Some Speculations about Semicompatibilism

Meghan Griffith

Davidson College - Davidson, NC

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

Correspondence | Meghan Griffith, Davidson College - Davidson, NC; Email: megriffith@davidson.edu

Citation | Griffith, M. (2017). Some Speculations about Semicompatibilism Science, Religion and Culture, 4(2): 4-14.

DOI | https://dx.doi.org/10.17582/journal.src/2017/4.2.4.14

At the very outset of his excellent Our Fate, John Martin Fischer suggests that arguments for the incompatibility of freedom to do otherwise and divine foreknowledge “are in important ways parallel to a more recent (“Modern”) argument for the incompatibility of causal determinism and freedom to do otherwise” (1). Comparisons between foreknowledge and determinism appear throughout the book, typically when Fischer is defending incompatibilist arguments regarding foreknowledge and freedom to do otherwise (i.e., alternatives). But in an intriguing discussion in the introductory chapter, Fischer draws a comparison between determinism and foreknowledge with respect to moral responsibility, arguing that, like determinism, foreknowledge is compatible with our having it. Thus, he offers a parallel to one of his landmark views, namely his semicompatibilism about determinism. Fischer defines semicompatibilism about determinism as “the doctrine that moral responsibility is compatible with causal determinism, even if causal determinism rules out freedom to do otherwise” (49). In Our Fate, he suggests that moral responsibility is compatible with foreknowledge, even if foreknowledge rules out this kind of freedom.

In setting up semicompatibilism between foreknowledge and responsibility, Fischer argues that determinism need not be true in order for there to be foreknowledge. He disagrees, for example, with those, like Patrick Todd, who hold that the reason foreknowledge rules out alternatives is because it is an indicator of determinism (40-41). Todd’s view is characterized, in part, as follows: “the only way God could have foreknowledge is by having sufficient evidence, and the only way God could have sufficient evidence is by having first-order evidence that entails the truth of the relevant future proposition, and this requires causal determinism” (41). But in an intriguing new argument (one that I unfortunately cannot really address here), Fischer argues that it is possible for God to have sufficient evidence under indeterminism.1 Thus, Fischer can say that foreknowledge rules out alternatives, not because of determinism, but for independent reasons. But of course his semicompatibilism suggests that this does not threaten moral responsibility. Setting things up this way allows Fischer’s semicompatibilism about foreknowledge to be consistent with incompatibilism about determinism and moral responsibility (so long as determinism rules out responsibility for reasons other than its ruling out alternatives). One might believe, for example, that foreknowledge does not rule out responsibility even though it rules out the ability to do otherwise, while also believing that determinism rules out responsibility by ruling out the proper kind of sourcehood (42). Fischer mentions David Hunt as one who holds such a view. Fischer suggests that semicompatibilism can be seen as “filling in” a view like Hunt’s (46).

But I assume that Fischer’s semicompatibilism is meant to accommodate both compatibilists and incompatibilists about determinism and responsibility. For example, a determinism-responsibility incompatibilist like Hunt can be a semicompatibilist about foreknowledge, but so can a thoroughgoing determinism-responsibility combatibilist like Fischer. The question of determinism is, in this sense, bracketed off. Determinism may or may not be problematic, but if it is, it is problematic for reasons that can be separated from the issue of foreknowledge. But might a semicompatibilist about foreknowledge and responsibility think this bracketing is not so clear-cut? Suppose we grant that foreknowledge is possible under indeterminism and therefore grant that there is a kind of independence between foreknowledge and determinism. Could there be another sense in which these issues are not independent? Hunt’s view is described in such a way as to regard determinism as problematic for reasons that seem unrelated to foreknowledge. Determinism is problematic because it rules out proper sourcehood, regardless of what we say about foreknowledge. But could a semicompatibilist, for example, think that determinism is problematic, not for independent reasons, but for reasons that arise from within the foreknowledge context?

I do not think Fischer explicitly addresses these questions in the book, but they are questions that arise from his fascinating discussions, and I would guess that they are of interest to those who care about these debates. Asking and considering these questions might illuminate certain aspects of the terrain. So my goal here is to raise them and to discuss how they might be answered. Note that in raising these questions here, I am not offering an objection to or critique of Fischer’s position. I am not, for example, claiming that these are questions that Fischer ought to have explicitly raised, nor am I claiming that answers to these questions will ultimately prove problematic for semicompatibilism. I am, instead, engaging in some speculations.

 

Freedom to do Otherwise

 

Before getting to these questions, though, it may be helpful to briefly discuss some of the broad contours of Fischer’s claims about foreknowledge and how they relate to his semicompatibilism. Recall that the semicompatibilist thinks that moral responsibility is compatible with foreknowledge even if foreknowledge rules out alternatives. On the one hand, this formulation of semicompatibilism means that alternatives just do not matter, so we could leave the relationship between alternatives and foreknowledge aside. But of course, we cannot really leave it aside! The traditional debate has been framed in terms of alternatives. Furthermore, if we ignore the issue of alternatives, we risk underselling the semicompatibilist view. The persuasiveness and power of arguments for the incompatibility of foreknowledge and alternatives demonstrates, among other things, what’s at stake. If foreknowledge rules out alternatives, what then? Semicompatibilism has an optimistic answer. Consistently with his semicompatibilism, Fischer spends the majority of the book defending incompatibilist arguments regarding foreknowledge and alternatives against compatibilist views. Interestingly, one of the most important positions he argues against, namely Ockhamism, tries to make its case for the compatibility of foreknowledge and alternatives by contrasting foreknowledge with determinism. The Ockhamist thinks that determinism rules out alternatives whereas foreknowledge does not. Looking briefly at Fischer’s response to Ockhamism provides some useful context.

The incompatibilist about freedom and foreknowledge is worried about the fixity of the past. God’s past belief about what I will do now seems like a fact about the past that is “over-and-done-with.”2 There does not seem to be anything I can do now to make it the case that this belief would be otherwise. The Ockhamist seeks to resolve the apparent incompatibility between foreknowledge and freedom (to do otherwise) by drawing a distinction between “hard” and “soft” facts. Roughly, hard facts are “temporally nonrelational” (131), meaning that they do not appeal to other times. Soft facts, on the other hand, are temporally relational and involve reference to times other than their own (131). Fischer gives the example of his alarm clock ringing at seven in the morning: “the fact that the alarm clock rang is a hard fact about seven. In contrast, the fact that the alarm clock rang two hours prior to my typing is a soft fact about seven” (131). The Ockhamist argues that it is only hard facts that we must worry about when thinking about whether we can do otherwise. God’s beliefs, argues the Ockhamist, are not hard facts. In making this case the Ockhamist distinguishes between the cases of determinism and foreknowledge:

The Ockhamist’s position, then, is that (a) facts which only appear to be strictly about the past but are really also about the future do not carry the necessity of the past, and (b) God’s beliefs are precisely this sort of fact. The Ockhamist then agrees with the incompatibilist [about determinism and freedom to do otherwise]…that facts which are strictly about the past (hard facts) are indeed now fixed and out of our control. But he claims that God’s prior beliefs are not in the class of such facts, and that there is no other reason stemming from the fixity of the past to deny that we can be free to do otherwise. (133-4)

Fischer argues (and, I think, persuasively) that Ockhamism is not successful. Here is the basic idea behind his position. Fischer points out that even if we concede to the Ockhamist that God’s beliefs about the future are soft facts, softness is not sufficient for securing the freedom to do otherwise.3 Not all soft facts are facts we can falsify—some soft facts are fixed (i.e., out of our control), in spite of their softness (134). And “God’s prior beliefs are in the class of soft facts which are nevertheless fixed” (134). There are different reasons why soft facts might be fixed. In the case of God’s beliefs, it is because there is some “hardness” involved (some element that is temporally nonrelational) and thus we must ultimately contend with the fixity of the past.4 Suppose God has a particular belief “at t1 that S will do X at t2” (139). God’s belief is a particular state of mind that “just is not counterfactually dependent on the future: it is not the case that one and the same state of mind would count as one belief given one future, and another belief (or no belief at all) given another future” (139). Distinguish this from the alarm example. The fact that the alarm rang two hours before typing is counterfactually dependent on the future (i.e., on whether the typing occurs two hours in the future). Thus, Fischer ultimately concludes that although “the pertinent past fact in the argument from causal determinism is a hard fact, whereas in the argument from God’s existence it is a soft fact,” nonetheless, “…in both cases it is plausible to say that an agent’s doing otherwise would require some hard (temporally nonrelational) feature of the past to be other than it actually was” (149). So on Fischer’s account, even if we allow an important difference between determinism and foreknowledge, namely that determinism poses its problem in terms of fully hard facts, this difference turns out not to matter. It does not matter because hardness infects God’s beliefs enough to thwart our alternatives. He claims that “at a deep level, the arguments are on a par” (149).

 

Moral Reponsibility

 

So it looks as if foreknowledge poses as much of a threat to alternatives as determinism does. But this, of course, is not going to mean that responsibility is ruled out according to the semicompatibilist. Why not? Fischer here appeals to his extraordinary work on moral responsibility. He discusses the famous Frankfurt Cases which purport to show that someone can be morally responsible in the absence of alternatives. These cases support his account of “guidance control” as the freedom-relevant element of moral responsibility. Fischer argues that an agent does not need “regulative control” (the ability to do otherwise), but only the kind of control that allows him to properly guide his action: “an individual exhibits guidance control to the extent that he acts from his own, suitably reasons-responsive mechanism” (47). Acting from one’s own mechanism means that the agent acts from a mechanism that he has taken responsibility for. This is a process that consists in the agent coming to understand the effects of his actions and coming to see himself as “a fair target of the reactive attitudes as a result of how he exercises this agency in certain contexts” (47-8). In order for the mechanism to be suitably responsive there must be moderate reasons-responsiveness. This means that there is an ability to recognize and respond to an appropriate pattern of sufficient reasons (including moral reasons) to do otherwise (48). Fischer claims that neither determinism nor God’s foreknowledge conflict with guidance control. There does not seem to be any reason to suppose that the agent could not be moderately reasons-responsive in either case. Nor does there seem to be reason to suppose that the agent could not take responsibility for the mechanism in question. Even if God’s beliefs have an element of hardness (as discussed above) such that we now lack regulative control, God’s beliefs do not rule out guidance control.

 

Semicompatibilism and Determinism

 

The foregoing should give the reader a general sense for why Fischer defends a semicompatibilist view with respect to foreknowledge. Now I turn to further considerations about its connection to determinism. As I noted at the outset, in an important way determinism is bracketed off from semicompatibilism about foreknowledge. Even so, discussions of determinism play various dialectical roles throughout the book, as we have already seen in the discussion of Ockhamism. Recall that Fischer compares determinism and foreknowledge with respect to alternatives and argues against the Ockhamist’s claim that determinism is uniquely threatening. If we already think the fixity of the past is a problem, as the Ockhamist does when he rejects determinism, then we also ought to worry about God’s past beliefs. Determinism’s incompatibility with alternatives seems like the easier case to make because of the full hardness of the relevant facts. But Fischer suggests that the step from here to the incompatibility of foreknowledge and alternatives is smaller than the Ockhamist thinks.

Interestingly, when it comes to compatibility with responsibility, the easier case to make, in Fischer’s view, is on the foreknowledge side. Recall that according to the guidance control “actual sequence” model of responsibility, foreknowledge poses no threats. Why might determinism? While Fischer obviously does not ultimately think that determinism does threaten responsibility, he notes the concerns of “source incompatibilists.” Some responsibility-determinism incompatibilists agree that alternatives do not matter but they think that determinism rules out the kind of sourcehood required for responsible agency (49). These source incompatibilists typically emphasize differences between Frankfurt Cases and determinism. In Frankfurt Cases the “unavoidability-generating factors” do not “play a role in the actual causal sequence” (50). This is not the case with causal determinism (50). The source incompatibilist thinks this makes the vital difference as to whether the agent can be a proper source. In a Frankfurt Case she can be such a source. Under determinism, she cannot.

On this score, could foreknowledge be on better footing? Fischer points out, crucially, that the sourcehood worry does not automatically arise in the case of foreknowledge “insofar as God’s beliefs are not conceptualized as bringing about human action” (49). God’s knowing what I will do does not appear to prevent me from being the source of my action in the required way. Determinism, on the other hand, raises at least prima facie trouble for sourcehood. On the other hand, Fischer implies that maybe if we reflect on this a bit more, we can use the foreknowledge case to see that sourcehood worries under determinism are overblown.

If one embraces semicompatibilism regarding foreknowledge, this might pave the way for semicompatibilism about determininism and responsibility. Fischer asks us why we are so concerned about the external sources at work in determinism (51). After all, as Fischer suggests, God’s foreknowledge might actually be viewed as a kind of external source insofar as it entails that “there is some condition that is entirely external to the agent (God’s belief) that is sufficient for the behavior in question” (50). Why does the difference in the kind of external source make such a difference (51)? Fischer puts it this way: “the interesting question, in my view, is whether this difference makes a difference” (51).

Thus, Fischer presents this question as a way of adding support to compatibilism between responsibility and determinism. If the incompatibilist about determinism and responsibility is not worried about foreknowledge, the burden is on the incompatibilist to say how determinism is different from foreknowledge in a way that is salient to responsibility. This goes to the heart of the debate about responsibility and determinism. But my point here is that Fischer’s fascinating discussions throughout the book also invite a subtly different question about determinism. In addition to wondering whether the difference makes a difference, we might want to know whether determinism makes a difference. In other words, does it make a difference from within the context in which we are discussing it—i.e., the context of divine foreknowledge?

 

The Difference with Determinism?

 

Fischer makes a convincing point, as he has in past work, that there is a significant burden on the incompatibilist (about determinism and responsibility) to explain why the difference (between the noncausal and causal sufficient conditions) is salient. It is difficult to articulate why external causal sources are more worrisome to us than other kinds of external sufficient conditions. One obvious attempted answer: control and causation are closely linked such that causal sources seem more apt as usurpers of control.5 But of course using this as a response is highly contentious, or even question-begging, given that compatibilist guidance control is supposed to rule out the circumvention of our causal control. On the other hand, it is hard to get away from the idea that causal sources have particular significance. What I would like to consider here is whether a semicompatibilist might argue that external causal sources are not necessarily more problematic than other kinds of external sources on their own or in themselves, but only when we are in the context of divine foreknowledge.

I begin with some very common “gut feelings” about God and human responsibility. 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.6 There is just the luck of the draw, and “the cards we are dealt” (to borrow a metaphor that Fischer has used in his other writings) are dealt by the laws of nature. There is no omniscient dealer behind our fortunes.

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. But why suppose that God is a manipulator? After all, Fischer seems correct that there is not anything about God’s beliefs that rule out guidance control. Like the counterfactual intervener in a Frankfurt Case, God’s knowledge of what I will do does not seem to in any way make me do it (even if it rules out my ability to do otherwise). Here is where I think adding determinism could matter. There seems to be a kind of additive effect when thinking about God’s relationship to our responsibility. This is not surprising. Theological problems are usually a package deal. It is the package that leads to the tension rather than one element in particular. So maybe manipulation worries arise due to the combination.

Derk Pereboom’s famous four-case manipulation argument comes to mind here (see Pereboom 2014, pp. 74-103). The four-case argument is meant for a different purpose than mine (it is meant to show that determinism-responsibility compatibilism is false), but it has some relevant features. The four-case argument begins with an obvious case of manipulation (complete with evil neuroscientists), in which most people would agree that the agent is not morally responsible, even though the example is set up such that the agent supposedly satisfies compatibilist conditions for responsibility. Pereboom proceeds through a series of cases, involving progressively less involvement from other persons, until he arrives at ordinary causal determinism. His point is to show that there is a burden on the compatibilist to explain why the obviously manipulated agent is not responsible while the ordinary determined agent is, when all four cases involve an agent who satisfies compatibilist conditions for responsibility.

Looking at some of Pereboom’s cases is useful for getting at the manipulation intuition that I refer to above. In case 1, neuroscientists are manipulating an agent’s (Plum’s) brain states “by pressing a button just before he begins to reason about his situation” (Pereboom 2014, 76). They produce the sort of reasoning process that is needed to deterministically cause him to do what they want him to do (kill White). But his reasoning process satisfies compatibilist conditions. Setting aside the occasional occasionalist, most theists can easily claim that this case does not resemble what is going on with God. God presumably does not swoop in just before we begin to reason in order to get our reasoning processes to be what is needed. But what about case 2? In case 2, Plum has been programmed “at the beginning of his life so that his reasoning is often but not always egoistic” such that in some circumstances his action will be causally determined by this sort of reasoning process (as is the case when he decides to kill White) (Pereboom 2014, 77). How might this case compare? There are obviously some important differences between the neuroscientists and God. But one might worry that the cases are too close for comfort. Even if God is not programming us in the literal sense specified in case 2, if determinism is true, then God is knowingly and intentionally actualizing a world in which causal laws, laws for which God is causally responsible, will produce our behavior.

In other work, Fischer has argued against the four-case argument. He argues that whereas the agent in case 1 is not responsible (because this agent does not, contrary to what Pereboom suggests, satisfy the guidance control requirements), the agent in case 2 is responsible in spite of the involvement of the neuroscientists. Interestingly for our purposes, one of the important differences Fischer cites between the first two cases involves the directness of the “manipulation.” He says:

 

It is salient, I think, that case 1 involves “hands-on”, direct manipulation, whereas case 2 does not; case 1 is a genuine case of manipulation, whereas case 2 is perhaps better described as a case of “initial design”. …. In contrast [to case 1], even though the neuroscientists “set up” Plum with a set of initial dispositions in case 2, they do not subse­quently intervene in a direct way, superceding his own mechanisms of practical reasoning. He has taken re­sponsibility for his mechanism of practical reasoning against a backdrop of his “given” initial endowments. (This contrasts with case 1, in which Plum has taken responsibility for his mechanism of practical reason­ing, but not for the “manipulation mechanism” incul­cated by the neuroscientists.) The situation in case 2 is not relevantly different from the ordinary situation in which we are simply “given” a set of dispositions to­ward feeling and action; and moral responsibility al­ways then is a matter of how one plays the cards that are dealt one, as it were. I think there is an important difference between a case of direct, hands-on manipu­lation, such as case 1, and a case of initial design, such as case 2; and I contend that this difference lies in the fact that Plum acts from his own mechanism in case 2 but not in case 1. (Fischer 2014, 205)

So Fischer would presumably argue that manipulation is not a worry with respect to the divine, given that the God case looks like what Fischer has in mind with respect to “initial design.” Whether “initial design” lines up with a theistic conception, though, might depend on how we are “given” these dispositions. Is there a personal being at the source or are these given to us by nature? We might think our answer matters with respect to our intuitions about manipulation. Even so, in response to a similar argument, Fischer explicitly addresses the idea that “initial design” by a personal knowing designer is still not any more problematic than determinism.7 He responds to Alfred Mele’s famous Zygote argument, according to which a goddess, Diana,

creates a zygote Z in Mary. She combines Z’s atoms as she does because she wants a certain event E to occur thirty years later. From her knowledge of the state of the universe just prior to her creating Z and the laws of nature of her deterministic universe, she deduces that a zygote with precisely Z’s constitution located in Mary will develop into an ideally self-controlled agent who, in thirty years, will judge, on the basis of rational deliberation, that it is best to A and will A on the basis of that judgment, thereby bringing about E. (Mele 2006, 188).

Fischer argues that ultimately, there is no salient difference between ‘initial design’ cases and “ordinary scenarios in which there is no special reason to doubt compatibilism” (Fischer 2011, 271). He tells a story about ordinary agents John and Mary who conceive Ernie in the ordinary manner. There is no reason to think that Ernie is not responsible, even though the act whereby John and Mary conceive him deterministically leads to his action 30 years later. Fischer argues that we can even keep adding to the scenario without getting to non-responsibility. Let’s suppose we make the story closer and closer to the Diana scenario. Let’s say that “John and Mary intended that their intercourse lead to Ernie’s performing A and bringing about E 30 years hence” (Fischer 2011, 268). Even so, argues Fischer:

 

The intentions of John and Mary, and their acting in the belief that they are providing (relative to the background) a sufficient condition for something they want in the future, do not in any way bear on the intuitive basis for Ernie’s moral responsibility in that context 30 years later. The basis for Ernie’s moral responsibility is more ‘local’ – or so it seems to me. (Fischer 2011, page 268)

There is considerable plausibility to Fischer’s responses. Just as Plum is able to act from and take responsibility for a moderately reasons-responsive mechanism in case 2, so does it seem that we are able to act from and take responsibility for ours, even if there is an omniscient designer. And why should we think that adding in an initial designer with beliefs and intentions makes the crucial difference between responsibility and non-responsibility? After all, the John and Mary scenario, (in which John and Mary believe and intend for their action to result in what Ernie does in 30 years) does not seem to preclude Ernie’s responsibility.8

On the other hand, we might still be inclined to think that the Diana and God scenarios are problematic. We might think so because we might think there are important differences between Diana and God on the one hand, and John-and-Mary on the other. 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. 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. For example, if I am a rational agent, I cannot intend to win an unrigged lottery—I can only intend to try to win it.9 This is because who wins the lottery is completely out of my control and I have no way of predicting it. There might be circumstances, however, in which I can intend to win. I can intend to win if I have sufficient reason to believe that it is rigged in my favor. Or I can intend to win if I am irrational and have an unfounded belief that winning is somehow within my control.10 In the cases in which I intend to win, if I do in fact win, we can say that I intended to win, and my action of buying a ticket in part fulfilled this intention. In the rigged case, we can say that I controlled the results. In the other case, however, it does not seem right to say that I controlled or manipulated the results of the lottery. It was really a lucky coincidence that I intended to win and then in fact won.11 Is this the case with John and Mary? Unless we fill in the justification for John and Mary’s belief, we are left wondering whether there is a substantive difference between their case and the case of a divine agent. 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 any more than the irrational lottery player’s intention can control his winning. 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.

These thoughts are by no means conclusive, and one might still reasonably argue that the addition of intention is not a strong enough justification for the difference between non-responsibility and responsibility. So ultimately, we might want to grant that the divine case is not akin to manipulation. But manipulation arguments and considerations regarding their similiarities to the divine case at least provide some reason for the semicompatibilist to worry.

 

What’s Determinism Got to Do with it?

 

But why suppose that determinism has anything to do with this worry? Perhaps manipulation is just a worry that arises from foreknowledge, omnipotence, and so on. To examine this further it might help to look at a view like Molinism which, as Fischer suggests, provides “a model of divine providence” (82). Molinism attempts to explain how God has a guiding hand in the world without our actions being causally determined. Like the Ockhamist, the Molinist holds that determinism rules out alternatives whereas foreknowledge does not (82). Molinism is often thought to provide a unique solution to freedom and foreknowledge, but as Fischer (and others) argue, Molinism does not provide its own response to the argument for theological fatalism. But it does provide an answer to how God is able to preserve “His providential powers” in light of our freedom (82). The Molinist believes that “knowledge of what creatures would freely do in all possible circumstances,” i.e.., God’s “middle knowledge,” is given to God (82). So there is an important sense in which what we do is not up to God, but is up to us. But according to Molinism, providence is preserved because “on the doctrine of middle knowledge, God can employ the relevant conditionals and His natural knowledge (of initial conditions) in the process of deciding which possible world to actualize” (82). Looking at Molinism is useful here because Molinism provides a model whereby our actions are not causally determined but in which we might still ask whether our actions are problematically manipulated. If they seem to be manipulated, then my hypothesis that determinism makes a difference is false. Why might one think that our actions are manipulated on this model? Well, we might think that God’s knowledge together with God’s control over circumstances is sufficient for manipulation. After all, God knows what we will do in certain circumstances and is the one who gets to decide whether these are the circumstances into which we will be placed.12 It seems that God intends that by actualizing the circumstances, we will perform particular actions.

Of course the Molinist will resist this claim about manipulation. God actualizes the appropriate circumstances so as to ensure that we carry out the divine plan, but God does not thereby force us to do anything. If we have in mind the counterfactuals of freedom, we realize that every one of these is ‘up to us’ and God is merely selecting from them by selecting which circumstances to actualize. When selecting these circumstances, God does intend that we perform certain actions. But this intention ensures our actions in virtue of God’s foreknowledge and not in virtue of their causal determination.

In order to do more to motivate the difference between the deterministic and indeterministic case, it is useful to look at a discussion in Fischer’s chapter on Molinism. Here, Fischer is considering an intriguing argument by Michael Bergmann that offers a kind of “multiple pasts compatibilism” with respect to alternatives and foreknowledge. The suggestion is that one might legitimately be such a compatibilist with respect to foreknowledge while rejecting such compatibilism for determinism. In an earlier chapter, Fischer explicates multiple pasts compatibilism (with respect to foreknowledge) with the example of Jane going to the movies on Tuesday. The multiple pasts compatibilist claims that:

Jane can do something on Tuesday (i.e., refrain from going to the movies) that is such that, if she were to do it, God would have known (and thus believed) on Monday that she would do it on Tuesday. (69)

So this is the sort of position that we might regard Bergmann as defending. But why be a compatibilist with respect to foreknowledge while rejecting it with respect to determinism? Bergmann suggests that the cases are not parallel. God’s beliefs about our future actions counterfactually depend on our actions, whereas the prior causes of a deterministic causal chain do not. In other words, God believes what he does because of what we do, whereas causal precursors to our actions are not the events that they are because of what we do. As Bergmann puts it, “’it turns out that past facts about events that causally determine my current acts are not what they are because of what I do now; rather, I do what I do now because of those past facts’” (95).

Fischer argues that if Bergmann wants to deny compatibilism with respect to determinism but not with respect to foreknowledge, he will have to restrict worries about the fixity of the past. On Bergmann’s view, the fixity of the past is only worrisome with respect to the events that come before (and cause) our actions. Fischer suggests that Bergmann’s argument must restrict the fixity of the past to “independent facts”—i.e., facts that don’t depend on our present actions. So it must just be in cases of “pastness plus independence [from our current behavior]” that we are unable to do something such that the past would have been otherwise (95). And this is the case under determinism. But Fischer worries that the distinction between the deterministic and foreknowledge cases starts to break down. Fischer points out that some of those who argue for multiple-pasts compatibilism with respect to determinism suggest that there is “counterfactual dependence of the past causal facts on the behavior in question” since “on some views of the relevant counterfactuals, if causal determinism is true and I actually perform some action X, the following “backtracker” is true: “If I were to refrain from X, the past would have been different all the way back”” (95). It is going to be difficult, then, to explain why “the relevant notion of ‘because of’ would be asymmetric” in Bergmann’s claim (quoted above) that “past facts…are not what they are because of what I do now; rather, I do what I do now because of those past facts’” (95).

I think Fischer’s argument is persuasive. On Bergmann’s view, pastness does not rule out alternatives, so everything rides on the dependence/independence distinction. But there does not seem to be a non-question-begging way to support the distinction. The multiple pasts compatibilist about determinism will say that past events do depend on our current behavior in the sense that were Jane, for example, to refrain from going to the movies, past events would have been different.

Nonetheless, maybe similar ideas to Bergmann’s are helpful for our purposes. There does seem to be an important distinction in the neighborhood of dependence and independence that we might still appeal to. Recall that we are not looking to argue about a difference between determinism and foreknowledge with respect to compatibilism. We are looking to see if there is a difference between a deterministic and non-deterministic case of foreknowledge. According to Molinism, God’s beliefs about what we do originate from God’s “middle knowledge” of all of the relevant counterfactuals of freedom. These counterfactuals seem to depend on us rather than on God. When we move away from multiple pasts compatibilism and away from a defense of alternatives, we do not need to characterize dependence in terms of the ability to act such that something in the past was different. The problem with characterizing dependence in terms of this ability is that it is then difficult to explain why the past does not counterfactually depend on our behavior in much the same way. Recall that according to Molinism, it is ‘up to Jane,’ for example, that the following counterfactual of freedom is true: if Jane is in circumstances C on Tuesday, she will go to the movies. Although it might be somewhat mysterious how the truth of this counterfactual is ‘up to Jane’, that it is ‘up to her’ does not need to entail that she can do something on Tuesday such that God’s belief on Monday is otherwise. And God’s belief on Monday about what Jane will in fact do on Tuesday is based (in part) on knowledge of this counterfactual of freedom. In this sense, then, God’s belief depends on Jane. There does not seem to be a parallel way to claim that the events that occur before and cause our actions are up to us. So the determinism case is not the same. And this may give us a rudimentary answer to our question about why adding determinism to foreknowledge makes a difference. It makes a difference, not by ruling out alternatives (these are already ruled out) but by adding a certain kind of external source. Recall that Fischer wonders why it matters what kind of external source is in play. Considerations about dependence and independence are meant to respond to a similar question: why it matters whether we add a certain kind of external source. Foreknowledge provides an external source, but this source is dependent. Determinism adds an external source that is independent. Maybe independent external sources are not particularly worrisome on their own, but they become so in the case of foreknowledge. This is because under foreknowledge, these external independent sources trace back to the intentions of another being.

One obvious question, though, is why this is not just a worry about external independent causes rather than determinism. After all, causal laws of any sort would seem to provide external independent sources.13 This is, I think, an important point. Why would the fact that the causal chain leading to my action is not fully deterministic give me the possibility of responsibility, given that God’s foreknowledge already rules out alternatives? After all, an indeterministic causal chain would still trace back to the intentions of another being. I am not ultimately sure what to say here. It seems that there are at least two possible lines of response that might be tried. The first is that the person worried about external independent sources might offer legitimate reasons for thinking that external independent indeterministic “sources” are not really sources. They are not, after all, the sufficient conditions Fischer has in mind. 14 Another response would be to concede that external causation of any kind is a problem (though there might still exist causal laws that do not cause our actions), but adopt a view whereby our actions need not have them. In this case, one might adopt noncausal libertarianism or agent-causal libertarianism concerning human action.15 According to both these libertarian views, the agent’s actions do not have external causal sources, either because these actions are uncaused or because the agent is herself the cause. These libertarian views have been met with considerable skepticism, but they are still interesting and worthwhile views that a semicompatibilist about foreknowledge and responsibility might find more plausible than a view according to which determinism is true under foreknowledge.

 

Conclusion

 

As mentioned at the outset, none of the foregoing is meant to provide an objection or critique of semicompatibilism. In fact, Fischer’s view is so persuasive, reasonable, and carefully defended that it is difficult to critique. It is also an important and thought-provoking view. Semicompatibilism has highly significant implications, given that the theist often wants to be able to preserve foreknowledge without sacrificing our responsibility. If I had to guess, I would venture that most, if not all, theists who defend alternatives in the face of foreknowledge are motivated in large part by concerns about responsibility. But the preservation of alternatives seems counterintuitive and problematic (as Fischer adeptly argues throughout the book). Semicompatibilism provides a response to the problem. But the question I have raised here is whether semicompatibilism is sufficient to solve the problem on its own, or whether a semicompatibilist has reason to worry about the truth of determinism. Fischer concedes that a semicompatibilist might worry about determinism for reasons of sourcehood. But then he challenges us to come up with a good reason that sourcehood should be more worrisome under determinism, given that we are not worried about it under foreknowledge. My thought here has been that it might be regarded as more worrisome not in itself, but when combined with foreknowledge. It is not clear to me that this combination is ultimately fatal to responsibility, but it does seem that the semicompatibilist has a reasonable worry. Or so I have speculated.

 

References

 

• • Clarke, Randolph (2003). Libertarian Accounts of Free Will. New York: Oxford University Press.
• • Dennett, Daniel C. (1984). Elbow Room: The Varieties of Free Will Worth Wanting. Cambridge, MA: MIT Press.
• • Fischer, John Martin (1982). “Responsibility and Control.” The Journal of Philosophy, Vol. 79, No. 1, pp. 24-40.
• • Fischer, John Martin (2011). “The Zygote Argument Remixed.” Analysis, Vol. 71, No. 2, pp. 267-272.
• • Fischer, John Martin (2014). Review of Free Will, Agency, and Meaning in Life, by Derk Pereboom. Science, Religion, and Culture, Vol. 1, Issue 3, pp. 202-208.
• • Fischer, John Martin (2017). Our Fate: Essays on God and Free Will. New York: Oxford University Press.
• • McKenna, Michael (2008). “A hard-line reply to Pereboom’s four-case argument.” Philosophy and Phenomenological Research 77: 142-59.
• • Mele, Alfred R. (2006). Free Will and Luck. New York: Oxford University Press.
• • Mele, Alfred R. and Paul K. Moser (1994). “Intentional Action,” Noûs , vol. 28, pp. 39–68.
• • Mickelson, Kristin (2017). “The Manipulation Argument,” in The Routledge Companion to Free Will, ed. Kevin Timpe, Meghan Griffith, and Neil Levy. New York: Routledge: 166-178.
• • Pereboom, Derk (2014). Free Will, Agency, and Meaning in Life. New York: Oxford University Press.
• • Velleman, J. David (1997). “How to Share an Intention,” Philosophy and Phenomenological Research, Vol. 57, No. 1, pp. 29-50.

 

End Notes

 

[1] See (Fischer 2017: 31-45).

[2] This phrase appears in various forms at various points in the book. I believe the first use of it is in a quote from David Widerker (see Fischer 2017: 23).

[3] Fischer expresses some skepticism that we should make this concession, but claims that the merit of his argument is that we can concede their softness and still support incompatibility (31).

[4] Fischer discusses two different ways of understanding these ‘hard’ elements: God’s beliefs could be “hard-core soft facts” or “hard-type soft facts.” I do not think the difference is relevant for our purposes here so I refer the interested reader to the text (see pages 134-9).

[5] For an important discussion of the connections between causation and control, see (Clarke 2003, Chapter 2), though Clarke is primarily discussing why control might require causation rather than how a causal source might usurp it.

[6] For example, in his highly influential Elbow Room, Daniel Dennett argues that worries about free will often stem from false “bogeymen.” Dennett, a compatibilist about freedom/responsibility and determinism, offers a number of examples designed to show that ordinary deterministic causes are not like bogeymen trying to control us. At one point in his argument, he suggests that “a jail without a Jailer is not a jail” (Dennett 1984: 8). He also suggests, appealing to a “nefarious neurosurgeon” example of Fischer’s (in Fischer 1982: 26), that:

What makes Fischer’s version more dreadful [than an ordinary deterministic case] is that Jones’ control of his own activities has been usurped by another controller, Dr. Black. A tumor might cause this or that in someone’s brain, and it would be terrible indeed to have a debilitating brain tumor, but it would take an awfully smart tumor to control someone’s brain. (8)

It’s important to note that in the article Dennett mentions, Fischer is not making a false appeal to bogeymen but is constructing a Frankfurt-style counterexample to the Principle of Alternate Possibilities. It is not clear to me whether Dennett is (falsely) accusing Fischer of this or just using his example to make a point. But in any case, Dennett’s main point, and the point for my purposes, is that there is something particularly disturbing to us about being controlled by another and this is separable on some level from our concerns about determinism. More will be said about this below.

[7] For more about the comparison between the four-case argument and the Zygote argument, see (Fischer 2011, 269, fn1) and (Mickelson 2017).

[8] See (McKenna 2008) for another important response to the manipulation argument. McKenna argues that the manipulated agent is in fact responsible.

[9] This point comes from David Velleman. He discusses the distinction between goals and intentions. Intentions involve what we can settle. He gives the example of getting a fellowship: “Of course, if you have the goal of getting the fellowship, then you can also form the intention of trying to get it, since whether you try is still up to you; but intending to make an attempt at doing something is not the same as intending to do it.” (Velleman 1997: 33).

[10] The lottery example comes from an article by Alfred Mele and Paul Moser. They argue that in a normal lottery case, the agent does not have “suitably reliable evidence” that she will win and therefore she does not intentionally win. If she did have such evidence, then she could intentionally win (Mele and Moser 1994: 60-61).

[11] These considerations about irrational belief are based on personal correspondence from Al Mele who was commenting on a different (unrelated) paper I had written. In his comments he suggests that the irrational lottery winner can intend to win, but if she wins, she does not win intentionally.

[12] Obviously Molinism is not the only possible model of providence, so perhaps there are other models that would garner different intuitions. But it seems that any traditional theistic model will have to contend with similar concerns.

[13] See (Mickelson 2017) for a discussion of how manipulation arguments involve “constitutive luck” and may raise problems for responsible action even without determinism.

[14] See also (Pereboom 2014). Pereboom responds to concerns from Alfred Mele that determinism is not doing the relevant work in the manipulation cases.

[15] Putting it roughly, noncausal libertarianism is the view that the relevant sort of freedom (the kind required for responsibility) is incompatible with determinism and our relevantly free actions are, or stem from, uncaused actions. Agent-causal libertarianism is the view that freedom is incompatible with determinism and our free actions are those that are caused by the agent, as a substance, rather than in virtue of events.