’s differential beliefs about Jones and Jenkins. This issues in a mystery. I concede this point. But I would claim that this sort of mystery—generated by the lack of a nuts-and-bolts explanation—is less disturbing than the mystery associated with the lack of a philosophical explanation of some phenomenon.

The situation is parallel in relevant respects to that of the atemporalist interpretation of God’s eternality. On this view, God is outside time (or outside of the temporal framework) in which we humans exist). Although some have contended that this interpretation is incoherent, I have been convinced by excellent work by contemporary philosophers, especially Eleonore Stump and Norman Kretzmann, which shows that we have a philosophical explanation of God’s atemporal eternality, and, in particular, His ability atemporally to know contingent truths about human behavior at all times, as well as His ability atemporally to will effects in the human temporal framework. (Stump and Kretzman 1981 and 1985) Stump and Kretzmann give a rigorous analysis of God’s atemporality, including a defense against the contention that it is incoherent (based in part on transitivity issues).

But still they do not offer nuts-and-bolts explanation of God’s atemporal willing of temporal effects, or His atemporal knowledge of continent truths involving human actions. For instance, when does He will the effect, given that there is no “when” for God, no way for the mode of His existence to relate to ours within the same framework? So how can God (atemporally) make the effect happen at a particular time in the human spatio-temporal world? One can posit inter-framework causation, and this might be philosophically coherent, but it still leaves unanswered the question of “how God does it”. How does God’s mind work to empower Him to make a certain effect happen at a specific time? It is hard to understand the nuts and bolts here—how God could have a recipe for creating effects at specific times in our human world.

So God’s atemporality faces a similar problem to the problem faced by the bootstrapping view: the lack of a nuts-and-bolts explanation (of the relevant phenomenon), but the presence of a philosophical explanation. Again, I think this is less troubling than the lack a philosophical explanation. Perhaps we just should not expect a nuts-and-bolts explanation of the workings of a being so different from us. In any case, I concede the mystery, but I still think the bootstrapping view should at least be taken seriously as a contender, just as we are willing to take seriously an atemporal interpretation of God’s eternality. The atemporal conception of God’s eternity has certainly been taken seriously for millennia. Some might conclude from the considerations offered above that the bootstrapping view is no better than the atemporal interpretation. I am inclined to agree, but I also hold that it is no worse. The glass is half-full.

As I wrote above, I believe in general that the lack of a nuts-and-bolts explanation is less worrisome than the lack of a philosophical explanation. Further, I wish to suggest that the stronger and more powerful the philosophical explanation, the easier it is to bite the bullet with respect to the lack of a nuts-and-bolts explanation. Recall that part of a philosophical explanation is identifying the implications of a view, and its philosophical connections. And this aspect of the philosophical explanation of the bootstrapping view is very powerful. First, it makes sense of the Ockhamist position—perhaps the dominant view of contemporary theological compatibilists—that, whereas causal determinism rules out human freedom to do otherwise, God’s foreknowledge does not. Additionally, as I’ve argued above, following David Hunt’s suggestion, God’s foreknowledge (on the bootstrapping view) provides a context in which we can vindicate the Frankfurt-Style Counterexamples (FSCs) to the Principle of Alternative Possibilities (PAP). We could thus make significant progress in defending compatibilism about God’s foreknowledge and human moral responsibility, and at least some progress in defending compatibilism about causal determinism and human moral responsibility. This is a big deal. Given these striking implications of the bootstrapping view, I can bite the bullet of the lack of a nuts-and-bolts explanation without intolerable philosophical indigestion.

Reference List

• • Fischer, John Martin. 1982. Responsibility and Control. Journal of Philosophy (79): 24-40.
• • Fischer, John Martin. 1994. The Metaphysics of Free Will: An Essay on Control. Oxford: Blackwell Publishers.
• • Fischer, John Martin and Mark Ravizza. 1998. Responsibility and Control: A Theory of Moral Responsibility. New York: Cambridge University Press.
• • Fischer, John Martin. 2010. The Frankfurt Cases: The Moral of the Stories. Philosophical Review 119: 315-36.
• • Hunt, David. 1996. The Frankfurt Cases: Some Comments on the Widerker-Fischer Debate. Faith and Philosophy 13: 395-401.
• • Hunt, David. 1999. On Augustine’s Way Out. Faith and Philosophy16: 3-26.
• • Stump, Eleonore and Norman Kretzmann. 1981. Eternity. Journal of Philosophy 78: 429-58.
• • Stump, Eleonore and Norman Kretzmann. 1985. Absolute Simplicity. Faith and Philosophy 2: 353-82.
• • Widerker, David. 2000. Theological Fatalism and the Frankfurt-Style Counterexamples to the Principle of Alternate Possibilities. Faith and Philosophy 17: 242-54.

 

 

Reply to Bernstein

I am grateful to Mark Bernstein for his very generous, and also challenging, comments. He focuses on my suggested principle of the fixity of the past, as formulated in terms of possible worlds:

(FP*) An agent can at T do X only if there exists some possible world with the same past relative to T as the actual world in which the agent does X at T.

I further refine (FP*), or provide the proper interpretation of it, by restricting the relevant facts about the past (or features of the past) to hard facts (or features). Here I will simplify the formulation by referring only to hard facts, leaving out other hard features:

(FP*) An agent can at T do X only if there exists some possible world with the same past (hard facts) relative to T as the actual world in which the agent does X at T.

On my approach fixity only applies to the hard facts about the past. The hard facts are those that are temporally nonrelational (as regards the future) relative to a time. Intuitively, the hard facts about the past are genuinely and solely about the past; they do not include a part that is about the future (in a genuine sense). It is very complicated and difficult to give a reductive analysis of the “genuine” sense at issue, and thus of hardness, but we don’t need to seek to do this for our purposes here. It is enough to keep in mind that fixity only applies to hard facts about the past, in the intuitive sense that we can recognize as genuinely about the past and not also about the future.

Additionally it is important to distinguish between the concepts of fixity (it being out of an agent’s control or power) and hardness (temporal nonrelationality). A compatibilist about causal determinism and freedom to do otherwise may deny (FP*); I have called such a compatibilist a “multiple-pasts” compatibilist. (Fischer 1994) So fixity and hardness are separate (although related) ideas. I find (FP*) intuitively plausible, and thus I adopt it for the sake of the analysis and evaluation of the theological incompatibilist’s argument. To put it in Carl Ginet’s terms, (FP*) expresses the very intuitive idea that our freedom is the power to extend the actual (hard) past.

Is (FP*) question-begging? Bernstein points out that a compatibilist might deny (FP*). As I observed above, a multiple-pasts compatibilist does just this (in the context of the debate about the relationship between causal determinism and human freedom, and, in particular, the consequence argument). I do not know how to establish (“beyond a reasonable doubt”) that (FP*) is valid. There are no knockdown arguments in this realm. But at least we should see that denying (FP*) is a considerable cost of some forms of compatibilism.

Bernstein gives a thorough and nice description of Newcomb’s Problem and my (purported) solution, and thus I won’t go into them in detail here. My solution employs the assumptions that God’s beliefs are hard (or have hard features) and (FP*). On my view, it is rational to take the one box, on the assumption that the predictor is infallible (and not merely inerrant). This is in part because whatever choice you make will be the only choice you can make, and you end up better off taking the one box. On the assumption of infallibility, there is no possible world in which the predictor predicts that you will take one box and you take two (or vice versa). The typical formulation of Newcomb’s Problem, in which it is supposed that you have two genuinely open options, would be incoherent, on this view. Still, it would be rational to choose the one box; this doesn’t imply that one has, or believes that one has, more than one metaphysically open alternative. The alternatives that are open are epistemic alternatives: options that are open, for all you know.

But on the assumption of mere inerrancy, there are possible worlds in which the predictor is wrong, and thus you can coherently be understood to have two genuinely open options: taking one box or two. Now a dominance argument implies that you should take the two boxes. You consider sequentially how you would fare if you were to choose the one box (under each scenario with respect to the predictor’s prior behavior) and then how you would fare if you were to choose the two boxes (similarly under each scenario pertaining to the predictor). You are better off under each scenario of the predictor’s prior behavior if you choose the two boxes.

This is an asymmetric solution (with respect to infallibility and inerrancy). It is rather unorthodox, but I think it is defensible, once one adopts (FP*). This approach eschews the various strategies that emphasize the importance of evaluating the relevant backtracking counterfactuals, which is notoriously difficult and leads to apparently unresolvable differences in intuitions. My solution points to the importance of the difference between infallibility and inerrancy. One might be struck by the significance (on my approach) of what might be thought to be a small difference in assumptions. But the analysis shows that this is not a small difference. Or perhaps it shows that small differences in initial assumptions make a big difference to outcomes; this could be considered a logical version of the “butterfly effect” (brought out by chaos theory).

I am fond of this strategy of thinking about Newcomb’s Problem and similar puzzles that involve “backtracking” (or a certain distinctive relationship between the present and past, and thus future). But Bernstein objects:

For Fischer’s resolution to work, we must take the possible worlds in which predictor is mistaken to have pasts identical to the actual world in which Predictor correctly predicts… This may seem unproblematic. … But what prevents someone from claiming that in the actual world the past contains the fact that I will make the selection that the Predictor predicted, or, equivalently, that the past contains the fact that the predictor’s prediction is (or will be) true? If this fact is considered a fact in the past where this is distinguished from the question of whether an action in the future makes this a fact or not (where, we might say, that a fact ‘in the past’ is distinguished from a fact ‘of the past’), then the pasts of the actual world and the pasts of the possible worlds in which Predictor makes mistaken predictions are not identical. … [Thus] (FP*) would imply that we cannot perform any action that we don’t perform, and, more to the point, that we could not make a selection other than the one we actually make. (Bernstein)

But. What prevents someone from “claiming that in the actual world the past contains the fact that I will make the selection that the predictor predicted, or, equivalently, that the predictor’s prediction is (or will be) true” is that the fact in question is not hard. It is a fact about the past—is “contained in” the actual past”—but it is not subject to (FP*)—it is not a fact only about the past. Clearly, the fact that the predictor’s prediction (about the future) is or will be true is not a hard fact about the past: it includes something about the future. This is a different fact from the fact that the predictor makes a prediction in the past. Similarly, the fact that God’s belief about the future will be true is not a hard fact about the past, but the fact that God held a certain belief is (arguably, at least) a hard fact about the past (or a fact with a hard aspect, but leave this complication aside). And the same point applies to the fact that I will make the selection that the predictor predicted (the fact Bernstein claims is equivalent to the fact that the prediction “is or will be true”). Note that human knowledge, insofar as it entails truth, is a soft fact about the past; but human belief, the fact that a particular human being had a particular belief—is a hard fact about the past.

Bernstein distinguishes a fact “in the past” from a fact “of the past,” but he doesn’t really offer an explanation of this distinction (or perhaps I simply don’t get it). He points out that some past facts are made true (made facts) by actions in the future. This is one way of understanding soft facts: they are dependent on the future in a distinctive way. But if this is so, then, again, we can see that the facts adverted to by Bernstein are soft facts: the fact that the prediction about the future will be true, or that I will make the selection the predictor predicted. They are thus not subject to (FP*). The same analysis applies to the fact that the Yankees won the 1927 World Series [if they did—Bernstein is more a scholar of Yankee lore than I!]. This is a hard fact about 1927, and thus (FP*) deems it fixed now. Neither Mark nor I nor anyone else can now do anything about the fact that the Yankees won the 1927 World Series: all possible worlds to which we have access (in the sense relevant to our powers) contain the 1927 Yankees World Series win in their pasts.

Bernstein points out that “the ‘can’ of personal capacity, ability, or power is just too elastic to be fully captured by (FP*)”. I fully agree. I distinguish between general capacities (powers and abilities) and a power referred to by “can, in the particular circumstances”. Alternatively, I distinguish general capacities from the power referred to by Austin’s term, “all-in sense of ‘can’”. Although the “can” of general capacity is relevant to moral responsibility, it is typically the “all-in can”, or “can in the circumstances”, that is directly relevant to moral responsibility (on the standard assumption that moral responsibility requires the power to do otherwise). If there is no piano in one’s vicinity (and one didn’t have any control over this fact), one cannot play, in the circumstances, and one is not morally responsible for failing to play. And yet one might have, and retain in the circumstances, the general capacity to play piano. Again: I agree with Bernstein that (FP*) does not give an adequate analysis of general capacities. It only purports to give a necessary condition of any adequate analysis of “can, in the particular circumstances”—the can most directly relevant to moral responsibility (if any “can” is). There is no dearth of proposals for understanding general capacities, a worthwhile and important task—but not the one I set for myself. (For just one example of this “new dispositionalist” project see Vihvelin 2013.)

Finally, Bernstein wonders whether one can sensibly deliberate in a context in which one knows that whatever one chooses will be the only thing she can choose, and where one also knows that one’s deliberations will be similarly necessitated. As he puts it, one knows one cannot escape the “metaphysical clutches” of the infallible predictor. Behind this worry is the picture of practical reasoning and deliberation according to which its point is to select among various paths genuinely open to the agent (and, indeed, known by the agent to be genuinely, metaphysically open). In contrast, I contend that what’s important to practical reasoning/deliberation is epistemic, rather than metaphysical, openness; it operates within the domain of epistemically open options (that is, paths that are open, for all the agent knows). I hold that the point of practical reasoning/deliberation is to figure out what the agent’s best or “strongest” reason is to follow an epistemically open path, and to seek to conform one’s “will” (choice, decision) to that evaluative judgment. If this is indeed the point of practical reasoning/deliberation, one can sensibly deliberate, even if one knows one has only one path genuinely metaphysically open to her. After all, this does not imply that the agent knows which path it is.

Suppose I am walking up a trail, and I come to a fork. I see on the right fork a rattlesnake, and no dangers on the left. Suppose, further, that I have taken an advanced undergraduate class on free will, and, combining the keen insights from this course with other information, I conclude that causal determinism is true, and that this is incompatible with freedom to do otherwise. Thus, I believe that, whatever I choose, this is the only choice I can make and the only path I can take. Given this knowledge, should I just flip a coin? Should I arbitrarily barrel forward, perhaps taking the rattlesnake path? I suppose I could say to myself that I just have one path open to me, so I should simply flip a coin or arbitrarily select a path—maybe the rattlesnake path. This would be crazy. It makes sense to pause to figure out that one has all-things-considered reason to take the left fork, and to choose this path. This is, you might say, the path of good sense, and even sanity.

Reference List

• • Fischer, John Martin. 1994. The Metaphysics of Free Will: An Essay on Control. Oxford: Blackwell Publishers.
• • Vihvelin, Kadri. Causes, Laws, and Free Will: Why Determinism Doesn’t Matter. Oxford: Oxford University Press.

 

 

Reply to Rogers

Kathrin Rogers contends that an Anselmian “does not find Pike’s version of the ‘Basic Argument’ for the incompatibilism [sic] of divine foreknowledge and a human ‘ability to do otherwise’ a problem.” Part of her piece is a summary of her interpretation of Anselm’s views about time and God’s knowledge of free human actions. The part that engages more directly with my collection of essays is her critique of Pike-style arguments for the incompatibility of God’s foreknowledge and human freedom to do otherwise. She finds the argument (and similar arguments for incompatibilism) obviously problematic and easy to dismiss. She holds that the argument doesn’t even get off the ground for “logical” reasons. Thus, as I interpret her, Rogers contends that the incompatibilist’s argument is just plain invalid, quite apart from Anselmianism. Her arguments would seem to me to imply that, for reasons that don’t depend on her development of Anselmian views, the incompatibilist’s argument is “not a problem”.

This would be a big deal, if it were right, but I can’t see why Rogers thinks she has dispatched this ancient argument so quickly and painlessly. She writes:

It is true that, if S chooses B at t2, S cannot fail to choose B at t2, but logic provides that consequence. If an “ability to choose otherwise” entails that an agent who chooses X at a time, not choose X at that time, then no one can choose otherwise, by the law of non-contradiction.” (Rogers)

First, note that this point would apply to action in general, and it has nothing to do with worries stemming from prior truths about putatively contingent truths about the future (fatalism), God’s foreknowledge, or causal determinism. For instance, even if God does not exist or does not have foreknowledge, the “logical” point would apply: if an agent does something at a time, she cannot do otherwise at that time. So it is an extremely blunt instrument, and has nothing specifically to do with God’s foreknowledge and its relationship to human freedom. Or so it seems to me.

Second, Rogers’ argument is specious. An ability to do otherwise (in the relevant sense of “ability”) does not entail that an agent who chooses X at a time, not choose X at that time. That is, it does not imply that the individual has it her power to bring about a situation in which she both chooses X and does not choose X at a given time. This would be logically problematic, but to suppose this is to fall into a well-known fallacy about the ability to do otherwise. The ability to do otherwise implies that, although the agent performed a certain action, she could have performed another action instead. It does not imply that the individual could both choose and not choose an action, or both perform and not perform it.

Perhaps I am misinterpreting Rogers, but I can’t see how to interpret her in a different, less problematic, way. An ability to do otherwise points us to a different possible world. If in the actual world S does X at T, then, if S has the ability to do otherwise, there must be a possible world (suitably related to the actual world) in which S does not do X at T. An ability to do otherwise (in the sense of Pike and the ancient argument, and, in particular, its contemporary regimentations) does not imply that there is a single possible world in which S both does and does not do X at T2. The mistake stems from a failure to see that the claim involves features of a structured set of possible worlds, rather than a single possible world.

Rephrasing this point: to say that an individual who does X has the ability (again, in the relevant sense) to do otherwise, does not imply that the individual has the power to bring it about that: he does X and does not do X. Rather, it is to suppose that such an individual, who actually does X, nevertheless has it in his power not to do X. There is no “logical” contradiction here. Of course, there are worries about the possession of this power that stem from various sources, but not the one invoked by Rogers. It is no surprise that a theorist who believes that no one could ever do otherwise, for logical reasons and quite apart from any assumption about God, would not find Pike’s argument a problem. But this is neither here nor there. In my opinion, any fair and reflective person who considers the incompatibilist’s argument should deem it a problem, even if, in the end, the problem is not insuperable.

Analyzing a part of the incompatibilist’s argument, as I have presented it (in its “conditional” form), and assuming that S chooses B at T2, Rogers considers the following three conditionals and reflects as follows (I employ the technical formalism of Rogers, in which the temporal variables are not capitalized):

1. 1. If S were to refrain from choosing B at t2, then God would have held a false belief at t1.
2. 2. If S were to refrain from choosing B at t2, then God would not have existed at t1.
3. 3. If S were to refrain from choosing B at t2, then God would have held a different belief from the one He actually held at t1, i.e., God would have believed at t1 that S would refrain from choosing B at t2.

Here is how the Anselmian would analyzes this argument: The three conditionals (1-3) are all true. … If S chooses B at t2 then, by the law of non-contradiction, S does not refrain from choosing B at t2. It is true that S cannot choose otherwise at t2, but it would be odd if one had to violate logic in order to be free. (Rogers)

Yes, it would be odd if freedom required the violation of logic, but, as above, it does not. And how could all three propositions be true? How could (1) be true? If God is essentially omniscient, He can’t have a false belief. It would seem then that the conditional would have to be false, given that its conditional is necessarily false. But is Rogers supposing that the antecedent is necessarily false, and therefore the consequent is true? But the antecedent is not necessarily false, even if we rewrite it as:

1*. If S actually does B at t2 but S were to refrain from choosing B at t2, then God would have held a different belief from the one He actually held at t1, i.e., God would have believed at t1 that S would refrain from choosing B at t2.

Howcould (2) be true? Is God’s existence not counterfactually independent of human action? This is the traditional idea of God’s aseity. Can I really determine whether or not God exists by my actions? Rogers simply writes that the Anselmian would consider all three conditionals true, without much (if any) explanation. If the whole point is simply the logical point—if this is enough on its own to show that the incompatibilist’s argument “is not a problem”—then why the discussion of these conditionals, in the various versions Rogers considers?

One final point about Rogers’ view that the incompatibilist’s argument (as presented by Pike and regimented in various ways in Our Fate) is “not a problem” for the Anselmian (of which club, I take it, she is a member). I pointed out that one could rephrase all of the versions of the argument by invoking a time-indexed ability indexed to a time prior to the action in question. It is well-known that precise attributions of (particularized) abilities to perform actions at times need to be double-indexed temporally: there has to be a temporal index for the ability as well as the action. So we might want to know if some agent S can (in the relevant sense) at T1 do X at T2. Of course, T1 might be identical to T2, but it need not be, and I suggested that if one holds that the present is fixed, this does not indicate that the incompatibilist’s argument rests on a logically incoherent supposition, given double-indexation. Rogers dismisses this (putatively) possible move:

And the move to what the agent can do ‘just prior to that time’ is not clear. No one can actually do something prior to the tine at which they actually do it. Say that the fixity of the present entails that if S chooses B at T2, the S cannot refrain from choosing B at T2. Can S refrain from choosing B at T2, at T1? It is hard to see how. (Rogers)

Of course, it would be ludicrous to suppose that anyone could do something prior to the time at which he or she actually does it (unless we are considering act-types). I do not suppose this. Rather, the relevant attribution must be double-indexed temporally. So, the idea is that someone might be able at a prior time to perform a certain action at a subsequent time. Rogers finds it hard to understand how it is possible that S can refrain from choosing B at T2, at T1. This would indeed be hard to understand, if it meant that S can engage in the same token omission at two different times. (Of course, an agent can omit to do B at T1 and then omit to do B at T2. There is nothing puzzling about this, where we are thinking of omission-types, rather than omission-tokens.) If one finds it hard to individuate omissions—or harder than actions—one can think here of acts of omission. The widely-accepted double temporal-indexation of freedom attributions does not entail anything straightforwardly incoherent or incomprehensible. (Lehrer 1976)

Although I have done my best to understand Rogers’ critique of the incompatibilist’s argument and also her Anselmian approach, I have the lingering worry that perhaps I have misinterpreted her or interpreted her uncharitably. If this is so, I very much hope that she will educate me and set me straight in her future work on these topics.

My main interest is in Rogers’ response to Pike’s argument, which, as far as I can see, is independent of the details of her interpretation of Anselm, I do have a thought about eternalism. On this view, all individuals existing at all times are ontologically on a par. Rogers complains that “[I=JMF] do not give it [eternalism, or in her preferred term, “isotemporalism”] the thoughtful treatment it deserves”. There is much discussion of this doctrine in contemporary philosophy of time, and it certainly deserves more attention than I could give it in my essays collected here. It is interesting to note that when Pike wrote and published his seminal essay, there was not much (if any) discussion of the various views in philosophy of time, including eternalism, presentism, and the growing-block theory, and the ontological status of individuals at various times. Nor was there much discussion of metaphysical grounding and how it interacts with the views about time. I think it would be helpful to think more deeply and carefully about these matters, and how they relate to Pike’s argument (and the early responses to it). (I and my co-editor make some efforts to begin along this path in: Fischer and Todd, eds. 2015.)

There are some obvious intuitive problems with eternalism. I don’t know if they are insuperable, and, of course, there are problems with every philosophical view about time. Then again I do not think we should sweep them under the rug. Rogers admits that “isotemporalism is radically weird.” She is not too worried about the implication that the dinosaurs are no less real than we are, but she is worried about the implication that the she of five minutes ago and the she of now are equally real (and “exist” equally). Frankly, the dinosaurs worry me more, since I do believe that the me of now is identical to the me of five minutes ago; otherwise, I don’t see how moral responsibility could make sense. I do however worry about this apparent implication of eternalism (and isotemporalism): the temporal stage or part of me now is identical to the temporal stage or part of me five minutes ago. She points out that many intuitive beliefs are inconsistent with contemporary physics, and it is not advisable always to stick with those intuitions. (She also concedes the limitations of this point, since so much of Anselmianism, and theism in general, is arguably inconsistent with contemporary physics.)

I (and my co-authors in the relevant article/chapter) simply wanted to point out that eternalism is indeed “radically weird”. We did not intend to infer that on balance it is to be rejected, only that its acceptance would come at a high philosophical cost. Once one departs from widely shared considered judgments or reflective intuitions about these matters, where does one stop? What is the check on philosophical theories? Only that they fit with the currently accepted science, especially when what is accepted in science is always changing, and we don’t currently have a “theory of everything”? Again, if this is the criterion, so much the worse for theism (as Rogers realizes).

If President John F. Kennedy and I are equally real and exist equally, then I suppose we don’t have to be sad about his assassination! What a relief, and consolation. But the beneficial effect is short-lived: we would have to lament that Hitler is still around, in the same sense that we are. I guess I (and my co-authors, and many others) don’t think we can blithely accept such results. If we have a choice between accepting a radically weird and unintuitive metaphysical view or “the almost ubiquitous teaching of Christendom,” I know what I would choose. I have no stake in defending Christendom, only fairly and reflectively evaluating its teachings—and going where the arguments take me.

Rogers writes that “… the parties to the freedom and foreknowledge debate tend to subscribe to the view that the one God is three persons in one nature, and that one of the persons has, not only a divine, but a human nature.” (Rogers) But I for one am respectfully skeptical about the existence of God (as are other prominent participants in the contemporary debates), and I must say that the need to accept a doctrine like the trinity doesn’t help. If I have a choice between a radically weird metaphysics or rejecting the doctrine of the trinity, guess what I would choose! If the only way to “save” a view of God that embraces the doctrine of the trinity, as well as the consistency of a perfect being with the nature and extent of evil in our world, is to adopt a radically conterintuitive metaphysics, I don’t think it is worth the price.

Reference List

• • Fischer, John Martin and Patrick Todd, eds. 2015. Freedom, Fatalism, and Foreknowledge. New York: Oxford University Press.
• • Lehrer, Keith. 1976. ‘Can’ in Theory and Practice: A Possible-Worlds Approach. In M. Brand and D. Walton, eds., Action Theory. Dordrecht: Reidel: 241-70.

 

 

Reply to Almeida

Before delving into the main points raised by Michael Almeida, I pause to note a couple of differences between the views of Rogers and his. Recall that Rogers claims that the Anselmian (of whom, I take it, she is one) holds that the three conditionals, (1) through (3), which are part of what I called the “conditional version” of the argument for theological incompatibilism, are all true. This is a fairly central point in her piece, “Anselm versus Fischer”. I am bumfluzzled by her claim, as I wrote above in my reply to Rogers. I note that Almeida writes:

It seems evident that the divine attributes entail that (1) and (2) are false. … (1) is false in virtue of God’s essential omniscience and (2) is false in virtue of God’s necessary existence. It is impossible that God believes a false proposition and it is impossible that God fails to exist. (Almeida)

In my view, Almeida is clearly correct in writing that (1) and (2) are false. Almeida obviously thinks that it would not be a “violation of logic” for an individual who actually does X at T to have the freedom to refrain from doing X at T (where these sorts of claims can be made more precise in the way indicated by Almeida). So the antecedents of the conditionals in question are not necessarily false. I agree that the relevant sort of freedom does not require a violation of logic, even if it is difficult to square with God’s foreknowledge. In reading Rogers and Almeida, one can feel a kind of philosophical whiplash.

The main thrust of Almeida’s contribution to this symposium is to offer an account of the nature of reality (the facts), and thus of God’s foreknowledge, that vindicates compatibilism, or at least insulates it from Pike’s argument. I commend this as an interesting—and even ingenious—move, although I hesitate to trumpet its virtues too much, since I myself presented something like it (and considered it) in Our Fate (Chapter Three). As I believe that the Consequence Argument (pertaining to the relationship between causal determinism and freedom to do otherwise) is importantly parallel to the argument for theological incompatibilism, I have also considered this strategy as a possible reply to the Consequence Argument. (Fischer 1996). Of course, Almeida develops this response in a different way than I do, and he emphasizes the importance of the doctrine of “endurantism” to the compatibilist reply to the incompatibilist’s argument.

I frankly do not see why endurantism is essential to the incompatibilist’s argument or the “world-indexation” strategy of response to Pike-style regimentation of the fundamental argument for incompatibilism. I don’t see why the arguments here couldn’t be interpreted in a way that invokes “temporal parts” of the same individual, and counterparts in other possible worlds. After all, Lewis (and his followers in this respect) think that modal claims about an actually existing individual (or an individual in a given possible world) are properly analyzed in terms of the non-modal properties of counterparts of that individual in other possible worlds.

But let’s leave these issues aside, as I think they are not the most interesting part of Almeida’s suggested strategy. The main idea is that all facts are world-indexed and all truths are thus world-indexed. So, when Jones does X at T2, it only seems like this is a truth about the world (whatever world in which Jones behaves in this way at this time), or, at least, a truth at the most basic level in that world. If the world in question is PW1, then the relevant (or real, fundamental) truth is expressed by a world-indexed sentence, “Jones does X at T1 in PW1”. Similarly, the fact is world-indexed: that Jones does X at T1 at PW1. There are no truths or facts that are not world-indexed.

The consequences of this sort of picture are initially attractive as a reply to the incompatibilist’s argument. Given that Jones does X at T2 in PW1, does he have the freedom at T2 (or earlier) to refrain from doing XI at T2? On the world-indexation strategy, Jones’s doing otherwise at T2 does not require that anything be different from the way it is in PW1. In PW1 there is a set of truths (and facts) about the past relative to T2 that include (as just two of a huge catalogue): Jones does X at T2 in PW1, and Jones does Y (incompatible with X) at T2 in PW2. In PW2 there is also (of course) a huge catalogue of truths and facts. The crucial point is that the catalogues are exactly the same in both possible worlds, given world-indexation. In PW2 it is also true that Jones does X at T2 in PW1. And similarly in PW2 it is true that S does Y at T2 in PW2. So in PW1 God believes at T1 that S will do X at T2 in PW1, and He also believes at T1 in PW1 that Jones will do Y at T2 in PW2. In PW2 God believes at T1 that S will do X at T2 in PW1, and in PW2 God believes at T1 that Jones will Y at T2 in PW2.

The set of truths, which are all world-indexed, is the same in both worlds. Given God’s essential omniscience, He holds the total catalogue of true beliefs in all possible worlds, and thus the set of His beliefs is exactly the same in PW1 and PW2. It follows that Jones’s doing otherwise at T2 does not require any change in the past, and it thus does not threaten the intuitive idea of the fixity of the past. The strategy is formulated so as to put God “in” worlds, but this is not essential; God could be outside of any particular possible world and still hold the total package of beliefs that are true at those worlds.

So, on the world-indexation approach suggested by Almeida as a promising way of addressing the incompatibilist’s argument, there is a huge catalogue of world-indexed facts (and associated truths). But, as in the song, is that all there is? Isn’t it a fact in the actual world (or, say, PW1) that Jones does X at T2? Why are there only world-indexed facts? Of course, the fact (in PW1) that Jones does X at T2 is a different fact from the fact that Jones does X at T2 in PW1. After all, the former is contingent, whereas the latter is necessary, if a fact at all. If it is a fact in PW1 that Jones does X at T2, then it follows that God believed at T1 that Jones will do X at T2. And, if so, then Jones’s doing something else, Y, would require that God held a different belief than He actually held at T1, namely, the belief that Jones would do Y at T2. So we are back to the drawing board, or so it seems.

At this point a proponent of the world-indexation strategy might dig in his heels and insist that there are only world-indexed facts (and truths), and thus that God has only beliefs whose contents are world-indexed. But there is a problem with this move: it would render all truths necessary! The distinction between necessary and contingent truths would be just a conceptual distinction, but in fact no truths would be on the contingent side of the ledger. This would be a big price to pay. Ironically, the attempt to save freedom would have issued in the obliteration of contingency—a bizarre result! The proponent of the world-indexation strategy, who holds that there are only world-indexed facts (and truths), would have to deny the following general principle: If “that S X’s at T in PW1” is a fact, then in PW1 “that S X’s at T is a fact.” But this principle seems undeniable (or, at the very least, unreasonable to deny)

Additionally, it seems that there is at least one truth about the past that Jones would render false by doing Y, instead of X, at T2. He actually does X, and we can assume that the actual world is PW1. His doing Y at T2 occurs in a different world, PW2. So if Jones has the power at T2 to do otherwise (i.e., to do Y rather than X), he has it in his power to make it the case that PW2 is the actual world, rather than PW1. So even on the world-indexation approach, there is at least one fact about the past—the fact about which world is the actual world—that would have to be different, if Jones were to do otherwise. It does not seem that Jones should be conceptualized as having the power so to act that a different possible world would have been the actual world (all along); the fact that PW1 was the actual world seems to be a fact about the past that Jones can’t (intuitively) do anything about at the time of his action.

What could a proponent of the world-indexation approach say about this worry? She might deny the coherence of the ontological picture presupposed by the objection. That is, she might deny that there is a single, unique actual world, and thus she could contend that there are no facts of the sort under consideration. For example, there is no fact that (say) PW1 is the actual world. There is no unique actual world; rather, the statement, “This is the actual world,” is true at every possible world. This is the “indexical approach” to actuality. David Lewis combines this with the view that each possible world is concrete and equally real. This is Lewis’s famous (or infamous) “indexical possibilism”. Each possible world is equally real; this is the modal parallel to Anselmian endurantism or “isotemporalism” (as discussed by Rogers).

On indexical possibilism, “This is the actual world” does not express a truth, just simpliciter and without a world-index or world-relativity. “This is the actual world,” uttered in PW1 is true relative to PW1. “This is the actual world,” uttered in PW2 is true relative to PW2, and so forth. But now, again, there is no problem about a violation of the fixity of the past; there is no non-world-relative truth about which world is the actual world, and thus no truth about the past that must be rendered false, if Jones were to do otherwise.

It is as though the ingenuity of the compatibilist of this sort is limitless. Whereas I salute this, I believe that it also shows the lengths such a compatibilist will (and perhaps must) go in order to defend his view. Clearly, and perhaps most importantly, the view that there is no unique actual world is very counterintuitive. Again, as with the collapse of modalities identified above (in which all truths become necessary), this is a huge philosophical price to pay, in order to reply to the incompatibilist’s argument.

Additionally, the view in question does not fit well with the overall theological picture. Most theists would argue that God chooses which possible world to make the unique actual world, and He does this guided by the idea that the world He selects is the best of all possible worlds. For just one example, the (quite influential) Molinist picture has it that God selects which (single and unique) world to actualize based in part on His knowledge of “counterfactuals of freedom.” Classically, Leibnitz held that God grasped all of the possible worlds, and He created this one because it is the best of all possible worlds. But on the view we are considering, this picture is fundamentally incorrect. In contrast to the more traditional view, God would have actualized all possible worlds equally, rendering them equally concrete and real.

Consider, also, how the view being evaluated fits with replies to the very difficult problem of evil. An important response to the problem of evil—and, thus, an important part of a theodicy--is to point out that God actualized this world because it is the uniquely best of all possible worlds. So even though we suffer, any other world would involve more suffering (more evil), or, at least, no countervailing good that would balance out the suffering. On the traditional view, God had a reason (or set of reasons) for actualizing this world; your suffering couldn’t be avoided without there being even more suffering, or sacrificing some great good. But on the view under consideration here, according to which there is no unique actual world, this move is unavailable. Your suffering is not justified as being part of the best of all possible worlds, as being necessary to avoid greater evil. In fact, God has actualized possible worlds in which you, or your counterpart in that world, suffers less, and no one else’s suffering is worse. That other world is equally real. So you might wonder, “Why am I suffering so much in this world, say PW1?”

And, in general, we might wonder why God made real so many possible worlds in which there is so much suffering. The people are just as real as our world, and in some worlds there is even more suffering than in ours, and no (plausible) countervailing goods. How could a perfectly good God (with the other divine attributes) create so much real evil and suffering-so much horrible pain and suffering by concerete and real persons throughout the galaxy of possible worlds?

I conclude that the view that denies that there is a unique actual world is highly implausible, especially from a theological point of view. (Again, see Chapter Three of Our Fate). Given this, it would seem initially that an agent’s doing otherwise would require some violation of the fixity of the past. Given, also, that it is implausible to collapse all modality into necessity (and to suppose that all truths are necessary truths), there would clearly be violations of the fixity of the past, if agents were conceptualized as having freedom to do otherwise, in a world in which God has foreknowledge. This would be true because even though there would be world-indexed truths in such a world, there would also be world-index-free truths.

Turning to a different point, Almeida wonders what can be said in defense of my view that if a conditional such as “If S were to do X at T2, then some hard fact about the past relative to T2 would not have been a fact” is true, then S cannot do X at T2. Almeida points out that David Lewis denies this view about the fixity of the past in his discussion of time travel. David Lewis is a compatibilist, and one shouldn’t be surprised that he would deny the fixity of the past, as it is interpreted by an incompatibilist. But I just don’t trust intuitions about time-travel; I do not think they should be given substantial weight. Perhaps I ought to defer to a philosopher as brilliant as Lewis, but these are very difficult and contentious issues, about which reasonable and smart people can disagree, and about which one can even disagree with David Lewis! (Of course, this would not be the only view of Lewis which many philosophers would dispute; for just one example, recall Lewis’s indexical possibilism, according to which, as he puts it, the possible worlds are all spread out in our universe “like raisins in a pudding ...”)

Allow me to take a stab at saying more in defense of my view. I take it that if it is a necessary condition of an agent’s doing something that he do something he intuitively cannot do (or bring about something he intuitively cannot bring about), then it is reasonable to conclude that the agent cannot do the thing in question. Imagine that it is a necessary condition of my keeping my promise to pick you up that I be at the airport in five minutes (I am now 100 miles from the airport). It follows that I cannot (given present technology) pick you up at the airport. Suppose, more generally, that we know that it is a necessary condition of my doing X that some natural law that actually obtains would not be a natural law. This is strong reason to conclude that I cannot do X. So it seems reasonable to think that if it is a necessary condition of an agent’s doing X at T2 that the past relative to T2 have been different from what it actually was, then the agent cannot do X at T2. If it is a necessary condition of doing something that one bring about something that it is unreasonable to suppose one can bring about, then it seems that one cannot do the thing in question. And it is unreasonable to suppose that one can change the past—in the sense that one can so act that the past would have been different than it actually was.

The fixity of the past is hard to “prove”. And, of course, there are different ways of interpreting the intuitive idea of the fixity of the past, some of which are more congenial to incompatibilism, and some of which fit well with compatibilism. At some point proof runs out, and we must rely on intuitions. I do not think it is too wild and radical to stick with the intuition that our freedom is always the power to add to the given past. Giving this idea up seems like an act of philosophical desperation.

Reference List

• • Fischer, John Martin. 1996. A New Compatibilism. Philosophical Topics 24: 49-66.

 

 

Reply to Swinburne

Swinburne agrees with many of the main theses of Our Fate. He accepts the conclusion of Pike-style arguments that a temporal God’s foreknowledge of human free actions, where “free” is construed to imply freedom to do otherwise, is impossible. He believes that we are indeed free in this sense, and because he rejects the atemporal picture of God’s omniscience, he adopts “open theism.” The open theist claims that God is omniscient in the sense that He knows all and only those propositions that can be known, but contingent truths about future human actions cannot now be known. The future is in this sense open, God remains omniscient, and humans are free to do otherwise. Of course, a downside (for some) of open theism is that it turns out that God has significantly less knowledge than might have been supposed and that is attributed to him in some traditions. It would be hard to see how God could have a “plan” for us, on the open theism view. On this view, God must take certain risks, if he indeed has such a plan, and intends to create the best of all possible worlds.

In contrast, whereas I agree with the incompatibilist’s conclusion about the relationship between God’s foreknowledge and human freedom to do otherwise, I go in a different direction. I contend that there are two kinds of freedom, one that requires or implicates freedom to do otherwise, and one that does not (acting freely). I hold that God’s foreknowledge is compatible with acting freely and moral responsibility. On this view—semicompatibilism—God is not only omniscient, but He possesses foreknowledge. Although human beings are not free to do otherwise, we have all the freedom—acting freely—that is necessary for moral responsibility.

Because much of Swinburne’s contribution is congenial to some of my main views, I’ll just pick a couple of points about which to offer some reflections. He writes:

There is an odd passage in Fischer’s book (2016, 190) in which he claims that some soft facts are fixed: ‘it is a soft fact about early this morning that the sun rose twenty-four hours prior to another sunrise, but presumably no one has a choice about this fact since no one can prevent the sun from rising tomorrow.’ But any powerful enough demigod could easily prevent the sun from rising tomorrow. (Swinburne)

A fair point. I should have made it clear that I was thinking of the powers of human agents, and the fixity of facts for any human being. I agree that it is not logically impossible to stop the sun from rising tomorrow, and presumably a demigod could do so. The fact in question is temporally relational and yet fixed for any human agent—out of any human agent’s power to affect. Perhaps one should qualify the claim about powerlessness further, to make it explicit that we are thinking about human powers, given current technology. My point is that mere temporal relationality does not entail a human power to affect the relevant fact. Softness does not rule out fixity for human beings.

Swinburne correctly observes that I characterize “hard fact” (and thus “soft fact”) in (at least) three different ways in the various essays in this book. I never intended to give a reductive analysis of hard facts; I would gladly offer such an analysis, if I had one! Rather, I presented what I take to be truths about hard facts (and soft facts), however we analyze them (and the distinction). It would be a problem if the putative truths in question were incompatible with one another, but, suitably interpreted, I do not think they are. The reductive analysis of these phenomena is notoriously difficult, and I am not sanguine about the project. It is similarly difficult reductively to analyze the distinction between intrinsic and extrinsic (or “mere Cambridge”) properties, of which the hard/soft fact distinction is the temporal analogue. But despite the difficulty of giving a reductive analysis, I believe that we have a tolerably clear grasp of the distinction, and that we should stick with attempting to identify claims that will be true of (say) hard facts, no matter what the ultimate analysis is.

Finally, Swinburne takes it that my “bootstrapping view” of God’s foreknowledge of propositions about human action is a way of responding to Pike’s argument. (Swinburne) But I did not intend to offer a strategy of replying to the incompatibilist’s argument by sketching the bootstrapping view. I referred to God’s foreknowledge of “contingent” claims about human action, but I did not refer to, or intend that my analysis apply to, free human actions, in the sense that implies freedom to do otherwise. They do not involve the claim of human freedom, but they are contingent in the sense that they are not necessary. In contrast, a proposition such as that expressed by “Jones does X at T2 or Jones refrains from doing X at T2” is both about human action and necessary.

Pike insists that his argument is driven by the fixity of the past, and not causation or causal determination. My approach here constitutes a defense of this claim against the worry that God’s foreknowledge presupposes causal determination, and thus causal determination ultimately drives even the argument for theological inompatibilism. If my suggestion (the bootstrapping view) is defensible, then not only do we have a way of separating the arguments for the incompatibility of God’s foreknowledge and human freedom and the incompatibility of causal determinism and human freedom, we would have a way of defending at least the coherence—if not the ultimate acceptability—of Ockhamism. The Ockhamist holds that whereas causal determinism is incompatible with human freedom to do otherwise, God’s foreknowledge is not. Even if one rejects Ockhamism as a way of defending this package of claims, one might want a way of defending its consistency. To use a perhaps infelicitous (but, I think, illuminating) neologism, we want to show that Ockhamism is “holdable”, if not tenable.

Swinburne’s interpretation of my strategy here depends on conflating acting freely with freedom to do otherwise. Of course, there is no knockdown and uncontroversial way of separating these two kinds of freedom. But I think that there really are two distinct kinds of freedom, and that acting freely is the freedom implicated in moral responsibility. Philosophers from Chrysippus to Locke to the rationalist philosophers to contemporary philosophers such as Harry Frankfurt and Robert Nozick have maintained the distinction between the two kinds of freedom, and have argued that acting freely is the freedom-component of moral responsibility. Frankfurt (and Nozick in lectures at Harvard) employed examples with a signature structure of pre-emptive overdetermination to make the case.

Of course, it would be question-begging blithely to assume that God can have foreknowledge of human free action, in the sense that requires freedom to do otherwise. The bootstrapping view does not do this; it certainly does not have the resources, in itself, to establish, or even argue for, compatibilism. It simply purports to offer a way in which God could know about future human actions (but not necessarily free actions) in a causally indeterministic world. More specifically, it exhibits a way in which God’s prior beliefs could count as infallible knowledge of future human actions. A defense of the compatibility of God’s foreknowledge with free human action would require addressing the Pike-style argument in a way that bootstrapping view does not seek to do.

 

 

Reply to Reichenbach

Reichenbach defends compatibilism against the incompatibilist’s argument, in its various formulations, arguing that its proponents fail to see ambiguities and thus elide crucial distinctions. Once these distinctions are made, according to Reichenbach, the argument can be seen to be specious. The true versions of the premises do not entail the argument’s conclusion, and the versions that entail the conclusion are not (uncontroversially) true. So the argument cannot be accepted as sound.

A crucial premise in the argument, as I contended in Our Fate and we have discussed at length in this symposium, is the premise that expresses the intuitive idea of the fixity of the past. Reichenbach writes, about what he takes to be an implication of the “conditional” version of the principle of the fixity of the past:

Compatibilists do not hold in the case of divine foreknowledge of human free actions that S being able to do Y requires some fact not to be a fact. That is, compatibilists do not hold that God had a belief at one time and then later believed its negation.

… If Jones does Y instead of X, then God would have believed that Jones does Y and not X. There would not have been a prior fact that God believed that Jones does X because Jones does Y, not X. … Because one cannot alter the past, God’s belief about Jones does not change: it is just that whatever Jones does, God believes. (Reichenbach)

If Reichenbach is correct here, the proponent of the incompatibilist’s argument is making a glaring blunder.. Since the consequence argument employs a similar fixity of the past idea, this widely-discussed, and influential argument would also be flawed in an obvious way. But, needless to say, I do not believe that Reichenbach is correct here (about either the argument for theological incompatibilism or the consequence argument).

Consider again the conditional version of the intuitive idea of the fixity of the past, which Reichenbach focuses on initially:

(FP) For any action Y, agent S, and time T, if it is true that if S were to do Y at T, some fact about the past relative to T would not have been a fact, then S cannot at (or just prior to) T do Y at T.

(FP) simply implies that a certain sort of conjunction of a “can-claim” and a backtracking counterfactual is unacceptable, given our considered judgments about freedom (and its necessary conditions), but not by virtue of logic alone. That is, it implies that if an agent’s doing Y at a given time T would require the past to be different all the way back, then the agent cannot do Y at T. Put in a slightly different way, (FP) implies that if it is a necessary condition of an agent’s doing something Y at a time T that the past would have been different (all the way back), then the agent can’t (in the sense relevant to freedom in a particular context) do Y. (FP) does not rely on the idea that we cannot “change” the past in the sense that we cannot bring it about that God held one belief at one time and another belief at another time. We cannot change the past in this way, but this is not the engine that drives the incompatibilist’s argument. Rather, it relies on the idea that if performing an action requires that something out of one’s control obtains (in this case, that the past be different all the way back), then one cannot perform the action in question.

Reasonable people can certainly deny (FP), and “multiple-pasts” compatibilists do precisely this, especially in the context of the consequence argument. (I introduce this term and discuss the view at some length in Fischer 1994.) There is no logical incoherence in denying (FP). But many find (FP) a plausible way of capturing the extremely widely shared intuitive idea of the fixity of the past. (FP) does not rely on the idea that an agent’s doing otherwise would require changing the past in the way indicated by Reichenbach (or the idea that if doing otherwise would require changing the past, then the agent couldn’t do otherwise). If it did, it would be a stunningly defective principle, but it doesn’t. If it is defective, it is not stunningly so (in the indicated way, or, in my view, any way).

Reichenbach adjusts his critique to apply to the possible-worlds version of the fixity of the past ([FP*]), as discussed above. Here he contends that the incompatibilist maintains that the possible-worlds version implies that an agent’s doing otherwise points us to a possible world in which God’s beliefs change over time. But the incompatibilist does not hold that a principle such as (FP*) has this sort of implication. As with the conditional version, the problem is not alleged to stem from a logically incoherent putative implication of God’s beliefs changing over time (say, in a given, single possible world). Rather, the intuitive engine driving the incompatibilist here is the simple and fundamental point that our freedom is always the power to add to (or extend) the actual past. I think this is a philosophical jet engine. Given that S actually does X at T, then the actual past includes God’s believing that S does X at T. So there is no possible world with the same past as the actual world in which S does Y at T. The problem is not that S’s freedom points us to a possible world in which God’s beliefs change over time; rather, the problem is that there is no possible world accessible to S at T in which S does Y at T. That is, there is no possible world with the same past as the actual world (given S’s actually doing X) in which S does Y at T. So S cannot do Y at T.

Reichenbach is not the only participant in this book symposium (and a range of contemporary debates) who contends that the incompatibilist’s argument is specious—logically defective in a fatal way. Throughout the history of discussions of the argument (in its various formulations), philosophers and theologians have made this charge, whereas others have defended the validity of the argument. The critics have pointed to a suite of putative non sequiturs. This view has a distinguished pedigree and a distinguished team of contemporary philosophers. But in 1965, Pike presented the argument in a way that made it salient that the argument is not invalid, although, of course, its soundness can be disputed.

At roughly the same time, philosophers such as David Wiggins, Carl Ginet, and Peter van Inwagen offered regimentations of the consequence argument that showed that it is also not invalid. Again, the consequence argument may not be sound, but it is valid. These relatively recent analyses (by both teams) have helped us to make progress in understanding and evaluating the arguments. Bottom line: the theological incompatibilist’s argument and the consequence argument can be regimented in valid ways. Those who interpret the arguments as valid are not making any errors that could be identified in Philosophy 1. In a graduate seminar, diligent students might well uncover a logical defect in some particular regimentation of the fundamental argument, but this does not in itself show that the argument cannot be regimented in a valid way.

Reichenbach helpfully presses on the conception of God’s beliefs that plays a crucial role in the incompatibilist’s argument. He points out correctly that the fixity of God’s prior beliefs is motivated by a particular conception of them, and more specifically, God’s belief-states, according to which they involve “representations” that present to God facts about future events (including human actions). I have argued that these formally defined elements of God’s belief-states have their specific content independently of what happens in the future. That is, it is not the case that one and the same state of God’s mind would have one content, if some event takes place in the future, and a different content, if a different event occurs in the future. Content is in this way “temporally internal.”

But I concede that it is unclear whether God’s beliefs are properly interpreted in this sort of way, as importantly parallel to human beliefs. As Reichenbach writes, perhaps God acquires and maintains His beliefs in a fundamentally different way from the way in which human beings do. This would not be surprising, given that we are physical creatures and God is not. How could God, a non-physical being, have mental states with “formal structure”? (I do not think it is obvious that a non-physical entity could not have something like formal structure: consider, for example, the set of natural numbers, which can be defined and whose properties can be exhibited in formal or “syntactic” ways.) But if a representational model of God’s beliefs is incorrect, then it becomes more difficult to defend the contention that they fall under the intuitive idea of the fixity of the past. So it becomes more difficult to defend the soundness of the incompatibilist’s argument.

As an alternative picture of God’s beliefs, and in particular God’s prior beliefs about future human actions, Reichenbach defends a kind of “immediate awareness” (or quasi-perceptual) model. On this approach, God is directly aware of future human actions, in a way that is parallel to perception of present events. This, or versions of it, has been called the “crystal ball” model of God’s foreknowledge. On this approach, one can apparently defend compatibilism against the Pike-style incompatibilist’s argument: the fixity of the past principle does not imply the premise stating that God’s prior beliefs are fixed and out of our control, because the principle doesn’t apply to God’s beliefs.

As I have been wont to contend (and we have discussed above), each plausible response to the incompatibilist’s argument has its pros and cons, and one should not precipitously reject any such response simply because it has cons. Ultimately, one’s evaluation of the various positions has to involve a philosophical cost-benefit analysis. The salient problem with the immediate awareness model (and, in particular, the version offered by Reichenbach) is that it is committed to significant and pervasive macroscopic backwards causation: future events backward-cause God’s prior belief-states. This seems problematic. If God’s prior belief is a direct awareness of the future event, then that mental state of God—His direct awareness—is backward-caused by the future event, even if God’s mental state is nonrepresentational.

Could one say that the future event does not backward cause God’s mental state, but is part of it? Russell’s theory of propositions has it that concrete individuals (and, presumably, events) are themselves parts of propositions. But this highly contentious and not widely accepted view has it that propositions (rather than the belief-states whose contents are the propositions) contain concrete entities; at best, then, this view could apply to God’s beliefs, construed as propositions believed. Sometimes we speak of beliefs as propositions believed, whereas on other occasions we are thinking of beliefs as belief-states. It is highly implausible that a future event could be part of God’s belief-state—the state of God’s mind (even if non-representational) in virtue of which He has awareness of the future event. I suppose the best interpretation of this view, if it has a chance of avoiding the fixity of the past principle, is to deny that God has beliefs in virtue of having belief-states—states of His mind—at all. But this is obscure.

If we construe God’s beliefs as representational, we can defend the relevant premise (stemming from the fixity of the past), but it is mysterious how God’s mind could contain formal representations. If we construe God’s beliefs as non-representational—as involving direct quasi-perceptual awareness of the events in question—it becomes more difficult to defend the incompatibilist’s argument. But one would then be committed to widespread macroscopic backward causation. Out of the frying pan and into the fire! This leaves us, in the end, weighing the advantages and disadvantages of the various views.

Reference List

• • Fischer, John Martin. 1994. The Metaphysics of Free Will: An Essay on Control. Oxford: Blackwell Publishers.

 

 

Reply to Rhoda

I am grateful for Rhoda’s subtle and challenging comments. I admit that my sketchy presentation of the bootstrapping approach in the introductory essay in Our Fate would need considerable development and defense, before it would be a serious candidate for acceptance. His careful formal presentation of the argument is helpful. Rhoda’s critique will push me to think about this approach more, and I offer some preliminary reflections here.

I wish to begin by noting that Rhoda defines “future contingency” in a way that requires indeterminism. I concede that this reflects to a considerable degree the history of usage of the term, but I’m not sure it is helpful, in all dialectical contexts, so to define the notion. I prefer to say that a future contingent proposition (relative to T) is a proposition about the future relative to T that is not (or does not correspond to) a necessary truth. Even if causal determinism were true, a proposition such as that Jones does X at T is not necessary (logically or metaphysically). Further, even if causal determinism were true, it would not uncontentiously follow that Jones cannot do otherwise at T. After all, a classical compatibilist would contend that causal determinism is fully compatible with an agent’s freedom to do otherwise. So a classical compatibilist could consider the proposition that S does X at T a future contingent (relative to a prior time), even if such contingency requires that the agent in question have freedom to do otherwise. Thus I don’t think we should define the notion of future contingency in a way that resolves these issues about the relationship between causal determinism and contingency (in the sense at issue here) by stipulation or by definition. But I don’t think this terminological issue makes a difference to the evaluation of the argument. I can defer to Rhoda’s preferred definition, and then we simply want to know whether God can know future contingent truths about human action.

Can He? I believe that the answer is yes. At the risk of some redundancy, I’ll sketch out the bootstrapping view in a slightly different way. Human beings can know enough about an individual’s settled character traits and the envisaged circumstances to meet the criteria for fallible knowledge about future contingents about that individual’s behavior. That is, a human being could be in a KCS with respect to an individual’s future behavior. So God could as well. But God knows that He is essentially omniscient. So He knows that His belief about the future is true; He believes this, and thus knows this, with certainty. Note: it is not as though God first is in a KCS that gives him fallible knowledge of the future, and then He infers that the belief in question must be true (that is, bootstraps to certain knowledge). It doesn’t work that way. These are not different temporal moments, but different logical or analytical moments, as it were. God simultaneously finds himself in a KCS with respect to the future contingent and recognizes His infallibility. So God does not bootstrap from knowledge, loosely construed, to knowledge, strictly construed. He does not go through a transition from fallible knowledge (knowledge loosely interpreted) to certain knowledge (knowledge strictly interpreted). So the argument does not commit the fallacy identified by Rhoda, as far as I can see. It does not slide illicitly from one notion of knowledge to another, more strict one.

I pause to note that I’m not sure that there are different “senses” or concepts of knowledge as claimed by Rhoda. Perhaps knowledge is univocal, but there are different evidential criteria in different contexts. It is not clear that an “analogical” account of knowledge is to be preferred to a context-sensitive or context-relative account. Rhoda writes that he hopes to defend an analogical account in future work, so I will let this pass here and look forward to this future work.

The bootstrapping can be put in terms of subjective and objective certainty. I just presented it (in sketchy form) in terms of God’s subjective states. That is, He is conceptualized as knowing that He is essentially omniscient, and this allows Him to know (and thus believe) with certainty the future contingent proposition. But we can easily convert it so that it invokes objective certainty. God finds himself in a KCS with respect to a future contingent pertaining to human action. Simultaneously, in virtue of His essential omniscience, His belief is (objectively) rendered certain. He never is in a situation in which He has fallible beliefs about the future; He is always in a belief-state that gives Him certain knowledge of the contingent truth about the future.
Again, there is no fallacy of equivocation here, as far as I can see.

My regimentation of the incompatibilist’s argument employs the following definition of essential omniscience (for a sempiternal God): For all propositions P and times T, it is necessarily true that God believes at T that P if and only if P is true at T. Rhoda asks whether future contingent propositions about human behavior can be true at prior times, since indeterminism must obtain (given his definition of future contingents), and thus the events specified by the propositions are not settled in advance. This is a good and tough question, and it brings out important presuppositions of the incompatibilist’s argument, as it has traditionally been discussed (especially as regimented by Pike). It also raises issues, such as metaphysical grounding, which were not in the forefront of philosophical discussion when Pike published his article, but which have taken center stage in metaphysics more recently.

I begin by observing that Pike did not believe that it is coherent to think of propositions as true at times. He thus did not employ the same account of divine omniscience as I do (above). Pike’s conception of omniscience is something like this: For all propositions P and times T, it is necessarily true that God believes at T that P if and only if P simpliciter. He implicitly relies on the idea that if (say) S does X at T2, then at T1 S will do X at T2. Thus (by Pike’s account of omniscience) it must be the case that at T1 God believes that S will X at T2. It is interesting to note that Peter Van Inwagen is also skeptical about the idea that propositions are true at times, so he develops the consequence argument (which, as I have been arguing, is parallel in important ways to the argument for theological incompatibilism) without presupposing that propositions are true at times. The two contemporary giants of incompatibilism maintain skepticism about “truth at a time.”

On Pike’s view, it follows from S’s actually doing X at T2 that at T1 S would do X at T2. I am not sure that this (without elaboration and defense) successfully addresses the basic worry behind Rhoda’s objection. How can it be that at T1 S will do X at T2, even though the proposition that S will do X at T2 it is not true at T1? Can it be the case at T1 that S will do X at T2 without its being settled at T1 that S will do X at T2?

I’m not sure how Pike, or a proponent of his version of the argument, would respond. I’m inclined to take a different tack. I’m inclined to contend that the proposition that S does X at T2 does indeed imply that it was true at T1 that S would do X at T2. But I think that this can be the case, even in an indeterministic world (in which S’s doing X at T2 is not causally determined). To suppose that truth in advance rules out contingency is to collapse the modalities; (prior) truth becomes necessity, and this is a mistake (in my view).

My view here is obviously not without problems (or, at least, challenges). Some of these stem from issues pertaining to metaphysical grounding, which have been brought out relatively recently. For instance, how can it be true at T1 that S will do X at T2 without this truth’s being grounded in conditions at T1? It would appear that I would have to defend my view by denying at least one of the following plausible and widely held views: that a proposition true at a time T must be grounded in (temporally nonrelational as regards the future) facts that obtain at T, that the facts that ground the truth of a proposition at a time must entail the truth of that proposition at the time in question, or that all true propositions, and, in particular, future contingents, must be grounded. Perhaps true future contingents need not be grounded at prior times, or perhaps the facts grounding a true proposition need not entail the truth of the proposition, or perhaps future contingents need not be grounded at all. Maybe, but it is not going to be easy to defend any of these options.

At the very least, Rhoda’s insightful comments illustrate the importance of “updating” the regimentations of the incompatiblist’s argument to take into account issues pertaining to metaphysical grounding. I hope that these issues, together with the relationship between the incompatibilist’s argument and different views in the philosophy of time (as raised by Rogers above), will be the subjects of future research by the community of scholars who still find the incompatibilist’s argument deeply fascinating.

 

Afterword(s)

 

First, I would like to thank Gregg Caruso for his support of this book symposium. His support and excellent help have been invaluable.

I am very grateful for the careful, insightful, and kind comments by all of the participants in this symposium. No author could expect more thoughtful engagement with his or her work. I have benefited from thinking about all of the contributions and trying to reply as best I can, although I certainly will not have assuaged all of the worries. I hope that this symposium will spur more work on these great issues of historical and contemporary interest. I am sure it will spur more work by me!

Finally, interested readers might have a look at another book symposium on Our Fate , which has appeared in European Journal for Philosophy of Religion vol. 9, no. 4, December 2017. Online at: https://philosophy-of-religion.eu/archive or https://philpapers.org/pub/2434