Causal histories and freedom

Linda Zagzebski proposes the plausible principle that one is able to ϕ only if ϕing is compatible with one’s causal history relevant to ϕing.

Suppose Alice is considering whether to rob a bank. While she is doing so, God loudly announces to her nearby friend Bob that Alice will not rob the bank. God’s announcement is in Swahili, which Bob knows and Alice used to know in childhood but completely forgot. But the sound of the language her loving parents spoke to her as a child leads to Alice putting more emphasis on virtue in her deliberation, and she freely decides not to rob the bank.

Since God cannot lie, and since God’s announcement is a part of Alice’s causal history in her deliberation, Alice’s robbing the bank is incompatible with causal history and by Zagzebski’s principle, Alice cannot rob the bank. Yet it is unclear how God’s announcement removes her freedom to rob. After all, had God announced in Swahili that Alice will have breakfast, that would have influenced her deliberation in the same way, and yet obviously she would still have been free to rob the bank. But since Alice doesn’t know Swahili, the content seems causally irrelevant.

I think there are two ways out of this. First, we might cut events very finely. There is (a) God’s saying something or other in Swahili and there is (b) God’s saying in Swahili that Alice will not rob the bank. To determine the causal history, we pare away from the events all that’s causally irrelevant, and so we include (a) but not (b) in the causal history.

Alternately, we might say this. Whether or not Alice knows Swahili, her decision is affected by the detailed facts about the sounds in God’s announcement. Indeed, by essentiality of origins, her deliberation is a numerically different process because of the difference of sounds. And now we can say that God cannot make the announcement, because doing so would result in a circularity in the explanatory order: God would be making the announcement because Alice is not going to rob and Alice is not going to rob because God is making the announcement because she is not going to rob. So it is not so much that the announcement takes away God’s freedom, but that God cannot produce explanatory circularities.

It’s worth noting that Molinism does not seem to help. Sure, the subjunctive conditional of free will

1. Were God to announce in Swahili that Alice won’t rob, Alice wouldn’t rob

is true. But it is necessarily true independently of Molinism!

A really bad moral dilemma

Here would be a really bad kind of moral dilemma:

• It is certain that unless you murder one innocent person now, you will freely become a mass murderer, but if you do murder that innocent person, you will freely repent of it later and live an exemplary life.

If compatibilism is true, such dilemmas are possible—the world could be so set up that these unfortunate free choices are inevitable. If compatibilism is false, such dilemmas are impossible, absent Molinism.

We might have a strong intuition that such dilemmas are impossible. If so, maybe that gives us another reason to reject compatibilism and Molinism.

The Church-Turing Thesis and generalized Molinism

The physical Church-Turing (PCT) thesis says that anything that can be physically computed can be computed by a Turing machine.

If generalized Molinism—the thesis that for any sufficiently precisely described counterfactual situation, there is a fact of the matter what would happen in that situation—is true, and indeterminism is true, then PCT seems very likely false. For imagine the function f from the natural numbers to {0, 1} such that f(n) is 1 if and only if the coin toss on day n would be heads, were I to live forever and daily toss a fair coin—with whatever other details need to be put in to get the "sufficiently precisely described". But only countably many functions are Turing computable, so with probability one, an infinite sequence of coin tosses would define a Turing non-computable function. But f is physically computable: I could just do the experiment.

But wait: I’m going to die, and even if there is an afterlife, it doesn’t seem right to characterize whatever happens in the afterlife as physical computation. So all I can compute is f(n) for n < 30000 or so.

Fair enough. But if we say this, then the PCT becomes trivial. For given finite life-spans of human beings and of any machinery in an expanding universe with increasing entropy, only finitely many values of any given function can be physically computed. And any function defined on a finite set can, of course, be trivially computed by a Turing machine via a lookup-table.

So, either we trivialize PCT by insisting on the facts of our physical universe that put a finite limit on our computations, or in our notion of “physically computed” we allow for idealizations that make it possible to go on forever. If we do allow for such idealizations, then my argument works: generalized Molinism makes PCT unlikely to be true.

Transworld depravity is false

Plantinga’s transworld depravity thesis holds that in every world that God is contingently capable of actualizing (i.e., every “feasible” world), either there is no significant freedom or there is at least one free wrong choice. I will argue that transworld depravity is in fact false, assuming Molinism.

But consider a possible situation A where the first significantly free choice runs as follows. Eve has a choice whether to eat a delicious apple or not, while knowing that God has forbidden her from eating the apple. Eve comes into the choice with a pretty decent character. In particular, she is so constructed that she is unable to take God’s prohibitions to be anything but reasons against an action and God’s commands to be anything but reasons for an action. Nonetheless, she is free: she can choose to eat the apple on account of its deliciousness, despite God’s prohibiting it.

By Molinism, if enough detail is built into the situation, either:

1. in A, Eve would eat the apple, or

2. in A, Eve would not eat the apple.

If (2) is true, then transworld depravity is false, because God could simply take away freedom after Eve’s first choice, and so we have a feasible world where there is exactly one significantly free choice, and it’s right.

Suppose then (1) is true. Now imagine a situation A^* where just before Eve is deliberating whether to eat the apple, God announces that the prohibition on eating the apple is now changed into a command to eat the apple. If in A, Eve would eat the apple on account of its deliciousness despite its being forbidden, she would a fortiori eat the apple if God were to command her to do so. Thus:

3. in A^*, Eve would eat the apple.

But then transworld depravity is false, because again God could take freedom away after Eve’s first choice.

The argument as it stands does not show that transworld depravity is necessarily false. I try to do that here with a similar but perhaps less compelling argument.

Subjective sameness of choice situations and Molinism

Suppose that Alice on a street corner sells Bob a “Rolex” for $15. Bob goes home and his wife Carla says: “You got scammed!” Bob takes the “Rolex” to a jeweller and finds that it is indeed a Rolex. He goes back to Carla and says: “No, I got a good deal!” But Carla says: “But if it was a fake, you would have bought it, too.”

Here is Carla’s reasoning behind her counterfactual. A typical fake would have looked the same as the real thing to Bob, and so the counterfactual situation where Bob is offered the fake would have been subjectively the same to Bob. And Carla subscribes to this principle:

1. If you were instead in a different situation that was subjectively identical to the one you were in, you would have chosen the same way.

I find (1) pretty plausible, and it seems to nicely come out as true on a Lewis-style account of counterfactuals in terms of similarity of worlds (with approximate match counting as similarity).

But I doubt a Molinist should accept (1). For the Molinist does not think that counterfactuals of free will are grounded in similarity of worlds (except of the trivial sort, where truth values of counterfactuals count as part of the similarity).

A bad argument from hiddenness

Consider the following variant of the argument from hiddenness:

1. If God exists, no mature human is ignorant of God’s existence through no fault of their own.

2. Some mature humans are ignorant of God’s existence through no fault of their own.

3. So, God doesn’t exist.

It’s occurred to me that premises like (3) are either nonsense, or trivially false, or far beyond our capacity to know to be true.

For to evaluate whether some x is ignorant of God’s existence through no fault of their own requires asking something like this:

4. Would x still have been ignorant of God’s existence had x lived a morally perfect life?

But it does not seem likely that there is a sensible positive answer to (4). Here is a quick argument for this. Those who deny Molinism are going to say that either the proposition asked about in (4) has no truth value or that it is trivially false. And even some Molinists will say this about (4), because Molinists are committed to there being conditionals of free will only when the antecedent is maximally specified, while “x lived a morally perfect life” is too unspecified. The question is much like:

5. Had Napoleon been born in South America, would he still have been a great military leader?

There are many ways for Napoleon to have been in South America, and they are apt to result in different answers to the question about whether he was a great military leader.

But even if (4) has a truth value, perhaps because (4) is to be interpreted in some probabilistic way or because we have an expansive version of Molinism that makes (4) make sense, it is far beyond our epistemic powers to know the answer to (4) to be true. Here is why. Our lives are full of wrongdoing. Our lives would likely be unrecognizable had they been morally perfect. To ask what we would have thought and known in the counterfactual scenario where we live a morally perfect life is to ask about a scenario further from actuality than Napoleon’s being born in South America.

Now, that said, there are times when we can evaluate counterfactuals that involve a massive change to the antecedent on the basis of certain generalities. For instance, while we have no answer to (5), we do have a negative answer to:

6. Had Napoleon suffered a massive head injury rendering him incapable of interpersonal communication, would he still have been a great military leader?

Similarly, if an atheist had suffered a head injury removing the capacity for higher level thought, the shape of their life would have been very different, but at least we can say that they wouldn’t have been an atheist, because they wouldn’t have had the concepts necessary to form the belief that there is no God. So, indeed, sometimes counterfactuals that take us far afield can be evaluated sensibly on general grounds.

But I do not think we have good general grounds for a positive answer to (4), unless we have independent grounds to doubt the truth and rationality of theism:

7. There are no good grounds for reasonably believing in God, and a person who lives a morally innocent life won’t believe things groundlessly, so they won’t believe in God.

8. There are people who grow up in societies where there is no concept of God, and they would not be aware of God no matter what the shape of their lives would have been.

Obviously, (7) requires independent grounds to doubt the rationality of theism. And if God exists, then for all we know, he has a general practice of making those who are morally perfect be aware of him, so (8) is dubious if God exists.

Of course, an atheist might think (7) is true, but this is unlikely to be a helpful move in an argument against the existence of God. After all, similarly, a theist might think the following is true:

9. God will ensure that every morally perfect mature human is aware of him.

Indeed, a typical Christian thinks that there have only ever been one or two people—Jesus and maybe Mary—who have been morally perfect, and both candidates were aware of God.

Molinism and sceptical theism

When we think of God’s reasons for permitting evils, we tend to think of fairly “natural” connections between evils and goods. But given Molinism, there could be some really weird connections. For instance, it could be that if Alice hadn’t been cut off by Bob in traffic today, Carl who witnessed this would have joined a terrorist organization. Not because there is any intrinsic connection between seeing someone get cut off in traffic and joining a terrorist organization, but just because that’s how the conditionals of free will worked out.

Indeed, a Molinist should expect there to be cases where the Molinist conditionals work out the opposite way to the “natural” connections. Thus, we can have cases where becoming more cowardly results in one’s behaving more courageously, just as a Molinist God might know that if the coin loaded in favor of heads would show tails in the next ten tosses while the coin loaded in favor of tails would show heads in the text ten tosses.

So there seems to me to be a very nice affinity between Molinism and sceptical theism.

It’s really too bad that Molinism is false.

A pre-established harmony with genuine mind-world causation

Molinists have the ability to give a distinctive pre-established harmony account of how the exceptionless truth of deterministic laws of nature could be made compatible with libertarianism.

Here is the story. God considers possible worlds where dualistic agents have the causal power to miraculously contradict the physical laws in their free choices, but where outside of exercises of free will, there are elegant deterministic mathematical laws governing the world. Call such worlds Candidate Worlds.

God then narrows his consideration to Finalist Worlds, which are Candidate Worlds that are feasible—i.e., compatible with the Molinist conditionals—and where as it happens the agents’ free choices accord with the elegant deterministic mathematical laws that govern the rest of the world.

And then God wisely and prudently chooses one of the Finalist Worlds for actualization.

On this story, there are elegant deterministic mathematical laws of nature which are true even of the agents’ choices, but they are true of the agents’ choices because the agents freely chose as they did. The agents had the causal power to violate, say, the conservation of momentum, but in fact freely did not do so.

There is an ambiguity in the concept of “exceptionless laws”. “Exceptionless laws” could mean: laws that allow no exception (they are pushy laws that are so strong as to make no exception possible) and laws that in fact have no exception. The deterministic laws in this story are exceptionless in the sense of having no exception, which is why I am talking of their exceptionless truth rather than their exceptionless power.

In this story, there is a dual explanation of the agents’ choices. On the one hand, there is a standard libertarian story about the agents’ free causality. On the other hand, the laws have explanatory power, because God chose the Finalist Worlds because they are worlds where the laws have exceptionless truth.

The big difficulty with the above story is Molinism. Also, it is worth noting that it is metaphysically possible that there turn out not to be any (feasible) Finalist Worlds: in that scenario, God wouldn’t be able to create a world where there is freedom and elegant deterministic mathematical laws holding exceptionlessly. But for reasons similar to why many people think Transworld Depravity is unlikely to be true, I think it is unlikely that there would be no Finalist Worlds.

It is also interesting to note that there are two more views that could be plugged into the story that can do the same job: Thomism and compatibilism.

On Thomism, God can use primary causation to make agents freely choose as he desires. Then we can suppose that God surveys the same Candidate Worlds as on the Molinist story. Then he chooses Finalist Worlds as those Candidate Worlds where the agents’ free choices in fact do not contradict the mathematical laws. And then God uses primary causation to actualize one of the Finalist Worlds. Again, the agents have the power to contradict the laws, but freely choose not to exercise it.

Finally we have straightforward compatibilism. A dualist can just as easily be a compatibilist as a materialist. On this story, we skip the Candidate Worlds, and the Finalist Worlds are worlds with compatibilist agents, with a deterministic non-physical mental life, who have the power of contradict the physical laws of the world but who are mentally determined, in a way compatible with freedom, never to exercise such a power. And then God chooses one of the Finalist Worlds. The agents then are as free as any compatibilist agents.

The compatibilist version of this story is close to Leibniz’s pre-established harmony, except that it has real mind-world causation, which is a big improvement.

Of course, the non-theist can’t make any of these moves. And, alas, neither can I, since my mere foreknowledge view denies Molinism, Thomism (about free will) and compatibilism.

Molinism and Thomism and control over others

1. It is not possible for a creature to exercise complete control over another person’s (non-derivatively) free action.

2. If Molinism is true, it is possible for a creature to exercise complete control over another person’s (non-derivatively) free action.

3. So, Molinism is false.

For, if Molinism is true, there will be a possible situation where God reveals to Alice that if she were to make a request of Bob while wearing blue gloves, Bob would acquiesce to the request, but if she were to make the request while wearing red gloves, Bob would turn down the request. In such a case, by controlling which gloves she wears, Alice could exercise complete control over whether Bob acquiesces to the request.

Interestingly the same argument works against Thomism. For on Thomism, God can use primary causation to determine Bob to freely acquiesce in the request and God can use primary causation to determine Bob to freely refuse the request. God could then promise Alice that he would hear her prayers as to whether Bob agrees or refuses, and then with her prayers, Alice would have complete control over Bob's decision.

The argument doesn't work against mere foreknowledge views, open theist views or compatibilist views. On mere foreknowledge and open theism, the analogue of (2) is false, while on compatibilism, (1) is not plausible.

God, probabilities and causal propensities

Suppose a poor and good person is forced to flip a fair and indeterministic coin in circumstances where heads means utter ruin and tails means financial redemption. If either Molinism or Thomism is true, we would expect that, even without taking into account miracles:

1. P(H)<P(T).

After all, God is good, and so he is more likely to try to get the good outcome for the person. (Of course, there are other considerations involved, so the boost in probability in favor of tails may be small.)

The Molinist can give this story. God knows how the coin would come out in various circumstances. He is more likely to ensure the occurrence of circumstances in which the subjunctive conditionals say that tails would comes up. The Thomist, on the other hand, will say that God’s primary causation determines what effect the secondary creaturely causation has, while at the same time ensuring that the secondary causation is genuinely doing its causal job.

But given (1), how can we say that the coin is fair? Here is a possibility. The probabilities in (1) take God’s dispositions into account. But we can also look simply at the causal propensities of the coin. The causal propensities of the coin are equibalanced between heads and tails. In addition to the probabilities in (1), which take everything including God into account, we can talk of coin-grounded causal chances, which are basically determined by the ratios of strength in the causal propensities. And the coin-grounded causal chances are 1/2 for heads and 1/2 for tails. But given Molinism or Thomism, these chances are not wholly determinative of the probabilities and the frequencies in repeat experiments, since the latter need to take into account the skewing due to God’s preference for the good.

So we get two sets of probabilities: The all-things-considered probabilities P that take God into account and that yield (1) and the creatures-only-considered probabilities P_c on which:

2. P_c(H)=P_c(T)=1/2.

Here, however, is something that I think is a little troubling about both the Molinist and Thomist lines. The creatures-only-considered probabilities are obviously close to the observed frequencies. Why? I think the Molinist and Thomist have to say this: They are close because God chooses to act in such ways that the actual frequencies are approximately proportional to the strengths of causal propensities that P_c is based on. But then the frequencies of coin toss outcomes are not directly due to the causal propensities of the coin, but only because God chooses to make the frequencies match. This doesn’t seem right and is a reason why I want to adopt neither Molinism nor Thomism but a version of mere foreknowledge.

Molinism and behavioral dispositions

There is some sort of a link between counterfactuals and dispositions, though there are lots of counterexamples to direct links. Here is a very weak principle affirming such a link:

1. Suppose that x is in state S at time t and that x’s being in S at t grounds x’s being disposed to behavior A after t. Then there is some maximally determinate categorical proposition p describing the world up to time t and logically compatible with x’s being in S at t such that it is false that were p true and x in state S, x would fail to engage in behavior A after t.

To put it very roughly, this messy principle says that if a disposition to a behavior is grounded in a state, then it’s not the case that no matter what one adds to the state, the behavior would not occur. Suppose that (1) is a necessary truth.

Add this:

2. It is possible for a human being to have an unactualized indeterministic disposition with respect to non-derivatively free behavior.

For instance, there is presumably a shade R of red that Jean Vanier has never met someone wearing, and yet he is disposed to behave non-derivatively freely kindly to persons wearing R.

What I have said so far does not, however, cohere with Molinism. For on Molinism, the conditionals of free will logically float free from the indeterministic dispositions of things. There is, for instance, a possible world where Jean Vanier still has the same kindly dispositions that he does in the actual world, but where the Molinist conditionals say that in the case of any of the appropriate maximally deterministic categorical strengthenings of the claim that he meets a person wearing R, if that strengthening were actual, he would behave unkindly to that person. This would violate (1).

[Note added later: This was, of course, written before the revelations about Jean Vanier's abusiveness. I would certainly have chosen a different example if I were writing this post now.]

Hiddenness and Molinism

Schellenberg claims that God cannot coexist with a non-resistant non-believer, since God being love would ensure that everyone who is non-resistant would be given the conditions necessary for a personal relationship with God.

It seems to me that a Molinist has a nice answer to this. A loving God would not want to compel people to have a particular kind of relationship with him, and would hence leave them free. But now imagine a particular non-resistant non-believer, Alice. God could know of Alice that if she believed in God and were free with respect to a relationship with God, she would freely choose a bad relationship with God. Then here are God’s main options with respect to Alice:

1. Not create Alice

2. Ensure Alice believes in God but make it impossible for her to have a bad relationship with God

3. Ensure Alice does not believe in God

4. Ensure Alice believes in God and allow her to have a bad relationship with God.

For Schellenberg’s case to work, (3) has to be an unacceptable option for a loving God. But (3) seems better than (1) and (4), and (2) seems contrary to the way that love requires respect for the freedom of the beloved. So while (3) is not ideal, it seems better than the alternatives.

And it could be—this is parallel to Trans-World Depravity—that in every feasible world there is someone like Alice.

I think the main response would be that a person who would have a bad relationship with God counts as resistant—i.e., a disposition to a bad relationship with God counts as resistance. However, this misses the Molinist point. Molinist conditionals of free will are not grounded in present character. One can be such that one would have a bad relationship with God if one believed in God, without having a disposition to such a relationship. One’s present character might, for instance, be neutral or even favoring of a good relationship with God, but given Molinism, it could be that were one to come to the decision point, one would decide against the relationship.

One could redefine non-resistance as being such that were one to believe, one would be in a good relationship with God. But because Molinist conditionals are ungrounded, we wouldn’t know whether a particular non-believer is resistant or not.

It's a pity that Molinism is false.

Molinism and prophecy

Here’s a curious puzzle. Every theist—including the Molinist and the Open Theist—will presumably agree that this conditional is true:

1. If God were to announce that Trump will freely refrain from tweeting tomorrow, then Trump would freely refrain from tweeting tomorrow.

After all, both presumably accept that God wouldn’t affirm what he didn’t know to be the case.

However, plainly, the truth of (1) isn’t enough to justify God in announcing that Trump will freely refrain from tweeting tomorrow. For, plainly, something like Molinist middle knowledge or mere foreknowledge or theistic compatibilism would be needed for God to be justified in issuing the prophecy. Something like (1) that holds independently of theories of divine foreknowledge is not going to do the trick.

Suppose now Molinism is true. It seems to be one of the advantages of Molinism that it can explain prophecy. But what relevant proposition beyond (1) does God know in the Molinist case that justifies his announcement?

Here is one possibility:

2. Trump will freely refrain from tweeting tomorrow.

But God’s knowing (2) is insufficient to justify God’s announcement. For imagine that the reason why Trump will refrain from tweeting tomorrow is that Trump likes to surprise people and nobody predicted that he wouldn’t tweet tomorrow. Then (2) can still be true—but if that’s the reason why (2) is true, then the truth of (2) won’t justify God in announcing that Trump won’t tweet.

I think what we want to say is that on Molinism what justifies God’s announcement is something like this:

3. Claim (1) holds not just because God’s announcements are always true.

But now here is the problem. If claim (1) holds not just because God’s announcements are always true, there must be some further explanation for why claim (1) is true other than just because God’s announcements are always true. But what is that explanation? Presumably it lies in the truth of some Molinist conditional. But it seems that the most relevant Molinist conditional is (1) itself, and that just won’t do.

Here’s another way of putting the point. The Molinist’s best response to the grounding objection is to say that Molinist conditionals are true but ungrounded. Such a Molinist has to say that the only reason (1) is ever true is that God doesn’t make untrue announcements. But, plausibly, if the only reason (1) is true is that God doesn’t make untrue announcements, then God isn’t justified in issuing the announcement. So God is never justified in issuing the announcement.

If I were a Molinist, I would say that God cannot make prophecies that end up being explanatorily prior to the prophesied actions. But if one makes that restriction, one might as well accept mere foreknowledge.

Deontology and future hypothetical wrongs

Molinism makes possible a curious kind of moral dilemma. God could reveal to Alice that if Alice doesn’t kill Bob today, she will kill Carl and David tomorrow (all these being innocents), and if she does kill Bob today, she won’t kill anyone tomorrow. Should she, thus, kill Bob today in order to prevent herself from murdering Carl and David tomorrow?

One might think that the possibility that Molinism allows for such a moral dilemma is a count against Molinism. But even without Molinism, one could have a probabilistic version of the dilemma where God reveals to Alice that if she doesn’t kill Bob today, she is very likely to kill both Carl and David tomorrow, and if she does kill Bob today, she is very unlikely to kill anyone tomorrow.

One way to make consequentialism fit with deontological intuitions is to set a high, perhaps infinite, disvalue on wrong action. That would imply that in the dilemma Alice should kill Bob in order to prevent the two murders tomorrow.

I think this is a mistake. Just as on deontological grounds it would be wrong for Alice to murder Bob to keep Eva from murdering Carl and David, so too it’s wrong for her to murder Bob to keep herself from murdering them. A eudaimonist may disagree here, holding that we should be promoting our own flourishing, so that when the choice is between committing two murders tomorrow and one today, we should go for the one today, but when the choice is between oneself committing one murder and another party committing two, we should let the other party commit the two. So much the worse, I say, for that kind of eudaimonist.

What makes it wrong for Alice to murder Bob is that the we shouldn’t perform bad acts. It’s not that we should minimize the number of bad acts performed, by others or oneself, but that we shouldn’t perform them. Of course, all other things being equal we should minimize the number of bad acts performed, by others and oneself, but a bad act is an act not to be done. And the lesson of deontology is that certain acts, such as intentionally killing without proper authority, are bad acts in virtue of their nature.

But isn’t killing Bob today the lesser evil?

Yet imagine Alice is debating whether she should eat ice cream, with its having been revealed to her that if she eats ice cream today, tomorrow she will kill Bob, and if she does not, then tomorrow she will kill Carl and David. In that case, it is clear: she should eat the ice cream. For the eating of ice cream isn’t the sort of act that is bad in virtue of its nature (unless a very strong form of moral veganism is true). Note, however, that if she eats the ice cream today, then her killing of Bob tomorrow is still wrong. (If you disagree, it may be simply because you disagree with Molinism, and you hold that the inevitability of her killing Bob takes away her freedom; if you think that, then go for a probabilistic version of the story.) This is true even though it is a lesser evil than her killing Carl and David.

In the original case, we can look at Alice doing two things when killing Bob:

1. Killing Bob

2. Bringing it about that she doesn’t kill Carl and David.

Her action is bad qua (1) and good qua (2). But we learn from Aquinas that for an action to be right, it must be right in every respect. So her action is wrong simpliciter.

On the other hand, in the ice cream version, in consuming the ice cream, Alice is doing two things:

3. Eating ice cream

4. Bringing it about that she doesn’t kill Carl and David.

Now her action is good or neutral qua (3) and good qua (4). In fact, it’s right in every respect. But her later killing of Bob is still wrong.

Molinism and the Principal Principle

Molinism says that there are non-trivial conditionals of free will and that God providentially decides what to create on the basis of his knowledge of them. I shall assume, for simplicity of examples, that what Molinism says about free will it says about other non-determined events. The Principal Principle says that when you know for sure that the objective chance of some future event is r, the epistemic probability you should assign to that event is r.

Suppose a dictator has set up a machine what will flip an indeterministic and fair coin. On heads, the coin will trigger a bomb that will kill almost everyone on earth. On tails, it will do nothing. Since the coin is fair, the objective chance of heads is 1/2. But suppose you are sure that Molinism is true. Then you should think: "Likely, God would only allow this situation to happen if he knew that the coin flipped in these circumstances would land tails. So, probably, the coin will land tails." Maybe you aren't too convinced by this argument--maybe God would allow the coin to land tails and then miraculously stop the bomb or maybe God is fed up with 99% of humankind. But clearly taking into account your beliefs about God and Molinism will introduce some bias in favor of tails.

This seems to be a violation of the Principal Principle: the objective chance of heads is 1/2 but the rational epistemic probability of heads is at least a little less than 1/2.

Not so fast! The "objective chances" need to be understood carefully in cases where foreknowledge of the future is involved. An assumption behind the Principal Principle is that our evidence only includes information about the past and present. If we know that a true prophet prophesied tails, then our credence in tails should be high even if the coin is fair. Given Molinism, the fact that God allowed the coin toss to take place is information about the future, since it indicates that the coin is less likely to land heads given the disastrous consequences.

So, Molinism is compatible with the Principal Principle, but it renders the Principal Principle inapplicable in cases where it matters to God how a random process will go. But everything that matters matters to God. So the Principal Principle is inapplicable in cases where the outcomes of the random process matter, if Molinism is true. This renders the Principal Principle not very useful. Yet it seems that we need the Principal Principle to reason about the future, and we need it precisely in the cases that matter. So we have a bit of a problem for Molinists.

An argument that Trans-World Depravity is unlikely to be true

Assume Molinism. Plantinga's Trans-World Depravity (TWD) is the thesis that every feasible world--world compatible with the conditionals of free will--that contains at least one significantly free choice contains at least one sin. I want to think about an argument that TWD is likely false.

For consider a world where God creates exactly one intelligent creature with the typical motivations and character of a typical morally upright adult human being. God then forbids the creature from imposing pointless pain on itself, and only ever gives the creature only one significantly free choice: to eat a nutritious food that it likes or to endure five hours of torture. Let's imagine the situation where God creates such a creature and it's about to make that one significantly free choice. Call this circumstances C. Given what we know about decent human beings and their motivations, the creature would very likely eat the nutritious food rather than be tortured. Very well. So very likely the conditionals of free will are such that the world where the creature eats the nutritious food is feasible. But if that world is feasible, then TWD is false.

That was too quick. I jumped between answers to two different probabilistic questions:

1. What is the epistemic probability of the Molinist conditional that were C to obtain, the creature would choose wrongly?
2. Were C to obtain, what would be the chance of the creature choosing wrongly?

It is clear that the answer to (2) is "Very low." But to argue that TWD is very likely false, I have to say that the answer to (1) is also "very low". This leads to a difficult set of questions about the relationship between Molinist conditionals and chances. Lewis's principal principle does imply that if we were to knowingly (with certainty) find ourselves in C, and if we were certain of Molinism, we would have to give the same answer to (1) and (2). The argument goes as follows: given C the Molinist conditional has the same epistemic probability as its consequent, but the epistemic probability of its consequent is the same as its chance by the principal principle. But the answer we should give to (2) in those circumstances where we were knowingly in C may not be the same as answer as we should actually give to (2). Consider this possibility. Our current epistemic probability of the Molinist conditional in (1) is 1/2, but God would be very unlikely to make C obtain unless the conditional were false. He just wouldn't want to create a world where the creature would freely wrongfully choose to endure the torture. In that case, if we were to learn that C obtains, that would give us information that the Molinist conditional is very likely false. And hence the answer to (1) is "1/2", the answer to (2) is "Very low", but were C to obtain, the answer to (1) would be "Very low" as well.

Maybe. But I think things may be even less clear. For the biased sampling involved in God's choosing what to create on the basis of conditionals of free will undercuts the principal principle, I think. I think more work is needed to be done to figure out whether or not there is a good argument against TWD here or not.

Two kinds of free will theodicies

I want to make a distinction that I've long thought is quite important, but which I think is not made enough. Suppose the theist responds to the problem of moral evil by saying that God's allowing moral evils is justified by the value of free will. There are two different ways that "the value of free will" could be invoked here.

First, the theodicy could depend on the intrinsic value of freedom itself. God allows me to freely choose the wrong thing, because my freely choosing something is intrinsically good qua free choice even when the thing I choose is bad.

Second, the theodicy could depend not on any value of the freedom in freely choosing the wrong thing, but only on the value that there would be in freely choosing the right thing were one to choose that. On this story, the only logically possible way God can have a chance of my freely choosing right over wrong in a given situation is by allowing me to have the possibility of choosing wrong over right in that situation. Notice that on this second story, if I end up freely choosing wrong over right, the evil of my choice might well be gratuitous in the sense that there is no compensating good for it. But even if the evil is gratuitous, God could be justified in permitting the evil to happen, since the only way he had to prevent it would have removed all possibility of my freely choosing right over wrong. This second story doesn't require there to be any intrinsic value in freedom as such. It only requires there to be value in freely choosing right over wrong.

I think the second story is superior to the first. But, interestingly, the second story is not available to a Molinist. For the second story only works when God cannot rely on knowing how you would choose in a situation when setting up the situation. (The story is available to a simple foreknowledge theorist if God cannot rely on knowledge of the future in ways that set up explanatory circularity.)

Procreation and love

Start with this consequence of the essentiality of origins:

1. The identity of an individual depends on the exact time of her coming-into-existence.

(It's very plausible that it depends on coarse-grained time of coming-into-existence: I couldn't have come into existence a hundred years earlier. But there is also a well-known argument from that for the claim that it depends on the exact time.) Now, it seems that no action people can take prior to the origination of an individual determines the exact time of her coming-into-existence. There isn't any true counterfactual of the form of the form: "If we were to do A, then an individual would come into existence exactly at t." In fact, for any action we can take, quantum indeterminism will ensure a cluster of infinitely (if time is continuous) many or at least very many close-together possible resulting conception times. Moreover, each exact time will have a very tiny, perhaps infinitesimal or zero, probability of being the actual time of origination. Hence it seems:

2. There is no true probabilistic counterfactual of the form: "If we were to do A, then an individual would be not unlikely to come into existence exactly at t."

(Molinists will, of course, disagree. But Molinism is false, I take it.) Thus:

3. People cannot procreate out of love for the particular individual who would likely come into existence from the procreation.

For there is no particular individual such that she would likely come into existence from the procreation.

Now, notice that God in creating an individual--either ex nihilo or in cooperating with the joining of an egg and sperm--is under no such constraints. Thus it is possible for God to act in such a way that were he to act so, an individual would come into existence at a precise time. And the same is true for other features of the origination of the individual besides the time of its occurrence. So God could create out of love for the particular individual who would come into existence from the creation.

This fact suggests a weakness in premise (2) of my above argument. It could be that God has already decided what individual would result if a couple chose to procreate, and has resolved in particular that sperm and egg would meet at a particular time, as long as the couple chooses the procreate. God, thus, is resolved to control all the quantum phenomena to ensure a particular circumstance of origination. Thus it's up to the couple whether someone comes into existence, but there is a particular someone who would exist if the couple procreated. It would not be surprising if this were to happen in special cases, and it would be possible for it to happen in all cases. I don't know if it does, though.

The upshot of the above argument, thus, is that unless special theological assumptions hold, it is not possible for a couple to decide whether to procreate out of love for the particular child that would result.

The mystical security guard

One objection to some solutions to the problem of evil, particularly to sceptical theism, is that if there are such great goods that flow from evils, then we shouldn't prevent evils. But consider the following parable.

I am an air traffic controller and I see two airplanes that will collide unless they are warned. I also see our odd security guard, Jane, standing around and looking at my instruments. Jane is super-smart and very knowledgeable, to the point that I've concluded long ago that she is in fact all-knowing. A number of interactions have driven me to concede that she is morally perfect. Finally, she is armed and muscular so she can take over the air traffic control station on a moment's notice.

Now suppose that I reason as follows:

• If I don't do anything, then either Jane will step in, take over the controls and prevent the crash, or she won't. If she does, all is well. If she doesn't, that'll be because in her wisdom she sees that the crash works out for the better in the long run. So, either way, I don't have good reason to prevent the crash.

This is fallacious as it assumes that Jane is thinking of only one factor, the crash and its consequences. But the mystical security guard, being morally perfect, is also thinking of me. Here are three relevant factors:

• C: the value of the crash
• J: the value of my doing my job
• p: the probability that I will warn the pilots if Jane doesn't step in.

Here, J>0. If Jane foresees that the crash will lead to on balance goods in the long run, then C>0; if common sense is right, then C<0. Based on these three factors, Jane may be calculating as follows:

• Expected value of non-intervention: pJ+(1−p)C
• Expected value of intervention: 0 (no crash and I don't do my job).

Let's suppose that common sense is right and C<0. Will Jane intervene? Not necessarily. If p is sufficiently close to 1, then pJ+(1−p)C>0 even if C is a very large negative number. So I cannot infer that if C<0, or even if C<<0, then Jane will intervene. She might just have a lot of confidence in me.

Suppose now that I don't warn the pilots, and Jane doesn't either, and so there is a crash. Can I conclude that I did the right thing? After all, Jane did the right thing—she is morally perfect—and I did the same thing as Jane, so surely I did the right thing. Not so. For Jane's decision not to intervene may be based on the fact that her intervention would prevent me from doing my job, while my own intervention would do no such thing.

Can I conclude that I was mistaken in thinking Jane to be as smart, as powerful or as good as I thought she was? Not necessarily. We live in a chaotic world. If a butterfly's wings can lead to an earthquake a thousand years down the road, think what an airplane crash could do! And Jane would take that sort of thing into account. One possibility was that Jane saw that it was on balance better for the crash to happen, i.e., C>0. But another possibility is that she saw that C<0, but that it wasn't so negative as to make pJ+(1−p)C come out negative.

Objection: If Jane really is all-knowing, her decision whether to intervene will be based not on probabilities but on certainties. She will know for sure whether I will warn the pilots or not.

Response: This is complicated, but what would be required to circumvent the need for probabilistic reasoning would be not mere knowledge of the future, but knowledge of conditionals of free will that say what I would freely do if she did not intervene. And even an all-knowing being wouldn't know those, because there aren't any true non-trivial such conditionals.

A thought about Molinism

Ordinary subjunctive conditionals "were A to hold, B would hold" tell us about how B depends on A. But if that's what Molinist conditionals did, then they would undercut freedom on incompatibilist grounds. So Molinist conditionals aren't the same as ordinary subjunctive conditionals. But if they aren't the same, then it is difficult to see how they are introduced in a meaningful way. Moreover, the Molinist conditionals are treated as if they were ordinary for the purposes of divine decision theory. So this is a problem.