By David Misialowski (2006)
It is Jan. 1, 2007. All night, my inconsiderate neighbor has been throwing a raucous New Year's Eve Party. What is especially galling is that he is a fan of 70s disco, and has been blasting the Bee Gees at top volume on his CD player, the din reverberating through the thin wall that separates his apartment from my own.
I decide that his liking the Bee Gees reflects poorly on his worldview, and calls into question the very existence of his soul. Moreover, I decide that if I am forced to listen yet again to "Night Fever", which he has already played 50 times, then I am going to strangle him. Sure enough, he plays "Night Fever" for the 51st time. I wait until the morning when all the guests have departed, break into his apartment and strangle him dead.
I am later arrested. It turns out that my neighbor had installed a videocamera in the wall of his apartment to record the party, and it made a videotape of me killing my neighbor. When I am brought to trial, my lawyer mounts an insanity defense. The repetitive throb of Night Fever, he argues, robbed me of my free will. I did not freely commit the murder, but rather was driven over the edge by the repetitive bombardment of lyrics like these:
Craftily, my lawyer plays "Night Fever" 51 times in a row for the jury. I am acquitted in 30 minutes. After the verdict is read, the members of the jury line up to shake my hand.
The insanity defense is often mounted in murder cases. It raises the general question of free will. Are all our acts free, or are they determined in some sense?
It is probably true that most of are acts are not entirely free, but let us, for the sake of discussion, stipulate that in some important sense, we do have free will: Options are available to us, and we can choose among them, exercising independent judgment.
Now let's return to my trial. Suppose, instead of mounting an insanity defense, my lawyer had argued like this: The reason I committed the murder is that I was being watched by a videocamera; and this camera entailed my action. In a like manner, my lawyer argues, the sun rises every morning because people watch it rise.
What's more, my lawyer continues, I committed the murder in the past. Hence, it is now an accidentally necessary fact about the past that I committed a murder! If it is now necessary that I killed my neighbor, it is not possible for me to have done otherwise. Hence, because of the presence of the videocamera, and because of the accidental necessity of the past, I had no choice - no free will - in killing my neighbor.
Do you think the jury would have bought an argument like that?
I don't think so, either.
I will now attempt to show why the argument to theological fatalism fails. What is that argument? It is that the existence of essentially omniscient agent (whom we shall take to be God) and human free will are incompatible. This argument is the topic of Robert P. Taylors's dissertation, and this essay, and the succeeding one, constitute a response to it.
Let's again consider my New Year's Day murder. Most people would see, intuitively, how bizarre it would be to argue that I committed the murder because I was being watched (by a videocamera) and that this watching entailed my act. Most people would also understand that it would be strange to argue that my act was unfree because it is now an accidentally necessary fact of the past.
But suppose my lawyer offered the following argument: God exists. God is essentially omniscient: he has infallible true beliefs about all contingent future propositions. It follows from this that God knew, before Jan. 1, 2007, that on Jan. 1, 2007, I would kill my neighbor. In fact, God knew that I would kill my neighbor on Jan. 1, 2007, before I was born. In fact, God has known that I would kill my neighbor on Jan. 1, 2007, for all eternity.
Question: if God knew for all eternity that I would kill my neighbor on Jan. 1, 2007, then how could I have done otherwise?
The question of theological fatalism seems academic to some. Many people say the problem is a non-starter because an essentially omniscient agent does not exist (there is no God). Others say that we lack free will even if there is no God - that determinism is true, for example.
However, such blasé stances assume atheism or determinism, and we are going to assume, for the sake of argument, that God exists and that humans are not governed by determinism. Although the discussion might be academic to the atheist, for the theist or the agnostic theist it has profound importance. That is because if I must kill my neighbor on Jan. 1, 2007, because God's infallible true beliefs about all future contingent events entail this act, then it seems to follow not only that I lack free will, but also that I lack moral responsibility. For how can I be morally responsible for an act that I cannot avoid doing? For the theist who believes in God's infallible foreknowledge but also believes in human free will and moral responsibility, defeating the argument to theological fatalism becomes an urgent necessity.
Most people would reject, with hardly a moment's thought, the idea that the presence of the videocamera entailed my committing murder, and most would also reject the argument that because that murder that I committed is (supposedly) an "accidentally necessary" fact of the past, I had no choice but to commit it. Yet somehow, foreknowledge of an act seems to be in a different category, and one has the uncomfortable sensation that if an act is infallibly foreknown, then it is entailed.
I argue that this is false. In fact, I hope to show that foreknowledge - even infallible foreknowledge - has no more bearing on the truth value of a proposition then does present knowledge of it (the videocamera example) or the memory of it.
Early in his dissertation, Robert writes: "An obvious escape for the theist is to claim that God does not know the truth-values of contingent future propositions, since such propositions have no truth-values." But, again, we just want to assume, for the sake of argument, the God does foreknow, infallibly, the future; otherwise there is no problem to consider! So we shall stipulate that contingent future propositions have truth values, and that God's prior knowledge of these values is infallible.
In his dissertation, Robert explores and discards several possible answers to the problem, and eventually gives a solution of his own. I would say he has made an admirable study of the problem. But I would like to suggest that all one needs to do is find a single solution early on (if possible), and if one is able to do that, then all other considerations evaporate. Of course, the solution could be wrong, but then it would be incumbent to demonstrate precisely why it is wrong.
The solution I have mind was expressed, in fiction, in my story The Omniscient Book. However, fiction is not an ideal vehicle for working out philosophy problems in detail, though it can be a vehicle for philosophy in general.
In the introduction to his dissertation, Robert writes:
This, in my view, captures the essence of the problem. The issue is that somewhere, entailment resides; the question is, where, exactly? There must be some kind of entailment somewhere, because God has infallible true beliefs about all future contingent events.
As I hope to show, and tried to demonstrate in my story, the argument to theological fatalism grasps the wrong end of the stick. It assumes that God's infallible true beliefs about all future contingent events entail those events, thus dooming free will. However, the events are contingent; how can contingent events be entailed? The right question, in my view, is not what makes x do p at some time t, but rather, what makes God's belief that x will do p at t true?
What makes God's belief in this matter true, I argue, is the fact that x freely does p at t; and if that is right, the problem of theological fatalism is dissolved. Under this account, it would turn out that God's infallible foreknowledge of an event - killing my noisy neighbor on Jan. 1, 2007, for instance - is just a special case, but not fundamentally different from, God having an infallible memory of this act, after it has happened. No one, I think, would argue that God's infallible memory would make a past event true, and I would hope to show that for the same reason, God's infallible foreknowledge does not make events true, either. I would argue instead that propositions like "I kill my neighbor on Jan. 1, 2007" bear truth values, and that these values are timelessly true, because propositions are abstract entities and hence do not exist in space and time. (This accords with the views expressed by Prof. Norman Swartz, who briefly joined our discussion in the thread, What Is Free Will?
However, to say a proposition like my homicide on New Year's Day is timelessly true, does not mean it is fated or predestined to be true. If that were the case, then we would have no free will even in the absence of an all-knowing God. Rather, my actions, while timelessly true, were chosen by me from a range of options; and had I chosen other options, then those acts would have been timelessly true instead. In the case of an omniscient God, He just sees, by this account, the truth of the choices that we make. But His foreknowledge, knowledge, or memory of this actions (or some other, atemporal kind of knowledge) does not force, necessitate or make propositions true, any more than to watch the sun rise, makes the sun rise; any more than a watching videocamera makes me kill my neighbor.
In the introduction to his thesis, Robert presents the formal argument this way:
- P1. Necessarily, if God knows I will do p, then I will do p.
- P2. God knows I will do p
- C. Necessarily, I will do p.
The conclusion does not follow from the premises. Why is that? Imparting necessity to the conclusion constitutes a fallacy of modal logic. As Robert writes: "...the assertion that God knows I will do p only allows one to conclude that God knows that I will do p contingently. Since I will do p contingently, it is not the case that I must do p, and since it is not the case that I must do p, my free will is not endangered."
It's worth looking at this a little more closely. For those who wish an extensive discussion of the modal fallacy, I direct them to Prof. Swartz's Lecture Notes on Free Will and Determinism, in which he dissects not just theological (or epistemic) determinism but the two other standard versions, causal and logical determinism, and concludes that they all fail for the same reason: they involve a fallacy in the use of modal logic.
To take one example due to Prof. Swartz, suppose we reason like this:
- If Paul has two sons and a daughter, then he has to have at least two children.
- Paul has two sons and a daughter.
- Paul has to have at least two children.
Compare the above argument to the one previously given for God's omniscience entailing a conclusion of, necessarily, I will do p. They both bear the same logical structure, and so if one is both valid and sound then so is the other. But in both cases, the conclusion does not follow from the premises. It might be easier to see why this is so in the case of Paul and his children.
Again, for a detailed discussion, I recommend Prof. Swartz's paper, but for now just notice that in the case of Paul, although he does have at least two children (in fact), there is no logical entailment that he must have at least two children. That is, it is not necessary ("has to" is false) that he have at least two children. Like anyone else, Paul could choose to have one child, or no children.
The fallacy of modal logic lies in drawing the false conclusion that Paul must have at least two children, from the premise that states that in fact he has two sons and a daughter (at least two children) The fallacy of modal logic (modal logic involving the study of modes of being) resides in imparting the mode of necessity to the conclusion, when in fact the conclusion involves contingency or possibility. In modal logic, a proposition can be necessary, or it can be contingent, but it cannot be both.
Consequently, the initial argument to theological fatalism must be rewritten as follows:
- P1. Necessarily, if God knows I will do p, then I will do p.
- P2. God knows I will do p
- C. I will do p.
The revised version of the argument is identical to the initial version of it, with one crucial exception: The word necessarily has been omitted from the conclusion. Now, it just happens to be the case that I will do p - that I will, for instance, murder my noisy neighbor on Jan. 1, 2007. But it's not necessary that I do this, even if God knew, before I was born, in fact for all eternity, that I would commit this act on Jan. 1, 2007.
But, if it wasn't necessary that I commit this act, then how could God have known for all eternity that I would commit it? The answer, as indicated, is that acts, or propositions, bear truth values (some dispute this). If that is the case, then God's foreknowledge of acts is irrelevant; or, it is no more relevant, to the act itself, then His memory of the act would be. If p stands for "Kill noisy neighbor on Jan. 1, 2007", let q stand for "Not-Kill-Neighbor". Under the modal account, if I declined to commit p, then the above argument can be recast as:
- P1. Necessarily, if God knows I will do q, then I will do q.
- P2. God knows I will do q
- C. I will do q.
We see that God's foreknowledge is still accurate: it must be accurate, because God has infallible true beliefs about all future contingent events. However, since what I do at any given time is in fact contingent (not necessary), then God's beliefs about future acts will just happen always to correspond with what I freely choose to do, given God's infallibility concerning His beliefs. There is a necessity entailment in this argument, but it is not in the conclusion. The necessity always resides in the conjoint state of affairs, as indicated in Premise 1. Necessarily (If God knows I will do p, then I will do p) is logical and correct. It denotes (given God's infallibility) the necessity of the compound relationship of God knowing what I will do, and I then doing what he foreknows. But what I actually do is entirely up to me, assuming that I do not lack free will for some other reason, a proposition that we are assuming to be true for the sake of argument.
At this point I am going to argue (though I will certainly consider Robert's other arguments, as well as his own proposed solution) that the whole problem is dissolved before it ever gets off the ground. Robert would not appear to agree, but if I can show that the argument to theological fatalism rests on a modal fallacy straight through, then I have no need to consider any further argument to reconcile foreknowledge and free will because the problem has been dissolved as a straightforward consequence of logical analysis.
For example, suppose we revise the original account to read like this:
- P1: Necessarily (If God knows p, then p)
- P2: Necessarily (God knows p).
- C: Therefore, Necessarily (p)
We have now imported necessity into P2, from which it seems to follow that the conclusion is necessary, and my free will is negated. But this is not, in fact, the case.
In this post in the What is Free Will? discussion here at the library, Prof. Swartz dissects the flaw in the above revised argument. He writes:
I think that the author of this argument is trying, in the second premise (P2), to capture the claim that God knows everything. (Or, to be more exact, that God knows everything that is TRUE.) But P2 does not say that God knows everything that is true.
Let's use the possible-worlds idiom to explicate necessity. In that idiom "necessarily" is explicated as "in all possible worlds".
The first premise, P1, is fine as written. It says that in all possible worlds, if God knows that p [is true], then p [is true].
But in those cases where p is contingent (i.e. true in some possible worlds and false in other possible worlds), P2 is false. For it says that in all possible worlds God knows that p is true. The falsity is easily laid bare by letting "p" stand for the proposition "In 2005, the entire ice cap in Greenland melts". The proposition, p, is actually false (i.e. is false in this, the actual, world); it is false (in this world) that God knows it to be true; and thus it is false that in all possible worlds gKp.
What, then, is the correct symbolization for the theological claim that God knows everything [that is true]? Simply, it is the inverse of P1, i.e. the corrected second premise should read:
- P2': Nec (if p, then gKp)
From P1 and P2', taken together, the conclusion, C, does not logically follow; i.e. such an argument, while having true premises, is invalid.
Now let's look at the transfer of necessity principle that Robert discusses in the introduction to his thesis.
Transfer of Necessity Principle.
At the conclusion of his introduction, Robert discusses the transfer of necessity principle, or TNP, which principle, he says, provides a rejoinder, or a stronger argument, to defeat the modal fallacy: i.e., it again raises the possibility that God's foreknowledge precludes human free will.
I do not think this to be the case. I think the TNP does not succeed in providing an argument for theological fatalism, either in the forms that Robert lays out or in the way it was considered by Zagzebski in her discussion of Divine Foreknowledge and Human Free Will at the Stanford Encyclopedia of Philosophy.
What is this supposed principle? It arises from the principle of the fixity of the past, which itself is easy to understand: the old saw, "There is no use crying over spilled milk" captures the essence of it. After the milk is spilled, there is nothing to be done about it except clean it up. You can't unspill it, because you can't change the past. The past is fixed.
In the case of God's foreknowledge, we can say that at T1, God has a certain infallibly true belief about some future proposition. Let's continue with my example of the murder that I will commit on Jan. 1, 2007, which I will call p. Now, at t2, it is the case that at t1, God believed that I would commit the murder (p) at, let us call it, t3. Due to the principle of the fixity of the past, we would say that at t2, it is now necessary that at t1, God believed p at t3. The argument seems to go that since God is infallible, and since it is now necessary that at t2 that he believed p at t1, then necessarily p at t3. The supposed necessity of God's belief is somehow transferred to the act at t3, making p at t3 necessary.
Another way to understand the situation is to invoke, as Robert does, the idea of accidental necessity. A necessary proposition must be true; a contingent one may possibly be true or false. So before an event takes place, it might have no truth value; but after it does it acquires such a value, and because the event is now in the past, it is accidentally necessary that it has just the value that it does, because the past is fixed (or so the argument goes). We would have to use this account of accidental necessity to replace modal contingency. As we have seen, the conclusion in the initial argument as presented was modally contingent, and it was a flaw of modal logic to assign necessity to it. The TNP, evidently, is offered as a way to get around this objection and reinstall necessity to the conclusion.
Robert offers two versions of TNP. They are:
- P1. f
- P2. (f -> y)
- C. y
- P1. f
- P2. (f <-> y)
- C. y
The symbol  denotes the condition of necessity; the symbol --> denotes "then", as in, if x happens, then y will take place; and the symbol < - > is the bi-conditional operator, which tells us that antecedent and consequent may each depend on the other. How does this translate into foreknowledge/free will talk? Under TNP 1, we might say as follows;
- Premise 1: Necessarily, God knows that I will do p.
- Premise 2: Necessarily, if God knows that I will do p, then I will do p.
- Conclusion: Necessarily, I will do p.
TNP2, as Robert explains it, is the same, except that Premise 2, which uses the bi-conditional operator, tells us that if I do p, God knows I do p and if God knows I do p, then I do p. In both cases, the reason why P1 is necessary, under TNP, is because the fixity of the past.
At this point I am going to skip over the details of the various arguments that Robert offers concerning TNP in his introduction, because I think all versions of TNP are wrong. So I'll cut to the chase.
At the conclusion of his Introduction, Robert writes that the full argument goes like this:
- P1. Necessarily, if God knows at t1 that I will do p at t3, then I will do p at t3
- P2. Necessarily, if I will do p at t3, then God knows at t1 that I will do p at t3
- P3. God knows, at t1, that I will do p at t3
- P4. It is accidentally necessary at t2 that God knows at t1 that I will do p at t3
- C. It is accidentally necessary at t2 that I will do p at t3
- P5. If my act at t3 is accidentally necessary, then it is not free
- C2. I do not freely do p at t3
Robert goes on:
The above argument looks pretty strong. It seems to circumvent the modal fallacy already discussed. But I argue that it does not. Premise 4 is false. P4 is false because it is incoherent. It is incoherent because it smuggles the modal fallacy into the picture through the back door. Let me state plainly: There is nothing necessary, accidental or otherwise, about the past, except those propositions that are in fact necessary, and such propositions are timelessly necessary.
Under modal logic, necessary propositions are said to be necessary at all possible worlds, while contingent propositions may occur at some worlds but fail to occur at others. If a proposition is necessary, it is necessary at all worlds and at all times. It is a timelessly necessary proposition that one cannot square a circle. It is timelessly necessary that 2 + 2 = 4. It is timelessly necessary that certain numbers are prime. And so on. At all possible worlds, such propositions are true.
Now let us look at a certain fact about history: The south lost the Civil War.
For us to accept the argument given above, involving the fixity of the past and the TNP allegedly defeating free will, we would have to say, by parity of reasoning, that it is now an accidentally necessary fact about the past that the south lost the Civil War. I hold that such a claim is incoherent, a misuse of language. (Think about the example of my trial, and how strange it would sound for my lawyer to argue that my committing murder was accidentally necessary, because of the fixity of the past.) For one thing, the Civil War isn't being fought now, it was fought in the 1860s. What we should say instead, is this: It is a contingent fact of history that the South lost the Civil War. Explicating the Civil War in possible worlds (modal) terminology, we just see that logically, the South could have won the war. This means that there are possible worlds at which the South prevailed. This contingency is timelessly true, and we see how it differs from a necessary proposition, such as 2 + 2= 4, which is true at all possible worlds and which is also timelessly true.
Consequently, I hold that we have to amend P4 to read as follows:
- P4. At t2 God knows at t1 that I will do p at t3
All talk of necessity is dropped! And it must be dropped, for the same reason that we dropped talk of necessity in the conclusion of the original argument to theological fatalism: such talk of necessity is a modal fallacy.
With P4 amended, we understand that God's knowledge at t1 that I will do p at t3 is a contingent fact of history, like the South losing the Civil War. And what is it contingent on? On what I freely (i.e., contingently) do at t3!
There is, however, a certain necessity in all these arguments. As I have already indicated, it resides (and only resides) in the conjoint state of affairs of God's foreknowledge corresponding 1:1 with what I in fact do. This must be the case, because an omniscient agent with infallible true beliefs about all future, contingent, truth-apt propositions cannot both know, and fail to know, the truth of these propositions. That would violate the Law of Noncontradiction.
Let me quote myself, from a post I made in the What is Free Will? thread:
The above is by way of introducing the atemporal solution, which Robert considers in the next part of his dissertation, and to which I will turn in the next essay. Strictly, though, in my judgment, there is no need to invoke the atemporal solution, or any other solution, to the foreknowledge/free will problem, because the problem has been solved, and no theist need fear the argument, heard so often from atheists intent on discrediting religious belief, that an omniscient God cancels human free will and moral responsibility. God's omniscience does neither, and the argument to theological fatalism is, I believe, a dead duck.