That we find the idea of travelling in time, and in particular travelling backwards in time, fascinating, is evidenced by the plethora of new science fictions shows depicting time travel that have recently hit our TV screens. I love time travel shows, and I can hardly keep up. In almost all cases these shows depict what philosophers call inconsistent time travel stories: stories that commit what my colleague Nick Smith at The University of Sydney calls the second time around fallacy. These stories depict individuals travelling back in time (or in some cases sending a signal back in time) in order to change some past event (or, in some cases, to try and mitigate the changes brought about by some other time travellers).

It’s easy to see why these storylines are captivating. They ask us to imaginatively entertain questions about the robustness of the present. We are asked to ponder to what extent the way things are now is the product of fluky events, so that had the past been ever so slightly different, the present would be very different indeed. These questions are intriguing because we all wonder to what extent our present selves are the result of choices we made, where we could so easily have made others, and to what extent who we are is most robust, resilient under small changes in our past decisions.

These are issues worthy of reflection, which is why these stories are entertaining, and often worth watching. Why, though, are they inconsistent? Because, at least according to philosophical orthodoxy, it is impossible to change the past. Things that look like changing the past are possible. It’s possible to travel to a parallel universe that is like ours up until some time that is a past time in our universe, and to act at that time so that the future in that universe is different than the future in our universe. So it’s possible to travel to a universe much like ours, kill a baby called “Hitler” and in so doing, make it the case that in that universe there is no World War II. That, however, is not changing the past, it is just moving from a universe in which Hitler grew up and started World War II, to a universe in which he is killed as a child: our universe remains one in which history remains as we know it to be.

Here is why philosophers typically agree that it’s not possible to change the past. In order for something to change, it needs to be one way, and then some other way. In ordinary cases of change things change when they are one way at one time, and some other way at some other time. Suppose that on Monday at 2 pm Annie’s mass is 21.2 kgs (Annie is a dog). Suppose she wants to change her mass. She won’t do this by trying to make her mass different at Monday at 2 pm. Instead, she will try and make her mass at times later than Monday at 2 pm, less, or more, than 21.2 kgs. Suppose, though, that instead of changing the properties of an object, like Annie, we want to change the properties of a time itself. Then that time needs to go from being one way, to being some other way. But how can it do that? A time can’t go from being one way at one time, to another way at another time, because no time exists at more than one time. At least, no time exists at more than one time in the same temporal dimension.

If we were to introduce a second temporal dimension – sometimes called meta-time – then a single time in one temporal dimension could exist at multiple times in a second temporal dimension. If we posit a second temporal dimension then changing the past amounts to making some time one way at one location in meta-time, and some other way at some other location in meta-time. We can now at least make sense of how a time could change. The “first way” the time is (the pre-change way) is how that time is at one time in meta-time. The “second way” the time is (the post-change way) is the way the time is at some other time in meta-time.

The problem is that most philosophers think there is no meta-time, and most think that even if there were, this still wouldn’t really be changing the past. That’s because the time, as it was before the change, still exists in meta-time. All that has happened is that another version of that time has been created in meta-time, at which different events occur. But the original events, at the original time in meta-time still exist. If one wanted to change the past in order to make it the case that those events never happened, then one has failed. All one has done is made it the case that in another version of that time, those events didn’t happen. So if erasing certain past events and replacing them with new events, is what changing the past would amount to, then even were there a second temporal dimension, we still could not change the past.

So far this makes it sound as if philosophers are a bunch of science fiction spoil sports. Not so! Although philosophers typically agree that we cannot change the past, most think it possible to causally affect the past. What’s the difference? Changing is altering a time from being one way, the first time around, to being some other way, the second time around (hence “the second time around fallacy”). Causally affecting a time is making a time the way that, in fact, it is. If anyone travels to the past they will causally affect the time to which they travel: they will breathe air, tread on bugs, talk to people, and so on. But nothing the time traveller does changes the past: instead, her actions are part of what makes that time the way that it is (and was).

Consider Tim, who despises his violent, nasty, grandfather by whom he was raised. Failing to understand that he cannot change the past, he travels back in time to try and kill his grandfather (despite the fact that, clearly, his grandfather did not die in the past, since he is alive in the present). Tim succeeds in shooting his grandfather in the head, causing significant brain damage. As it turns out, it is this brain damage that later causes his grandfather to be violent, which, in turn is what motivates Tim to try and shoot him. This is a consistent time travel story. Tim causally affects the past: he shoots someone in the head. But he doesn’t change the past; he is simply causally responsible for the past being the way it has, always, been.

Given that time travellers can causally affect, but not change, the past, it seems reasonable to ask: what sorts of things ought time travellers to the past to try and do? Put it like this. Given that Tim’s grandfather is alive in 2017, not only do we know that Tim will not kill his grandfather in 1980, but Tim is also in a position to know that he will not kill his grandfather in 1980. Here’s where things get interesting. Plausibly, someone can only deliberate about whether to do X, if that individual does not know whether or not X. For instance, I cannot deliberate about whether to turn blue, because I know I cannot turn blue. So suppose I know exactly what happened on some day in the past, including knowing that my future-self time travels to that day, puts on a duck suit, and marches in the “free the ducks” parade. Then I cannot deliberate about whether I will travel back in time to that day and put on the duck suit, any more than I can deliberate about whether to turn blue. That’s because I know what I will do, and knowing what I will do precludes me from deliberating about what to do.

It is, however, rare that we know everything about some past time. So that leaves plenty of room for potential time travellers to deliberate about what to do in the past. Some things, no doubt, will be ruled out as deliberation worthy. Insofar as I am certain that Hitler grew to adulthood, I cannot deliberate about whether to travel back in time to kill him in his youth, since I know that I do not do any such thing. But we have greater, and lesser, amounts of evidence about the past. Sometimes evidence is weak. In cases where evidence is weak, or where the stakes are high, I might rationally deliberate about acting in the past so as to bring about some desirable consequence.

If I have weak evidence that ten thousand people died of famine at some past time I might decide to travel back to give them crop advice, reasoning that the weak evidence I have that they died of famine may be misleading, and by travelling back in time and offering advice I might save the ten thousand lives: after all, perhaps it is now true that 10 000 people did not die of famine, though they almost did, and perhaps that is true, because I prevented the famine.

Indeed, as my colleagues Andy Egan (Rutgers University) and David Braddon-Mitchell (The University of Sydney) have pointed out, it might seem reasonable to manufacture evidence about the past. Suppose that rather than there being weak evidence that there was a famine, there is really a fair amount of historical evidence. Then either the famine occurred, in which case there is no point my travelling back in time to try and prevent it, or the famine did not occur, perhaps because I prevented it, and the evidence in the historical records is misleading. So I seem to have reason to try and fabricate historical evidence of a famine. Here is why. Suppose I am considering whether to travel back in time to try and prevent the famine. I have some reason not to do this, given evidence that the famine occurred. But suppose I travel back to times after when the famine would have occurred, if it did, and I tell a bunch of historians that there was a terrible famine that killed ten thousand people. If I do that, then I have an excellent explanation for why there is evidence of a famine: my future-self created it through time travel. And if I have reason to think that the evidence of famine I have is misleading, I am no longer sure whether or not the famine happened, and that makes it rational to travel back in time to try and prevent the famine.

So it seems that I ought to first travel back in time and plant evidence that there was a famine, then travel back even further and make sure the famine does not occur. That seems odd. For, you might think, it cannot be reasonable to manipulate evidence in this way: after all, either the famine happened or it didn’t; either I prevented it, or I didn’t, and no amount of manipulating the historical records can change that. So how can it be rational to manipulate those records, in order to make it rational to try to prevent the famine? Yet it seems that the more reason I have to think that evidence of the famine is misleading, the more reason I have to travel back in time to try and prevent the famine.

In fact, I think that what holds for the past also holds for the future. If I already know what I will do in the future – say, tomorrow – (by having already decided, or by being told by a prophet or a super computer) then I cannot deliberate about what to do. Of course, I typically don’t know what I will do tomorrow (other than by first deliberating about it) whereas I might know what I did do last year. So my ignorance is not symmetrical. I typically know a lot more about my past actions than my future ones. That means in general it will be rational for me to deliberate about what to do in the future, but not what to do in the past. But that’s not a difference in kind, it’s just a difference in degree: in how much each of us knows about the future compared to the past.

If I have excellent evidence that tomorrow I will rob a bank, then the more certain I become that that is what will happen, the less it will make sense to deliberate about whether or not to rob the bank. If my evidence comes from future newspapers, a computer prediction, a reliable oracle, or some such, then if I am able to travel into the future, it seems reasonable for me to try to create false evidence that I rob a bank. For that can explain why the newspapers, computer predictions, and oracles, all tell me that I will rob a bank. After all, if I know there is false evidence, because I planted it, then this undermines my certainty that I will rob a bank, and seems to make it rational for me to deliberate about whether to rob a bank, in the very same way that planting false evidence that there was a past famine seems to make it rational for me to deliberate about whether to try and prevent the famine.

It doesn’t really matter whether you’re travelling into the past, or the future; what matters is how much you know about the time to which you travel. If you know a lot about what you did, or will do, or if you know a lot about what did, or will, happen, then your capacity to deliberate about what to do is severely undermined. But if you don’t know what you did, or will do, and if you don’t know what did, or will, happen, then your capacity to deliberate remains intact. And if you think you know what did, or will, happen, and you’d prefer that it didn’t, or won’t, happen, then it might even be rational to produce misleading evidence as of that thing happening, in order to make it rational to travel in time to try and prevent that thing from happening. That all goes to show that time travel, and decisions of time travellers, are plenty interesting enough without needing to think that changing the past is possible.

Kristie Miller is associate professor in philosophy at the University of Sydney, and joint director for the centre for time; she is author of Issues in Theoretical Diversity: Persistence, Composition and Time as well as many research papers in metaphysics. 

Image: Brian Pirie via Flickr Creative Commons