The purpose of the defense is to show that omniscient, omnipotent and benevolent God is consistent with the existence of evil in creation. The most popular version of the defense is due to Alvin Plantinga, see How does Plantinga's defense of free will align with omniscience. It relies on omnipotence being subject to logical necessity (but no other limitations), the libertarianist concept of free will (which means that one can take more than one possible course of action in a given set of circumstances), and the free will being a greater good than prevention of evil. These can be questioned but I am inclined to accept them, and I still do not see how Plantinga's defense succeeds.

Plantinga's starting point is that to have free will God creates ("actualizes") only creatures and situations, not complete worlds, then the creatures "complete" the creation with their free choices. The key to the defense is the notion of transworld depravity. A creature in a possible world is called transworld depraved if there is a situation in that world such that if this world is created the creature would choose "bad" in this situation. Then, says Plantinga, "it is possible that every free creature is transworld depraved". If that is the case then it is possible that God, albeit omnipotent, can not create a world with free creatures but without evil. Such worlds are possible, but they can not be "actualized". This supposedly shows that omnipotent, omniscient and benevolent God is logically consistent with the existence of evil, because creation of free creatures is paramount.

But does it? "It is possible that every free creature is transworld depraved"? Well, I am not sure that it is. If that were so then all free creatures in all possible worlds would be transworld depraved, or else we did not cover all possible worlds. But then every free creature is necessarily transworld depraved, in other words transworld depravity is a logical consequence of being a free creature? This goes against the libertarian concept of free will, I think. Indeed, it would mean that creatures that would always happen to choose "good" in a possible world, if it were created, can not be considered free. I can only see this following if one assumes that full foreknowledge of choices implies causation, and hence eliminates free will. Otherwise, omniscient God can identify just such a world, omnipotent God can actualize it without breaking logical necessities, and benevolent one will. So how does the defense work?

P.S. Plantinga seems to operate with some meta-modality, where he can have multiple possible universes of all (!) possible worlds, but with "omnipotent" God bound to one of those meta-modal universes (where every possible free creature is transworld depraved). I think this is another way to state Geirsson and Losonsky's objection to Plantinga's defense, but Wikipedia says that Plantinga's argument is still considered credible by many. So what am I missing?

  • 1
    "...in other words transworld depravity is a logical consequence of being a free creature? This goes against the libertarian concept of free will, I think." - Is it that hard to believe that part of the built-in essence of having a free will, of self existing, is that we are selfish and place ourselves ahead of God and others? For that is the essence of moral failing, and the root of evil in this world. – LightCC Mar 13 '17 at 6:29
  • It works by means of agreement regarding (often woefully unexamined) weltanschauung rather than advancing by confirmation of falsifiable and verifiable hypotheses; or reasoning to a valid and sound conclusion; or by constructing status functions obliging desire-independent reason to action in accordance with weltanschauung. So, what ever in the world is it you mean by your use of the term "God"? – Mr. Kennedy Mar 13 '17 at 7:04
  • @LightCC Alas, it is too easy to believe from observing our "wretched world". But that is an empirical surmise, perhaps reinforced by (limited) imagination. What I do not see is how transworld depravity is logically contained in the freedom to choose. And if Plantinga means containment with something like "metaphysical necessity", which is murky enough for us to be uncertain, it will not suffice for his defense. Omnipotent God is not bound by metaphysical constraints, only by logical ones. This is another weakness, I think, in addition to somewhat question-begging definition of free will. – Conifold Mar 18 '17 at 20:19
  • 1
    @Conifold I don't believe from the defense that TD must logically be required by free will - it only needs to be a possibility that it is. The whole defense rests on there being no proof against that possibility, if I understand correctly. I'm curious what you mean by the definition of free will begging the question - is that easy to explain or do you have a link on it? – LightCC Mar 19 '17 at 4:57
  • @LightCC What Plantinga calls "libertarian free will" isn't what most libertarians mean by it (although Molinists call it that). To them it means, roughly, the ability to choose different actions in the same circumstances, which rules out the "counterfactuals of freedom" having truth values. To quote Felt again, "they posit the determinate outcome of a free agent’s acting while excluding the acting! Thus, futuribles are metaphysically inconsistent fictions which cannot form an object of anyone’s knowledge, not even God’s". – Conifold Mar 19 '17 at 18:22

Geirsson and Losonsky, in Plantinga and the Problem of Evil (from your "P.S." statement) fail at least two ways in attempting to address Plantinga's argument:

  1. Counter-Factuals (CFs) are possible actions. But all possible actions are not guaranteed to be actualizable actions.
  2. They miss a piece of Plantinga's argument which they even quote: ".. he will take at least one wrong action"

For those who don't wish to read all of Plantinga's argument and the response from Geirsson and Losonsky, the particular passage I refer to here is regarding an example from Plantinga in which a person with free will, Curley, is presented with an opportunity to accept a bribe, and he accepts. This is considered a moral failing (evil in the world) by Plantinga, but you could substitute any other moral failing or free will action that you believe brings evil into the world.

Point 1: Counter-Factual Actualization

Regarding possible worlds and CF actualization: both your question and Geirsson and Losonsky are assuming that all CFs are actualizable. In other words, there is an assumption in the argument against Plantinga that there is at least one possible world where all CF possibilities could actually occur.

Consider Plantinga's example of Curley and whether or not he will accept the bribe: Plantinga is stating that a weakly actualized world in which Curley is presented with the same opportunity to accept the bribe will result in Curley always accepting the bribe. That is the free will choice that he makes, as a morally free agent. He calculates the moral situation and chooses wrong. Presented with the same circumstances and the same ability to freely choose between good and evil in this particular situation, he will always choose evil and accept the bribe. Free will doesn't mean we can willy-nilly choose a random action from among those that are available. It simply makes us the agent that does the choosing between the right and the wrong way without compulsion to choose one or the other by circumstance or God forcing our hand. It doesn't mean that given the same circumstances in another possible world that we would choose differently! If we did, that would make us a different person - a different moral actor.

While there is a possible CF action where Curley does not accept the bribe, Plantinga's argument relies on the fact that Curley, given the same circumstance, will always accept the bribe - his will/nature does not and won't change unless his free will is destroyed through strong actualization.

Edit: Plantinga's word for this is "Essence". Curley's Essence leads him to make the free will choice of always accepting the bribe in that circumstance. "Essence" is distinguished from "Person" as explained here: Alvin Plantinga's Free Will Defense

Therefore, while there is a possible CF action, there is not a realizable CF action. The only way of obtaining a possible world where Curley does not accept the bribe, according to Plantinga's argument, is for God to strongly actualize a world with the circumstance's changed in a way which removes Curley's freedom of action to the point that God has, in reality, removed Curley's free moral choice, his free will. God must either not allow him to choose by removing the circumstance so there is no choice, or God must directly force him to choose differently.

The long and the short of it is, that according to Plantinga's argument, there is no possible world where Curley both has free moral agency, and does not take the bribe, due to Curley's nature. Now, there are plenty of people other then Curley, who put in the same circumstance, will not take the bribe. So there are possible worlds where we can find a person with free will who will not take the bribe - I will address this in the second point, but for the first point, we are only dealing with Curley, and his free will choice, given Plantinga's example, is that with the circumstances placed before him, his free will choice is to take the bribe. No CFs exist that can be actualized in another possible world for Curley, in particular, where he does not take the bribe, without God changing the circumstances in a way that does violence to his free will choice.

In order to find a possible world where Curley does not take the bribe, you must remove his free moral agency - you must change his nature, his will, so that he no longer desires the same outcome from the situation. Such a possible world does not exist while keeping his free will intact.

The only other option is for God to not create Curley. For than, see point 2.

Point 2: ".. he will take at least one wrong action"

The second point is that Plantinga doesn't require for a free will agent to take the bribe in the given scenario in order for them to be Transworld Depraved (TD). While we know that Curley is TD because he does (and always will) accept that bribe given the same circumstance, but that doesn't mean that all moral agents must also accept this same bribe to be TD. After all, if every moral agent only accepted the bribe, then it would be easy to argue that free will didn't truly exist.

But Plantinga doesn't require that every moral agent accept the bribe, given the same circumstances. Plantinga only requires that any given moral agent must take at least one wrong action to be TD (quoted from the end of his example).

God could choose to actualize a world without Curley, because God knows that Curley will bring evil into the world by accepting the bribe. He can certainly find a possible world with free moral agents other than Curley who will not accept the bribe. The key, though is that if these other actors fail morally in any other choice, then they also will be TD. To pull this off and create a world with both free will and no evil, he needs to find a world with moral actors that won't choose any wrong action. Plantinga's argument only requires that any given free will moral actor fail on one possible moral choice for them to be TD.

So, let's take Mary, who chooses the good and does not accept the bribe. That only shows that it is possible that she is not TD. It does not prove that she is not TD. We must put every other possible situation in front of her, and if she fails morally at one of them, then she has taken (or would take) at least one wrong action, and is TD. Plantinga's argument only requires the possibility that all moral actors are TD. To defeat his argument, it would have to be proven that this is not possible - that there must be a possible world where there are free moral agents will not undertake a single wrong moral action.

I actually believe that Plantinga's defense would stand up even if it can just be shown that a single moral actor would always choose wrong; but it is a stronger argument if it remains a possibility that all moral actors must be TD.

Bonus Point: Time Segmentation? What the...

The remaining thrust of Geirsson and Losonsky's argument against Plantinga's defense relates to actualizing segment's of time. While other's appear to have accepted this at its face and addressed this directly (see Transworld Depravity and Unobtainable Worlds by Richard Otte), I believe time segmentation and trying to build a perfect good world is quite irrelevant. Can we build a world without evil by creating Adam and then ending the world before he sins? Sure, but that destroys the point completely. If the point is to defend the God of the Bible (or an all-powerful, all-knowing, always good God who creates man to have a relationship with God, if you prefer), then attempting this time segmentation by cutting the timeline short does violence to God's purpose with respect to creation - to create free moral beings that are truly capable of love and relationship with him. If that is the ultimate good that required free will moral actors to be created, then it should also be a part of the equation.

I understand that in the end the time-segmentation argument is an attempt to string together all the time segments in which only good occurs into a possible world where there is no evil done - but again, that cannot be achieved because it relies on the assumption that all CFs are actualizable for a given moral agent with free will. This argument explodes because it forces a person to be different then they choose to be in order to have a possible world where they choose differently. It is just a disguised form of strong actualization - trying to do piecemeal what Plantinga's base argument disallows in its definition of strong actualization.

  • I think your analysis is correct, Geirsson-Losonsky miss the point of transworld depravity. The problem is not in the argument but in the premises: "Plantinga's argument relies on the fact that Curley, given the same circumstance, will always accept the bribe". Felt put it well, I think, "Like Molina, Leibniz tries to have it both ways: God is supposed to know the determinate outcome of an agent’s free action, but without the acting of the agent". Plantinga's (Molinist) notion of "free will" defeats the purpose of his defense. – Conifold Mar 14 '17 at 21:45
  • Thanks Conifold. I read on a little bit after answering this, and found that Plantinga distinguishes between a "person" and an "essence" with respect to whether Curley would always accept the bribe. What I called Curley's "nature" here is what Plantinga would call his "essence" - I think. – LightCC Mar 17 '17 at 2:51

But then every free creature is necessarily transworld depraved, in other words transworld depravity is a logical consequence of being a free creature. This doesn't work even under the libertarianist concept of free will. Indeed, it would mean that creatures in a possible world that would always happen to choose "good" if it were created, can not be considered free. I can only see this following if one assumes that full foreknowledge of choices implies causation, and hence eliminates free will, which is not in the premises.

It is not clear what precisely you mean by "would always happen"—why? There needs to be a mechanism for avoiding evil choices, no? That mechanism would ban all free creatures from being actualized in the first place, because such creatures would be transworld depraved. And so, precisely the world which is actualized would a priori contain no free creatures.

You seem to be suggesting some world be actualized with creatures which are guaranteed to not make any evil choices, and then complain that, hey!, how can we say that those creatures aren't free creatures? But given transworld depravity, you can create only these two kinds of worlds:

     (1) an all-good-choice world with no free creatures
     (2) free creatures who inevitably make evil choices

And so, your world-picking procedure has the freedom property smuggled into it, since transworld depravity logically combines freedom and your criterion for picking the world.

  • No, there need not be such a mechanism, not logically. There are possible worlds with identical parts caused by God, in some of which creatures chose some bad, and in others they always chose good contingently (as Plantinga admits). Omnipotent God is not subject to temporal restrictions, he doesn't cause his part and then "waits" for creatures to complete theirs, so there is no need for a "guarantee". He simply actualizes the latter, and knows that it is the latter because he is omniscient. – Conifold Oct 2 '14 at 21:10
  • We do not need to know how omniscience works, the point is that there is nothing in the definition of foreknowledge that logically necessitates causation. Sure, it usually works that way as far as we know, but that's empirical. But nothing short of logical necessity suffices for Plantinga's defense because logical necessity is all he has to limit omnipotence. If God's foreknowledge requires causation then Plantinga's God is already not omnipotent, and the transworld depravity detour is just a distraction. – Conifold Oct 2 '14 at 21:10
  • @Conifold: Nothing I mentioned requires time (logical sequence/predication ⇏ temporal sequence), and nothing I mentioned has to do with omniscience. Last night, I did skim my copy of God, Freedom, and Evil, but really drawing out the answer you want in a more rigorous fashion than I attempted... I'm not sure I have the energy for it. :-| Suffice it to say that you would be finding an objection to Plantinga that no other academic scholars had—possible, but not likely. – labreuer Oct 3 '14 at 0:34
  • 1
    Well, thanks for trying (+1). But I don't think I am inventing something new. Geirsson and Losonsky are academics and they construct in the paper I linked an explicit counterexample where no creature is transworld depraved using God's knowledge of counterfactuals of freedom (which Plantinga himself uses). I was looking for a response to that but couldn't find. I only replaced knowledge of counterfactuals of freedom with more generic foreknowledge without causation, which suffices for their reasoning but not for expicit construction. – Conifold Oct 3 '14 at 15:56
  • @Conifold: Huh, somehow I missed that. Perhaps I will take a deeper look at God, Freedom, and Evil, match it against that paper, and see what arises. Perhaps I also need to request The Nature of Necessity from my library, to see the more rigorous form and perhaps supporting thought on how modal logic works. This has been on my list to do; you've just upped the priority. :-) It'll probably be a few days, though. – labreuer Oct 3 '14 at 16:36

Alvin Plantinga describes transworld depravity as follows: (page 48)

A person P suffers from transworJd depravity if and only if the following holds: for every world W such that P is significantly free in W and P does only what is right in W, there is an action A and a maximal world segment S' such that

  1. S' includes A's being morally significant for P
  2. S' includes P's being free with respect to A
  3. S' is included in W and includes neither P's performing A nor P's refraining from performing A


  1. If S' were actual, P would go wrong with respect to A.

To see how this works consider the example Plantinga provides of Curley who is currently in the habit of taking bribes. Go back in time before Curley made his first morally significant free act that was wrong, call that act A.

At that point there exists a possible world W in which Curley always does good. Let S' be the maximal world segment for W up to A allowing Curley freedom as in (2) of the definition. Once Curley makes his choice to go wrong, as he would since we know his history, it is no longer possible for God to actualize world W.

This world W becomes an example of a logically possible world that nonetheless God cannot actualize, because of Curley's free choice, any more than a coherent definition of God's omnipotence allows him to make a square circle.

Before considering Heimir Geirsson and Michael Losonsky's objection to this consider what Plantinga is trying to do. Plantinga's Free World Defense is not a theodicy. Although it is based on Augustine's Free Will Theodicy (pages 26-7), unlike Augustine he is not claiming his description of God is true, only that it is possible. His goal is to create a model that shows the consistency of God's omnipotence, omniscience and benevolence with an actual world that contains evil (page 28).

For Geirsson and Losonsky to succeed in their argument they must show that Plantinga's model is inconsistent. They try to do so by claiming that God, in spite of Plantinga's claim, still has the power to actualize Curley's possible world W where he always did good without destroying Curley's free will. (page 5)

In fact, clearly it is possible that God has this power. God surveys all the possible worlds, including how they are completed by free beings, and has a choice between worlds that are completed by free beings in such a way that there is no moral evil and worlds that are completed by free being in such a way that there is moral evil. A perfect God, namely one that is omniscient, omnipotent and wholly good, will weakly actualize the world that is completed with no moral evil.

For Plantinga, the maximal world segment S' for Curley contains at least two paths, a wrong one and a right one. Curley, faced with A, chose wrong. For Geirsson and Losonsky S' can be chosen so Curley only chooses right. They appear to be accepting a determinism that would destroy Curley's freedom and a compatibilism that would justify saying that in spite of this determinism Curley was still free.

Plantinga rejects compatibilism, but all that is necessary to note is that compatibilism is not necessarily true. That implies that Geirsson and Losonsky's objection does not make Plantinga's model inconsistent. Hence Plantinga's model, whether he or anyone else accepts this model's view of God, allows for God's omnipotence, omniscience and benevolence to be consistent with the existence of evil.

Geirsson, H., & Losonsky, M. (2006, February). Plantinga and the Problem of Evil. In The Proceedings of the Twenty-First World Congress of Philosophy (Vol. 8, pp. 109-113). https://geirsson.public.iastate.edu/pdf/Plantinga%20and%20the%20Problem%20of%20Evil,%20World%20Congress.pdf

Plantinga, A. (1977). God, freedom, and evil. Wm. B. Eerdmans Publishing.

  • Frank, Plantinga's model is not inconsistent, but its "free will" is of the same fake variety as the compatibilist one. His notion is Molinist, and although he calls that "libertarian" it is not what most libertarians have in mind. For example, Plantinga's God knows in advance what Curley will "freely choose" before he does the choosing (Molina talked about God's "superintuition" into free will), the whole idea of transworld depravity hangs on this. With libertarian free will Plantinga's defense is redundant, there is nothing for God to know as to the outcomes of free acts before they happen. – Conifold May 28 at 19:25
  • @Conifold I originally had problems with Plantinga's argument, but as I see it now it is merely a possible model demonstrating that evil is consistent with God's omniscience, omnipotence and benevolence. This model need not be anyone's actual view of God. It particular it need not be Molinist nor libertarian. For this to work, to claim that God does not know what Curley will choose is enough for this model. Does that preserve God's omniscience? I think it does: God knows everything there is to know. What Curley will chose is not something there is to know. – Frank Hubeny May 28 at 19:43
  • It does need to be Molinist, Plantinga himself is explicit about it. And it is rather obvious, without middle knowledge (a.k.a. "counterfactuals of freedom") the transworld depravity does not make sense. I agree that it demonstrates how God's omniscience, omnipotence and benevolence is consistent (as far as we know) with Molinist "free will", but it is not the kind of free will that many subscribe to. The libertarian one does not need Plantinga's defense, but the omnipotence and omniscience it is compatible with are much weaker, God has little control/knowledge of outcomes. – Conifold May 28 at 20:13

Informally, the idea that I believe Plantinga's argument is trying to capture formally, is that free will isn't worth much without consequential choices.

If all the choices are between the good and the good, are they really choices? It's like your mother saying "do you want to go to bed now, or in five minutes?" It has the form of a choice, but your agency is illusionary.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.