Explaining Molinist conditionals
April 12, 2013 — 20:55

Author: Alexander Pruss  Category: Molinism  Tags: ,   Comments: 16

I remember David Manley (who I think was a first year grad student at the time) querying Al Plantinga over a meal whether counterfactuals of creaturely freedom (CCFs) could be explained. I think Al didn’t have an answer but thought it was a really good question.

I may finally have an answer to David’s question. I think that the Molinist should answer in the affirmative if and only if non-derivatively free actions have explanations.

Suppose w0 is the actual world. Consider the conditional C→A, where C says that Curley has such-and-such character and is offered a $5000 bribe at t0, and A says that he freely accepts the bribe at t0. Suppose w1 is a sufficiently close-by world where C and A are true. Now let’s put ourselves in w1. So, Curley freely accepts the $5000 bribe. Does this have an explanation? If not, then a fortiori I think we should not have said in w0 that C→A had an explanation. After all, if it has an explanation in w0, it surely doesn’t lose one in w1, just because C holds there. But it would be just too weird that in w1, C→A has an explanation but A does not, especially if, as will at least typically be the case, C has an explanation.

Conversely, suppose that in w1, A has an explanation. What kind of an explanation is that? The most plausible candidate for an explanation of a free action is in terms of non-necessitating reasons and character. Maybe, in w1, what explains A is that Curley is very greedy. But that Curley is very greedy is a part of C. So it seems very reasonable to say at w0 that what explains C→A is that were C to hold, Curley would be very greedy (a necessary truth, since C includes a description of Curley’s character). Now you might say: Yeah, but that he would be greedy in C doesn’t entail or maybe even make likely that he would take the bribe. But the very same point holds in w1: that he is greedy doesn’t entail or maybe even make likely that he takes the bribe–yet, we supposed, it explains it. If we accepted the explanation of the categorical claim in w1, we should accept the corresponding explanation of the conditional claim in w0, if w1 is close enough to w0.

  • Matt

    Is there an explanation for why Curly accepted the bribe in one world but not in another? Supposedly for his action to be free, the past and the laws would have to be identical in each world right up until the point of his decision. If there is no difference between the two worlds before the decision, then to what would we point to to explain why he accepted the bribe in one world while not in another? (This is basically Mele’s problem of cross world luck)

    April 14, 2013 — 6:10
  • Alexander Pruss

    Well, when you ask why something happens at world w1, that’s necessary: “p is true at w1” is always necessary or impossible.
    Maybe you’re just asking the question of contrastive explanation. I think appropriate descriptions of the reasons can provide such, for a certain sense of “contrastive”. Namely we can give an explanation for p that wouldn’t explain not-p.

    April 14, 2013 — 6:23
  • Mike Almeida

    That all sounds right. But the worry is w2, where Curley has the same character, is made the same offer, and does not take the bribe. Even supposing that w2 is not as close to w0 as w1, it will be Curley’s character that explains why he didn’t take the bribe. Could that be right?

    April 14, 2013 — 12:53
  • Mike:
    Are you questioning whether free actions are explained by character, or whether CCFs are explained by counterfactual character?

    April 16, 2013 — 8:49
  • Nathaniel Hagthorpe

    Alexander, i just wanted to clarify a bit, because i am somewhat confused, too. Does your proposal mean, that if Curley’s Character (C) is the same in all examined worlds, an in some he take the bribe, but in others refuses, we stall can cite C as the explanation for his action in ALL the worlds, regardless if he has taken the bribe or refused?

    April 16, 2013 — 9:15
  • Maybe and maybe not.
    One proposal is that the explanation is the same in all the worlds.
    Another proposal is that although the character is the same in all the worlds, the explanation differs, because it exhibits different aspects of the character. Thus, in some worlds the explanation is that Curley is greedy. In others, the explanation is that Curley is afraid of going to jail. In other words, which part of the character explains the action depends on what action was actually chosen.

    April 16, 2013 — 14:30
  • Joshua Rasmussen

    Is your answer substantially different from the one I gave

    April 17, 2013 — 17:41
  • Nope. You scooped me. Are you going to write this up? I have something written up.

    April 17, 2013 — 21:55
  • Keith Brian Johnson

    I’m puzzled. Is Curley’s being greedy supposed to explain his action in the sense of being a sufficient condition for his taking the bribe, or is it only supposed to explain his action in the sense of making it reasonably likely that he will take the bribe?
    In the first case, the truth of C->A in w(1) is explained by Curley’s being greedy, and also at any other possible world at which Curley has the same character C->A will be true and will be explained by Curley’s greed. At any world, like w(0), at which A is false, C (if it includes Curley’s being greedy) will be false, but anyone living in such a world will still explain the truth of C->A at w(1) the same way as before.
    In the second case, not all possible worlds at which C->A is true (and in which C includes Curley’s greediness) will also be ones at which A is true. What will be explained is not C->A but rather C->Pr(A) (using “Pr(A)” to mean “probably, A”). C->A, considered as a material conditional, will be straightforwardly true at w(1), but C->A considered as the ordinary sort of presuppositional conditional we use in daily speech–“imagine C were true; would A follow from it?”–will not have a definite truth-value; instead only C->Pr(A) will be true, and true at all possible worlds, because A will be true at a high percentage of possible worlds at which C is true. Then the likelihood of Curley’s accepting the bribe will be explained by his being greedy in just that sense of explanation.
    So, what is intended here by explanation–sufficiency-explanation or likelihood-explanation? (And, for that matter, by greediness–sufficiency-greediness or disposition-greediness?)

    April 20, 2013 — 22:24
  • The explanation of C→A works just as the explanation of A does. If you think the latter is in terms of sufficient conditions, because you’re a soft determinist, we can say the same about C→A.
    If you think the latter is in terms of probabilities, then the same is true for C→A. Moreover, just as in the world where A and C both happen, Curley’s greed explains A, and not just P(A) being high, so too C→(Curley is greedy) explains C→A and not just C→(P(A) is high). If you find this implausible, that’s probably because you find the explanation of libertarian free action implausible. But if libertarian free action can’t be explained, it’s no surprise that the conditionals of free will can’t be either.

    April 20, 2013 — 23:06
  • Dianelos Georgoudis

    If there is an explanation of A in the actual world then there it is also possible to predict that given C, A will obtain. But if that were so then Curley would have no freedom of choice in the matter.
    I believe that C only explains and predicts the probability of A obtaining. It is the person’s free, sovereign, intrinsically creative will which fixes, within that distribution, the actual choice. Thus if C entails that Curley is greedy, believes it is very unlikely that she will be caught, and so on, then this explains/predicts that A has (say) a 0.8 probability. But if, say, A does not obtain then there is no explanation for Curley’s unlikely choice to not accept the bribe – beyond that this was her choice.
    Modal thinking about possible worlds only helps for visualizing what p(C -> A)=0.8 means. Thus we can imagine observing Curley accepting the bribe in 80% of the possible worlds. But in fact freedom is only exercised in the actual world. Only actual choices are choices. Curley, the person identified by the name “Curley”, chose to not take the bribe – and that’s the only choice she made.
    If for simplicity’s sake we want to visualize choices existing in possible worlds, then we should think like this: The person named “Curley” in all possible worlds in which she does take the bribe is not Curley, but only somebody who goes by the same name. We should not confuse agents with objects, for their ontology is different. The same glass may be empty or full, but the same person may not have made or not have made a particular choice. For what a person is depends on her choices. A possible world in which a person with Hitler’s DNA has chosen to become a pacifist is not the person we mean when we speak of Hitler.

    April 22, 2013 — 10:09
  • When the airplane has a strut that fails and the plane crashes, we don’t say: “The strut failure explains why the plane had, say, a 98% chance of crashing.” We say: “The strut failure explains why the plane crashed.”

    April 23, 2013 — 8:38
  • Dianelos Georgoudis

    Right. What I wanted to point out though is an ontological difference between agents and objects: In general agents are partly free, objects are partly random.
    The previous conditions C (her character, state of affairs, etc) only predict or explain why Curley will accept the bribe with probability 0.8. That she nonetheless did not accept the bribe admits of no other explanation but that she freely chose to do so.
    With the airplane it’s different. Here the previous conditions (which includes a failed strut) predict or explain why the airplane will crash with probability 0.98. If the plane does not crash there will be no other explanation but lucky chance.

    April 23, 2013 — 16:26
  • That he nonetheless did not accept the bribe admits of explanation on the view that character explains action: It is due to his recognition of the wrongness, or due to his fear of punishment, or some combination of these.
    But in any case, this is irrelevant to what I am arguing for. I am arguing for a biconditional that Molinist conditionals are explicable iff ordinary free action is. You deny the latter and the former as far as I can tell. So you agree with the biconditional (at least if we read it as material).

    April 23, 2013 — 21:18
  • Dianelos Georgoudis

    In my understanding Curley’s recognition of the wrongness or his fear of punishment explain that there is a 0.2 probability that he will not accept the bribe. But does not explain why Curley does not in fact accept the bribe.
    Incidentally that understanding is grounded in my own experience of freedom. At the moment I choose I usually feel a force pulling me one way or the other, which force does affect but does not decide my choice. And I find that if I want to change my future behavior in some significant way I must work now to change my character (or else change the state of affairs I’ll have to deal with). I find that this natural understanding fits well both with ethics and with the soul-building theodicy I happen to find convincing. Moreover I find that physical law is such so as to make it possible for that understanding to obtain without violating the physical closure or the naturalistic elegance of the universe. For all these reasons I feel quite confident that this understanding of freedom is the right one.
    Now, even though I know little about molinism, my understanding is that molinist conditionals do not require the existence of explanation. The molinist, I take it, does not think “God knows the reason I freely choose in the actual world, thus knows the reason I would freely choose in any possible world, thus knows counterfactuals of freedom”. Rather, I take it, the molinist thinks “There is no logical contradiction in God knowing counterfactuals of freedom, therefore, given God’s omniscience, God does know all counterfactuals of freedom. Whether our choices can or cannot be explained is irrelevant”. I disagree with molinism, first because I do not accept molinism’s implicit understanding of omniscience as knowing all that can possibly be known, but mainly because I think it is in fact impossible to know counterfactuals of freedom, for freedom only obtains in the actual world. Only actual choices are choices, thus one cannot properly speak of people’s choices in possible worlds, and therefore there is no such thing as counterfactuals of freedom to know something about.
    In general, since possible worlds do not actually exist, we should be careful about how we think about them, and not simply project reason from the actual world to possible worlds. For example, what does it mean to say that “2+2=4” is a necessary truth because it obtains in all possible worlds? I can easily conceive of a world in which counting does not exist, and in which therefore “2+2=4” is meaningless, and thus not even wrong.

    April 24, 2013 — 23:57
  • A paper by Josh Rasmussen and me building on this is forthcoming in Religious Studies.

    July 22, 2013 — 13:05
  • Leave a Reply to Alexander Pruss Cancel reply

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