A Refutation of Simple Foreknowledge Revisited

By | August 31, 2021

In my previous post, I develop what I take to be a refutation of the simple foreknowledge (SF) view. In this post I want to look at little more carefully at that refutation, consider how a proponent of SF might try to get around it, and show that this response does nothing to resolve the contradiction.

SF refutation recap

The gist of my refutation is that SF requires that contradictory information be available prior to the resolution of a future contingency. On the one hand, SF requires that there be determinate information to the effect that some future contingent event will happen or will not happen prior (in the temporal sequence) to the actual occurrence of that event. In other words, it is a settled fact that the future contingency is resolved in the precise way that it eventually is. On the other hand, because SF is committed to a preventable future response to fatalism, according to which God’s foreknowledge is explanatorily and, indeed, ontologically dependent on the actual occurrences of future contingent events, determinate information about a future contingent can’t exist in the past until the event actually occurs. In other words, because of the ontological dependency relation, the intrinsic indeterminacy of future contingents is automatically propagated to whatever ontologically depends on their resolution. Thus, there is now and at all prior moments an an objective indeterminacy as to how any given future contingent is resolved. Whether the contingency is resolved this way or that way remains an open question. But now we have a problem. Given SF it follows that at all past moments, the state of information regarding future contingents is both determinate (a settled fact) and indeterminate (an open question). This is a contradictory state of affairs because the same information (that a future contingent is resolved in some particular way) can’t both exist and not exist at the same time. Hence, SF is necessarily false.

A SF rejoinder—counterfactual dependency and ersatz eternalism

How do SF proponents get around this problem? Well, they don’t, or at least I’ve never seen an SF response to an objection of this sort. The reality of the situation, I think, is that SF proponents haven’t ever really considered an objection of this sort because they uncritically employ two assumptions that, in effect, blind them to the reality of the problem.

The first assumption is that the kind of dependency of the past on the future that preventable futurism requires can be adequately understood as mere counterfactual dependence. To revisit my example from the previous post, if I am about to make a libertarian free choice between chocolate and vanilla ice cream, then I now have it in my power (before the choice) to do either. Suppose I go on to choose chocolate. In that case the SF proponent will say that God has always infallibly believed that I will choose chocolate. And if (counterfactually) I had chosen vanilla, then God would have always infallibly believed that I will choose vanilla. So whichever choice I make, the SF proponent will say that God’s foreknowledge is counterfactually dependent on my actual choice.

How does this assumption help the SF proponent? Well, counterfactual dependence is a much weaker relation than explanatory, causal, or ontological dependence. The latter are asymmetric (or at least anti-symmetric), whereas counterfactual dependence is not. Counterfactual dependence entails only a kind of necessary correlation between two things. It doesn’t imply that one of those things is dependent in some more robust asymmetrical sense on the other, and so it won’t propagate indeterminacy from future contingents into the past. Without that, my refutation doesn’t work.

The second assumption that SF proponents typically make is the omnitemporality of tenseless truth (OTT). Given a dynamic ontology of time, which SF proponents endorse, there are no tensed truths that are true at all times. What will happen (future tense) at some point will be happening (present tense) and later will have happened (past tense). So tensed propositions obviously have to be able to change their truth value. But it has often been supposed (usually with little or no argument) that tenseless truths are different and that they can’t change their truth values. The supposition here is that a kind of ersatz eternalism holds at the tenseless level of description. Ersatz eternalism and OTT go hand-in-hand with the common idea (among analytic metaphysicians) that there is such a thing as a unique “actual world” that contains a complete past, present, and future.

How does OTT help the SF proponent? It helps by giving them a way to transfer information from future times to past times. Once a future contingency is resolved (i.e., it actually occurs this way and not that way), the SF proponent will say that that event grounds a corresponding tenseless truth. And if tenseless truths are omnitemporally true, then we can shift that tenseless truth to the past and then reinstantiate it, so to speak, as a tensed truth about the future. For example, suppose it is now T0 and that at future time T1 I freely choose chocolate ice cream over vanilla. At that time, my actual choice will ground the tenseless truth Alan chooses chocolate at T1. Given OTT, if a tenseless truth is ever true, then it is always true. Since it is true at T1, it is therefore true at T–1000 (a past time) as well. And since it’s true at that past time, we can reintroduce tense and say that it’s true at T–1000 that Alan will choose chocolate at T1. God, being omniscient, would of course have to know that future tense truth at T–1000, and this supposedly explains how God could have foreknowledge of that future contingent event.

Rebutting the rejoinder

Unfortunately for the SF proponent, these two assumptions don’t rescue SF from my refutation.

In the first place, it is easy to demonstrate that mere counterfactual dependency is not sufficient for preventable futurism to work. A more robust notion of dependency is required if fatalism is to be avoided. The easiest way to show this is by considering the openly fatalist (at the creaturely level) position of theistic determinism. According to the theistic determinist, God is the ultimate sufficient cause of “whatsoever comes to pass,” to quote the Westminster Confession. There is thus a necessary correlation between what comes to pass and what God causes to happen. If I choose chocolate, then God will have always known that I was going to choose chocolate, and He will have known that because that’s what He caused to occur. And if I choose vanilla, then God will have always known that I was going to choose vanilla, and He will have known that because that’s what He caused to occur. Hence, on theistic determinism, God’s causing me to choose one way or the other is counterfactually dependent on what I do; nevertheless, the arrow of explanatory dependence goes only one way, from God to creatures. To block fatalism, therefore, mere counterfactual dependence is not enough. The preventable futurist needs the explanatory arrow to run unambiguously from the actual occurrences of future contingent events to God’s foreknowledge and not vice-versa. This dependence has to be asymmetric, and that allows my refutation to get up and running, because it means that determinate information about how future contingencies are resolved can’t be in place until those contingencies are in fact resolved.

In the second place, once the refutation is up and running, the notions of OTT and of ersatz eternalism do nothing to assuage it precisely because the contradiction is between two tensed truths. OTT may give the SF proponent a halfway plausible story to tell as to how it can be determinate at T–1000 that Alan will choose chocolate at T1, but given the asymmetrical dependence of past truths on future contingent events that comes with preventable futurism, it must also be indeterminate at T–1000 whether Alan chooses chocolate at T1. Because that future contingency has not yet been resolved as of T–1000, neither of the determinate tensed propositions <Alan will choose chocolate at T1> nor <Alan will not choose chocolate at T1> can be true at T–1000. (Indeed, if either of those were true then, then at T1 I could falsify God’s foreknowledge simply by choosing otherwise.) And if the SF proponent appeals to OTT to establish that one of those tensed propositions is true at T–1000, then we again arrive at the contradiction that is the heart of my refutation of SF. For we will have to say that one and the same tensed proposition is both true and not-true at the same time—true because derived from ersatz eternalism via OTT, and not-true because falsely representing an objectively indeterminate future as though it were already determinate.

In sum, then, appeals to mere counterfactual dependence and the supposed existence of a true, tenseless description of past, present, and future reality—what I’ve called ‘ersatz eternalism’—cannot save SF from my refutation based on incompatible past information about future contingents.

Postscript: arguing against OTT

While it’s not necessary for my argument, I believe we can argue cogently that OTT is false. It is not the case that tenseless truth is omnitemporal. The reason for this is that tenseless truth, like all truth, or at least all logically contingent truth, supervenes on being. That is to say, truth has to be grounded in reality. For ersatz eternalism to be correct, for there to be this unique, complete, true, tenseless description of a logically contingent reality there has to be a reality that backs up that description. But what sort of reality could do this? If all of history were somehow metaphysically or causally necessary, à la theistic determinism, then we could ground the truth of that description in those necessitating conditions. But SF affirms future contingency and so must reject any that sort of grounding for tenseless truths about future contingents. Alternatively, if metaphysical eternalism were true, that is, if all past, present, and future events eternally and tenselessly exist, then we could ground the truth of that description in the events themselves. But SF proponents reject eternalism in favor of a dynamic theory of time. That’s why God on SF is temporally situated. It’s also why God is thought to need temporally prior fore-knowledge as opposed to timelessly knowing the “future” from the viewpoint of a timeless eternity. Now the only quasi-dynamic version of eternalism is the “moving spotlight” theory, which combines metaphysical eternalism with an additional tensed fact (the “moving spotlight”) specifying what time it is “now”. The problem here is that the asymmetric dependency relation required for preventable futurism is incompatible with any sort of metaphysical eternalism. (I won’t develop the argument for this here, but the gist of it that the resolution of future contingents requires real becoming and a much more dynamic theory of time than the moving spotlight will allow.) In sum, there seems to be no “reality” that could ground ersatz eternalism in a way that is consistent with SF. And so we have a grounding argument against OTT. If reality is robustly dynamic and causally open, then what tenseless truths there are can change over time. That Alan (tenselessly) chooses chocolate at T1 is not true prior to my choice, but can become true at T1 (if I make that choice) and would then remain true thereafter.

One thought on “A Refutation of Simple Foreknowledge Revisited

  1. Pingback: Philosophical Essays against Open Theism – ch. 3: Arbour – Open Future

Leave a Reply

Your email address will not be published. Required fields are marked *