Fischer and Fate

Mark Bernstein

Purdue University, West Lafayette, IN, 47907, USA

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

Correspondence | Mark Bernstein, Purdue University, West Lafayette, IN, 47907,USA; Email: [email protected]

Citation | Bernstein, M. (2017). Fischer and Fate. Science, Religion and Culture, 4(2): 27-37.

DOI | https://dx.doi.org/10.17582/journal.src/2017/4.2.27.37

“In philosophy it is can in particular that we seem so often to recover , just as we had thought some problem settled, grinning residually up at us like a frog at the bottom of the beer mug.”

Ifs and Cans, J.L. Austin

The dust jacket for John Martin Fischer’s memorable collection, Our Fate: Essays on God and Free Will, features a picture of a desolate, dusty road of an apparently long, but indeterminate, length. The road has no forks, lending the impression that as travels begin, there is only one direction available to the sojourner. Once on the road, we are fated to follow it; our desires, hopes, and aspirations to go elsewhere rendered utterly ineffectual. While the barren landscape of the picture presents a certain type of pure and pristine beauty, what it represents is deflating and depressing. Where our ordinary self-conception includes the freedom to pursue a host of alternative paths during our lives, the illustration serves as a catalyst to the terrifying idea that our unquestioned confidence in a world offering us optional ways of leading our lives is unwarranted.

Those readers of Our Fate seeking reassurance for our status as free-acting persons will be somewhat disappointed. The good news is that even enmeshed in this alternative-free world, we remain, in general, morally responsible for our behavior. The bad news is that if we are reluctant to abdicate either our belief in the truth of causal determinism or our belief in the existence of a traditional Judeo-Christian God, the path we follow will continue unremittingly branchless. Fischer’s semi-compatibilism has us living morally responsible, unfree lives. Here I will focus on only the tension that results from our belief in a divine being and our conviction that we possess free agency. While I share Fischer’s incompatibilist vision, we may yet have some reason to have a bit more hope in a harmonious relationship between the two beliefs than Fischer’s sophisticated and ingenious arguments suggest.

 

1

Most religious Jews and Christians conceive of God as a perfect being. Minimally, God is said to be necessarily existent (existing in all ‘possible worlds’) omnipotent (all-powerful), omnibenevolent (all-good), and omniscient (all-knowing). But even attributing omniscience to this very remarkable being does not do Him justice. Not only does God never err, he cannot ever be mistaken. There is no possibility that God blunders; there is no ‘possible world’ that contains a false belief of God. He is necessarily omniscient or infallible.

From these rather spare definitional points of departure, we can develop an argument that God’s foreknowledge is incompatible with human free action, where we understand an agent’s freedom regarding an action requiring that she have alternative courses of action available to her at the time she performs that action; somewhat more formally, S’s free action A at time T requires that S at T could have refrained from performing A. Suppose now, for the purposes of a reductio that Tom freely attends a baseball game at some time, t2. By virtue of His omniscience, God knows at an earlier time, t1, that Tom will attend the baseball game at t2. From God’s infallibility, He cannot be mistaken. In virtue of our analysis of a free action, that Tom freely attends the baseball game at t2 entails that Tom, at t2, can (has the freedom to) refrain from attending the baseball game at t2. If Tom exercises this capacity at t2, and so refrains from attending the baseball game at t2, God’s belief at t1 that Tom will attend the baseball game at t2 would be mistaken. Since God cannot have a mistaken belief, it must be that Tom cannot at t2 refrain from attending the baseball game at t2. So, pace our starting assumption that Tom freely attends a baseball game at t2, Tom lacks the freedom to attend the baseball game at t2. So, the existence of a traditional Judeo-Christian God or, more to the point, the necessary existence of any infallible being, is incompatible with human free will. In worlds that include infallible individuals, the only acts that humans can perform are those they actually do perform.

Though requiring tweaking and refinement, this encapsulates the reasoning that grounds the incompatibilist view that divine foreknowledge and human free will cannot co-exist. Fischer endorses incompatibilism and, as we will discover, for reasons not very different from those provided in this original argument. Compatibilists, on the other hand, believe that we can assuage whatever worries this argument elicits; we can maintain our self-conception as free agents in the face of an infallible deity. Especially for Christians, compatibilism is the thesis of choice. The narrative of salvation demands both human freedom and divine infallibility. Remove the former, and we are left as purposeless creatures mulling about in a meaningless world; erase the latter, and God loses His insurpassibility, and with it, His claim as the most perfect conceivable being.

 

2

Fischer begins his defense of incompatibilism by endorsing our firm opinion that the past is inviolable; relative to any moment in time, we are completely impotent to affect or influence the constitution of the past. The aphorism that there’s no point in crying over spilt milk captures the idea that what has already happened is ‘over-and-done with’, and that trying to access the past or to make it available for alteration is a fool’s errand. This attitude reverses when we think of the future, where we believe that our present actions frequently influence what will subsequently occur. While we now are incapable of shedding the horrors of Nazism, the assassination of Gandhi, and the Yankee victory in the 1927 World Series, we believe that whether we now (freely) drop a glass or not will likely have a decisive effect on the constitution of the future. Dropping the cheap glass from 100 feet above the concrete sidewalk will likely have the future include many shards of glass at the location of contact; simply holding onto the glass will almost certainly result in a shardless sidewalk. While we think that we are impotent regarding some segment of the future’s constitution (e.g., 2+2 will still equal 4 regardless of what acts we now perform), we are convinced that we are incapable of making any alterations to the constitution of the past; the future is, in large measure ‘open’ while the past is completely ‘closed’ or ‘fixed’.

The ‘fixity of the past’ will play a prominent role in our discussion (no surprise since the alleged incompatibility involves divine foreknowledge) so it is eminently reasonable for compatibilists to ask for some codification of what is, after all, presumably a metaphor. Fischer suggests the very plausible and philosophically popular

(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 wouldn’t have been a fact, then S can’t at T do Y at T.1

Reverting to the 1927 World Series victory by the Yankees, our intuition about our present incapacity to change this fact in any way seems perfectly captured by (FP). Try now (or any time subsequent to the moment of the Yankee victory) to perform some act that, as a consequence of its performance, eliminates the Yankee victory from the annals of the past. You will fail, and for good reason; as (FP) informs us such an action cannot be performed on pain of violating the ‘fixity’ of the past. Suppose, on a more somber note, that it’s a fact that 6 million Jews were slaughtered by the Nazis in the 1930s and 40s. Suppose that Miriam were to ride her bike now, it would not be a fact that these 6 million were murdered. Since Miriam’s current bike-riding cannot have this salutary consequence in virtue of the inviolability or ‘fixity’ of the past, Miriam now cannot ride her bike. (FP) provides us with a commonsensical articulation of what it means to characterize the past as ‘fixed’.

Incompatibilists now implement (FP) to their advantage. Recall that our necessarily omniscient deity had at t1 the belief that Tom will ride his bike at t2. If Tom’s ride was freely performed then, at t2, he can refrain from riding his bike at t2. Let Tom exercise his capability, and so not ride his bike at t2. But if this were so, the fact that God, at t1, had the belief that Tom will ride his bike at t2 would not have been a fact. (FP) disallows this, and consequently Tom cannot refrain from riding his bike at t2. And this entails that our assumption that Tom freely rode his bike at t2 cannot be sustained.

 

3

What’s a compatibilist to do? Well, perhaps the initial lure of (FP) is meretricious, and when placed under closer scrutiny fails to live up to its billing. The most direct and transparent way to demonstrate the unsuitability of (FP) is to present clear cases in which both

1. i. S can do Y at T.
2. ii. If S were to do Y at T, some fact about the past relative to T would not have been a fact.

Compatibilists have constructed quite a few examples that purport to manifest the compatibility of (i) with (ii). In fact, Fischer himself has contributed such an example which we will describe after discussing perhaps the most famous of this group. We turn first to Alvin Plantinga’s carpenter ants.

(Carpenter Ants) Let us suppose that a colony of carpenter ants moved into Paul’s yard last Saturday. Since this colony has not yet had a chance to get properly established, its new home is still a bit fragile. In particular, if the ants were to remain and Paul were to mow his lawn this afternoon, the colony would be destroyed. Although nothing remarkable about these ants is visible to the naked eye, God, for reasons of his own, intends that it be preserved. Now as a matter of fact, Paul will not mow his lawn this afternoon. God who is essentially omniscient, knew in advance, of course, that Paul will not mow his lawn this afternoon; but if he had foreknown instead that Paul would mow this afternoon, then he would have prevented the ants from moving in. The facts of the matter, therefore, are these: if Paul were to mow his lawn this afternoon, then God would have foreknown that Paul would mow his lawn this afternoon; and if God had foreknown that Paul would mow his lawn this afternoon, then God would have prevented the ants from moving in. So, if Paul were to mow his lawn this afternoon, then the ants would not have moved in last Saturday. But it is within Paul’s power to mow this afternoon. There is therefore an action he can perform such that if he were to perform it, then the proposition that the colony of carpenter ants moved into Paul’s yard last Saturday would have been false.2

And now let’s examine a close paraphrase of Fischer’s very similar story about the salty old seadog.3

(Seadog) Each morning at 9 am, for the past 40 years, the salty old seadog has called the weather service to ascertain the weather at noon. If the weatherman says at 9 am that the weather will be fair at noon, the seadog goes sailing at noon. And if the weatherman says that the weather won’t be fair at noon, the seadog never goes sailing at noon. The seadog has certain extremely regular patterns of behavior and stable psychological dispositions- he is careful to find out the weather forecast, is not forgetful, confused, or psychologically erratic, and whereas he loves to go sailing in sunshine, he detests sailing in bad weather.

It is now noon, and at 9 am this morning theseadog called the weather service which reported that the weather at noon would be horrible. The seadog is healthy and alert, and his sailboat is ready to go. Bearing in mind the weather forecast, he decides at noon not to go sailing. Yet he can go sailing at noon; there’s nothing hindering him from acting out of character (e.g., nobody is holding a gun next to his head threatening to shoot him unless he doesn’t go sailing). And given his hyper-reliable history and psychological capacities, it’s reasonable to think that if the seadog were to go sailing at noon, the weather report at 9 am would have predicted fair weather at noon. So, if the seadog had acted in a way that he was free to act at noon (i.e., had he gone sailing at noon), then the past would have been different from the way it actually was (i.e., the past would have consisted of a report at 9 am predicting fair weather rather than horrible weather.)

We can easily deify this story. God believes at 9 am that the seadog will not go sailing at noon. But if the seadog had- uncharacteristically- gone sailing at noon, then God would have had a belief at 9 am (viz., the belief that the seadog will go sailing at noon) that would be different from the belief that God actually had. Just as in (Carpenter Ants), we have a case where, had an individual acted differently at t2 (a course of action that he is (allegedly) free to take) from the way he actually acted at t2, God would have had a different belief at t1 from the belief that he actually had at t1. As with (Carpenter Ants), the requirements set out in (i) and (ii) for a tale that would demonstrate the complaint against (FP) have been satisfied.

To avoid misunderstanding, I should emphasize that neither Paul nor the salty old seadog have the power to change the actual past. It is not- to take the example of seadog- as if the past relative to t2 really included the weather report predicting horrible weather and then, by virtue of the seadog freely going sailing at t2, the past changes to now including the fact that the weather report reports fair weather. There is not, that is, a particular constitution of a past that is replaced by a different constitution of the past. Rather, the tales of (Carpenter Ants) and (Seadog) are intended to show that agents have the counterfactual power to affect the constitution of the past. The power is ‘counterfactual’ because in (Seadog), for example, the weather report at t1 predicted poor weather at t2 but- contrary to fact- if the salty old seadog had gone sailing at noon, the past would not have included a weather report predicting poor weather at noon, but instead would have contained a weather report predicting good weather at noon.

 

4

I speculate that most people not steeped in this debate concerning the (in)compatibility of divine foreknowledge and free-willed actions would find the stories of (Carpenter Ants) and (Seadog) gimmickry, almost as if they serve as bases of bad Borscht Belt jokes.

So my girlfriend and I are looking out the window and we see some guy jump from the twentieth story of our hotel room. In a couple of seconds we see him splattered on the pavement. I turn to my girlfriend and tell her if I’d ever jump from the twentieth floor, nothing bad would happen to me. She asks ‘why’s that’, and I say ‘whaddya think I’m crazy. The only way I’d ever jump if there was a net waiting for me’. Now- and this tells you why we broke up- she says ‘yeah, well why don’t you test your idea and jump.’ (Big rim shot.)

Of course, I wouldn’t take her up on her challenge, and I suspect neither would you. None of us would think that, rational as we are, a net would magically appear on our downward trek. And yet, there seems to be something to the idea that if I’m really sane, have the psychological stability to ward off any suicidal thoughts, and am expectantly waiting to see next week’s Knick game, then I couldn’t freely choose to make the leap. Of course, someone could push me (my girlfriend?) or I can slip on the wet kitchen floor, but these and other similar provocations would make my journey unfree, an act that didn’t issue from my will after consideration of various courses of action.

The point- which Fischer makes in several places (albeit far less humorously) is that we should not blithely agree that in (Carpenter Ants) and (Seadog) that it is self-evident that either John could mow his lawn on Saturday or that old Salty could have sailed at noon. It’s true, of course, that the usual constraints that compatibilists conceive as freedom-reducing, if not freedom-extinguishing, are absent. Paul has his lawnmower directly in front of him, it’s in excellent working order, Paul has not just experienced a stroke, and so forth, but even a compatibilist must concede that even if these typical restraints on free-willed action are missing, we are not dealing with anything approaching an ordinary case. On the face of it, there’s a world of difference between conditions that are sufficient to impede us from freely acting in a way that will affect the constitution of the future, and conditions that suffice for us to have counterfactual power over the past. None of this is to say that neither Paul nor Salty can act freely in their circumstances- perhaps Paul can freely mow his lawn on Saturday, and maybe Salty can freely sail at noon-but it is to say that the satisfaction of (i) is not as obvious as the narratives would have us believe. We need some argument where none has been provided.

Fischer, however, manifesting his characteristic generosity allows, arguendo, examples such as (Carpenter Ants) and (Seadog) to successfully meet their aims. He reaches this point by two magnanimous gestures. First, he allows that the correct analyses of counterfactual conditionals (‘if Paul had mowed his lawn today, God would have prevented the ants from forming a colony on his lawn last Saturday’; ‘if Salty had gone sailing at noon, the weather report would have predicted fair weather’) are still not matters of settled law, and so it would be hubristic to believe that we currently have a good handle on the truth-conditions of ‘backtracking’ conditionals. Secondly, he provides a ‘possible-world’ analysis of both (i) and (ii), in which the truth-conditions for the ‘can’ claim and the subjunctive conditional are compatible. The upshot is that, with all these concessions, we would be dogmatic to continue our allegiance to (FP).

At this point one might think that Fischer has given away the store. After all, with the truth of (i) and (ii), it seems as though the thought-experiments of (Carpenter Ants) and (Seadog) are coherent stories that accomplish their intended compatibilist goals; we can, after all, have free-willed actions co-exist in a world with an infallible deity. This conclusion, however, is too quick. If we can supply a different, yet intuitively plausible principle that expresses our firm belief in the fixity of the past that remains unthreatened from tales such as (Carpenter Ants) and (Seadog), we perhaps could establish a distinct and compelling incompatibilist argument.

 

5

The suggested new ‘fixity of the past’ principle is:

(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.4

(FP*) is a claim the limits an agent’s capability to act by insisting that it be possible for that act to be an extension of the actual past holding the laws of nature fixed. Sometimes Fischer makes this point in terms of what worlds an agent can actualize; an agent can actualize (i.e., bring into existence) only those worlds with a past identical to the past that exists in the world in which the agent is attempting to actualize.

Fischer approvingly quotes Carl Ginet whom he credits for having the same idea:

If I have it open to me now to make the world contain a certain event after now, then I have it open to me now to make the world contain everything that has happened before now plus that event after now. We might call this the principle that freedom is freedom to add to the given past.5

Let’s re-visit Tom who refrains from attending the baseball game at t2. Given God’s omniscience, God believes at t1 that Tom would now refrain from attending the baseball game at t2. We now invoke (FP*). If Tom can at t2 attend the baseball game at t2, then there exists a possible world with the same past as the actual world relative to t2 in which Tom attends the baseball game at t2. But any possible world with the same past as the actual world contains God believing at t1 that Tom will refrain from attending the baseball game at t2, since God’s belief at t1 that Tom will refrain from attending the baseball game is part of the actual past. So, if at t2 Tom can attend the baseball game at t2, there is a possible world in which both (a) God believed at t1 that Tom would refrain at t2 from attending the baseball game at t2 and (b) Tom attends the baseball game at t2. But in virtue of God’s infallibility, there can’t be such a world, and so Tom cannot attend the baseball game at t2. So, there are no worlds in which, i.e., it is impossible that, humans are capable of doing other than what they actually do in the presence of a divinely omniscient deity. Incompatibilism is true!6

(FP*) is equally applicable to (Seadog). If Salty failed to go sailing at noon, then God knows at 9 am that Salty will refrain from sailing at noon. And given God’s infallibility, His belief that Salty will fail to go sailing at noon entails that Salty will not go sailing at noon. Given (FP*), Salty can go sailing at noon only if his sailing at noon can be an extension of the actual past. But God’s infallible knowledge at 9 am that Salty will not go sailing at noon is part of the actual past, and if Salty can go sailing at noon God would lack infallible knowledge at 9 am that Salty will not sail at noon. So, Salty’s sailing at noon cannot be an extension of the actual world, holding the laws of nature fixed. Consequently, Salty cannot go sailing at noon. Again, incompatibilism reigns.

I have grave doubts that Plantinga and other compatibilists would accept (FP*) as the appropriate vehicle to capture what we mean by speaking of the past as being fixed. We need here to be especially wary of question-begging. Plantinga and his ilk cannot claim that (FP*) is unsatisfactory just because stories such as (Carpenter Ants) are possible yet, on the other hand, the mere assertion by Fischer and other incompatibilists that (FP*) seems ‘reasonable’, ‘intuitive’, or the like, may be viewed by compatibilists suspiciously like question-begging as well. Regardless of this speculation, I think we have some justified worries with (FP*). We start with examining whether it has quite the power that Fischer attributes to it.

 

6

Introduced to the philosophical world by Robert Nozick in 1969, Newcomb’s Problem7 asks us to suppose

… a being in whose power to predict your choices you have enormous confidence. There are two boxes, (B1) and (B2). (B1) contains $1000. (B2) contains either $1,000,000 ($M) or nothing… You have a choice between two actions: (1) taking what is in both boxes, (2) taking only what is in the second box. Furthermore, and you know this, the being knows that you know this, ……

1. I. If the being predicts that you will take what is in both boxes, he does not put the $M in the second box.
2. II. If the being predicts that you will take only what is in the second box, he put the $M in the second box.

First, the being makes its prediction. Then you make your choice. What do

you do?

The ‘expected-utility maximization’ strategy, tells us to choose just Box (B2). If I choose two boxes, the predictor will have almost certainly predicted this, and so would have put nothing in (B2), and so by picking two boxes, I would only net $1000. However, if I choose just (B2), the predictor will have almost certainly predicted this, and so would have put ($M) in (B2). So I would get ($M), a 1000 times profit.

The ‘dominance’ strategy tells us to choose both boxes. After all, prior to making our choice the predictor has already either placed the $M in (B2) or placed nothing in (B2). If the $M is in (B2), then, by picking both boxes, I receive the $1,000 that he placed in (B1) and the $M he placed in (B2), and so receive 1,001,000. If the predictor did not place $M in (B2), and I take both boxes, I receive the $1000 that he placed in (B1), whereas if I choose only (B2), I receive nothing. My best choice, then, is to take both boxes.

Both strategies seem to employ good reasoning, and yet they suggest different courses of action. Therein lies the problem in Newcomb’s Problem.

It is not difficult to see why Fischer would think that his responses to compatibilist tales such as (Carpenter Ants) and (Seadog) would be relevant to attempts to solve Newcomb’s Problem. Although his final word on the (in)compatibilist debate regarding free-willed action and divine foreknowledge is intentionally independent of one’s position on determining when certain conditionals ‘backtrack’ and, if they do, how to evaluate their truth value, most of the literature assumes that an adequate account of ‘backtracking’ conditionals is the key to resolving the (in)compatibilist dispute. And in Newcomb’s Problem it certainly seems, at least at first blush, that the correct answer to the puzzle lies in figuring out the proper way to treat counterfactual and ‘backtracking’ conditionals. Correctly analyzing statements such as ‘if I were to take both boxes, the predictor would have placed no money in B2’ and ‘if I were to take just B2, the predictor would have placed $M in B2’ initially appear to be essential to solving the Problem. So we should expect that, just as he distances himself from a view on counterfactual claims when he advances his novel solution to the (in)compatibilist debate between free will and divine foreknowledge, Fischer will operate similarly when he has a go at Newcomb’s Problem.

Fischer begins by distinguishing between a merely inerrant predictor and an infallible predictor, and asks what we should do (what is most rational to do, what do we have most reason to do) in each case. A merely inerrant predictor is one who will not make a mistake, but is capable of making an error; the nature of an inerrant predictor is such that there are some possible worlds in which she makes an erroneous prediction. An infallible predictor is incapable of issuing a mistaken prediction; in virtue of her nature, there are no possible worlds in which her prediction is wrong.

Let’s begin with identifying the selection we should make when confronting an infallible predictor. Unsurprisingly- given his suspicion that the ‘can’ clause in both (Carpenter Ants) and (Seadog) are far from obviously true- Fischer questions whether the Puzzle conditions are coherent. Since Predictor is infallible and so cannot make a mistaken prediction, we cannot assume that we have the option to take one or two boxes. Since Predictor must accurately predict- and so there is no possible world in which Predictor makes an inaccurate prediction- we have a situation significantly different from the case where we supposed Predictor merely inerrant. Now we no longer have possible worlds that allow for either of our choices to be extensions of the actual world. Since we cannot in advance know which prediction Predictor made, we cannot know in advance what the actual past consists in. Nevertheless, despite this epistemic uncertainty, there is no metaphysical possibility that Predictor make an inaccurate prediction, and so there’s no metaphysical possibility for us to choose other than what we, in fact, choose.

In order to ensure intelligibility, we need to tweak the parameters of Newcomb’s Problem. While we don’t have the option to select either one box or two, we can still ask ourselves what course of action is rational to pursue. And here Fischer’s answer is that we ought to select just one box, B2. While we don’t know in advance of our prediction whether Predictor placed the $M in B2 or not, we do know in advance of our decision that we would be far better off if the predictor predicted that we take just one box—and so this is what we have most reason to do.

Fischer is acutely aware of the obvious rejoinder: if we cannot select other than the way we actually do select, then any deliberation in which we participate in deciding which choice to make can be nothing but a sham. Since there are no alternative courses of action open to us- necessarily, if Predictor predicted that we would select one box then we will select one box, and, necessarily, if Predictor predicted that we would select two boxes then we will select both boxes- and deliberation requires that there be, metaphysically, options that we can instantiate- it would appear as if we cannot deliberate about whether we should select one box or two.

Fischer may not deny that we cannot deliberate about whether to choose one box or two boxes. Rather, the object of the deliberation is to identify what we have most reason to do. But in the case at hand, identifying what we have most reason to do is simply identifying which selection we have most reason make. At this point, the question is whether deliberation about what selection we have most reason to choose makes sense even if the choice one makes is fated. Alternatively, the challenge is to show that when there is only one choice available to us it is not farcical to speak of practical reasoning terminating with a decision about what choice to make.

In his reply to J.H. Sobel, who effectively raises this objection, Fischer insists that, although ‘having a choice’ requires open alternatives, ‘making a choice’ does not.8 Although, it may well be true that I have no choice about whether to select one box or two, I can still intelligibly ask myself whether I have better reasons to choose one box or two boxes. Fischer adds that I have better reasons for making the choice of one box, where these ‘better reasons’ consist in the consideration that I’d be better off (i.e., richer) if the one choice I have (indeed, the unique choice that I must have given an infallible predictor) is to select one box rather than two. So, if at bottom Newcomb’s Puzzle is a puzzle about what is rational to choose, then the answer is, in the case of an infallible predictor, selecting just B2. As long as we don’t know which of the two choices are necessitated- that of choosing two boxes or choosing one- the fact that one of these two are necessitated does not render practical reason about what choice to make nugatory.

But we should not underestimate the scope and power of a necessarily infallible predictor. Just as she knows in advance of our decision what our choice will be (i.e., whether we will choose one box or two), she also knows in advance of our deliberative process exactly which choice we will come to believe to have the most reason to perform. Since the infallibility of the predictor serves as the ground for characterizing the selection (selecting either one box or two) as metaphysically necessary, the predictor’s infallible foreknowledge of both our deliberative process in ‘making our choice’ as well as its the practical terminus (i.e., the actual choice that is made) should also confer metaphysically necessary on both the process and its culmination. In the face of a necessarily omniscient being, the fact that we cannot (metaphysically) perform any actions other than ones we do perform applies to everything we do including, trivially, ‘making a choice’. In brief, there are doubts whether the distinction between ‘having a choice’ and ‘making a choice’ really does the work Fischer assigns to it.

The temptation is to climb to a meta-level. So, while begrudgingly admitting the problem of free deliberation- that there cannot be any such process in a world occupied with infallibly omniscient being- we still can freely deliberate about freely deliberating; we can still ‘make a choice’ to ‘make a choice’. But even cursory reflection shows that this tactic is a non-starter. Just as we applied the power of an infallibly omniscient being to our (first-level) deliberative process, we can apply it to any level meta-process. There is no escaping from the metaphysical clutches of such a remarkable individual. Still, it must be noted that this is scarcely a victory for the orthodox Christian. For even if this line of reasoning is persuasive, and the spirit of Sobel’s criticism survives, Fischer’s incompatibilism remains intact. Free-willed creatures still cannot co-exist with a necessarily omniscient being; additionally, we now know substantive practical reasoning also cannot survive in such an environment.

We turn now to Fischer’s resolution of the Problem when we face a merely inerrant predictor. Assume, first, that Predictor did place $M in B2. By invoking (FP*), we know that, although the only decisions I can make are those that can be extensions of the actual world, i.e., extensions of the world that includes the Predictor placing $M in B2, since the (merely) inerrant predictor can be mistaken, either of my possible selections- choosing both boxes and choosing just one box- can be extensions of the actual world. Thus, I really do have the option of taking both boxes or just one box. I should choose both boxes since I get more money than if I picked only B2 (I would receive $1,000,1000, rather than just $M).

Since Fischer’s reasoning here will be crucial to what follows and the argument is a bit compressed, it serves us well to expand upon what I take to be his thinking. First, we should emphasize that Predictor is inerrant. She will not make a mistaken prediction in the actual world; we can rest assured that in the world in which we make our prediction, Predictor will have predicted correctly. Next, we should emphasize the mere inerrancy of Predictor. She can make a mistake; equivalently, there is a possible world in which she does make a mistaken prediction.

There are exactly two ways that Predictor’s mistaken prediction can occur. If we chose both boxes, Predictor could have predicted that we would choose one box. In this possible world, Predictor would have placed $M in B2 reflecting the fact that in this possible world her prediction is wrong. In this possible world, we would collect $1000. The second way Predictor could be mistaken is by predicting that we would choose 2 boxes but we choose only box B2. In this possible world, Predictor fails to place $M in B2. In this possible world, we collect no money at all.

It might be thought that since the truth of (FP*) is Fischer’s working assumption- and so the only choices I can make are those that are extensions of the actual world- and Predictor is inerrant (i.e., never mistaken in the actual world), that I cannot select differently from what the Predictor predicts. True, I don’t know in advance of my selection which choice is the only choice that is an extension of the past in the actual world, but this epistemic uncertainty is irrelevant to the (alleged) metaphysical truth expressed by (FP*). So, how can Fischer allow- under his own diagnosis- that I actually have (i.e., have in the actual world) two choices available to me? How can he allow that in the actual world, I can select either both boxes or one box?

Here’s where the mere inerrancy of Predictor crucially enters the picture. For in the actual world, I have access to a world (i.e., I can actualize or bring about a world) in which the mistake is manifested. I have access to both a world in which I select one box and Predictor predicts that I will select two boxes, and a world in which I select two boxes and Predictor predicts that I will select just B2. That Predictor can be mistaken- that there are possible worlds (albeit, worlds that do not include the actual world) in which Predictor is mistaken- is what makes it (metaphysically) possible for me to choose either both boxes or just one. Whichever choice I make can be an extension of the actual world or, equivalently, either choice I make is an extension of some possible world to which I have access.

In the case, then, where Predictor is merely inerrant, a ‘dominance’ solution presents itself. I actually have two available choices and so we can apply the aforementioned reasoning that supported a ‘dominance’ rather than an ‘expected value’ strategy. No matter what Predictor predicted, I am better off choosing both boxes rather than one, a conclusion Fischer reaches, we should highlight, with no allusions to controversial ‘backtracking’ conditionals. All we need is a clever application of the (purportedly) less controversial (FP*).

My concern here is that a challenge can come from those who seek more clarity about what it is for two or more worlds to have identical pasts. 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. (Obviously, the pasts of the two non-actual possible worlds would also be identical.) This may seem unproblematic. In all worlds, the Predictor makes a prediction, and then- in the worlds’ future- the prediction becomes either true or false by virtue, presumably, of what selection we freely make. 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. And, further assuming that these three worlds (the actual world and the two possible worlds in which Predictor made mistaken predictions) are the only relevant possible worlds, (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.

None of this assumes that the pasts of the two possible worlds in which Predictor is mistaken contain Predictor’s mistaken prediction; this can be considered an open question. Moreover, none of this concern quite implies fatalism. But it does assume that when we speak of the pasts of possible worlds, it is legitimate to include facts that make reference to future state of affairs. It may be that our judgment that pasts of worlds are identical is a more complex matter than we ordinarily believe.

The final worry I have with Fischer’s treatment of Newcomb’s Problem is proposed most tentatively. My concern is that if the reasoning implemented in his resolution of the merely inerrant predictor is correct, we are committed to a most bizarre result, one which both Fischer and I would consider unacceptable.

Consider the 1927 World Series victory of the Yankees. Fischer commonsensically conceives this fact as ‘fixed’; we are now powerless to undo any aspect of this past event. In terms of Fischer’s favored ‘principle of a fixed past’, (FP*), any action that we can now perform- any world that we can now actualize- must have this Yankee victory as part of its past. Just as the inerrant predictor’s prediction is in the past relative to the time at which we make our choice, the Yankee 1927 World Series victory is now past. In the case of a merely inerrant predictor where I can access worlds in which Predictor’s prediction is incorrect, it seems that the contingent truth of the 1927 Yankees World Series victory would allow me access worlds in which the Yankees lost the ’27 World Series. So, if I can actualize worlds in which the inerrant predictor makes an inaccurate prediction- although, of course, Predictor will not in fact make an incorrect prediction- I’m not sure why I cannot actualize worlds in which the ’27 Yankees did not win the World Series. It’s true that I won’t actualize such worlds- that I won’t do anything to prevent the Yankee victory from occurring- just as I won’t actualize any world in which Predictor wrongly predicts, but this is consistent with me having the power to do so.

The story radically changes if the past event is necessary. In the case of the truth that yesterday 2+2=4, I cannot, let alone will not, now actualize any world in which this truth is not true. Here the analogy is with the limits of my decision-making when confronting an infallible predictor. Just as confronting an infallible predictor leaves me with no possible worlds in which my decision is other than that I make in the actual world, there are no worlds I can access that include the falsity of any necessary truth, and a fortiori, I cannot actualize any world in which 2+2 does not equal four. Just as infallibility or necessary omniscience filled all the logical space of alternative courses of action, leaving me incapable of actualizing a world that contains her mistaken prediction of my choice, the necessity of some past event disables me from now actualizing any world where the Yankees did not prove victorious in the 1927 Series.

Effectively, I am analogizing the effects that an infallible predictor have on my possible actions, with the effects that a necessary truth have on my behavioral options. Here Fischer and I agree; we both believe that in the case of meeting an infallible predictor we cannot make any prediction other than the one we actually make. I am also analogizing the effects of my behavior when in contact with a merely inerrant behavior to the alternative courses of action open to me when dealing with contingent truths. And I am with great hesitance suggesting that if Fischer’s reasoning regarding the merely inerrant predictor is correct- and so I can access worlds in which such a prediction is mistaken (i.e., not true), it would seem to apply mutadis mutandis when we confront past contingent truths. But when so applying this reasoning we have the presumably unacceptable result that I do now have the capacity to actualize worlds in which these contingent truths- such as the Yankee 1927 victory- are false. Left with this unacceptable result, we would have little choice but to believe that Fischer’s resolution of the Newcomb Problem involving a merely inerrant predictor must be rejected.

 

7

We may have good reasons not to restrict the domain of the actions we can perform at a certain time to only those that are capable of being extensions of the actual world, holding the laws fixed. (FP*) may be too strong a condition to place on performable actions.

(Basketball) You and a friend are watching the NBA playoffs on television. It’s the seventh game of the Finals, and Steph Curry is on the foul line with no time remaining, needing to sink both his foul shots to tie the game. Just as he misses his first free throw, you turn to your friend and say “I can make this shot”. Now, suppose your friend says “No way, there is no chance of you making that shot”. This surprises you since you are a fantastic basketball player, and an especially accomplished free throw shooter, facts of which your friend is well aware, and so you ask why he believes that you can’t make that foul shot. He responds by pointing out that Curry has already missed the free throw that you claimed you could have made, and so when insisting that you could make that free throw you would be attributing access to a world that, at that time, you had no access to: at the time of your boast you had no access to a world in which both Curry missed the first free throw and you made the foul shot.

Presumably, you would recognize this as a (poor) attempt at humor. Of course, when you said that you can make the foul shot, you didn’t mean that at this very moment in time, with the past as it is, i.e., including Curry’s missed foul shot, that you could have made that free throw. What you meant, and how the remark would be understood by anyone in his right mind, is roughly that if you were in in the situation Curry was in just a moment ago, you would be able to (i.e., could) make the shot. The use of ‘can’ seems common and appropriate, and yet does not require that the past be congruent to that of the actual world.

There are other imaginable scenarios that complement this narrative. He might say that you can’t at this very moment make the shot because you’re nowhere near a basketball, or because you’re now sitting down and don’t have the strength to toss the ball the required 15 feet. And so forth. However, (FP*) tells us that I can’t make the shot because I can only make the free throw if there is a possible world which has an identical past to the actual world in which I do make the shot. There is, we agree, no such world, but yet it seems wrong to say that I can’t make the free throw that Curry (who missed by the way) took.

The discussion could have taken a different course. Your friend understands what you mean but still adamantly thinks that you couldn’t make the shot. Why not? The pressure would be too great. Missing the shot means losing the championship, and millions of people are watching you, with the fate of millions of dollars being decided by your making or missing the free throw. Reminded of these circumstances, you might agree with your friend- or you might not. But the point is that not only are there occasions when the truth of a ‘can’ statement does not require that we consider only those possible worlds with pasts identical to the actual world (pace (FP*)), but that the truth of the claim requires a past that is distinct from that of the actual world. The ‘can’ of personal capacity, ability, or power is just too elastic to be fully captured by (FP*).

A very different kind of case that may be worrisome to advocates of (FP*) occurs in circumstances where we have reason to distinguish between two persons on the basis of their respective capacities to act despite the fact that both lack the same (first-order) ability to act. We frequently say that one person can do something that another cannot in virtue of the former possessing and the latter lacking the relevant meta-ability (‘generalized ability’).

(Garden) I’m walking with a friend through someone’s garden, and say that I can garden while denying that my friend has this ability. Although neither of us know a darn about gardening, I alone have the temperament necessary for gardening. I’m patient, meticulous, careful, and eager to learn this craft, while my friend has none of these characteristics. Moreover- and now we go beyond characterological requirements- I have a good color sense and a good sense of proportion, both of which my friend also lacks. Armed with this talent and attitude, I would be a wonderful gardener if only I had taken horticulture classes when I was an undergraduate. My friend, however, given both his lack of talent and his philistine character would not be a capable gardener even if he had taken a host of relevant classes many years ago. I have ‘what it takes’ to be a gardener, while he doesn’t.

In (Garden), my claim to attribute the ability to garden only to myself amounts to insisting that I have the ability to learn gardening techniques while my friend doesn’t. Of course, this does not mean that you should hire me now to prune your shrubs. At this time, I cannot tell the difference between pruning shares and a hoe. But it does mean that (only) if the past had been different from the way it actually was, I, unlike my friend, in virtue of my meta-ability to garden, would now have the ability to prune shrubs and plant fruits and vegetables in their appropriate seasons. At this moment I can do something my friend cannot although manifesting this capacity would require a past different from the actual past. To the extent (Garden) is persuasive, (FP*) is problematic.

 

8

Our Fate is a predictably superb collection of essays. For more than 3 decades, Fischer has written some of the most trenchant essays relating the timeless subjects of free will, God, and moral responsibility. Invariably these papers are written with charm, wit, clarity, accessibility, and humility. We can only hope for (at least) another 3 decades of work from such a coruscating mind.

Endnotes

 

[1] This formulation of (FP) appears in several of Fischer’s papers both in Our Fate and other works. Cf., p.118 in Our Fate for one instance.

[2] Alvin Plantinga, “On Ockham’s Way Out”, Faith and Philosophy 3 (1986) p.254.

[3] Our Fate, p.103 is one instance where this tale is narrated.

[4] Our Fate, p.110. In other places, Fischer adds the qualification that a necessary condition for assessing an agent as performing a free action is that we hold the natural laws fixed. (See, for example, The Metaphysics of Free Will: An Essay on Control (Blackwell: Cambridge, MA), 1994, p.88. For our purposes, this variance in articulating (FP*) will play no factor.

[5] Carl Ginet, On Action (Cambridge: Cambridge University Press), 1990, p.231.

[6] One might think that ‘indexing’ God’s beliefs to possible worlds may help compatibilism. See, Our Fate, p.72ff. for Fischer’s response to this proposal.

[7] Robert Nozick, “Newcomb’s Problem and Two Principles of Choice”, in N. Rescher’s, ed. Essays in Honor of Carl G. Hempel (Dordrecht: D.Reidel, 1969), pp.114-46. I use the abridged version that appears in Nozick’s “Newcomb’s Problem and Two Principles of Choice” in Campbell and Sowden, eds. Rationality, Cooperation, and Paradox: The Prisoner’s Dilemma and Newcomb’s Problem (Vancouver, University of British Columbia Press, 1985), 107-33. Fischer avails himself of this abbreviated version in his critical review of J.H. Sobel’s Puzzle for the Will in the Canadian Journal of Philosophy, vol.31, no.3 (Sept. 2001), pp. 427-44.

[8] See Fischer’s Critical Review of Sobel’s Puzzles for the Will (cf. n.7), pp. 439ff.