Philosophical Studies

, Volume 164, Issue 3, pp 829–844

Soft facts and ontological dependence

Authors

    • Institut für Christliche PhilosophieUniversität Innsbruck
Article

DOI: 10.1007/s11098-012-9917-4

Cite this article as:
Todd, P. Philos Stud (2013) 164: 829. doi:10.1007/s11098-012-9917-4

Abstract

In the literature on free will, fatalism, and determinism, a distinction is commonly made between temporally intrinsic (‘hard’) and temporally relational (‘soft’) facts at times; determinism, for instance, is the thesis that the temporally intrinsic state of the world at some given past time, together with the laws, entails a unique future (relative to that time). Further, it is commonly supposed by incompatibilists that only the ‘hard facts’ about the past are fixed and beyond our control, whereas the ‘soft facts’ about the past needn’t be. A substantial literature arose in connection with this distinction, though no consensus emerged as to the proper way to analyze it. It is time, I believe, to revisit these issues. The central claim of this paper is that the attempts to analyze the hard/soft fact distinction got off on fundamentally the wrong track. The crucial feature of soft facts is that they (in some sense) depend on the future. Following recent work on the notion of dependence, however, I argue that the literature on the soft/hard distinction has failed to capture the sense of dependence at stake. This is because such attempts have tried to capture softness in terms of purely modal notions like entailment and necessitation. As I hope to show, however, such notions cannot capture the sort of asymmetrical dependence relevant to soft facthood. Arguing for this claim is the first goal of this paper. My second goal is to gesture towards what an adequate account of soft facthood will really look like.

Keywords

Free willDeterminismOntological dependenceForeknowledgeOckhamism

1 Introduction

Almost 50 years ago, Nelson Pike published his seminal paper, “Divine Omniscience and Voluntary Action”, arguing for the incompatibility of divine foreknowledge and human freedom to do otherwise.1 That paper provoked a veritable avalanche of work of the issues of foreknowledge and freedom. Perhaps the driving intuition behind Pike’s argument (and its successors) became known as the “Principle of the Fixity of the Past”—roughly, in the words of Carl Ginet, that our freedom is the freedom to add to the “given past”—that is, the past, intrinsically considered.2 And perhaps the main reply to Pike-style arguments discussed in the subsequent literature became known as “Ockhamism”. The Ockhamist agrees that if we can perform an action, our performing it must be consistent with the “given past”. However, the Ockhamist’s core contention is that God’s relevant past beliefs (concerning our future actions) do not, contra Pike, belong to the “given past” after all. Instead, in Pike’s universally adopted terminology, the fact that God had the relevant beliefs is a so-called “soft” (temporally extrinsic or relational) rather than “hard” (temporally intrinsic) fact about the past. The dispute between proponents of Pike-style arguments and Ockhamists was thus primarily a dispute how we should characterize (and what should be included in a statement of) the temporally intrinsic state of the world at a time. A substantial literature arose in connection with these distinctions, though no consensus emerged as to the proper way to analyze them, and memory of the vigorous dispute is—perhaps fortunately, according to some—fading.

It is time, I believe, to revisit these issues. The central claim of this paper is that the attempts to analyze the hard/soft fact distinction got off on fundamentally the wrong track. The crucial feature of soft facts is that they (in some sense) depend on or hold in virtue of the future. Following recent work on the notion of dependence, however, I argue that the literature on the soft/hard distinction has failed to capture the sense of dependence at stake. This is because such attempts have tried to capture softness in terms of purely modal notions like entailment and necessitation. As I hope to show, however, such notions cannot capture the sort of asymmetrical dependence relevant to soft facthood. Arguing for this claim is the first goal of this paper. My second goal is to gesture towards what an adequate account of soft facthood will really look like.

Perhaps the primary upshot of my paper is simple: Ockhamist defenders of the compatibility of foreknowledge and freedom at the very least have more work to do; in fact, I ultimately conclude that the Ockhamist program is a failure. However, the soft/hard fact distinction has various important applications beyond the foreknowledge dispute. For instance, determinism (standardly) is the thesis that the temporally intrinsic state of the world at some given past time, together with the laws, entails a unique future (relative to that time). Philosophers attempting to define determinism, however, have (as far as I have seen) simply set aside the problem of characterizing what should and shouldn’t appear in a statement of “the state of the world at a past time” relevant to these definitions.3 The topics of this paper should thus be of more general metaphysical interest.

The plan of the paper is as follows. First, I introduce the notion of a soft fact by showing how this notion is needed to give a proper interpretation of the fixity of the past. I then turn to attempts in the literature to analyze the hard/soft distinction, and argue that they all share the same fatal defect. I then suggest a new characterization of the distinction.

2 Soft facts and the fixity of the past

It is widely agreed that one can perform some given action only if there is some possible world suitably related to (and thus ‘accessible from’) the actual world in which one does perform the action. However, just which worlds are accessible from the actual world in this way is highly controversial. According to Pike (and, for instance, incompatibilists about causal determinism and freedom), one can perform an action only if (as a first approximation) there is a possible world with the same past as the actual world (up to the time just prior to one’s action) in which one performs it. This is the thesis of the fixity of the past. Pike employed the idea that the past is fixed to argue that divine foreknowledge is incompatible with freedom. Suppose God believed 1,000 years ago that Jones would sit at (some future time) t. Since God is essentially infallible, God’s having had this belief 1,000 years ago entails that Jones sits at t. According to the fixity of the past, however, Jones can do otherwise than sit at t only if there is a possible world with the same past as the actual world in which he does not sit at t. However, given that God had the relevant belief 1,000 years ago, there is no such world. For any world with the same past as our world will include the fact that God had that belief 1,000 years ago, and in all such worlds, Jones sits at t. Generalizing, divine foreknowledge is incompatible with human freedom.4

The above definition of the fixity of the past appeals to the notion of a world with the same past as our world. But the incompatibilist would wish to point out that this definition has an implicit restriction. After all, critics have argued that the principle seemingly requires too much. Suppose the incompatibilist said that what’s needed is a possible world with the same past, where “same past” is unrestricted in any way. Then facts such as the following could be included in a statement of the “given past”, or in any world with the “same past” as our world:
  1. (1)

    John F. Kennedy was being shot in Dallas in 1963, 49 years prior to my writing this paper in 2012.

     
  2. (2)

    Smith correctly predicted yesterday at t1 that Jones would sit tomorrow at t3.

     
  3. (3)

    Smith truly believed yesterday at t1 that Jones would sit tomorrow at t3.

     
And so on.5 A little reflection will tell us that if our performing an action must be consistent with “the past” in this sense, then the result will be logical fatalism—contrary to the intention of the incompatibilist advancing the argument. When the incompatibilist says that all we can do is add to the “given past”, she means that all we can do is add to the past intrinsically considered, where “the past” thus restricted will not include such facts as (1)–(3). Whereas (1)–(3) express soft facts about the past, however, (1*)–(3*) plausibly express hard facts about the past:
  1. (1*)

    John F. Kennedy was being shot in Dallas in 1963.

     
  2. (2*)

    Smith predicted yesterday at t1 that Jones would sit tomorrow at t3.

     
  3. (3*)

    Smith believed yesterday at t1 that Jones would sit tomorrow at t3.

     
But the Ockhamist now senses an opening for a reply to the argument given above. Once we see that the fixity of the past only applies to hard facts about the past, perhaps there is room to contend that facts about God’s past beliefs (about our future decisions) should ultimately be classified as soft (with (1)–(3)) rather than hard (with (1*)–(3*)).

For our purposes, it is crucial to see why one might think that soft facts about the past needn’t be fixed for us at later times. Suppose yesterday Smith predicted that Jones would sit tomorrow at t3. Intuitively, whereas (at least according to the fixity of the past) there is nothing Jones could now do about that, other things being equal, there may indeed be something Jones can do about whether Smith’s prediction was correct; perhaps Smith has it is his power to sit or stand at t3. After all, whether Smith’s prediction was correct or not depends (in an important way) on what Jones does at t3; if Smith’s prediction was correct, then it was then correct in virtue of how the future was (relative to the time of his prediction)—in particular, in virtue of Jones’s sitting at t3. Thus, so the thought goes, even if Smith’s prediction was correct (when he made it), this fact about the past needn’t be held fixed when evaluating what Jones can do at t3. In short, soft facts about the past express facts that once held because of or in virtue of how the future was (relative to the given times), and sometimes in virtue of our own future decisions. Most thus contend that at least some soft facts about the past needn’t be fixed.

At this point, however, the proponent of the incompatibility of foreknowledge and freedom contends that the fact that God had the relevant beliefs 1,000 years ago is a hard fact about 1,000 years ago. Ockhamists, however, have given various characterizations of soft facthood on which it turns out that God’s relevant past beliefs are soft. However, as I see it, these accounts have not captured the particular way in which soft facts crucially depend on the future. To these issues I now turn.

3 The entailment view of soft facts

So far I have just been speaking in the standard way of hard and soft facts about the past. Given various problems with the notion of ‘aboutness’, however, my preferred unit of analysis will instead be what it is to be a soft fact at a time.6 But we can translate talk of soft and hard facts at past times into talk of soft and hard facts about the past as follows. It was a hard fact at the given moment in 1963 that Kennedy was then being shot; ‘Kennedy is being shot’ was a hard fact at this time. We can thus say that it is a hard fact about the past that ‘Kennedy is being shot’ was then a fact. In general, if F was a hard (or soft) fact at given past time, then it is a hard (or soft) fact about the past that F was then a fact.

Marilyn Adams was the first to propose an analysis of the soft/hard fact distinction, and her analysis proved to be a trendsetter. The basic idea of Adams’ approach is this. A fact at a time is soft if and only if something’s happening or not happening in the future (relative to that time) is a logically necessary condition for the fact’s being a fact. In other words, a fact at a time is soft if it entails something about the future.7 After all, the most obvious difference between (1)–(3) and (1*)–(3*) is that the former all express facts that entail something about the future (relative to the given times), whereas the latter do not. For instance, the fact that Kennedy was being shot 49 years prior to my writing this paper entails something about the future (relative to 1963), namely, that I write this paper.

We might motivate Adams’ approach as follows. When a fact obtains necessarily only if some other fact obtains, the former fact would seem to exhibit a certain sort of dependence on the latter fact. Soft facts at times, then, are those facts that have logically necessary conditions in the future; they require certain future happenings in order to be facts, and thus depend on those happenings. On the other hand, hard, temporally intrinsic facts at times will be those facts that lack such necessary conditions in the future, and thus do not depend on the future. Further, Ockhamists would wish to point out that, on this criterion of soft facthood, it turns out that God’s having had the relevant belief in the past is a soft fact about the past. For, necessarily, God believed 1,000 years ago that Jones would sit at t only if Jones sits at t. On this approach, God’s having had this belief would then depend on Jones’ sitting in the sense relevant to soft facthood.

Unfortunately, Adams’ analysis had certain technical defects. As John Martin Fischer pointed out, for instance, the (obviously hard) fact at t1 that Jones is sitting entails this much about the future (relative to t1): that he does not sit for the first time at those times.8 Despite Fischer’s criticisms, it appears to have struck many that, though technically deficient, Adams’ analysis was nevertheless on the right track. So philosophers did not subsequently simply abandon Adams’ approach. Rather, quite naturally, they tried to modify it.

I move quickly from here. Hoffman and Rosenkrantz required that a soft fact entail something “unrestrictedly repeatable” about the future; this condition was (inter alia) meant to block Fischer’s ‘sitting for the first time’ counterexample.9 Freddoso invoked those facts that are “temporally indifferent” in the sense that (roughly) they could obtain at either the first moment, an intermediate moment, or the last moment of time; these are the hard facts at a time.10 Zemach and Widerker employed the notion of the set of facts compatible with the world’s ending at a time; these are the hard facts at that time.11 Similarly, Hasker’s analysis appealed to “future indifferent” propositions, which, he says, “must permit, but not require, that the entire universe should disappear and there be nothing at all” after the relevant time.12 Kvanvig’s analysis invokes those facts that are “independent” of later times, with the result that no fact that entails a certain sort of future fact will come out hard.13

Commenting on these prospective analyses, Fischer remarks that what he calls the “Entailment Criterion of Soft Facthood” appears to be “the engine driving the various approaches.”14 Similarly, Zagzebski comments that the core intuition driving these “more and more elaborate and sophisticated definitions” is that it is “compatible with the hard facts at t that the world go out of existence immediately after t, whereas the soft facts at t entail that the world continues in existence after t.”15 And this seems right; anyway, it is enough for my purposes to note that all of the proposed analyses would have it that if a fact at t entails a certain sort of ‘basic’ fact about the future (relative to t), such as that Jones mows his lawn at t10, then it is a soft fact at that time.16 Moreover, it seems clear that these analyses would also endorse the claim that a fact is soft only if it has such an entailment; no soft fact fails to entail something ‘basic’ about the future. Thus, we get the following accounts of soft and hard facts:

(ENT-S) A fact F at a time t is soft if and only if there is some ‘basic’ fact F* about the future relative to t such that, necessarily, F is a fact only if F* is a fact.

(ENT-H) A fact F at a time t is hard if and only if it is not the case that there is some ‘basic’ fact F* about the future relative to t such that, necessarily, F is a fact only if F* is a fact.

These are the fundamental claims of the literature on the hard/soft fact distinction. I believe, however, that they are mistaken.

4 Against the entailment view

Unfortunately, ENT-S does not seem to capture the sort of dependence relevant to soft facthood. Consider, instead of God’s past beliefs about the future, God’s past unconditional decrees (if such there be) concerning the future. If God has decreed that Jones is to sit at t, this entails that Jones sits at t—God’s decrees infallibly come to pass. Given this fact, it follows that God has decreed that Jones will sit at t necessarily only if Jones sits at t. Moreover, that Jones sits at t is obviously a ‘basic’ fact about t. However, it seems plainly incorrect to say that God’s past decrees concerning the future depend on the future (relative to the time the decrees were made). In other words, consider the theological model (broadly speaking) of the late Augustine and Leibniz. On their model, God has issued what we might call a ‘creative decree’ that in some way determines how the future of the world unfolds. Plainly, on this view, God’s having so decreed in the past is incompatible with the non-obtaining of what God decreed to come to pass. Thus, it will turn out that, necessarily, God has decreed that Jones sits at t only if Jones sits at t. However, again, it seems obvious that God’s decree does not obtain in virtue of Jones’ sitting at t; indeed, if anything, the dependence seems to go the other way around. (I return to this point below.)

In order to provide a counterexample to ENT-S, what is needed is a fact that obtains at a time that (1) necessitates some intuitively ‘basic’ fact about the future relative to that time but that (2) does not hold because of or in virtue of the fact it necessitates. Clearly, no paradigmatic soft fact will provide such a case; whereas the fact that Kennedy was being shot 47 years prior to my writing this paper necessitates my writing it, it nevertheless seems open to say that this fact obtained because of or in virtue of my writing the paper. However, God’s decrees meet both conditions. God’s decrees necessitate future events, but do not obtain because of those events. Though God’s decrees entail certain ‘basic’ facts about the future, what we should conclude here is not that facts about God’s past decrees are soft, but that, contra ENT-S and ENT-H, the intrinsic state of the world at a past time may nevertheless necessitate the intrinsic state of the world at a future time. In short, whereas all soft facts (such as (1)–(3)) will entail something ‘basic’ about the future, it is (very plausibly) not in virtue of having these entailments alone that these facts count as soft; in principle, hard facts also may have such entailments. So the entailment view fails.17

5 A comparison: ontological dependence and modality

I think this example shows that past accounts of the soft/hard fact distinction got off on fundamentally the wrong track. What this example reveals, I believe, is that the notion of entailment is insufficiently discriminating to capture the relevant notion of dependence. From the mere fact that X entails Y, we cannot conclude that X obtains because of Y. Philosophers working on the notion of ontological dependence have noticed precisely the same thing. Intuitively, some objects exist in virtue of the existence of other objects, and thus depend for their existence on those objects. For instance, sets depend for their existence on their members and friendships depend on their constituent friends. This notion seems to be of central importance to metaphysics, especially if you are inclined to think, to use Jonathan Schaffer’s recent phrase, that metaphysics is importantly about what grounds what.18

It has been commonplace to attempt to explain ontological dependence modally: one object depends for its existence on another just in case the former could not exist without the latter; sets cannot exist without their members, nor friendships without their friends. Following Kit Fine’s article, “Ontological Dependence”, however, it has been widely noted that, while this modal fact is (plausibly) necessary for the relevant sort of dependence, it is not sufficient.19 Fine pointed out that, necessarily, Socrates exists only if the set consisting solely of Socrates—singleton Socrates—exists; Socrates and singleton Socrates necessarily coexist, so neither could exist without the other. However, it is in virtue of Socrates that singleton Socrates exists, and not the other way around. Fine concluded that modal notions like entailment cannot capture the asymmetrical sort of dependence that holds when one object exists in virtue of another.

I have exploited a similar point here. Necessarily, God’s decree that Jones sits at t obtains only if Jones sits at t, but God’s decree does not obtain in virtue of Jones’ sitting at t. So an account of soft facthood in terms of entailment fails.20 In a sense, the primary aim of this paper is simply to bring the relatively new work on the notion of dependence to bear on the older, notoriously vexed literature on the hard/soft distinction. Something similar has been underway for some time concerning the definition of a substance. A traditional definition has it that a substance is something which depends on nothing else for its existence. Given the modal account of dependence, we thus get the following analysis of a substance, which I borrow from E. J. Lowe:

(SUB) x is a substance if and only if there is nothing y such that y is not identical with x and, necessarily, x exists only if y exists.21

Notably, SUB is an exact definitional analogue of ENT-H; Adams and company would have us define a hard fact as SUB defines a substance.22 However, as the example of Socrates and singleton Socrates brings out, SUB fails as a definition of substance. Socrates may or may not be a genuine substance, but he should not be demoted to nonsubstancehood merely because necessarily, he exists only if singleton Socrates does. Given Fine’s results, one might suspect, as I do, that there will be counterexamples to any purely modal account of any philosophical term or thesis which involves one thing existing (or holding, or…) in virtue of another. That is, one might suspect that we should simply give up on trying to capture this relationship modally. Here I have sought to confirm such suspicions at least with respect to one important philosophical distinction, the distinction between hard and soft facts about the past.
In the introduction to his 1989 volume of collected works on Ockhamism, Fischer presciently wrote the following:

Perhaps the Entailment Criterion could be justified as follows. When a fact’s obtaining at a time depends on a fact’s obtaining at some future time, then we think that the first fact is a soft fact about the time at which it obtains; we say that the first fact obtains in virtue of the obtaining of the future fact. And the notion of entailment might be thought to capture the relevant sort of dependence. When p entails q, the truth of p depends on the truth of q.

Whereas this might be the underlying rationale for the Entailment Criterion, it is not evident to me that it is adequate. This is because it is not evident to me that the notion of entailment captures the idea of dependence relevant to soft facthood.23

Fischer left it at that. I have argued that Fischer could have concluded with something much stronger here: that it is evident that the notion of entailment does not capture the sort of dependence relevant to soft facthood.

6 Soft facts, essence, and identity-dependence

Recall Fine’s example concerning Socrates and singleton Socrates. Neither can exist without the other. However, it is in virtue of Socrates that singleton Socrates exists, and not vice versa. Fine thus argued that we need to invoke the idea of an object’s essence to account for the relevant sort of dependence. According to Fine, it is no part of the essence of Socrates that singleton Socrates exists, whereas it is the essence of singleton Socrates to be the set consisting solely of Socrates.24 Lowe has made this same point using the example of Socrates and the temporally extended event of his ‘life’, and has drawn similar conclusions.25 In particular, Lowe has suggested—and I am inclined to agree—that the relevant sort of dependence is best understood in terms of identity-dependence; roughly, a thing exists in virtue of on another thing just in case the identity of the former is at least in part determined by the latter. For example, the identities of sets are determined by their members; the identity of singleton Socrates—which set that set is—is (in this case wholly) determined by its being the singleton set of Socrates. What this suggests, I believe, is that what it is for one fact to obtain at a time in virtue of a fact’s (or event’s, or…) obtaining a later time must likewise appeal to identity—namely, the identities of the facts in question.

The thought that soft facts will in some sense depend for their identities on the future has been suggested before, though not under this description. Consider this important passage from Fischer, which has been cited approvingly by Pike:

Consider the fact that Caesar died 2,009 years prior to Saunders’s writing his paper. What lies behind our view that this is not a hard fact about 44 B.C.? We might say that it is a soft fact about 44 B.C. because one and the same physical process would have counted as Caesar’s dying 2,009 years prior to Saunders’s writing his paper, if Saunders wrote his paper in 1965, and would not have counted as Caesar’s dying 2,009 years prior to Saunders’s writing his paper, if Saunders had not written his paper in 1965. This captures the ‘future dependence’ of soft facts; a soft fact is a fact in virtue of events which occur in the future.

Similarly, suppose that Smith knew at tl that Jones would do X at t2. Smith’s knowledge is a soft fact about tl because one and the same state of Smith’s mind (at tl) would count as knowledge if Jones did X at t2, and would not count as knowledge if Jones did not do X at t2. Exactly the same sort of future dependence explains why both facts - the fact about Caesar’s death and the fact about Smith’s knowledge - are soft facts.

Thus the incompatibilist [about divine foreknowledge and human freedom] might insist on the following sort of constraint on an account of the hard/soft distinction: the only way in which God’s belief at tl about Jones at t2 could be a soft fact about the past relative to t2 would be if one and the same state of mind of the person who was God at tl would count as one belief if Jones did X at t2 but a different belief…if Jones did not do X at t2. But it is implausible to suppose that one and the same state of mind of the person who is God at tl would count as different beliefs given different behavior of Jones at t2.26

Here we have, according to Pike, “as clear a statement as one could have of the intuition underpinning the hard/soft distinction.”27 Notably, Fischer is here talking about what counts as what. Whether the given physical process counts as taking place 2009 years prior to Saunders’ writing his paper is (at least in part) determined by whether Saunders writes his paper.28 Returning to examples (2) and (3) above, whether Smith’s prediction (or Smith’s belief) concerning Jones counts as correct or incorrect (or true or false) is (at least in part) determined by whether Jones sits at t. In this we have, I believe, the beginnings of an analysis of the hard/soft fact distinction—anyway, if the distinction can be analyzed at all. A fully general formal definition of soft facthood capturing these ideas is difficult to formulate, but suppose we started with:

(IDT-S) A fact F at a time t is soft if and only if F specifies an entity E as having a property P at t, and whether E counts as having P at t is at least in part determined by whether there exists an event or events in the future relative to t.

For instance: ‘Kennedy is being shot 49 years prior to Todd’s writing his paper’ was a soft fact at the given time t in 1963. This fact then specified a certain entity (in this case presumably an event or a state of affairs): Kennedy’s being shot. And it specified this entity as having the property of taking place 49 years prior to Todd’s writing his paper. But whether this entity counts as having this property is at least partially determined by whether there exists an event in the future relative to t, namely, my writing this paper. Similarly for Smith’s true belief about Jones: the relevant fact specifies a certain entity: Smith’s belief. (To what ontological category does a belief belong? It’s hard to say; this is why I have used the categorically-neutral ‘entity’ above.29) And it specifies it as having this property: being true. But whether this entity counts as having this property (at t) is at least in part determined by whether there exists an event in the future relative to t.

I won’t attempt a full defense of the proposal here. Suffice it to say that a range of questions could be raised about IDT-S that I must set aside, both because I lack the space to address them, and because I don’t really know how to do so. For instance, what is it for a fact to specify an entity? A hard question.30 Moreover, IDT-S (itself already an analysis of what it is to be a certain sort of fact) includes reference to a motley assortment of properties, events, and ‘entities’, which can presumably be from any ontological category you’d like. But perhaps the more pressing worry is whether the notion of identity-dependence (or something’s counting as something) underlying IDT-S is itself clear enough to serve as an informative analysis of soft-facthood in the first place. To this question, I am inclined to say that it is. Consider other, more mundane examples of extrinsicality (or relationality). Whether punishment counts as being just is at least in part determined by whether the person committed the given crime; whether the mark on my leg counts as a mosquito-bite is at least in part determined by whether it was produced by a mosquito; whether an assassination took place is at least in part determined by who was killed. It may not be easy to give a formal specification of this relation, but it seems clear enough. And the core suggestion is simple: when something in the future is doing this sort of determining, then we have soft-facthood.31 And when not, not.

7 Ockhamism and the fixity of the past: a reappraisal

I think we can further unpack IDT-S and see its virtues if we return to the issue that gave rise to these debates in the first place: the fixity of the past. Here we might employ as a heuristic the distinction between determination and discovery. Recall Smith’s belief about Jones. Intuitively, other things being equal, it need not merely fall to Jones to discover whether Smith’s belief was true; rather, whether it was true is determined by what he does. However, there would seem to be nothing that Smith could do that could determine whether Smith had the belief in the first place, for whether someone counts as having a belief (or that belief) at a time (very plausibly) is not (even in part) determinedby what happens at later times. In particular, one does not count as believing that something will occur even in part in virtue of the obtaining of what one believes will occur; if that were so, then it would follow that no one has ever had a false belief about the future, since one counts as having such beliefs partially in virtue of the very occurrence of the events one believes will occur. Similar remarks apply to Smith’s past prediction; one does not count as predicting that X will occur even in part in virtue of the future occurrence of X. As I see it, it is for precisely this reason that the fact that Smith made the given prediction belongs to the “hard” past, and further explains why (intuitively) Jones has no choice about whether he made it: nothing he does or doesn’t do could determine whether he made it.32

But now consider the case of God’s past decrees. Suppose in the past God has somehow decreed everything that shall happen in the future. Intuitively, if this is the case, then all I can do now is to discover (or reveal) what God’s decrees once were; by performing a given action, it follows that God decreed that I shall perform it. But by performing these actions, I do not thereby determine that God once decreed these actions; it is not in virtue of my doing X that some given past divine activity counts as decreeing that I shall do X. Nothing that happens now could determine whether God once made the given decrees; this explains why such decrees belong to the “hard” past, and further explains why (intuitively) there is nothing we can now do about whether God made them.33

And now we can come to crucial case of God’s past beliefs. Suppose that God had comprehensive beliefs in the past regarding what we do in the future. Now, in doing what we do, do we thereby determine what God’s beliefs once were, or is it that what we do merely reveals what they once were? In particular, does God count as having believed that you would perform a given action even in part in virtue of your performing it? The answer seems clearly to be ‘no’. As we saw above, one does not count as believing that something will happen even in part in virtue of the occurrence of that very thing; again, if that were so, then no one has ever held a false belief about the future. In short, not even God could count as believing that you will perform an action partially in virtue of your performing it; to maintain otherwise, I contend, would ipso facto be to deny that God really has the relevant belief. At this point, however, a curious question arises. For I seem to be treating the case of God’s beliefs precisely as I treated the case of Smith’s belief about Jones, and I treated the case of Smith’s belief about Jones as a paradigm case of a hard fact at time. Why then have I used the decrees case as a counterexample to the entailment view?—why not simply use the case of God’s beliefs?

Well, for one thing, given the history of the debate concerning Ockhamism, it may seem tendentious at best simply to offer the case of God’s beliefs as a counterexample to the entailment view. (Though Peter van Inwagen does precisely that.34) But the deeper answer goes to the heart of why there is a foreknowledge dispute in the first place. There is a long history of compatibilists about foreknowledge and freedom insisting on the following point: God knows (or believes) that you will perform an action because you will, in fact, perform it; and it isn’t the case that you perform the actions you perform because God knew (or believed) that you would perform them. Trenton Merricks, for instance, has recently been at pains to emphasize this point, and he himself traces this view as far back as Origen.35 The fact is, then, that according to many, God’s beliefs about the future share a significant commonality with soft facts at times; that God believes that something will happen “because it will happen” seems to suggest that God believes that it will happen because of its happening. And that is certainly very close to the notion of soft-facthood. It is for this reason that the case of God’s beliefs is so difficult—and also why the case of God’s decrees is a more effective counterexample to the entailment view. In that case, it is obvious that God does not decree that the event shall occur (in any sense) because that event (‘already’, so to speak) will occur, and so I don’t think anyone would be tempted to argue that facts about God’s decrees should come out soft.

My concluding advice to compatibilists about foreknowledge and freedom is thus the following: do not be Ockhamists. Again, the Ockhamist grants the fixity of the past: all we can do is to add to the given (“hard”, temporally intrinsic) past. So she denies that God’s past beliefs really belong to the “given past”. But the hard/soft distinction is, I believe, a matter of what counts as what; it is a matter of things at future times determining features of the world at earlier times. And it simply is not plausible that God’s beliefs could be like this; it is not plausible that God could count as believing that you will perform an action in virtue of your performing it. Belief just isn’t like this. And so Ockhamism was the wrong strategy to begin with. Rather, compatibilists about foreknowledge and freedom should admit that God’s beliefs—like beliefs must—belong to the given past; of course they do. They should instead maintain that God’s past beliefs—even though “fully past”—nevertheless substantively depend on (and were held in some sense because of) what we do, thus undergirding the key anti-fatalist thought that they needn’t be held fixed when evaluating what we can do. Of course, this sort of dependence will not be the sort of dependence at issue in soft-facthood; God will not count as having the given belief because of what one does. Nevertheless, God will still have it somehow because of what one does. The project for the compatibilist is thus to specify the relevant sort of dependence, and explain how it could obtain between God’s beliefs and our (future) actions in such a way that could make it plausible that we could have a choice about God’s (fully past) beliefs. This will not be an easy project. And any such compatibilist project will have to give up the very plausible picture of our powers as captured by the thesis of the fixity of the given past. My advice to compatibilists (which will, no doubt, be embraced with alacrity) is thus ultimately the following: become incompatibilists!

Footnotes
1

Pike (1965).

 
2

Ginet (1990, chap. 5).

 
3

For instance, in his chapter, “Defining Determinism”, in his book, A Primer on Determinism, John Earman says, concerning the challenge of distinguishing hard and soft facts, that he “will simply assume that the challenge can be met.” See his 1986, p. 15. And in his encyclopedia entry on the topic, Carl Hoefer (2010) invokes the “state of world at a time”, but does not address the distinction between soft and hard facts. See further van Inwagen (1983, p. 59).

 
4

This is Pike’s presentation of the argument in his 1977, which differs somewhat from his 1965.

 
5

Another standard candidate would be: (4) It was true yesterday at t1 that Jones would sit tomorrow at t3. There is, in general, no “canonical” way to refer to or express soft facts about the past, and I won’t detain us by trying to develop one here. How one will wish to more formally regiment the soft/hard distinction will substantially depend one one’s metaphysical framework concerning the status of facts, propositions, states of affairs, times, and the like. I am confident, however, that the basic points I develop below could be made mutatis mutandis within any such framework.

 
6

I thank Fritz McDonald in particular for helping me to see the need for this change.

 
7

See Adams (1967).

 
8

See Fischer (1983) in Fischer (1989).

 
9

See Hoffman and Rosenkrantz (1984).

 
10

See Freddoso (1983).

 
11

See Zemach and Widerker (1988).

 
12

See Hasker (1988).

 
13

See Kvanvig (1986, chap. 3).

 
14

Fischer (1989, p. 43).

 
15

Zagzebski (1991, pp. 72–73).

 
16

Plantinga, certainly one very important Ockhamist, denies that he is giving a “criterion” for soft facthood, but nevertheless maintains that “no proposition that entails [that Paul will mow his lawn in 1999] is a hard fact about the past.” Plantinga (1986) in Fischer (1989, p. 193).

 
17

Widerker (1990) also employs the example of God’s past decrees (and God’s past promises; the seemingly hard fact that God once promised to bring something about entails that the thing comes about) against the entailment view of soft facts. As Widerker says, and I agree, such decrees and promises seem “fully accomplished” and “over and done with”. In the literature on the hard/soft distinction, it is common for philosophers to talk of hard facts being (in Pike’s phrase) “over and done with.” In my view, what seems “over and done with” is best thought of as a heuristic in determining whether a fact about the past is hard or soft. My points here can thus be seen as trying to bring out the deeper reasons why the entailment view fails, and not just establishing (as I am inclined to think Widerker did establish) that it does fail.

 
18

See Schaffer (2009).

 
19

See Fine (1995).

 
20

Theists, of course, should be inclined to grant this example. But I think it would be a mistake even for non-theists to respond to the decrees case by saying that there is no God, or that the existence of someone who could make such determining decrees is impossible. What’s important is that the example is conceptually coherent. Similarly, I take it that it would be a mistake to respond to Fine’s singleton-set case by denying that there are (or even could be) sets. For a similar “neutrality” approach concerning ontological dependence, see Correia (2005, pp. 9–10).

 
21

Lowe (1998, p. 138).

 
22

SUB is sometimes also glossed as the idea that a substance is something that could exist all alone in the world. Intuitively, this idea is exactly parallel to the theme from Freddoso, Zemach and Widerker, and Hasker that hard facts are those facts which could obtain at the last moment of time or are compatible with the world’s ending. Further, these accounts of hard facts seem to be exactly parallel to David Lewis’ (subsequently much refined) suggestion that a thing’s intrinsic properties are those properties it could have compatibly with loneliness. See Lewis (1983) and Langton and Lewis (1998). There are substantial parallels between these two literatures that I lack the space to bring out.

 
23

Fischer (1989, p. 43).

 
24

See Fine (1995).

 
25

Lowe (1998, chap. 6). But see also Lowe (2006, chap. 3).

 
26

Fischer (1983) in Fischer (1989, pp. 93–94).

 
27

Pike (1993, p. 135).

 
28

Of course, Fischer only puts the point in terms of counterfactuals, but I think the core idea Fischer’s passage suggests is that whether the process counts as the process thus specified is (at least in part) determined by (and not merely counterfactually dependent on) Saunders’ writing his paper. Further, the counterfactual formulation suggested by Fischer seemingly runs into trouble with respect to God’s beliefs (and decrees); Fischer is in effect suggesting that the state of God’s mind at t1 which counts as his belief that Jones will do X at t2 would still count as that belief even if Jones didn’t do X at t2. I think this is ‘getting at’ something true, but it nevertheless relies on a (best avoided) per impossible conditional. I thank John Fischer for helping me to formulate this point about per impossible conditionals.

 
29

Or consider the fact at t1 that ‘Jones is bleeding to death’. Whether Jones is bleeding to death at t1 is—so one might maintain—in part determined by whether there exists in the future the event of Jones’ dying. Here, Jones himself (presumably a substance?) is the relevant entity.

 
30

For instance, one might suppose that ‘Kennedy is being shot 49 years prior to Todd’s writing his paper’ specifies the following entity: Kennedy’s being shot 49 years prior to Todd’s writing his paper. What would decide between this proposal and the one just given? My tentative thought is simply to say: it specifies both—but so long as it specifies an entity with the relevant features, then it should come out soft, and does so under my proposal. Of course, it’s tempting to define ‘specifies an entity’ as ‘entails that there exists an entity’, but, as I have been at pains to emphasize above, entailment is a rather blunt instrument, and though I can’t address these questions here, I have my doubts that it is fit for the needed task. (I thank Kenny Boyce for raising this issue.)

 
31

It is worth noting at this point that the relation here is not causal. The criminal’s having committed the crime (plausibly) does not cause the punishment to be just; rather, it helps to constitute its being just. And the mark’s having been produced by a mosquito does not cause it to be a mosquito bite, but helps to constitute its being a mosquito bite. Similarly, Jones’ sitting at t3 does not cause Smith to have been correct at t1; what’s at issue in soft-facthood is not backwards causation, but something more like backwards constitution (or determination). There are, of course, substantive issues concerning what (if any) ontology of time such a relation presupposes, but I must set these issues aside.

 
32

It is important to see that, just as it is no part of the concept of a soft fact about the past that someone has a choice about it, it is no part of the concept of a hard fact about the past that no one has a choice about it; if one thinks one has a choice about some past fact, one does not (or need not) ipso facto deny that it is hard. The “Multiple Pasts Compatibilist”, for instance, responds to the Consequence Argument by denying that even (uncontroversially) hard facts about the past must be fixed. For more on this issue, see Fischer (1994).

 
33

These points also apply to the case of causal determinism (the laws and the past plausibly don’t count as being what they are or were in virtue of one’s actions) and suggest a partial account what is “up to us”, viz. that whether something obtains is “up to one” only if one’s actions could determine that it obtains. But a full development of these points must lie outside the scope of this paper.

 
34

Referring obliquely to the debate about the hard/soft fact distinction, van Inwagen writes that “If some philosopher’s definition or analysis of ‘about the past’ has the consequence that the proposition that God believed in 1900 that I was going to lie at a certain moment is not ‘about the past’, that proposition constitutes a counterexample to that philosopher’s definition.” Van Inwagen (2008, p. 218).

 
35

See Merricks (2009); for a reply, see Fischer and Todd (2011), and in turn Merricks (2011).

 

Acknowledgments

This paper has a long history, and I have many people to thank, both for comments on previous drafts and for helpful conversations. In particular, I’d like to thank the members (or former members) of the Agency Group at the University of California, Riverside, where the first version of this paper was presented, Garrett Pendergraft, Chris Franklin, Ben Mitchell-Yellin, Justin Coates, Philip Swenson, Michael Nelson, and especially Neal Tognazzini. I’d also like to thank Phil Bricker (whose probing early comments led—I hope—to substantial improvements), Bill Hasker, Eleonore Stump, Dean Zimmerman, Michael Rea, Alan Rhoda, E. J. Lowe, Kathrin Koslicki, Alex Arnold, and especially Kenny Boyce and Andrew Bailey. Finally, and most importantly, John Fischer’s influence on the paper has been enormous, and I thank him for comments on countless drafts; John has been (as usual) incredibly supportive of this project from the very beginning, and for this I’m deeply thankful. A short version of the paper was presented at the Eastern meeting of the SCP at Wake Forest University (March 2010), the Alabama Philosophical Society, in Pensacola, Florida (September 2010), the 2010 Joint Meeting of the Illinois and the Indiana Philosophical Associations, in Charleston, Illinois, and finally at the 2011 Central APA in Minneapolis. Thanks to audiences at these conferences, and especially to Fritz McDonald for his excellent comments at the APA. Thanks also to Alvin Plantinga and Marilyn McCord Adams for comments at this session. This project/publication was made possible through the support of grant ID#15571 from the John Templeton Foundation. The opinions expressed in this publication are those of the author(s) and do not necessarily reflect the views of the John Templeton Foundation.

Copyright information

© Springer Science+Business Media B.V. 2012