If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Metaphysics: Problem of Free Will

Richard discusses one of the classic philosophical problem of free will --- that is, the question of whether we decide things for ourselves, or are forced to go one way or another. He distinguishes between two different worries. One worry is that the laws of physics, plus facts about the past over which we have no control, determine what we will do, and that means we’re not free. Another worry is that because the laws and the past determine what we’ll do, someone smart enough could know what we would do ahead of time, so we can’t be free. He says the second worry is much worse than the first, but argues that the second doesn’t follow from the first.

Speaker: Dr. Richard Holton, Professor of Philosophy, Massachusetts Institute of Technology.
Created by Gaurav Vazirani.

Want to join the conversation?

  • leaf green style avatar for user Tomasz Stachowiak
    It is clear that anyone subject to the experiment can in fact predict the state of the light bulb. So this is not a counter example to the implication: determinism => foreknowledge. It merely shows, that one can set up the experiment to frustrate the communication of the foreknowledge, but this is not really the question. In essence, it is like a similar setup in which you can tell me the outcome with words, but beforehand I cut your tongue out...
    (19 votes)
    Default Khan Academy avatar avatar for user
  • hopper cool style avatar for user oriramikad
    Why can't I just smash the lightbulb on the top of the box and then predict that the lightbulb will be off?
    (4 votes)
    Default Khan Academy avatar avatar for user
    • leaf green style avatar for user Agent Smith
      The point of the device is to illustrate that foreknowledge does not follow from determinism provided that we could be "frustrators" capable of acting against predictions.
      Of course you could break the bulb and make your predictions but that would be missing the point of the experiment.
      (5 votes)
  • hopper jumping style avatar for user carter
    The light bulb example is irrelevant because with all the computing power in the world you could send a signal to the simple machine to make it do the exact opposite of what it is supposed to do.
    (3 votes)
    Default Khan Academy avatar avatar for user
    • duskpin ultimate style avatar for user Lyle Gonzalez
      This misses the point of hypothetical questions. Yes, you could do that in real life but this is a hypothetical situation so you should assume you can't do that since it wasn't specified. The point is that it is logically possible for there to be scenarios where you are stopped from being able to communicate an accurate prediction even though you have perfect knowledge of the system. The light bulb example isn't meant to be a challenge to find all the loopholes. Otherwise, why not just get up and smash the light bulb? He didn't say you couldn't do that either.
      (5 votes)
  • spunky sam blue style avatar for user Max
    The book of life is just a s problematic as the light bulb. Say the book says you will close the book, but you don't. The book, obviously, either cannot account for free will through determinism, or it is just wrong. However, at Richard says the book is 'Cataloged perfectly' This is obviously wrong in this case. Since the book is perfectly deterministic, this could not have happened, as the book would have seen it coming and changed to reflect that (well, technically not since it will have always been that way). This may come to a state like the light bulb's, where nothing that could be written at a point would be true. This is a paradox on the side of the book, not free will, or lack thereof. The existence of a perfect and complete prediction of a universe that the universe the prediction is of can use is impossible, not because of free will, but because the prediction existing will cause anything that can read it to act differently than it would otherwise, so the prediction would acknowledge that, and it would affect the out come differently, so it be re-written again , and so on. The prediction is impossible because it affects itself by being part of the universe, free will or not.
    (4 votes)
    Default Khan Academy avatar avatar for user
  • leaf green style avatar for user Gregory Greif
    Does quantum mechanics make the existence of free will more likely, considering that it disproves determinism, or does it merely shift the problem from humans being influenced by an initial state to humans being influenced by random quantum fluctuations that we cannot control?
    (3 votes)
    Default Khan Academy avatar avatar for user
    • blobby green style avatar for user Pavlos Kanellakis
      Quantum physics has no influence on free will because, so far, all it has shown is an inability of ours to predict certain events and states at the quantum level. This inability stems from our lack of understanding, therefore any argument with this at the base is ultimately an argument from ignorance.

      Furthermore, so far the explanation given involves randomness. If this holds true then all it shows is that at the quantum level nature (or certain parts of it) operates outside the causal chain. We, however, still do not. Our choices continue to be determined by our nature and nurture. Those do, in fact, follow a causal chain. To suggested that quantum theory eradicates determinism you would have to claim as a prerequisite that thoughts appear in our heads just as randomly and completely independently of any events prior to the thought. This means (in a simple analogy) that I could "randomly" have a thought in a language I do not speak about a subject I have never experienced and have no knowledge of. If this were true and possible then it raises two important questions: 1) Why has this never been recorded as happening? and 2) Why doesn't it happen constantly? The second question is more important here because if we had such random thoughts we would not be capable of functioning. The only way to get around this problem is to re-introduce determinism and replace randomness with it. In other words, it only seems random to us because we are unaware of the underlying cause.
      (3 votes)
  • blobby green style avatar for user Chris Matthews
    Assuming that foreknowledge is like the device introduce at is kind of like assuming you have only two options in every choice in life isn't it? If you read that next page in the book of life and see that you will light the book of life on fire, does that mean that there is only the option of burning it or leaving it alone? Did you not have the option to close the book of life and walk away? Or to keep reading? If the actual ability to foresee what the future holds was out there, wouldn't the device be more likely to be multiple light bulbs on the box and you have to predict the light bulb that would be turn on? Given the possibilities in your life are not limited by anything but your own level of intelligence and experience, among other things, which would really give each decision endless outcomes. You have to decide what to do next after high school, you can either turn the light on (go to college), or turn the light off (get a job). But there are more than one light bulbs you could turn on. You could stay at home with Mom be irresponsible. You could get a job AND go to college. So wouldn't it make more sense to have more than one light bulb in the metaphor and if so, would't that mean that our idea of "foreknowledge" is different than what it really is? Couldn't it be that there are mulltiple ways for things to happen and the "demon" that can see what is going to happen can really only see all the possible decisions that you COULD make but not necessarily know exactly which one you WILL make? And we could also assume that this "demon" is fairly experienced and can make rather accurate educated guesses?
    And another thing is, seeing as we do not KNOW for fact that there is someone with foreknowledge wouldn't that take the "Frustrator" out of the picture? Because you can only intentionally choose a different option to "frustrate" the system, if you know what option the "forseer" has picked. In this metaphor the 'forseer" (us) must light the light bulb to signify his/her predictions, and the device's light sensor takes that signal to make its decision. But in our reality we have no signal as to what our predetermined moves are. So how is anyone able to be a frustrator?
    (2 votes)
    Default Khan Academy avatar avatar for user
  • male robot hal style avatar for user Corey Piper
    All of the initial conditions other than that it has a light sensor right?
    (2 votes)
    Default Khan Academy avatar avatar for user
  • mr pink red style avatar for user Canibalizm
    (Playing devil's advocate here) This is not fair. You're making the only way of me expressing my prediction to be through a lightbulb. I'd rather just tell you through my mouth or text you what my prediction is. I think this thought experiment is flawed because the medium of communication is a frustrator. If something else was a frustrator then I'll reconsider this thought experiment. Is there a different version of this experiment? Also, you are assuming that a frustrator exists in a system that supposes free-will.
    (2 votes)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user dickydawg9
    How is it possible too in better ones life habits
    (1 vote)
    Default Khan Academy avatar avatar for user
  • piceratops ultimate style avatar for user OcculusSlime
    You certainly can make accurate predictions with a frustrator in it. You can predict that the frustrator will frustrate.
    (1 vote)
    Default Khan Academy avatar avatar for user

Video transcript

(intro music) My name is Richard[br]Holton, and I teach at MIT in the Linguistics and[br]Philosophy Department. And today I'd like to talk about the problem of free will. Or, more precisely, about[br]a couple of problems surrounding free will, which I think tend to[br]get muddled up together. And part of what I want to do today is to draw those two problems apart. So the first problem of free will is probably the one[br]most people think about. And that is the idea that if all the laws of[br]nature are deterministic, then if the initial conditions are fixed, everything that will[br]happen thereafter is fixed, by those conditions and[br]by the laws of nature. People have thought,[br]understandably enough, that that puts a real[br]problem on free will. Because part of the intuitive[br]idea of free will is that when I'm faced with some choice, decision, I could choose one way and act that way, or I could choose the[br]other way and act that way. And those two choices are free. But let me mention,[br]now, the second problem. Which I think gets tied[br]up with this first one. So the first one is a metaphysical issue, it's about determinism. The second one is basically[br]an epistemological issue, that is it's an issue[br]about what we can know. And this is actually an idea[br]that goes back much further. It goes back certainly to the stoics, and it's come up time and[br]again in Christian thought. So this is the problem,[br]really, of foreknowledge. That is, if things are determined in[br]the way we talked about before, then, if we know the initial conditions, and we know the laws of nature, then it looks as though[br]we'll be able to predict everything that happens. So Laplace phrased this very neatly, and came up with this idea[br]called "Laplace's demon," who is able to predict everything[br]that you're going to do. So there you are, thinking[br]you're acting freely, and this little demon[br]knows what you're gonna do. And you can make the problem even worse. So, suppose you could get a book in which all this was written down. a Book of Life as it's sometimes called. You can open the book and you can look at your history and you can see everything you've done,[br]it's catalogued perfectly. You flick through the[br]pages, you get to today, you run your finger down the ledger, and you get to the entry which says, "You are reading the Book of Life." It's the end of the page. And you think, "If I turn over the page, I'm gonna know exactly[br]what I'm going to do. How can I hang onto any[br]conception that I'm free, if when I turn the page, I'm going to see everything that I'm going to do." This second problem, I think, is in many ways worse than the first. What I want to suggest is actually that the second idea doesn't[br]follow from the first. And in order to illustrate[br]that difference, I want to think about a[br]very, very simple device, in a very, very simple game. This is a challenge I offer to you. I have this device, and[br]on the top of the device, is a light bulb. And what I want you to do is to predict whether or not that light[br]bulb will be on or off at exactly noon today. And I'm going to make[br]it rather easy for you. I'm going to give you a[br]lot of resources to tell whether or not that[br]light bulb is on or off. I'm going to tell you exactly[br]what the configuration of the machine under the light bulb is. I can tell you all about the circuitry, all about the power coming into it, so you'll know exactly how that works. And in fact, I'm gonna tell you about anything else in the universe[br]you might want to know. So I'll tell you all of[br]the initial conditions. And I'm gonna tell you[br]all of the laws of nature. And if that's not good enough, I'm going to give you something else. I'm going to give you all the computing power you can ask for. And my challenge to you then is, using all of this, predict whether or not the light bulb will be on or off at noon today. Small catch though. In fact, there are two small catches. First catch is this. You have to make your prediction before 12:00. You have to make it at[br]exactly five minutes to 12:00. Secondly, you have to make your prediction by putting on a light bulb of your own, which is quite close to[br]my light bulb on my box. If you put that light bulb on, then that's an indicator that you believe the light bulb on the[br]box will be on at noon. If you leave it off, that's an indicator that you believe the light bulb will be[br]off on the box at noon. So that's the challenge. Sounds easy, right? You've got all that power,[br]you've got all that knowledge, you should be able to predict whether the light bulb's going to be on or off. But now let me tell you what's in the box. Let me reveal some of the wiring. What's in the box is[br]a little light sensor. And it's directed at your light bulb. So if your light bulb goes on, that sensor will sense that it's on, and that will trip a switch, which will ensure that the light bulb on the top of the box will be off. Reversely, if your light bulb is off, the sensor will sense that, and that will trip a switch[br]which will ensure that at noon the light on the top of the box will be on. Now it doesn't seem so easy[br]to win the competition. Because what this machine[br]is, is a frustrator. It's designed to frustrate[br]whatever prediction you make, it's designed to make it go false. But now think what we've got. I've given you all the initial knowledge, I've given you all of the resources to compute, and yet you still can't come up with an accurate prediction once you're in a world[br]with a frustrator in it. Two things to say about this. First thing to say: it seems to me that many many things, but particularly human beings, may well be frustrators. In fact, you might be[br]a frustrator yourself. You might look at what's[br]predicted for you to do when you turn that page[br]in the Book of Life and you see what's predicted for you, and you might do the opposite, because you want to show[br]that you're no puppet of what's happening in the Book of Life. Secondly, of course,[br]someone outside the system, if they have all the predictive abilities, they could predict what would happen. But of course what they'd be predicting is just that you were[br]going to lose the game when you are stuck inside the system. So what conclusion can we draw? Well, I think the conclusion to draw is that if we are creatures[br]who are inside the system, which in some obvious sense we are, and if we or others[br]around us are predictors, then we can't infer from[br]the truth of determinism to the possibility of foreknowledge. That doesn't show, I think, that free will is home and dry, but I think it does remove one of the most worrying aspects to free will. Subtitles by the Amara.org community