Suppose you wake up in the hospital tomorrow morning, and in your groggy state you can’t remember whether you had a painful operation yesterday, or whether instead the painful operation is to be performed tomorrow. Suppose the operation is equally painful whether you had it yesterday, or are still to have it tomorrow. It turns out that most people would prefer to learn that they had the operation yesterday, rather than that they are to have it tomorrow. This phenomenon is known as future bias.
Recently empirical work from the Centre for Time shows that not only do most people prefer to have had the painful operation yesterday, rather than tomorrow, but, more startlingly, Greene, Latham, Miller and Norton found that most people preferred to learn that they had the painful operation in the past, rather than the future, even when the past operation was ten times more painful than the future operation would be.
The moral: we devalue the pains of our past selves relative to the pains of our future selves. We would prefer that a past self underwent an operation instead of our future self having to undergo an operation, even if operation undergone by the past self would have been ten times more painful than the operation the future self would undergo.
Put yourself in the position of the person in the hospital bed who has just woken up groggy and cannot remember whether they already had an extremely painful operation, or are due to have a significantly less painful operation. You might share the intuition that you would prefer to learn that you already had the very painful operation, rather than that you are still to have the much less painful operation.
But now imagine things from a different perspective. Imagine you were allowed to choose a life to live. You can pick a life off a rack, and that life will be yours. Suppose you are trying to decide between two lives. The two lives are exactly the same in all respects, except that in one of them you undergo an extremely operation on a Monday, and in the other life you instead undergo a much less painful operation on Tuesday (i.e. one day later). So the first life contains overall more pain than the second (since except for the location and amount of pain of the operation, they are exactly the same). Which life do you choose? You might think it would be deeply irrational to choose the first life over the second. But that intuition seems to be in conflict with the intuition many people have, that when they wake up in hospital they would prefer that their past self underwent a very painful operation, rather than that their future self undergo a less painful operation.
So what’s going on here?
Why are we so awful to our past selves compared to our future selves? Why would we prefer a life that is overall worse, (has the more painful operation) when we are in the hospital bed, but we wouldn't make that choice when we are choosing lives?
Why, in the former scenario, would we burden our past selves with more pain, to avoid our future selves having less pain? After all, they’re all our selves.
One possibility is that we are never really in a position to choose whether, say, a painful operation was had yesterday, or will be had tomorrow. If you had the painful operation yesterday, then you had it yesterday: but if you didn't have it yesterday there’s nothing you can do now, to make it the case that your operation was yesterday. So even if we would prefer that awful things happened to our past selves, rather than somewhat less awful things being going to happen to our future selves, this isn't going to result in our choosing overall worse lives, because those preferences don't determine our choices. To put it another way, if you have these preferences they are sort of harmless, because you can’t act on them. Indeed, if you are in a position of having a choice—such as you would be if you were able to choose between different lives—you wouldn’t choose in such a way as to make yourself overall worse off by choosing the life with the more painful operation.
The idea behind this thought has led to the hypothesis that what explains our having future-biased preferences is that the past is practically irrelevant. We can’t causally influence past events: we can’t choose to have the painful operation in the past. So, the thought goes, past events can’t be relevant in our making of present choices the way future events can. Recent experimental work by Latham, Miller, Norton and Tarsney suggests that this is at least part of the explanation. They came up with an ingenious experimental set up to test this idea.
They reasoned that if the reason we devalue the pains of past selves is because those pains are causally irrelevant in our decision making, then if we make some past events causally relevant, this will diminish (or eliminate) future bias with respect to those events.
They ran an experiment with 1462 participants. They divided participants into groups. Four of these groups are relevant. All participants saw a vignette that described a very similar state of affairs. But two groups of participants were asked to make a decision about what to do, and were told that their decision would causally affect what happened to them in the past. The other two groups of participants were asked what outcome they preferred, where that preference would not causally affect what happened to them in the past.
Participants in the first group read a vignette which described how they (the participant) have been abducted by an alien species called the Predictors. They were told the following: The Predictors have decided, out of scientific curiosity, to subject you to a lengthy series of painful electric shocks. You try to count the shocks, but you quickly lose track. This continues for several days. Finally, the Predictors are ready to release you. Feeling regretful for the suffering they have caused, they will repair any physical harms of the experiment, and give you a pill that will reverse any traumatic psychological effects (though without erasing any of your memories). Thus, you will suffer no further physical or psychological harm from the experiment after you are released. But before releasing you, the Predictors give you a choice: You may choose to voluntarily undergo one final electric shock before you are released. The Predictors inform you that they had already predicted your response to this offer before the experiment began.
One group of these participants were then told that if the predictors predicted that you WOULD choose to undergo the final shock, then they shocked you exactly 10,000 times since your kidnapping. If they predicted that you WOULD NOT choose to undergo the final shock, they shocked you 10,100 times. As their name suggests, the Predictors are extremely good at “predicting” human behavior: In fact, they have developed the technology for time travel, and upon observing your choice, one of the Predictors will travel a week into the past and report your choice so that the Predictors can shock you the appropriate number of times. Thus, if you choose to undergo the final shock, you will leave the Predictors’ spaceship having been shocked 10,001 times in total, while if you do not, you will leave having been shocked 10,100 times. So if you choose to undergo the further shock you will receive a total of 10,001 shocks, and you choose not to undergo the further shock you will receive a total of 10, 100 shocks: an extra 99 shocks.
The second group of these participants were told much the same, except they were told that the predictors would have shocked them 10 000 times in the past if they choose the extra voluntary shock, and 10 001 times in the past if they do not accept the voluntary shock. So participants in this group receive 10 001 shocks regardless of whether or not they choose the voluntary shock. So even though their choice is causally effects how many shocks they had in the past, it doesn’t affect how many shocks they are given, overall.
For participants in the first group, then, choosing to accept one future shock will thereby prevent 100 past shocks.
Finally, two other groups of participants saw similar vignettes, except that they are not faced with any decision. They are simply told that the Predictors randomly assigned them to one of two schedules. If they are on Schedule 1, then they were shocked 10,000 times in the last several days, and before their release they will be given one final shocks (leaving them shocked 10 001 times in total). If they are on schedule 2, they received 10,100 shocks in the past, but will not receive any further shocks.
Participants in these groups were then asked which schedule they would prefer to discover they are on. They can only either prefer that they receive no shocks in the future, but more in the past (schedule 2) or that they receive one more future shock, which means they received fewer past shocks, and fewer shocks in total (schedule1).
So what happened? The study found that people were less future-biased when they were given a decision, and that decision was causally relevant. So participants in the first two groups were less future-biased than participants in the second groups. The study also found, perhaps unsurprisingly, that participants were more likely to choose to accept the future shock when doing so would prevent the 100 past shock than they were to choose to accept the future shock when it made no difference to how many shocks they would get.
In all, this suggests that the reason we seem to care a good deal less for our past selves than our future ones, is that under most conditions only our future selves can be harmed or benefited by our present decisions.
Still, it is salutary to note that although future bias diminished when the past was casually accessible, there was still future bias present: some people chose to make their past selves suffer 100 more shocks to prevent their future self from getting a single shock. Understanding why that is, even under those conditions, will be important in understanding how we think about our past selves.
Greene, P., Latham, A. J., Miller, K., and Norton, J. (ms) “On preferring that Overall, Things are Worse: Future-Bias and Unequal Payoffs”. https://philpapers.org/rec/GREOPT-2
Latham, A. J., Miller, K, Norton, J, and Tarsney, C. (2020). “Future bias in action: Does the past matter more when you can affect it”.