Open Theism and Omniscience

Open theism is the view that God does not have comprehensive knowledge of the future. In particular, and most especially, God does not have knowledge of future contingent facts (true future contingent statements, where a future contingent statement is a contingent statement about the future such as, e.g., "There will be a sea battle tomorrow"), especially those concerning the future free actions of creatures. Prominent proponents of this view include William Hasker, Dean Zimmerman, Richard Swinburne, Peter van Inwagen, and Gregory Boyd. The primary philosophical motivation for open theism is the affirmation of human freedom along with the alleged incompatibility between comprehensive divine foreknowledge and human freedom. A common criticism of open theism is that it is incompatible with divine omniscience since it seems to straightforwardly deny that God knows all truths, particularly all the truths about the future.

In response to this criticism, open theists fall into two general camps. In the first camp are open theists who hold that there are true future contingents, and God does not know them. From here, such open theists either argue that God is nevertheless omniscient according to a proper account of omniscience or accept the implication that God is not omniscient. In the second camp are open theists who hold that there are no true future contingents in the first place and that thus God's omniscience is not even apparently challenged by the fact that He does not know future contingents.

Open theists in this second camp hold to the openness of the future according to which there are multiple possible futures (where the possibility in question is typically taken to be causal possibility), and there is no privileged possible future that is marked out as the actual future. Which future becomes actual is indeterminate, i.e., there is no settled fact of the matter as to which future becomes actual. Such an open future view clearly presupposes either presentism (according to which the present alone is real while the past and future are unreal) or the growing block theory of time (according to which the past and present are real while the future is unreal). Eternalist theories of time according to which the past, present, and future are equally real are incompatible with the thesis of an open future. As such, debates about the general nature of time are very much relevant to the consideration of whether the future is open. (It should also be said that while an open future presupposes presentism or growing block, neither of these theories requires an open future. William Lane Craig, for instance, is an example of a presentist who nevertheless rejects the thesis of an open future).

In a previous post, I sketched the following argument against open theism:

1. Future tense, contingent propositions have no truth values, or they are uniformly false, or they are uniformly true, or they are variously true or false and correspond to the reality of what will happen in the future.
2. It can’t be the case that future tense, contingent propositions have no truth values (because then tautologies like “there will or will not be a sea battle tomorrow” couldn’t be true, and they are true).
3. It can’t be the case that future tense, contingent propositions are uniformly false (because then contradictions could be derived like “it is false that there will be a sea battle tomorrow and it is false that there will not be a sea battle tomorrow”).
4. It can’t be the case that future tense, contingent propositions are uniformly true (because then contradictions could be derived like “it is true that there will be a sea battle tomorrow and it is true that there will not be a sea battle tomorrow”).
5. Therefore, future tense, contingent propositions are variously true or false and correspond to the reality of what will happen in the future (1, 2, 3, 4).
6. If future tense, contingent propositions are variously true or false and correspond to the reality of what will happen in the future, then there are contingent facts about the future.
7. Therefore, there are contingent facts about the future (5, 6).
8. If God is omniscient, then He knows all facts.
9. God is omniscient.
10. Therefore, God knows all facts (8, 9).
11. Therefore, God knows the contingent facts about the future (7, 10).
12. If God knows the contingent facts about the future, then God has foreknowledge.
13. Therefore, God has foreknowledge (11, 12).
14. If God has foreknowledge, then Open Theism is false.
15. Therefore, Open Theism is false (13, 14).

In this post, I want to consider various ways that open theists can resist this argument. Open theists in the first camp will reject either (8) or (9), while open theists in the second camp will reject either (2) or (3). In my view, any being that is not omniscient is simply not God. As such, I don't think that rejecting (9) is tenable, and few open theists reject (9) in any case. Those who reject (8) reject the standard account of omniscience and substitute an alternative account in its place. The standard account of omniscience is as follows (cf. Wierenga, "Omniscience," pp. 130; in The Oxford Handbook of Philosophical Theology):

(O1) S is omniscient if and only if S knows all truths.

A typical alternative definition of omniscience offered on behalf of open theists in the first camp comes from William Hasker (see "A Philosophical Perspective," pp. 136; in The Openness of God: A Biblical Challenge to the Traditional Understanding of God), who has proposed the following:

(O2) S is omniscient if and only if S knows everything that it is possible to know.

This definition admits of the following counterexample: It is possible to know that unicorns exist (since possibly, unicorns exist). From this definition, therefore, it would follow that God (being omniscient) knows that unicorns exist. But unicorns do not exist; hence, God “knows” a falsehood, which is absurd.

Clearly, then, this proposed redefinition of omniscience needs refinement. Here is a natural suggestion:

(O3) S is omniscient if and only if S knows every truth that it is possible to know.

In other words, omniscience is knowledge of all knowable truths. With this account of omniscience, open theists can argue that future contingents are not knowable and so the lack of knowledge of future contingents on God's part does not count against His omniscience. Thus, such open theists can accept (9) but reject (8). What should we make of this? William Lane Craig complains that such a definition of omniscience is objectionably "cooked." What he means by this is that O3 seems to be contrived so as to enable the open theist to continue to say that God is omniscient even though He might lack knowledge of various truths, and this conflicts with our clear pre-theoretical intuitions about what it means to be omniscient. According to those pre-theoretical intuitions, omniscience is a categorical notion, not a modal one; omniscience is simply knowledge of all truths, not knowledge of all knowable truths (Time and Eternity, pg. 254-255; see also "The Middle-Knowledge View," pp. 138; in Divine Foreknowledge: Four Views).

In response, it seems to me that this is not a compelling objection. For it is hard to complain that omniscience shouldn't be constrained by possibility. From the standpoint of perfect being theology, God is the greatest possible being. If an attribute is impossible, then it is no mark against God's greatness that He does not possess it. Insofar, then, as perfect being theology ought to lead us to affirm that God is omniscient, omniscience must be possibly had. Given this, construing omniscience as knowledge of all knowable truths seems quite legitimate. It is hard to fault a being for not knowing what cannot be known. Despite this, however, there is a problem with the idea that O3 is an alternative to O1. For there is what I take to be a fairly knockdown argument that O3 in fact entails O1 (i.e., knowing all knowable truths entails knowing all truths). This argument comes from Alex Pruss, though I have changed the specific details.

Suppose that it is in fact true that Colin will (freely) eat Cheerios for breakfast tomorrow. Let p = "Trump is not the president, or Colin will eat Cheerios for breakfast tomorrow." Then, God can know p because there is a possible world in which Trump is not the president, and God knows that Trump is not the president in that world (open theists grant that God knows all past and present truths). Further, p is true. It then follows by O3 that God knows p. Now, God knows (in the actual world) that Trump is the president. It thus follows that God knows that I will eat Cheerios for breakfast tomorrow. Why? Because it is true that Trump is the president, and God knows this. Further, p is true, and God knows p. But the first disjunct of p (viz., "Trump is not the president") is false, and God (of course) knows this. Hence, by disjunctive syllogism, it follows that since the whole disjunction is true, the second disjunct (viz., "Colin will eat Cheerios for breakfast tomorrow") must be true, and God knows this (surely God knows how disjunctive syllogism works). Hence, given the eminently plausible thesis that knowledge is closed under entailment, it follows that God knows that "Colin will eat Cheerios for breakfast tomorrow." At the very least, God is in a position to know that "Colin will eat Cheerios for breakfast tomorrow" (i.e., that "Colin will eat Cheerios for breakfast tomorrow" is knowable to God), and by O3, it then follows that God knows it. This reasoning generalizes to all future contingents and in fact to all truths, period. So, O3 entails O1, and O1 entails (8).

Thus, the open theist is still committed to the truth of (8) even given this redefinition of omniscience. As such, it seems to me that rejecting (8) or (9) is not a promising move for an open theist to make. Of course, further alternative accounts of omniscience can conceivably be offered by the open theist that might allow for the rejection of (8), but for the purposes of this post, I leave it to open theists who want to reject (8) to identify and defend the plausibility of such accounts. The ball is in their court.

This brings us to (2) and (3). Open theists of the second camp (open futurists) can accept (8) and (9) while rejecting either (2) or (3). Their strategy is to adopt an open future view according to which there is no unique actual future that will obtain and therefore no true future contingents. This involves either adopting an Aristotelian view according to which future contingents are neither true nor false or a view defended by Patrick Todd (and others) according to which all future contingents are false (see “Future Contingents are All False! On Behalf of a Russellian Open Future”). The former approach rejects (2), while the latter rejects (3). The Aristotelian view can be seen as the Strawsonian counterpart to Todd’s Russellian view. The idea that Todd proposes is that there is significant parallelism between future contingents and definite descriptions. A definite description is a phrase of the form “the such-and-such” that picks out an individual person, place, or thing. “The author of War and Peace” is an example of a definite description.

Bertrand Russell and P.F. Strawson famously disagreed about how to analyze statements containing definite descriptions, particularly definite descriptions that fail to refer. The most famously discussed example of such a statement is “The present king of France is bald.” On Russell’s analysis, this statement comes out false, while on Strawson’s, it comes out neither true nor false. According to Russell, the statement should be analyzed as a shorthand expression of the conjunction of the following three claims:

• There is a present king of France. [existence]
• At most one individual is a present king of France. [uniqueness]
• Everything that is a present king of France is bald. [maximality]

Since there is no present king of France, the first conjunct (i.e., the existence claim) is false and thus the whole conjunction is false. Consequently, “The present king of France is bald” comes out false. By contrast, in Strawson’s view, the existence of the present king of France is not asserted by the statement but is rather presupposed by it so that if this presupposition fails, then the statement is neither true nor false. Since, then, there is no present king of France, “The present king of France is bald” incurs a presupposition failure and thus fails to be either true or false.

The Aristotelian view of future contingents is parallel to the Strawsonian view of definite descriptions, while Todd’s “all false” view is parallel to the Russellian view. On the Aristotelian view, the statement that “There will be a sea battle tomorrow” presupposes that there is a unique actual future (specifically, a unique actual tomorrow) that either contains a sea battle or does not. Combined with the view that the future is open and that there therefore is no unique actual future, the statement incurs a presupposition failure and thus fails to be either true or false. By contrast, on Todd’s Russellian view, the statement “There will be a sea battle tomorrow” is to be translated in a manner parallel to Russell’s theory of definite descriptions. In particular, it is to be translated roughly as follows: there exists a unique actual future (specifically, a unique actual tomorrow), and that future features a sea battle. Combined with the view that the future is open and that there therefore is no unique actual future, the statement comes out false since the first conjunct is false.

On either view, the open theist thesis that God does not know future contingents is compatible with the standard definition of omniscience (i.e., O1). For according to both views, there are no true future contingents to be known in the first place. A significant price of either view, though, is that apparent tautologies turn out not to be tautologies after all. In particular, although the surface grammar of “there will be a sea battle tomorrow or there will not be a sea battle tomorrow” suggests that it is a tautology, it is not.

On the Aristotelian view, neither disjunct is true or false, and so the disjunction is also neither true nor false. A common way to formally flesh out this view is to adopt a three-valued logic that recognizes the truth values of true, false, and null. On every three-valued logic of which I am aware, if both disjuncts of a disjunction are null, then the whole disjunction is null. As such, the disjunction “there will be a sea battle tomorrow or there will not be a sea battle tomorrow” turns out to be null. It is not a tautology because the law of excluded middle (for any proposition p, either p or not-p) does not hold in three-valued logic. This fact by itself will be deemed by many to be a bridge too far (it certainly is for me, anyway). In my view, the law of excluded middle is more obviously true than any premise in an argument for the incompatibility of comprehensive divine foreknowledge and human freedom, and it is precisely such an alleged incompatibility (along with a commitment to divine omniscience) that is supposed to motivate the open future version of open theism in the first place. Even more implausible, in my view, is the alternative suggestion that we adopt a three-valued logic according to which “there will be a sea battle tomorrow or there will not be a sea battle tomorrow” is a tautology and yet neither disjunct is true. For how can a disjunction be true if neither disjunct is true?

On Todd’s Russellian view, both disjuncts are false and so the whole disjunction is false. As stated above, the statement “there will be a sea battle tomorrow” is to be translated in a manner parallel to Russell’s theory of definite descriptions. In particular, it is to be translated roughly as follows: there exists a unique actual future (in this case, specifically a unique actual tomorrow), and that future features a sea battle. Combined with the view that the future is open and that there therefore is no unique actual future, the statement comes out false since the first conjunct is false. The same is true, however, of the statement “there will not be a sea battle tomorrow.” On Todd’s view, this statement is to be translated roughly as follows: there exists a unique actual future (specifically, a unique actual tomorrow), and that future does not feature a sea battle. Once again, given an open future, the first conjunct is false, and so the whole statement is false.

So, both statements come out false on this analysis, and the disjunction “there will be a sea battle tomorrow or there will not be a sea battle tomorrow” turns out, on Todd’s Russellian analysis, to not be a tautology. Appearances notwithstanding, it is not a disjunction of the form p or not-p. Additionally, “there will not be a sea battle tomorrow” turns out not to be the negation of “there will be a sea battle tomorrow.” Both of these consequences seem rather implausible, although they do have the virtue of allowing Todd's theory to be compatible with the law of excluded middle and with there being only two truth values (viz., true and false), in contrast to the Aristotelian view. Yet another problem is that if I said yesterday that “there will be a sea battle tomorrow” and there is a sea battle today, it seems very implausible to think that I should retract my assertion on the grounds that it was false. On the contrary, it seems that I should regard my claim as having been true. Yet Todd's view implies otherwise. For these reasons, Todd’s “all false” view seems very implausible to me. Indeed, similarly to my aforementioned comment about the law of excluded middle, that I should regard my claim as having been true seems more obviously true to me than any premise in an argument for the incompatibility between comprehensive divine foreknowledge and human freedom, and it is precisely such an argument that is supposed to motivate open theism in the first place. In short, the cure is worse than the disease.

It is worth noting that the theologian Gregory Boyd also holds to an "all false" view of future contingents, but his account is importantly different from Todd's Russellian account. Like Todd, Boyd maintains that "there will be a sea battle tomorrow" and "there will not be a sea battle tomorrow" are not contradictories but are instead contraries (i.e., they cannot both be true, but they can both be false). But the reason for this according to Boyd is not that future contingents should be given the Russell treatment. Instead, the reason is this: the contradictory of "x will occur" is (according to Boyd) "x might not occur," and the contradictory of "x will not occur" is "x might occur." Further, if x is genuinely contingent, then it is true that x might occur and that x might not occur (see "God Limits His Control," pp. 197-198; in Four Views on Divine Providence). It follows in such a case that both "x will occur" and "x will not occur" are false. Thus, all future contingents are false.

Boyd, along with co-authors Alan Rhoda and Thomas Belt, expands on this theory in the paper "Open Theism, Omniscience, and the Nature of the Future." The authors seem to endorse the following semantics (which are basically what Arthur Prior called the "Peircean" semantics for the temporal operators "will" and "will not" after C.S. Peirce):

• "x will occur" is true if and only if x occurs in all causally possible futures.
• "x might not occur" is true if and only if it is not the case that x occurs in all causally possible futures.
• "x will not occur" is true if and only if x does not occur in any causally possible futures.
• "x might occur" is true if and only if x occurs in some causally possible futures.

The causally possible futures are the metaphysically possible ways things could go from the present that are consistent with the past and the natural laws. Given these semantics, it is clear that "will" and "might not" are contradictories as are "will not" and "might" and that "will" and "will not" are contraries (at least on the assumption that there are causally possible futures), while "might" and "might not" are (again, on the assumption that there are causally possible futures) sub-contraries (i.e., they can both be true, but they can't both be false). Further, it is clear that if x is causally contingent (i.e., not causally determined to happen) and so is something that both might and might not happen, then given these semantics, it is false that x will happen and false that x will not happen. Thus, these semantics more precisely flesh out Boyd's "all false" view of future contingents.

This view suffers from the same implausible consequences as Todd's view, and it faces an additional implausibility as well that is brought out in a criticism that Wiliam Lane Craig levels against Boyd. Craig criticizes Boyd for proposing "a reform of the English language" ("Response to Gregory A. Boyd," pp. 229; in Four Views on Divine Providence). Boyd's "will" is simply not the "will" of common English. The "will" of common English simply indicates that an event occurs in the future whether or not it is causally necessitated to occur, whereas Boyd's "will" indicates that an event is causally necessitated to occur in the future. In this way, Boyd's "will" would seem to be more accurately phrased as "will necessarily," where the necessity in question is causal necessity. In other words, "x will occur" means that x is causally determined to happen. If we use "will" according to normal English usage and not as a synonym for "will necessarily," then "x will occur" and "x might not occur" are not contradictories, and neither are "x will not occur" and x might occur." Consequently, the truth of "x might occur and x might not occur" does not entail the falsity of both "x will occur" and "x will not occur." Further, Craig thinks that future contingents involve the use of the normal English "will" and "will not."

From this, Craig concludes that "insofar as 'will' has its normal English meaning of expressing simply the future tense of a verb, on Boyd's view the principle of bivalence must fail for future contingent statements" (ibid.). Note that according to the principle of bivalence, for any proposition p, either p is true or p is false, i.e., there are only two truth values and there are no truth value gaps, which is something that is denied by the Aristotelian three-valued logic view but is supposed to be affirmed by Boyd's view (see "God Limit's His Control," pp. 198, footnote 29; in Four Views on Divine Providence).

This line of criticism is similar to Todd's criticism of Peircean semantics for being "rigged" and "causally loaded" in contrast to the causally unloaded, merely predictive sense of "will" that is formalized by the alternative "Ockhamist" semantics that his "all false" theory of future contingents adopts according to which, roughly, "x will occur" is true if and only x occurs in the (unique actual) future. (“Future Contingents are All False! On Behalf of a Russellian Open Future," pp. 777, 782-784). This sense of "will" seems to be more in line with ordinary language usage, and it is logically compatible with a causally open future so that (as far as logic alone is concerned) it can be true both that "x will occur" and that "x might not occur" (in the foregoing sense of "might"). Nevertheless, as was seen above, on Todd's Russellian analysis, if there is no unique actual future (as the open future view holds), then even on the Ockhamist semantics, all future contingents turn out false. In this way, Todd's view seems more plausible than Boyd's as it is less semantically revisionary. In addition, however, and in light of this, Craig's claim that "bivalence must fail for future contingent statements" if "will" has its ordinary meaning is evidently too quick, at least if it is granted that the Ockhamist semantics accurately capture the "normal English meaning" of "will." In order for Craig to make good on his claim, he would need to rebut the proposed Russellian analysis of the Ockhamist semantics.

In any case, since Boyd's view suffers from the same problems that Todd's view does and is additionally even less plausible due to it being more semantically revisionary, if we have good reason to reject Todd's view (as I think we do), then we likewise have good reason to reject Boyd's view. In the end, then, I think that open future views are implausible and should therefore be rejected. So, I do not think that rejecting (2) or (3) is a viable strategy either. As such, we should admit that there are true future contingents and that God, being omniscient, knows them. Consequently, we should reject open theism.

Of course, much ink has been spilled on these issues, and so this post can hardly be said to constitute anything like a definitive resolution to the philosophical debate over open theism. My goal has simply been to consider various moves that open theists can make to resist the premises of an argument against open theism that I offered in the past as well as to sketch the reasons that move me to conclude that open theism is philosophically implausible. A fuller discussion would involve a consideration of the problem of the compatibility between comprehensive divine foreknowledge and human freedom, which is the chief philosophical motivation for affirming open theism in the first place, and such a consideration is beyond the scope of this post. Theologically speaking, of course, it is clear that open theism is false as it is incompatible with de fide teachings of the Catholic Church (though non-Catholics will obviously not be moved by this consideration).

Continuing with the theological side of things, I would like to conclude with what I think is an interesting historical note: It is arguable that the claim that "future contingent propositions are to be evaluated according to Peircean semantics" falls under a condemnation contained in a papal bull issued by Pope Sixtus IV in 1474 called "Ad Christi vicarii." In the bull, the following proposition of Peter de Rivo (which was later revoked by Peter in writing) is condemned: "Item, quod non sufficit ad veritatem propositionis de future, quod res erit, sed requiritur, quod inimpedibiliter erit" [Likewise, that it does not suffice for the truth of the proposition concerning the future, that the thing will be, but there is required that it will be unpreventably]" (Denzinger, Enchiridion Symbolorum, no. 1394). If a thing occurring unpreventably is understood as being causally determined to happen, then Peter's condemned proposition amounts to the claim that a future tense statement is true only if the event it reports obtains in all causally possible futures. The falsity of this claim entails the falsity of Peircean semantics for future contingents. I'm not sure what the theological note of this condemnation is, but it is still historically interesting regardless. In any case, the falsity of Peircean semantics for future contingents is surely entailed by dogmas pertaining to divine foreknowledge and human freedom. The same is true of the Aristotelian view of future contingents according to which future contingent statements are neither true nor false.

References

Beilby, J. K., & Eddy, P. R. (Eds.). (2001). Divine Foreknowledge: Four Views. IVP Academic.

Craig, W. L. (2001). Time and Eternity: Exploring God’s Relationship to Time. Crossway Books.

Denzinger, H. (2012). Enchiridion Symbolorum: A Compendium of Creeds, Definitions, and Declarations of the Catholic Church (P. Hunermann, Ed.). Ignatius Press.

Jowers, D. W. (Ed.). (2011). Four Views on Divine Providence. Zondervan.

Pinnock, C., Rice, R., Sanders, J., Hasker, W., & Basinger, D. (1994). The Openness of God: A Biblical Challenge to the Traditional Understanding of God. IVP Academic.

Rhoda, A. R., Boyd, G. A., & Belt, T. G. (2006). Open Theism, Omniscience, and the Nature of the Future. Faith and Philosophy, 23(4), 432–459. https://doi.org/10.5840/faithphil200623436

Todd, P. (2016). Future Contingents are All False! On Behalf of a Russellian Open Future. Mind, 125(499), 775–798. https://doi.org/10.1093/mind/fzv170

Wierenga, E. (2009). Omniscience. In T. P. Flint & M. C. Rea (Eds.), The Oxford Handbook of Philosophical Theology (pp. 129–144). essay, Oxford University Press.