Anselmian freedom and ontological necessity

In order to have libertarian freedom you must be able to choose between alternative possibilities. But divine foreknowledge poses a problem for libertarian freedom in that it entails a kind of necessity about future choices. It entails at least – to adopt the terminology of David Kyle Johnson – ‘ontological’ necessity, the necessity generated by the state of determinate reality; if it is the case today that you will choose X tomorrow, then it is necessary that you will choose X tomorrow. Ontological necessity is more than merely analytic and the foreknowledge dilemma is not solved simply by noting (as Augustine did in On Free Will) that it is illicit to move from ‘It is necessary that (if I choose X tomorrow, then I choose X tomorrow)’ to ‘If I choose X tomorrow, then it is necessary that (I choose X tomorrow)’. Ontological necessity is generated by the facts about the world. So, for example, it is common to hold that there is an ontological necessity about the past because, from our perspective in the present, the actual past cannot be other than it was. We should not cry over spilt milk because there is nothing we can do to undo what is done. If you chose X yesterday, then your choice is now a ‘fixed’ and unalterable part of reality. If you chose X yesterday then, today, it is ontologically necessary that you chose X yesterday, and you are now helpless to alter that fact. But divine foreknowledge entails the same ‘fixity’ regarding the future. God's present knowledge of what you will choose tomorrow entails that there is a truth today about what you will choose tomorrow. Today your choice tomorrow is part of determinate reality. So, by ontological necessity, you cannot fail to choose X tomorrow. If the future is, at present, as fixed as the past, doesn't this undermine the possibility of libertarian freedom?

We grant that divine foreknowledge does entail the ontological necessity of the foreknown choice. But we argue that this is the only sort of necessity entailed by divine foreknowledge if we adopt the Anselmian position that God is ‘outside’ of time and simply ‘sees’ your future free choice as you make it and because you make it.Footnote ¹ And we defend the claim that choices made with ontological necessity may nevertheless allow the sort of alternative possibilities and the aseity required for freedom by responding to arguments developed by William Hasker and, more recently, by David Kyle Johnson.

According to the Anselmian, the solution to the dilemma of freedom and foreknowledge requires adopting the correct theory of time – the theory which asserts the falsity of ‘presentism’, the view that the present moment and what it contains is all that is real.Footnote ² Rather, what we now perceive as past and future moments are equally real. All the moments of time (or the entire duration of time – we can set aside the issue of whether time is a sequence of moments or a duration) have the same ontological status, and what we call past, present, and future are relative to a given perceiver at a given time. God, the ideal knower, who sees things as they are, sees all times just ‘there’ in one act of knowing.Footnote ³ This theory goes by many names in the contemporary literature, none of them quite satisfactory.Footnote ⁴ What we want is a term which captures the notion that all times (or all of time) are ontologically equal. We propose that the term ‘isotemporalism’ be adopted, from the Greek iso- meaning ‘same’ or ‘equal’.Footnote ⁵

If we posit an isotemporal universe we can explain divine foreknowledge this way: what, from our temporal perspective, is now, and what, from our temporal perspective, is some future moment, are both immediately ‘there’, equally present to God. Though past, present, and future are relative to the temporal perceiver, before and after are objectively real, so posit two moments, t1 and t2.Footnote ⁶ Suppose your free choice occurs at t2. The defender of divine foreknowledge says that it is true at t1 that God knows what happens at t2. (We do not mean to suggest that God exists ‘at t1’ as a temporally limited being. Nonetheless, the thesis of divine foreknowledge entails that at t1 it is true that God knows what happens at t2.Footnote ⁷) But God knows this because, from the divine perspective, t2 is simply ‘present’, as is t1. If we hold that God knows that you choose X at t2 because you choose X at t2, and He ‘sees’ you choose X, then God knows at t1 that you choose X at t2, because you choose X at t2. At t1 (and at every time) it is true, and is known by God, that ‘You choose X at t2’. That being so, it is true at t1 (and at every time) that, in one sense, you cannot possibly fail to choose X at t2. Divine foreknowledge does entail that a sort of necessity attaches to your choosing X at t2 – ontological necessity, the necessity generated by the determinate state of the world. But does this ontological necessity undermine freedom?

For the purposes of this article the freedom which we take to be consistent with ontological necessity is libertarian freedom under an ‘Anselmian’ description (Rogers (Reference Rogers2008), ch. 4). Free will, for the creature, requires indeterminism. A free choice involves alternative possibilities. We are subject to competing motives and so we face genuinely open options in that we are not causally determined by anything at all to choose one over the other. Which option we choose is entirely up to us. We have ‘ultimate responsibility’ for our choice.Footnote ⁸ And this is the more crucial element in Anselm's view. In choosing freely we have aseity –from-oneself-ness. Without indeterminism, the causally open options, we could not, in a God-made universe, have aseity. But having the options is a means to the end of aseity. It is the aseity itself which helps to constitute the image of God in the created agent.

Given this understanding of Anselmian freedom, if you freely choose X at t2 it is the event of you actually choosing X at t2 on which the truth of ‘You choose X at t2’ depends.Footnote ⁹ So you make it true that ‘You choose X at t2’ by choosing X at t2.Footnote ¹⁰ If we put the point in terms of truthmakers, the truthmaker for ‘You choose X at t2’ is your choosing X at t2. Oppose this to the position of the presentist who would defend divine foreknowledge. He must locate the truthmaker for the foreknown proposition in the present, since all that is is in the present. But if what makes the proposition about the future true is some fact about the present then the more obvious choices like preceding physical events or God's knowledge of His own intentions lead to determinism. But combine isotemporalism with Anselmian freedom, and libertarianism can be reconciled with divine foreknowledge. Suppose you make an Anselmian choice, choosing X at t2. You meet the two criteria of confronting causally open options and choosing with genuine aseity. On isotemporalism God knows at t1 that you choose X at t2, because you choose X at t2.You, the free agent, are the ultimate cause of God's knowledge. In one sense, it is true at t1 that you cannot do other than choose X at t2, but this is only because, by choosing X at t2, you have made it an aspect of determinate reality that you choose X at t2. The ultimate responsibility is yours, and so the necessity involved does not conflict with freedom.

Some have argued that this move fails. William Hasker has consistently held that if by ‘freedom’ we intend libertarian freedom, then positing an eternal God and an isotemporal universe does not solve the dilemma. In Hasker's view, libertarianism requires that the agent choose between alternative possibilities, where ‘alternative possibilities’ means something more than the causally indeterminate options that Anselm had in mind. Hasker writes that for an agent to have libertarian freedom, ‘at the moment when she chooses it must be really possible, all things considered, for her to make a different choice than the one she actually makes’. He asks us to consider Annie, whose choice of Y exists at t2 in an isotemporal universe. ‘But in order for her choice to be free it must have been really possible, really within Annie's power, to choose X instead of Y.’ But then it must be the case at t1 that Annie could choose X at t2. But that is impossible since it would entail that ‘there are future actions of [Annie's] which timelessly exist in the divine eternity which are such that it is in [Annie's] power, now, to bring about that those actions do not exist in eternity’ (Rogers and Hasker (Reference Rogers, Hasker and Jordan2011), 15–16).Footnote ¹¹ The timing of the different elements – the powers, the alternative possibilities, and the choices – is a little puzzling here, but we take it that the argument is roughly this: if Annie makes a libertarian free choice for Y at t2, then it is possible at t1 that Annie choose X at t2. But if God knows at t1 that Annie chooses Y at t2, then it is not possible at t1 that Annie choose X at t2. And the Anselmian grants the point, on a certain understanding of ‘not possible’: if the universe is isotemporal and Annie chooses Y at t2 then it is always ontologically necessary, part of determinate reality, that Annie chooses Y at t2. We may say that it is ‘not ontologically possible’ at t1 that Annie choose X at t2. In Hasker's view this rules out the sort of alternative possibilities required for libertarian freedom. Anselm's requirement that the agent engage in non-determined, a se, choice between motivationally open options is necessary, but not sufficient. Hasker holds that it must also be the case that the agent confront ‘ontologically open’ alternative possibilities. But these are negated by divine foreknowledge which entails ontological necessity. Hasker's conclusion is that we should abandon divine foreknowledge and adopt an ‘open’ theism.

But if Hasker's argument is that any choice which is ontologically necessary cannot, for that reason, be a free choice, then there seems an obvious rebuttal. Ontological necessity attaches to any choice at all. That is, if you choose X now, then your choosing X now is part of determinate reality. By ontological necessity you cannot possibly fail to choose X now. If Hasker's Annie chooses Y now then it is not in her power to choose X now. Hasker, as we noted above, writes that in order for Annie to choose freely, ‘at the moment when she chooses it must be really possible, all things considered, for her to make a different choice than the one she actually makes’. But Annie's choice is ontologically necessary when she makes it, and surely we do not want to say that this ontological necessity means that her choice is not free. On the Anselmian analysis, the necessity that is involved with the foreknown choice is only an ontological necessity and it would be absurd to argue that any ontologically necessary choice – which would mean any choice at the time it is made – ipso facto cannot be free.

But perhaps the argument is really that it is the ontological necessity of a choice before the choice that conflicts with libertarian freedom. David Kyle Johnson develops this case in a recent contribution to the debate. He argues that a temporally preceding ontological necessity conflicting with the freedom of your future choice would be generated, not only by divine foreknowledge, but also by isotemporalism by itself (Johnson (Reference Johnson2009), 435–454). In brief, Johnson's argument concerning isotemporalism goes like this: Libertarian freedom entails that, ‘An action is free only if not performing the action is ontologically possible before the action is performed.’ Johnson explains ontological possibility,

But isotemporalism entails that what we call the future and all it contains, along with the past and the present, exists in ‘determinate reality’. It does not exist ‘today’, but it exists simpliciter. ‘Today’ is indexed to the temporal perceiver, and the ideal observer, God, for example, sees that all times have the same ontological status. So if you perform action X tomorrow, and tomorrow exists, it is not ontologically possible today that you not perform X tomorrow. And so no libertarian freedom. Johnson concludes that libertarian freedom requires presentism – only the present moment exists. It also requires a denial of the principle of bivalence regarding propositions about future free choices since that would render the same result as isotemporalism. And God, if there is a God, does not know future free choices (Johnson (Reference Johnson2009), 448–452).

Johnson holds that his characterization of the sort of alternative possibilities required for freedom – open ontological possibilities before the choice – is intuitively obvious, but he does present two arguments in defence of what he takes to be an obvious intuition. His first defence begins by noting that his opponent might propose that the intuition is undermined by the fact that a free action is ontologically necessary while it is being performed. That is, the opponent might say what we said above against Hasker: if you are doing X at t2, then you cannot possibly not be doing X at t2, but surely this, by itself, is not a reason to deny that you do X at t2 freely. And so, questioning the intuition, the opponent may ask why a free action cannot be ontologically necessary beforehand given that a free action is ontologically necessary while it is being performed. But Johnson responds that an action ‘attains its ontological necessity, by attaining its ontological status, at the moment it is performed. But if it had that status and necessity before it was performed, it would not be attaining that status and necessity at that moment.’ So the intuition is not undermined. Johnson's second defence of the intuition that a choice must be ontologically open before it is made has to do with a key aspect of the experience of choice. As we deliberate, ‘we assume that whether or not the action is going to be performed is yet to be determined … if the action is ontologically necessary before it is performed, the central assumption we make in deliberation is grossly mistaken’ (Johnson (Reference Johnson2009), 440–441). If possible one wants to avoid claiming that a ubiquitous assumption is grossly mistaken, and so we should allow the intuition. In response, we first try to show that, as regards common intuitions, it is unlikely that we ordinarily take alternative possibilities to require ontological openness as Johnson has spelled it out rather than the sort of causal openness that Anselm takes to be sufficient. Then we address his two arguments in defence of his understanding of the intuition about alternative possibilities.

A first point concerning our intuitions about alternative possibilities has to do with looking back at a past action. The question about alternative possibilities arises in connection with libertarian freedom, a view ordinarily advanced as an appropriate theory within which to ground moral responsibility. Praise and blame are central to this discussion, and praise and blame, from our human perspective, necessarily involve looking backward in time. You have already done the deed when the time comes for us to praise or blame you. We often say, and find it intuitively plausible, that we hold you responsible when we believe that you could have done otherwise. But, when it comes time to praise or blame you for some deed you have done, the judgement that ‘you could have done otherwise’ is always wrong if it is supposed to mean that it is consistent with determinate reality, i.e. it is now-ontologically possible, that you did otherwise. At t2 you did X, so at t3 it is not ontologically possible that you not have done X at t2, and so you could not have done otherwise. Johnson might well respond that perhaps what we really mean is that, although now at t3 your doing X at t2 is ontologically necessary, it was true at t1 that you could have done otherwise. But ordinary language does not insist upon this qualification. It allows us, after the fact, to say simply that ‘you could have done otherwise’. This suggests that our intuitions about what counts as ‘being able to do otherwise’ do not encompass beliefs about ontological necessity and possibility.

We can reinforce the suspicion regarding Johnson's take on our intuitions about alternative possibilities through a courtroom example. Say that you have committed crime X at t2. We, the jury have strong intuitions that responsibility requires libertarian freedom and we are thoroughly convinced that your choice perfectly fit the description of Anselmian freedom. You had alternative possibilities in that you were struggling to pursue two conflicting motives. And it was entirely up to you to choose, with no determining factors at all, which motive to pursue. You bear ultimate responsibility for the choice. That being the case, we are ready to convict you – but wait! Your clever attorney has one last argument to make. He presents the view that we are living in an isotemporal universe. (Perhaps he calls a physicist to the stand to offer the argument from relativity and a metaphysician to make the ‘truthmaker’ case.) And now he adds Johnson's point that, from the beginning of time, you choosing X at t2 is part of determinate reality. You could not have done otherwise, and so you are not really free in the way required for libertarian freedom.

If your lawyer had been able to show that your deed – though in outward appearance the same as a libertarian choice – had actually been necessitated by preceding causes by demonstrating that ours is in fact a determinist universe, we, the jury, might change our minds. But is it intuitively obvious that we, the jury, should acquit you due to the ontological necessity of your crime? All that has changed is our theory of time. Every fact about you, the accused, and your deed, remains the same – your motives, your not being determined, what it was you actually did. If you are guilty in a presentist universe, and you didn't do anything different in an isotemporal universe, it seems bizarre to convict you in one, but not the other. We praise you or blame you for what you have done, and competing theories about the nature of time just do not seem to affect that issue. If the point of positing libertarian freedom is to explain how moral responsibility can be grounded, then it seems wildly counterintuitive to insist on a version of libertarianism which renders the bizarre result that you may effectively plead not-guilty owing to isotemporalism.

So it is not at all clear that ordinary intuition supports Johnson's understanding of the alternative possibilities required for libertarian freedom. We turn now to the arguments with which Johnson supports his reading of the intuition. Johnson argues first that an action, your choosing X let's say, attains its ontological necessity at the moment it attains its ontological status. Your choosing X at t2 becomes part of determinate reality at t2. And the Anselmian seems required to accept this since he insists on the aseity of choice. Your choice – including whatever true propositions it grounds and whatever ontological status it has – is entirely up to you. Even divine foreknowledge is dependent upon the fact of your actually making the choice you make. But then, Johnson argues, contrary to what isotemporalism entails, ‘You choose X at t2’ could not have been ontologically necessary before t2 because, ‘if it had that status and necessity before it was performed, it would not be attaining that status and necessity at that moment’ (Johnson (2009), 440). Thus, Johnson holds, the argument we made above against Hasker is ultimately unsuccessful. We argued above that it cannot be the case that ontological necessity conflicts with free choice, since every choice, at the time it is made, is ontologically necessary. You can't fail to choose X as you choose X. But Johnson responds that there is a relevant distinction between a choice being ontologically necessary at the time it is made and a choice being ontologically necessary before it is made. It is the ontological openness beforehand that counts, and isotemporalism conflicts with this preceding openness.

But in stating that the truth of your choice could not have ontological necessity before it gains ontological status Johnson begs the question against the isotemporalist. The isotemporalist can happily agree that you choosing X at t2 attains its ontological status and hence necessity when it is performed at t2. The isotemporalist will understand by ‘attains its ontological status’ to mean roughly is made to exist (or caused to exist, in some broad sense of ‘caused’). As the isotemporalist considering your free choice understands things, it is indeed what you do at t2 that makes (or causes) your choosing X at t2 to exist and be ontologically necessary. (Or, if ‘makes’ or ‘causes’ suggest the wrong relationship, it might suffice to say that the existence and ontological necessity of your choosing X at t2 ‘depends’ upon your choosing X at t2.) In the isotemporal universe your choosing X at t2 has existence and ontological necessity at all times, but it has them because of the action you take at t2, so it attains them at t2. Johnson's claim that your choosing X at t2 cannot have existence and ontological necessity temporally before it attains it, that is, before the event happens which makes it the case, simply begs the question against the isotemporalist.Footnote ¹² True, in ordinary speech it sounds odd to say that P has r before P attains r, but ordinary speech – while it seems legitimate evidence regarding common intuitions – does not provide much guidance when it comes to analysing the nature of time and properties like ‘now-ontological necessity’.Footnote ¹³

Anselmian freedom and deliberation

Johnson's second argument supporting his understanding about our intuitive view of alternative possibilities deals with our experience of choice. He writes that ‘an action having ontological necessity before it occurs is contrary to the central assumption we will make when deliberating about whether or not to perform the action’ (Johnson (Reference Johnson2009), 440–441). We believe this argument fails.

Johnson argues as follows:

1. (DKJ1) A person P deliberates on action A at time t only if P believes at t that action A is undetermined.

2. (DKJ2) P truly believes that [action A is undetermined] only if is A is now-ontologically possible and ∼A is now-ontologically possible.

   Therefore, when P deliberates on action A, either P believes falsely or A and ∼A are now-ontologically possible.

There is a high cost to accepting that every time we deliberate we believe falsely, and so Johnson's argument is potent.

One might think we are being unfair in our reconstruction of the argument. Is not DKJ2 just the claim that the relevant type of indeterminism in question is that of being now-ontologically possible? And doesn't that make the entire argument circular? We do not think so. DKJ2 is a claim about what we believe when we act. For, we are willing to grant DKJ1 for now; there is some sense of ‘undetermined’ that we presuppose when deliberating. What DKJ2 provides is the claim that what goes on in our mind presupposes now-ontological possibility. It is this claim that we deny.

What are some of the options for fleshing out what we believe when we act? The following is a good start:

1. (B1) P consistently believes that A and ∼A are now-ontologically possible.

2. (B2) P consistently believes that there are two possible states of affairs that include parts of the past up to time t, the laws of nature, A or ∼A respectively, and the fact that no miracle was required in bringing about A.

3. (B3) P does not believe some fact F such that F makes it impossible for P to believe consistently that there are two possible states of affairs that include the parts of the past up to time t, the laws of nature, A or ∼A respectively, and the fact that no miracle was required in bringing about A.

The propositions B1–B3 are given in order from strongest to weakest. If P consistently believes A is now-ontologically possible (satisfying B1) then there are two open possibilities in which P can believe (satisfying B2). If B2 is true, then P cannot consistently believe a fact F that would falsify P's belief in the possibility of either state of affairs. Note that B3 does not entail B2. B3 is merely a claim about P's ignorance. B3 does not entail that P believes anything at all about P's future.

Which of B1–B3 should we think is closest to the truth? Here is a story a philosopherFootnote ¹⁴ used to tell (the story may or may not be true, but it is interesting nonetheless): For a number of years while David Lewis was at Princeton he would receive invitations from universities in Australia to come and visit during the summer break at Princeton. Lewis would make a great show about deciding whether or not he was going to accept the invitations. However, this philosopher was of the mind that Lewis's actions were not deliberations. It was common knowledge, to Lewis and everyone else, that Lewis would end up accepting the invitations as he had done each year for a number of years. Applying the lesson of Lewis's invitations, the philosopher went on to suggest that God cannot freely act. After all, for any action A, God knows whether or not He will perform that action.

There are two crucial claims in the story. The first is that free action requires deliberation. This is not obviously true, but it is not what the current dispute is about. Even if we claim that free action does not require deliberation, we would agree that we do sometimes successfully deliberate about a free action. It would be a strange position that held that all our actions are non-deliberative free ones. Thus, we will ignore the first claim. The second claim is that deliberation requires that the agent not believe that he will act in a certain way. Such a belief somehow destroys the deliberative process. We will now suggest the way in which this happens.

Deliberation seems to be primarily an exercise in imagination. It is the comparing of two different ways things could be and then deciding to actualize one of them. It is here that talk of our freedom being made in the image of God is so compelling. In choosing a path we are mini-co-creators with God, bringing about the future with our act. For example, suppose a couple purchases a new piece of furniture for their living room. When the piece of furniture arrives at their house they must decide where to put it. In this household deliberation usually takes two stages. First, they move the pieces of furniture around the room physically so as to picture what the room would be like in certain configurations. This stage does not last very long. The floors are damaged with too much moving, the furniture heavy, and their wills weak. After just a little rearranging one partner starts to insist more and more that they merely imagine what the room would be like in various configurations. In order to make their imagination more vivid, perhaps he will hold his arms apart in an attempt to mimic the size of the new piece of furniture. He will then walk around the living room exclaiming things like, ‘Well, it seems as if it would fit here’, or ‘I just can't really picture it this close to the TV’, etc.

Note that there are some things we can imagine and some things we cannot imagine. We can picture a new piece of furniture just inches from the screen of the TV. We find the state of affairs disagreeable and do not want to actualize it. On the other hand, we cannot imagine a state of affairs where the TV and piece of furniture occupy the same space. We know enough about the way medium sized physical objects work that such a state of affairs is impossible.Footnote ¹⁵ In this case, our knowledge destroys our ability to deliberate about a certain course of action. If you doubt this, ask yourself whether or not you can imagine a round square. As we've told countless classes of undergraduates, we don't think you can.

When we talk about imagining something in our mind we are sometimes referring to the action of bringing forth a visual representation of a state of affairs in our consciousness. At other times, it refers to a less concrete action. When you read a story you picture it in your mind, but often, if you were pressed for the details of what you were picturing, you would be unable to provide any. That is because imagining a state of affairs is a strange phenomenon, full of ghostly images, thin representations, and vague feelings of consistency, implication, etc. While we can produce a much more specific scene when we imagine, we often do not need to. We are rightly content with surface-level imagining. It is only when the comfortable link between conceivability and possibility is challenged that we have to go further. When pressed about whether or not a course of action is truly possible, a common response is to provide more details about the imagined case so as to dispel doubts about its possibility.

Having considered the familiar case of moving furniture, let us consider an abstract case of person P deliberating on whether or not to perform action A. P pictures in her mind the state of affairs that contains A and at least one other state of affairs that contains ∼A. If P believes that A occurs, then P is unable successfully to imagine the state of affairs containing ∼A. This is not because P thinks a ∼A state of affairs is impossible in the broadly logical sense, but because P is forced to take the belief that A will occur into consideration when determining what future to bring about. This is just the very same action we perform if, when considering where to go on vacation, we are told that it will rain where we are heading. It is not impossible for it not to rain, but belief that it will rain forces itself into our different scenarios. This mechanism supports at the very least B3 and explains why Lewis could not deliberate about going to Australia. Belief about what one will do prevents one from consistently imagining doing something else in the future.

Additionally, the suggestions we have made here about imagination and deliberation support an adoption of B2. To deliberate is to have in one's mind's eye ways that the world could be, but it is also not to have a complete idea of the way the world could be. That is why we talk of parts of the past in B2.

We do not think that this imagination requires us to adopt B1. Consider the following case. While P is deliberating between A and ∼A, she is interrupted by someone she knows to be reliable. This person claims to know what P will decide, writes down P's decision on a slip of paper, and places it in the envelope. The interrupter then leaves. Suppose P believes that whatever is written on the paper is accurate. P believes the following: Either A or ∼A is now-ontologically impossible. If P looks into the envelope, she'll come to believe that either A is now-ontologically impossible, or that ∼A is now-ontologically impossible. This will prevent her from deliberating. But what if she doesn't look? She can still successfully deliberate in good faith. P can imagine two distinct consistent states of affairs, each holding different futures for her. She can modify, ask questions of, and evaluate each state of affairs. In short, the mere belief that Either A or ∼A is now-ontologically impossible does not short circuit the imaginative process that is crucial to deliberation. When we deliberate we are engaged in a process of self-discovery, simultaneously causing and learning the truth about the future.

A likely objection is as follows: you're not giving P enough credit. Believing that Either A or ∼A is now-ontologically impossible P will also come to believe that one of the states of affairs must be impossible, thus violating B2. P won't know which state of affairs is inconsistent but will believe one is and thus will be unable to deliberate in good faith. We disagree with this claim, because we think that each state of affairs P is imagining is possible in the right way. Just because A is now-ontologically impossible doesn't mean that A is impossible in every relevant sense.

Recall the thinness of our imagination and the lack of depth to the picturing when we imagine. For deliberation to play the role it does there must be some sort of relevant information that imagination provides. Peter van Inwagen makes these same points with a different moral when he writes on modal epistemology in his article aptly titled, ‘Modal epistemology’ (1998). There he argues that our lack of power in conceiving is such so as to render the conceivability–possibility hypothesis false. He claims we can come nowhere close to imagining a possible world and so we are in no position to ascribe possibility to whatever anaemic states of affairs we do picture. We agree with van Inwagen that our powers of imagination are weak but we disagree with him about the claim that they are useless for determining what is possible. Our powers of imagination provide us with a fallible sense of what is possible in a limited domain of consideration. For example, we do use imagination to determine where to put the piece of furniture. What we learn from imagining the various arrangements is that with respect to the arrangements of furniture in the room the chair could go here, but not over there. We do not expect our musing about furniture arrangement to include, among other things, detailed imaginings of whether or not God will prevent such placement. Our imagining of the furniture in a certain arrangement is not support for the belief that there is a possible world in which the furniture is arranged in a certain way. It is support for the claim that there is some smaller, less ambitious, less complete, state of affairs in which the furniture is arranged and which is explicitly consistent.

The same is true when we deliberate about our actions. To imagine doing A we do not take into account everything. We can't do that. Rather, we cheat. We take into account the relevant parts of the world and try to imagine a consistent state of affairs in which A occurs. Provided that we are successful, we deem the action possible. Further, we know that we can't take into account everything about the past, every truth that there is, when we deliberate. We very consciously leave some things out. When P deliberates about action A, believing that either A or ∼A is now-ontologically impossible, she is not acting in bad faith. She is quite properly ignoring the fact that it is true now that she'll do one or the other in the future. Her deliberation is no less useful to her than our incomplete imaginings of where the furniture will fit.

The word ‘possible’ is context sensitive and that contributes much to our debate about open theism. There is some sense in which given truths about the future we can't act freely and there are other senses in which the truths about the future matter not a bit when it comes to the possibility of our actions. Let us suggest that the solution to this problem mirrors what some of us do when faced with what we believe to be the context sensitivity of ‘know’ and ‘reliable’. The epistemic contextualist learns to ignore the Cartesian complaint about certainty; the standard is too high for everyday cases of knowing.Footnote ¹⁶ The same is true with open theism. The type of possibility open theists demand of us is not only impossible (because there are truths about the future) but not to be desired. And that leaves the Anselmian solution as a strong contender for the best answer to the dilemma of freedom and divine foreknowledge.Footnote ¹⁷