An inquiry into the logical structure of time travel, foreknowledge, future contingents and free will
Fatalism and Free Will
Determinism and fatalism are not quite the same, and there are different supposed versions of each. One version of fatalism is often expressed by the saying, “What will be, will be.”
From this it follows that what is, is; and what was, was.
But these statements cannot capture the essence of fatalism. For what is the opposite of that statement, “What will be, will be”?
The opposite is, “What will be, will not be,” which describes a logically contradictory state of affairs.
Regardless of whether fatalism holds or free will holds, it is trivially true that “what will be, will be.” There is only one actual history. The counterfactual histories that could have been but weren’t are virtually infinite in number, as compared with the one way that history actually is.
Necessarily, everything that is actual is possible. But obviously, not every possibility becomes an actuality.
The fatalist who denies free will really wants to say: “What will be, will have to be; from which it follows that “what is, has to be,” and, “what was, had to have been.”
That is, the fatalist must believe that there are no contingent facts about the world at all. Contingency would have to be an illusion. Everything that has ever occurred, or ever will occur, would have to be either actually necessary, or necessarily impossible.
But why should we believe that contingency is an illusion?
One worry, dating back to Aristotle, is the problem of future contingents.
If a statement today about the future is truth-apt – if it has a truth value now – then intuition suggests that the event that the proposition describes has to happen. If that is the case, then fatalism holds, and free will is impossible.
The problem of future contingents parallels the problem of divine omniscience. The worry is that if God knows infallibly ahead of time what you will do, then you must do it.
A third parallel is the problem of backward time travel.
Given that, even before you step into a time machine, you were already in the past as judged by the external time, then it logically follows that you will do in the past, just what you in fact already did in the past.
It certainly sounds as if a time traveler has no free will, and that a form of predestination holds.
The problems and paradoxes of future contingents and free will, of foreknowledge and free will, and of time travel and free will, would therefore all seem to have an identical logical structure.
It follows that if these problems and paradoxes with respect to free will can be adequately resolved for one of scenarios described above, then they can be adequately resolved for all three of them.
Imagine a time telescope. It looks into the past and future. Further stipulate that the scope is absolutely accurate, and never shows something other than what happened, or what will happen.
A cartoon on the Internet depicts a man named Bob (actually a flower named Bob, but that’s immaterial) peering into a time telescope, and watching his future self say the word “pancakes.”
The Bob cartoon seems to be the strongest possible formulation of the alleged threat to free will posed by future contingents, by God’s foreknowledge, and by the seeming inevitability of a time traveler’s actions in the past.
For not only is it true in advance of Bob’s saying “pancakes” that he will say “pancakes,” Bob actually knows that it is true, for he is peering ahead in time with his time telescope, which hereafter we will call “timescope” for brevity’s sake. We shall imagine that the timescope, like God, is infallible; it offers a true picture of the future.
We shall proceed on the assumption that Bob has free will, prior to the introduction of the timescope. He may be unfree for some other reason, but that does not concern us, for if he is, then the entire exercise of evaluating his freedom vis a vis the scope is moot. So this is a “for the sake of argument” approach. We shall define free will as: the ability to choose among genuinely available alternative options.
By assumption, too, it is plain that future contingents in the scenario that we are considering must be truth apt prior to the events that they describe, because otherwise the timescope could not function. Many argue that propositions about future events can’t be truth apt, but that’s irrelevant to this argument. We will assume that they are, to see what the implications of this assumption are.
The idea is to see whether truth-apt future contingents and free will are in conflict, logical or otherwise. If they are not, then identical logical “problems” associated with foreknowledge and time travel will also evaporate.
At time 1 — t1 — the t1 version of Bob peers into the future, and sees at t2 the t2 version, his future self, utter, “pancakes.”
Bob worries that if it is true now that later on he says “pancakes,” it follows that he must say pancakes, which in turn means that fatalism is true.
To confute this fatalist idea in the cartoon under consideration, Bob intends to wait until the appointed time, when he says “pancakes,” and say “ass” instead. Saying “ass,” he imagines, will validate his free will.
But there is a problem. The timescope, recall, is infallible, like God. This means that there is no possible world at which what the timescope foresees, and what Bob actually does, fail to match.
Given this curious state of affairs, our intuition leads us to believe that fatalism holds and that Bob lacks free will.
To simplify, we can narrow Bob’s choices down to two: Either he will say “pancakes,” or he will not say “pancakes.”
If he says anything other than the word “pancakes” – if he says the word “ass,” for instance, or any other conceivable word, or if he just remains silent – then those choices fall under the rubric of “not saying pancakes.”
So there are two choices. Either he will say pancakes, or not say pancakes. Bob will either SP (say pancakes) or NSP (not say pancakes).
We readily see that his choices are mutually exclusive and jointly exhaustive, which means that:
There is a possible world at which Bob will SP.
There is a possible world at which Bob will NSP.
There is no possible world at which Bob will both SP and NSP, because to do so would be logically contradictory.
And there is no possible world at which Bob fails to instantiate either option. One or the other must obtain.
However, there are other possible worlds in this scenario.
There is a possible world at which (Timescope shows Bob SP, and Bob SP).
There is a possible world at which (Timescope shows Bob NSP, and Bob NSP)
Given that the timescope is infallible, the two possible worlds above can be classified, under modal logic, as:
Necessarily, (if timescope shows Bob SP, then Bob will SP)
Necessarily, (if timescope shows Bob NSP, then Bob will NSP)
From the above, it now logically follows that there is no possible world at which:
(Timescope shows Bob will SP and Bob will NSP)
And there is no possible world at which:
(Timescope shows Bob will NSP and Bob will SP)
1. Necessarily, (if timescope shows Bob SP, then Bob will SP)
It becomes tempting to rewrite the statement as:
2. If timescope shows Bob SP, then Bob MUST SP.
But 2. does not logically follow from 1. The statements are not equivalent.
Statement 1 is logically correct, but Statement 2 commits the modal fallacy. It illicitly imparts necessity to a contingent event. It confuses necessity with truth.
It is, was, and forever shall be the case that whether Bob SP or NSP, either outcome is contingent.
What is necessary, however, on the assumption of the infallibility of the timescope, is that what the timescope shows, and what Bob does, must match.
By logical parity of reasoning:
What an infallible God foreknows, and what Bob does, must match, and NOT, “what God foreknows, Bob must do.
And by further logical parity of reasoning:
What a time traveler did in the past (as judged by the external time), and what he will do as he time travels to that past location (in his own personal future), must match and NOT, “what the time traveler will do, he must do.”
And by further logical parity of reasoning:
“If it true today that x will do A tomorrow, then x will (not MUST!) do A tomorrow;” i.e., the truth value of a future contingent now, and the event that the proposition describes, must match. It seems we are now making an observation that is trivially true, which is:
“True statements are true.”
For how can true statements be false?
Abstracting out our Gods, our timescopes and our time travelers, this entire alleged “problem” of free will boils down to this: true statements are true.
It seems weird to infer that because true statements are true, then free will is somehow compromised. So I consider that under the scenarios mooted, it is not.
But what makes true statements true?
At t2, Bob says “pancakes.”
What makes him say that?
Well, on the assumption of free will, he says it because he chooses to say it.
But now we have the apparent contradiction that Bob, in an effort to vindicate his free will, is going to try to NSP instead of SP as predicted by the infallible timescope.
More abstractly, we can ask: If a proposition today is true about an event tomorrow, what makes the proposition true?
It seems there is only one coherent answer. The proposition is made true by the event it describes. Propositions take their truth from the way the world is, and not the other way around.
If that is right, then what makes it true that Bob says “pancakes” is that Bob freely says “pancakes.”
The Timescope doesn’t make him say “pancakes.” How could it?
Nor, obviously, does a true proposition make an event happen. This gets the flow of truthmaking backward. An event happening, is what makes the proposition describing the event true.
But we have an apparent quandary with Pancake Bob. Nothing obviously causes him to say “pancakes” except his own free exercise of will.
But now, knowing in advance that he will (freely) say “pancakes,” Bob wants, freely, to say “ass” (or something else) instead.
In other words, even though Bob himself at t2 makes it be true at t1 (and at all times) that he says “pancakes,” Bob is determined to say something other than what he freely says.
That is, Bob wants to do the logically impossible. Bob wants to SP and NSP at the same time. Bob wants to make A and ~ A true simultaneously. He wants to have his cake and eat it.
He can’t do this, on pain of logical contradiction.
But, Bob at t1 is mistaken to suppose that he must be able to say “ass” to vindicate his free will, given that it happens contingently to be true at t2 that Bob says “pancakes” and does not say “ass.” It is Bob himself at t2 who freely actualizes the “pancake” world and renders all other worlds possible but non-actual.
Bob at t1 is actually asking, not for free will at t2, which he has, but rather for the power at t2 to both do and not do something at the same time.
He wants to make a true statement that he by his own free act makes true, untrue.
Put another way, he wants the statement to be true and to be untrue at the same time, which is logically impossible.
In the seventh panel of the cartoon, after Bob in the future of t2, at t3, has looked back in time and seen himself saying “pancakes,” at t2, even though, at t1, he was determined to say “ass” after finding out that he would say “pancakes” at t2, he is plunged into gloom. He thinks he has no free will.
His reasoning goes like this:
1. “Given that I said ‘pancakes’ at t2, the timescope foresaw that at t1 I would say ‘pancakes.’ This means I could not have done other than what the timescope foresaw.”
2. “From this it follows that I could not have done, other then what I did; and free will is an illusion.”
But statement 2 does not validly follow from Statement 1. The two statements are expressing distinctly different ideas.
It is logically invalid to argue that because Bob could not have said other than what the timescope showed him saying, then he could not have said other than what he said.
He could have chosen NSP. But he chose SP instead.
Had he chosen NSP, then the timescope would have shown him saying NSP.
Bob could have said, other than what he said. He just could not have said, other than what the timescope showed him saying. That is, while his act was free, he could not escape detection of his act, even before he carried it out.
When Bob worries about “genuine possibility,” he is illicitly and illogically conflating the notion of choosing SP or choosing NSP, both of which he can do, with the notion of doing other than what the timescope shows. It is true that he is unable to do other than what the timescope shows. It is not true that he is unable to say other than what he says.
Bob could have said “ass” instead of pancakes.
But in that case, the artist would have to redo the cartoon.
He’d have to show Bob in the opening panel spying on his later self saying “ass” and then pacing around scheming to say “pancakes” instead.
There is no pleasing Bob!
Objection one: Bob “really” has no option.
Once an earlier version of Bob spies on a later version saying “pancakes,” Bob knows, from then on, that no matter how much care he takes, he will say “pancakes.” The logical argument mooted above notwithstanding, the objection goes, is that Bob really has only one option. It’s not so much that he must say “pancakes,” which notion commits the modal fallacy, but that he will say “pancakes” because he literally has no other option. Thus he cannot have true libertarian free will, under which agents have genuinely available options.
This objection strictly avoids the modal fallacy, and the deep peculiarity of the timescope scenario makes it beguiling. But it does not hold up.
Bob’s timescope/pancakes conundrum is just a time travel scenario running forward in time instead of backward. In David Lewis’s classic 1976 paper, “The Paradoxes of Time Travel,” Tim travels backward in time with the objective of killing his grandfather before grandfather reproduces. If Tim is successful, he will bring about the Grandfather’s Paradox: He will ensure that he himself is not born, hence making it impossible that he went back in time to kill his grandfather.
But if Tim is wise, he will realize that no account of free will, libertarian, compatabalist or otherwise, admits of an agent bringing about a logical contradiction. The grandfather’s paradox describes a logically contradictory state of affairs, which is why it is a paradox. Given that Tim exists to travel back in time to begin with, it is logically guaranteed that neither he nor anyone else, in the past, killed his grandfather before grandfather began the line that led to Tim.
More precisely, Tim will not do anything in the past different from what he in fact did. For even as Tim is traveling to the past as judged by his personal time, that past is over with as judged by the external time: the forward-direction of time that everyone else is experiencing, and which Tim also experienced before he entered his time machine.
This means that when Tim goes back to the past, it is logically guaranteed that he will do there, what he in fact already (freely) did even before he was born, as judged by the external time.
To desire to do otherwise is to demand the ability to bring about a logical contradiction. It would be a logical contradiction for Tim to not do something that he did in the past, or to do something that he did not do in the past. So Tim will do in the past, what he in fact already (freely) did before he was even born.
Objections that Tim has no genuine options available to him are beside the point. Given that there can be only one actual past, present and future, the past, present and future cannot be changed by anyone, moving in any time direction. One cannot change the present or the future, either. This is true whether presentism or eternalism is the correct philosophy of time. The present subjective moment changes under both presentism and eternalism; it’s just that the presentist claims that the present moment is all that there is, whereas the eternalist holds that past, present and future moments all exist: are all ontologically on par. But under either doctrine, one’s experience of the subjective present is constantly changing.
But while moments change, and while one can change light bulbs, diapers, hair styles, religious affiliations or political beliefs, no one can change history. As J.J.C. Smart noted, if one moves one’s arm, one does not change the present or the future; one makes the present and future be what they in fact are.
Free will allows one to actualize the past, present and future, but not change them.
Therefore, Bob cannot change the future that he has espied, anymore than Tim the time traveler can change the past. The mistake is to suppose that this restriction invalidates libertarian free will. It does not. Free will admits of one actualizing past, present and future from a menu of realistic options; not changing them.
Bob in the pancake story and Tim in the time traveler story want a do-over. They want it to be A when ~A is the actual case — the actual case, made actual by their own free acts. Again, no account of free will requires as a pre-condition for its exercise the ability to break the laws of logic.
One is, of course, troubled by that mystifying interval of time between when Bob looks ahead and sees himself saying “pancakes” and decides to say “ass” instead at the big moment.
The logical arguments to one side, what, after all, will stop a sufficiently motivated Bob from uttering the word “ass” (or any other word under the all-inclusive rubric NSP)? Will Logic Man, like a dues ex machina, descend from the sky in fiery flowers of logical smoke at the crucial moment and contrive to force Bob to say “pancakes” to keep the world logically intact? Lewis asked the same question in his time travel paper about what would stop a sufficiently motivated Tim from killing his grandfather, and concluded that this idea of Logic stepping in and stopping him was a “boring evasion.”
Why doesn’t Tim kill grandfather? “For some mundane reason,” Lewis notes.
And so it is for Bob.
We can imagine he changes his mind and decides to say “pancakes” after all. Or maybe he says “ass” a but someone right next to him, at the breakfast diner, loudly says “pancakes.” This means that Bob would have interpreted the intelligence in the timescope incorrectly at t1, but the scope itself would have been correct, as stipulated by its infallibility.
Maybe he’s at the diner the next morning, when the big moment approaches, and he is getting ready to say “ass.” Just then the waitress comes by. Bob is a regular, and the waitress knows that he always orders either pancakes or eggs. As the moment arrives, and Bob readies his vocal cords and molds his lip to say “ass,” the waitress appears at his shoulder and says, bubblegum snapping: “So what’ll it be this morning, hon? eggs or pancakes?” And Bob, without even thinking about, blurts, “Pancakes!” And in so doing, what he says and what the scope shows match.
One may object that in the final example, Bob didn’t freely say “pancakes” after all. He just blurted it out. But that’s no argument against free will in the context of the timescope. Or, if it is, it’s an argument against free will without the scope. We often say and do things without giving them much thought beforehand.
Under the best of circumstances, free will is constrained and sporadic. As has been stressed, no account of free will permits one to bring about a logical contradiction.
But there are other constraints.
We cannot will ourselves to do that which is physically impossible: flap our arms and fly to the moon, for instance. We can’t freely do many things that fail to be compossible with a wider, more inclusive set of facts: a plumber is not free to wake up tomorrow and become a surgeon. He is free to try to do that, but failure is guaranteed by virtue of wider, more inclusive facts such as: he has no training as a surgeon; he has no aptitude for surgery; he lacks the requisite surgical tools; no one will hire him to do surgery and no one will agree to be his patient. So the plumber, even if he tries to be a surgeon, will fail. But failure does not impeach freedom. We try, and fail, to do things all the time, and no one complains that failure entails the lack of free will.
Finally, the idea that knowing in advance what we will do somehow constrains our free will is itself deeply problematic. In fact, we very often know exactly what we are going to do, before we do it (unless we blurt something out, as noted earlier). Indeed, one might say that the ability to know in advance what one will do is a pre-requisite of free will. For if one could not plan and foresee one’s actions, how would they be freely done? Either they’d be random acts, or they’d be the outcomes of mindless, causally deterministic processes, and neither account is consistent with free will.
Objection Two: The Argument to Ontological Fatalism
Arguments to fatalism generally come in three forms: causal fatalism/determinism, logical fatalism/determinism, and epistemic fatalism/determinism. The first claim has not been considered here, but is false just owing to the observed fact of quantum indeterminacy (and it is also false if one subscribes to a regularity view of physical law).
The second two have been adequately dealt with, if the arguments here go through. Epistemic determinism — the alleged problem of God’s foreknowledge of human acts precluding their freedom — is just a proper subset of logical fatalism, the issue first mooted by Aristotle in his sea battle argument. This essay has tried to show, via the example of Bob and the time telescope, why such fears are unfounded.
But a more pressing objection might be to ontological fatalism. One might say, “Look, the very ontology of a world that permitted the existence of time telescopes — of devices that could look into the future and infallibly foresee what will happen — by itself rules out free will. This is because in such a world, everything that ever was, is or will be, is simply fixed in place, a bug-in-amber world of frozen acts and events. Nothing, before it “happens,” is actually unsettled or indeterminate; one could go further and say, as do some advocates of this view, that nothing actually happens at all in such a world; everything just is. It is a world without actual change or motion; both of which would be illusions. Such would be a Parmenidean world in which the very idea of free will is incoherent.
For a time telescope to work, the world would have to be eternalist in nature: just as every location in space exists, every location in time would exist. Socrates exists, just not now; and future people exist (including all the future versions of you and me) just not “now” as we who write and read this count “now.”
The argument to ontological fatalism does not go through. I suspect that the “bug-in-amber” picture is an analogic and reification fallacy. There is simply no logical reason to believe that a future version of you cannot be acting freely, even as a past version of you did act freely. Under eternalism, no time is special; so if you were free in the past then you will be free in the future too, even if past, present and future are all “set.” After all, there is only one actual history, and it can only be one way. So if past, present and future are set, it does not follow that they are necessary. This, again, is to confuse truth with necessity.
Indeed, because there is a pattern to our lives, a pattern that fulfills a cause-and-effect-like succession, or sequence of events, it is reasonable to suppose that in many cases when and where we are able, it is we ourselves who contribute to the creation of that pattern across spacetime. The objection that if the future exists, it is unavoidable and we cannot change what is to come, is to wrongly assume that free will entails that we be able to change anything.
But as discussed earlier, even as one can’t change the past, one can’t change the present or future, either. Instead, one makes the past, present and future be what they were, are and will be. Even if your future is scripted before you “arrive” there, it is you yourself who contributed, at all the times of your life, to the writing of that script.