The argument for the deliberation restriction

Our question, then, is whether we need to restrict the concept of omniscience so as to exclude certain truths about deliberation from the set of truths known by an omniscient being.Footnote ⁶ (And since the relevant omniscient being is usually God, I will take our question to be synonymous with the question of whether we need to restrict the set of truths known by God.) My answer is that we do not need such a restriction, because it seems plausible that there are no such truths; and even if there are, they can still be known. To focus the discussion, let us consider a representative argument (from Peterson et al. (Reference Peterson, Hasker, Reichenbach and Basinger2009), 81) in favour of a deliberation restriction on God's knowledge. We'll begin with a simplifying assumption that is presumably endorsed by most proponents of the deliberation restriction, namely that God makes decisions in time.Footnote ⁷ We can now reconstruct their argument as follows:

1. (1) If God knows all true propositions at a given time, then at that time he will know the truth about the result of his deliberations.

2. (2) God makes decisions.

3. (3) If God makes decisions, then there will be some time t during which he deliberates about some decision.

4. (4) There will be some time t during which God deliberates about some decision. (2, 3)

5. (5) It is impossible for anyone, including God, to deliberate about some decision while knowing what the result of that deliberation will be.

6. (6) There will be some time t during which God does not know some true propositions (namely, those involving the results of his deliberation at t). (4, 5)

7. (7) Therefore, there will be some time t during which God does not know all true propositions. (1, 6)

This argument appears to provide compelling reason – at least for someone who is committed to a relatively traditional version of theismFootnote ⁸ – to restrict the concept of omniscience such that a being can count as omniscient even without knowing any truths about the results of deliberation (at least while that deliberation is underway).Footnote ⁹ I will argue, however, that appearances here are deceptive. More specifically, I will argue that we should reject both the fourth and fifth premises. Let us begin with the fourth.

Non-deliberative divine behaviour

The argument for the deliberation restriction depends on the claim that God deliberates, which is captured in (4) above. But if (4) is false, then (5), even if true, is irrelevant – and so is the corresponding deliberation restriction on the concept of omniscience. And one reason for thinking that (4) is false is the claim that deliberation is an inappropriate activity to be attributing to God in the first place. Bruce Reichenbach (Reference Reichenbach1984) takes this line when he argues that God can act intentionally without deliberating. David Basinger (Reference Basinger1986) offers the only direct response to this manoeuvre, in which he argues that removing deliberative activity from God's intentional repertoire has unorthodox (or at least unsettling) consequences.Footnote ¹⁰ His response emphasizes one apparently troublesome implication of Reichenbach's view:

Basinger then claims that most theists would reject this implication, i.e. they would not endorse the idea of a God who has never engaged in deliberative decision-making. Although it is not immediately clear why most theists would reject a non-deliberative God, Basinger does support this contention with two representative examples of divine behaviour: the act of creation and response to prayer. I will consider response to prayer first.

Basinger offers two models of how God might respond to prayer. On the first model, God's responses resemble that of a father who, when asked by his son for a glass of water, complies without deliberating over the matter. On the second model, God's responses are instead the end result of a serious weighing of alternatives. Basinger claims that it is the latter, deliberative model that most closely approximates the thinking of most theists (ibid., 171). Although I disagree with Basinger on this point (and will attempt to show that the non-deliberative model turns out to be superior), his description of the deliberative model does suggest a first (perhaps a zeroth) approximation of an account of deliberation. On this model, deliberation is any genuine consideration of alternatives so as to determine the best course of action. In other words, this model suggests at least two necessary conditions on deliberation: available alternatives and a genuine weighing of those alternatives.Footnote ¹¹

In the case of prayer, God certainly has alternative responses available to him,Footnote ¹² but does he weigh them seriously (or at all)? In our own case, the only reason to weigh alternatives is in order to determine which course of action will most effectively realize our goals.Footnote ¹³ Deliberation, e.g. over whether or not to have an additional cup of coffee, comes about as a result of uncertainty regarding which answer is best, all things considered. Since God is not similarly ignorant of which of the alternatives available to him is (or are) most effective for realizing his ends, the activity by which he realizes those ends does not exhibit at least one of the necessary conditions for deliberation. Thus it would be inapt to attribute deliberation to him, and the non-deliberative model is to be preferred. Contrary to Basinger's reading of the average theist, this model is the one that seems like it should be more satisfying to the theist, in so far as a cognitively perfect being should not have to weigh alternatives in order to discern which one is superior.Footnote ¹⁴

Let us now examine the other example of divine behaviour. Basinger asks:

Basinger apparently thinks that the answer to the question posed in the first sentence is clearly ‘No’. But, as we saw above, deliberation about which creative option to actualize implies uncertainty as to which option is best; and presumably God is not subject to such uncertainty. Thus, whether or not most theists would grant that God never deliberates about which creative option to actualize, they should.

Perhaps, however, I have conflated two distinct phases of God's creation activity. Perhaps God doesn't need to deliberate while implementing his goals for creation, but does need to deliberate while forming those goals. As far as I can see, the only line of reasoning that supports this claim runs as follows. We want to say that God chooses freely, but we can only say that if there is some point during (or at) which he deliberates. This deliberation might occur while God is forming his goals, or while he's implementing the goals formed. And since we have seen that God doesn't deliberate while implementing his goals, he must deliberate while forming them. As I explain in the next section, I think this line of reasoning goes wrong at the first step.

Non-deliberative divine freedom

God, as we have seen, does not need to deliberate in order to create or in order to answer prayer effectively. Perhaps the need for divine deliberation is not related to God's interactions with his creation, but instead is required for him to remain free in his choices. This is Basinger's second problem with the Troublesome Implication, namely that it appears to be inconsistent with the ‘common theistic contention that God's actions are free in an indeterministic sense’ (ibid.). In motivating this particular rejoinder to the notion that all divine actions are the result of non-deliberative decisions, Basinger argues that if God has never deliberated over a decision then ‘there has never been a time when two creative options were still equally plausible alternatives in his mind’, and thus he has never chosen freely (ibid.).

The argument here seems to be something like the following:

1. (8) If God doesn't deliberate, then there is no time at which two options are equally preferable alternatives in his mind.Footnote ¹⁵

2. (9) If there is no time at which two options are equally preferable alternatives in God's mind, then he doesn't choose freely.

3. (10) Therefore, if God doesn't deliberate, then he doesn't choose freely.

The problem with this argument is that Basinger has given us no reason to think that (9) is true. Notice that (9) is equivalent to the following necessary condition on divine freedom: God chooses freely only if there is at least one situation in which there are equally preferable alternatives in his mind. Is this condition on divine freedom an instantiation of a more general requirement for choosing freely, or is it a requirement that applies to God but not other agents? I can't see any reason why such a requirement would apply to God alone, so I will assume that Basinger intends (9) to follow from a more general requirement for choosing freely. Call it the Equal Preferability Requirement:

Although several philosophers have defended a version of the Equal Preferability Requirement,Footnote ¹⁶ it is at best a controversial claim.Footnote ¹⁷ And I would argue that it is not just controversial, but false; for it implies that slight changes in the reasons favouring an alternative can make the difference between a free choice and an unfree choice. Consider a situation in which an agent, Silas, is fully exercising his rational and agential capacities in order to make a choice between two alternatives – between, say, having that extra cup of coffee or drinking water instead. We'll stipulate that both alternatives are equally preferable, and that there are no factors (such as coercion or manipulation) that would undermine Silas's freedom. Suppose that Silas chooses the coffee. Now alter the situation ever so slightly, such that the available reasons favour, but just barely, having the coffee. According to the Equal Preferability Requirement, we now have a situation in which Silas's choice – which, remember, otherwise manifests all of his rational and agential capacities – is not free. And this fact about Silas's choice has nothing to do with his abilities or capacities, or with external forces or states of affairs that are typically thought to influence whether a choice is free. Instead, this fact is simply a fact about the relative strength of his preferences. This result strikes me as highly implausible, and a good reason to reject the Equal Preferability Requirement.

To recap: (9) is true only if the Equal Preferability Requirement is true, and the Equal Preferability Requirement is arguably false. (It is at best a contentious claim.) Thus I conclude that, absent some additional argumentation in its favour, (9) is not a firm foundation on which to build an argument that deliberation is required for divine freedom.

Of course, even if the above argument fails to establish the necessity of deliberation for divine freedom, there may be other arguments for that conclusion. Thus a complete defence of the possibility of a God who is free and yet does not deliberate would require showing that no such arguments are available. I won't attempt to show that here, but I will provide two quick gestures in that direction.

One way to show that there are no successful arguments for the conclusion that divine freedom requires deliberation is to examine various plausible accounts of divine freedom, and then ask whether deliberation is required by any of those accounts. To consider just one brief example, Richard Swinburne provides an account according to which God is free if and only if ‘no external cause influences which purposes God forms: God acts only in so far as he sees reason for acting’ (Swinburne (Reference Swinburne1996), 43).Footnote ¹⁸ It seems pretty clear that neither having reasons for acting nor being free of external causal influences requires deliberation. Thus, to the extent that this conception of God's freedom is a satisfying one, the deliberation requirement is too strict.

Another way to argue that deliberation is not required for divine freedom is to argue that it's not required for freedom in general, and that nothing about the shift from freedom in general to divine freedom introduces such a requirement. Given what has been established above, namely that an omniscient being cannot be uncertain about which available alternatives are best (and thus cannot genuinely weigh those alternatives), it's hard to imagine that divine freedom has a deliberation requirement even if freedom in general does not. Thus, arguing that deliberation is not required for freedom in general should suffice to show that it's not required for divine freedom either. Here is a sketch of such an argument.

It is plausible to claim that someone has a will if she has some power by which she exercises decisive control over her behaviour.Footnote ¹⁹ There are, of course, all kinds of further questions about what it means to exercise decisive control over one's behaviour, but it's hard to see why deliberation would be required for such control.Footnote ²⁰ If this is right, then having a will doesn't require deliberation. Moreover, it's not clear why we would need to add deliberation in order to get from merely having a will to having a free will. There are plenty of candidates for that thing (or set of things) that need to be added in order to turn a will into a free will, but deliberation is not one of them. Perhaps something else needs to be true of the agent (e.g. that she has certain rational or normative capacities), or perhaps something else needs to be true of the world (e.g. that it's indeterministic), but in neither case is deliberation relevant.

To sum up: I have argued that God does not (and in fact cannot) deliberate, because an omniscient being cannot be uncertain about which alternative course of action is best and thus cannot genuinely weigh those alternatives. Some have argued that deliberation is required for divine freedom, but I have shown that these arguments fail – at least in so far as they rely on the Equal Preferability Requirement. Perhaps there are alternative arguments for a deliberation requirement on divine freedom, but I provided some reason to think that none of those arguments will succeed either. Thus we can plausibly maintain that God is omniscient in a robust sense even if omniscience and deliberation are incompatible.

However, someone might insist that God must deliberate – not for the sake of freedom, or genuine interaction with the world and the people in it, but simply because he is an agent, and the best possible being, and it's better to be an agent that deliberates than an agent that doesn't deliberate. If this line of reasoning is correct, and the argument above is sound, then the proper response is not to revise the claim that God is omniscient, but simply to abandon it.Footnote ²¹

I don't think that line of reasoning is correct. But I can accommodate it nonetheless, because the argument for the incompatibility of omniscience and deliberation is flawed in more than one way. In particular, it is flawed when it claims that it is impossible to deliberate about some decision while knowing what the result of that deliberation will be. Thus, as I argue in the next section, even if we assume that God does deliberate, we still do not need to revise the concept of omniscience (much less abandon it entirely).

The difference between practical settling and epistemic settling

I have argued that the deliberation restriction is irrelevant because God does not deliberate (and thus that (4) in the argument above is false). But even if we grant that (4) is true, the argument for the deliberation restriction on omniscience still fails. That is because, as I argue in this section, (5) is false.

Many philosophers have assumed, along the lines of premise (5) above, that it is impossible to deliberate while knowing the results of that very deliberation. And while there have been numerous recent endorsements of this supposition, it can be traced back at least as far as Richard Taylor (Reference Taylor1964), who claims that deliberation presupposes ignorance; without ignorance, deliberation is a sham. Seemingly deviant statements such as a governor's saying, ‘I am, as a result of my forthcoming deliberations, going to reprieve Smith’ illustrate this point, for it seems clear that in this case knowledge of the future reprieve precludes deliberation (ibid., 73). Taylor also offers a more elaborate example in which a groom (let's call him Gavin) observes the various trappings of his imminent wedding (ibid., 75). If such trappings are considered by him to be reliable indicators of the fact that he is about to be married, then he is unable to deliberate about what to do. On the other hand, if he is deliberating over whether to get married, then he cannot view the signs of his approaching wedding as reliable. Taylor is, in short, highlighting an essential difference between practical deliberation about what to do and theoretical deliberation about what to believe. To bring one's practical deliberations to a close is to settle the matter, practically speaking, by forming an intention to act (or refrain from acting). To bring one's theoretical deliberations to a close is also to settle the matter, but in an epistemic sense.Footnote ²²

Although Taylor highlights the difference between practical deliberation and theoretical deliberation in support of his claim that deliberation requires a certain kind of ignorance, the distinction between the two kinds of deliberation actually supports the opposing point. Since they are different kinds of deliberation, the default view should be that they could on occasion come apart, such that one kind concludes while the other kind continues. To reject this default view, we would need an argument that the two kinds cannot come apart – not just an example (as above) in which they don't. On the other hand, if we can construct a case in which they do come apart, then we can infer that no such argument is in the offing. Here is such a case.

Let us imagine that Gavin (the indecision of the wedding now behind him) is one of the plaintiffs in a knockdown class-action lawsuit.Footnote ²³ The defendants, anxious to get this business behind them, invite each of the plaintiffs, in turn, to consider a now-or-never settlement offer. As it happens, Gavin is the last to receive the offer, and as such is able to benefit from his knowledge of the choices of the previous plaintiffs. Oddly enough, each and every individual before him accepts the settlement (though they are bound by oath not to disclose the terms). Gavin, who is not particularly interested in the details of the case or even the amount of money at stake, has no reason to believe that his response will be any different than the numerous other plaintiffs who have gone before. Hence, Gavin eventually comes by the knowledge that he, too, will accept the offer of settlement. Now consider him in the negotiating room: upon receiving the offer, will he still be able to deliberate about what to do? It seems that he will indeed. There is no real barrier to him weighing the relevant alternatives (accept the settlement, or reject it) while at the same time retaining his knowledge that he will in fact accept the settlement.

Or consider a variation on the case.Footnote ²⁴ This time Gavin doesn't know what the others in the plaintiff class have decided, and he has been given twenty-four hours to think about whether he wants to accept the settlement offer. He shares the terms with his wife (to whom he's been married for some years now), and she, in some ways knowing him better than he knows himself, tells him that he will accept the offer. He thus comes to know, on the basis of testimony, that he will accept the settlement. And yet, again, it does not seem as though any sort of barrier has been erected that now prevents him from (practically) deliberating. The issue may have been settled epistemically (through his coming to know what he will do), but this is different from practically settling the issue.

If this is correct, then why does a view like Taylor's initially seem so plausible? (Why do statements such as the governor's ‘I am, as a result of my forthcoming deliberations, going to reprieve Smith’ seem so deviant?) The simple answer is that in the vast majority of cases, the way we come to know what we're going to do is by forming an intention, on the basis of deliberation, about what to do. Epistemic settling typically follows immediately after practical settling. Thus, when we're considering a situation involving deliberation, it's usually safe to assume that the deliberator will come to know what she's going to do in the typical way. But, as noted above (and as elaborated on below), epistemic settling is not by itself a barrier to deliberation and there can be unusual cases in which the epistemic settling occurs first.

There are, of course, alternative ways to explain what's happening in the Gavin cases. For example, someone could resist my characterization of the case by claiming that Gavin knows he will accept the offer but is not deliberating about whether to accept; instead he is seeking an explanation of why he will accept. On this view Gavin can examine alternatives, but not in order to choose between them: at most he can examine them in order to figure out why he chose the way he did. The problem with this way of looking at the case is that it implies an overly strong connection between evidence (in favour of a proposition) and choice (of action on the basis of the apparent truth of that proposition). If Gavin's examination of the alternatives can at most provide an explanation of why he chose the way he did, then his choice must have occurred at some point before he was made aware of the details of the settlement. This strikes me as an odd thing to have to say. There are certainly some cases in which an accumulation of evidence produces a decision indirectly (perhaps even without us being aware of the decision), but to say that this must happen in all such cases is implausible. In some cases there will be less than complete overlap between the set of facts that is essential for deliberation (which in Gavin's case would be the set of facts including the terms of the legal settlement) and the set of facts that constitutes potential evidence for the truth of a proposition describing the results of the deliberation process. When the overlap between these two sets is only partial, then it's possible that the deliberator could acquire evidence that is strong enough to give him knowledge of the results of the deliberation process, even though that process cannot start until the essential facts become available.

Here is another alternative explanation.Footnote ²⁵ Most of us have experienced a situation in which we were certain that we were going to make some choice C, but when the moment of decision arrived we actually made a different choice D. (Perhaps the evidence in this situation was even stronger than the evidence cited in the Gavin case.) In this type of situation the question of which choice we'll make is apparently epistemically settled, but not actually epistemically settled. So it's possible that deliberation is compatible with apparent epistemic settling but not with actual epistemic settling, and also that the Gavin case only seems plausible because of the compatibility between deliberation and apparent epistemic settling.

The problem with this alternative explanation is that if it's going to generate an objection to my claim (that epistemic settling and practical settling can come apart), it needs to come with an argument that all cases in which practical settling and epistemic settling come apart are actually just cases of apparent epistemic settling. I'm not sure what such an argument would look like, but here's one reason for thinking that no such argument exists. If such an argument did exist, then we would expect there to be some sort of correlation between practical settling and genuine epistemic settling, such that when epistemic settling follows practical settling it is typically genuine, but when epistemic settling precedes practical settling it is typically not genuine. However, no such correlation exists. This is partly because it is exceedingly rare for epistemic settling to precede practical settling, but also – and more importantly – because almost every case of merely apparent epistemic settling is one in which the practical settling comes first. Almost every such case is one in which we deliberate about what to do, and then decide what to do, and thereby (apparently) come to know what we're going to do; but then at the moment of choice we either reopen our deliberations or simply choose in a way that is contrary to what we thought we were going to do. (For example, perhaps you've had the experience of deciding to let a choice be dictated by a coin flip; but then, having flipped the coin and having seen how it came up, you find yourself unable to make the dictated choice. So you make the other choice instead.) In short, neither the presence nor the force of practical settling provides any clues as to when epistemic settling is genuine. So there seems to be no correlation of the type needed to support the claim that all instances of the two kinds of settling coming apart are instances in which the epistemic settling is merely apparent.

Thus, with the distinction between practical settling and epistemic settling firmly in hand, we can see our way clear to separating the practical question of whether Gavin should accept the settlement from the epistemic question of whether he will accept it. While it would indeed be impossible for him genuinely to wonder whether or not he will accept the settlement (since he already knows that he will), nothing about his knowledge of that acceptance precludes him from genuinely weighing the alternatives so as to determine the best course of action.Footnote ²⁶ It's possible, of course, that there are other necessary conditions on deliberation that are not satisfied in the example – but I'm not aware of any proposals in the literature involving a condition that is both necessary for deliberation and not satisfied in the cases described above.

Alternative constraints on deliberation

The examples above deliver a surprising result: it is after all possible for someone to deliberate while knowing how that deliberation will turn out. In other words, assuming that this result transfers from the human case to the divine case,Footnote ²⁷ (5) is false and thus we have no reason (at least no reason stemming from considerations about deliberation) to accept (6):

1. (6) There will be some time t in God's life during which he does not know some true propositions (namely, those involving the results of his deliberation at t).

And since (6) is an essential component of the argument for the deliberation restriction on omniscience, that restriction is unmotivated.

But perhaps the defender of the deliberation restriction can motivate it a different way, by reformulating the constraint on deliberation suggested in (5). For example, even if knowledge of the results doesn't preclude deliberation, perhaps certainty about the results does.Footnote ²⁸ This might be a promising strategy – and there has been quite a bit of interesting recent work on the question of what exactly deliberation requires and involvesFootnote ²⁹ – but it needs further development. The initial steps seem plausible. Assume some fallibilist account of knowledge, and then point out that the examples above are all cases in which the deliberator knows the results of his deliberation, but is not certain of them. Claim that if he were certain of the results, he would not be able to deliberate. And then argue that whereas human knowledge rarely (if ever) involves certainty, God's knowledge always does.

The problem with this suggestion is that replacing ‘knowledge’ with ‘certainty’ doesn't make the proposed deliberation constraint any more viable. If it really is possible to acquire knowledge about the results of deliberation on the basis of induction, or testimony, or some other generally reliable process, and yet still be able to deliberate, then it's hard to see how changing the process by which that knowledge is acquired (or the strength of the justification for the belief involved) changes anything about the relevant agent's capacity for deliberation. There are some things about which we're certain, and with respect to which we cannot deliberate; but I would argue that these are all things over which we have no control. It is that fact that precludes deliberation, and not anything about the strength of our justification for the relevant belief.

Barriers to deliberation

To elaborate on the previous point (and in further support of this idea that epistemic settling does not entail practical settling), consider some of the genuine barriers to deliberation: ignorance, impossibility, and inefficacy.Footnote ³⁰ Certain kinds of ignorance, for example, preclude deliberation: if I'm not aware of an alternative, then I cannot consider it. (If I'm not aware that there's a lift nearby, then I can't deliberate about whether to take the lift or the stairs.) So does impossibility: if some state of affairs is impossible (and I'm aware that it's impossible), then I cannot deliberate about whether to bring it about or prevent it. (If it's impossible for anything to travel faster than the speed of light, then I cannot deliberate about whether to make something travel faster than the speed of light.Footnote ³¹) And finally, if my deliberation would not be effective in producing a decision (and I'm aware of this inefficacy), then I cannot deliberate about whether to make that decision. (If I'm aware that an evil neuroscientist will force me to vote for a particular presidential candidate, then I cannot deliberate about which candidate to vote for.Footnote ³²)

There may also be psychological barriers to deliberation. Someone who is in the throes of drug addiction may not be able to deliberate about whether to take the drug. Alternatively, I don't think anyone in normal circumstances could deliberate about whether to kill a loved one in exchange for a dollar. (I will admit that I'm not confident about this last example; perhaps deliberation remains possible even when all of the alternatives are what we would otherwise describe as unthinkable.)

Ignorance, impossibility, and inefficacy (together with certain psychological barriers) thus represent the kind of factors that preclude deliberation. Notice, however, that none of these factors need be present in a case in which someone knows what the results of his deliberation will be. It should be clear, in other words, that knowledge of what one is going to do does not produce anything resembling the most common and uncontroversial barriers to deliberation. (This is another reason why the attempt to rescue the deliberation constraint by shifting from knowledge of the results to certainty about the results fails. Adding certainty to the equation doesn't introduce any factors that resemble barriers to deliberation.) Thus I conclude that it is possible to deliberate while knowing what the results of that deliberation will be.