I think I finally understand free will. I’d like to share my thoughts.
I won’t be making any definitive, be-all-end-all claims about free will here—I think the words “choice” and “free will” and “compatibilism” mean different enough things to different people to render any such attempt fruitless. Instead, I’ll focus on addressing common confusions surrounding free will. Through personal experience and many discussions, I’ve found that most such confusions can be phrased in one of three ways:
- How can I make any real choices, if any choice I make is already fixed in the future?
- If the future is fixed, how could any of my actions possibly change it? If I can’t change the future, then how could I be said to have control over it?
- How can I be said to be the source of the actions I make, if my actions ultimately result from atoms abiding by the laws of physics?
I will address each of these questions in this post.
How can I make any real choices, if any choice I make is already fixed in the future?
This question is confusing because we have two similar, but subtly different notions of possibility. In one sense, something can happen if it doesn’t violate the laws of physics for it to happen. In a deterministic universe, where the laws of physics completely determine all future possible states of the universe, saying something can happen is the same as saying it actually will happen.1
In another sense, we say that someone can perform an action if, should they genuinely try to do it, they will succeed in doing so.2, 3 In this sense, I can jump an inch in the air, but I can’t jump 10 feet in the air—if I try to jump an inch in the air, I’ll succeed, but no matter how hard I try I will not jump 10 feet in the air.
I’ll refer to the first sense as “canp” (p for “physical”) and the second as “cana” (a for “ability”).
We can illustrate the distinction in the two senses by considering a chess program trying to pick its next move. If the program is following a deterministic algorithm, its move will have been determined by the algorithm beforehand, so it canp not pick a move different from the one it will make. At the same time, when it’s trying to decide on a move, there are clearly a number of moves it cana take (namely, all of the legal moves—these are the moves it would successfully make if it decided to).
It happens that “cana” is the sense of “can” we almost universally refer to when talking about ability. When we say “Billy can run a 4-minute mile!” we mean that if he eats his pasta and warms up and runs a mile while someone times him, he’ll clock under 4 minutes—that is, if he tries to run a 4-minute mile he’ll succeed. We don’t mean to say that there is some possible future state of the universe in which he does run a 4-minute mile.
Let’s return to the original question: How can any of my choices be real, if any choice I make is already fixed in the future? A choice only feels real if we have multiple different possibilities to choose from. But if our choice is fixed in the future, that means we can only choose one possibility. The question can thus be rephrased: When I am making a choice, how can I have multiple different possibilities to choose from, when there is only one possibility I can choose? And now it’s clear why this question is confusing—it seems to ask how a direct logical contradiction could be possible!
But to say that no alternative choice is possible is to say that no other choice canp be made, while to say that we have the ability to choose is to say that we cana make choices. And there’s nothing contradictory about saying, I cana choose from multiple different possibilities even when there is only one possibility I canp choose. That’s how it is for chess programs, at least! Once we delineate the separate senses of “can” appearing in the question, the apparent contradiction vanishes.
Fundamentally, it’s the same sort of contradiction in “The more cheese, the more holes; the more holes, the less cheese.” It just gets taken a lot more seriously.
If the future is fixed, how could any of my actions possibly change it? If I can’t change the future, then how could I be said to have control over it?
Your actions don’t change the future, but they do determine it, and when you are responsible for determining the future, you have control over your future.
Consider the following scenario. A student is pondering whether to study for her test tomorrow. She knows that if she studies, she will get an A, and that if she doesn’t, she will get an F. Suddenly, God appears from the heavens, tells her that he saw into the future, and gives her a transparent glass case with a folded sheet of paper inside, with her future grade written on it. He then tells her that the case will only open after the test.
The student reasons that since her grade on the test is already fixed (it’s written on that slip of paper already, after all!) she can’t change her grade, so there’s no point in studying. Feeling smug about philosophizing her way out of studying, she parties all night, gets an F, opens the case, and finds an F on the sheet of paper. She shrugs. “Guess it was always an F, nothing I could’ve done about it!”
There’s something off about her reasoning. She couldhave studied, in which case she’d have gotten an A, and the grade written on the paper would always have been an A.4 It is only because she chose not to study that the grade on the paper had always been an F.
As the contents of the paper never changed, the student never changed her future by deciding whether to study. But her future did depend on her actions. In other words, even though she couldn’t change the future, she still had control over it.
If my actions result from atoms abiding by the laws of physics, how can I be said to be the source of the actions I make?
I am the source of the actions I make because I am the atoms resulting in my actions.
How can it be said that my hand opened a bottle of water, when it was really just my fingers and palm? How can it be said that I read a message from my friend, when all he sent me were a bunch of pen marks on a sheet of paper? How can it be said that I picked up my food with a fork, when all I used was really just a bunch of atoms?
It can be said because my hand is my fingers and my palm; because the message is the pen marks on the sheet of paper; because the fork is just a bunch of atoms. I, likewise, am the atoms and chemical processes in my brain responsible for everything that I do.
To see this another way, suppose that a bunch of crazy scientists implant a bunch of electrodes in my brain that make me always desire cherry pie over apple pie. Certainly, whenever I choose cherry pie over apple pie, the scientists are responsible for my decision to eat the cherry pie. But this does not mean I am not responsible for my decision as well! The electrodes are a part of my brain, a part of of my decision-making process, a part of who I am. I ate the pie because I wanted to—the fact that the electrodes caused me to want the pie is irrelevant.
1. If our universe is deterministic, then any state of the universe determines all possible future states of the universe, so to reach a configuration of the universe not among these possible future states we’d need to violate the laws of physics.
2. It would be more technically correct to say that an observer with all relevant information would predict the agent would successfully cause the outcome, in a universe in which they do try. If someone is in a prison cell and really desires to leave, we probably wouldn’t say she had the ability to leave if her warden unexpectedly happened to come by at some point and let her walk out. On the other hand, if she knew the warden were coming along, we would say she had the ability to leave (whenever the warden came along).
3. With cana, we are considering what would happen in a hypothetical world which is logically valid to reason about, even if such a world cannot be reached from the laws of physics. There may be some concern that it is logical nonsense to talk about physically impossible worlds. I think it still makes sense to talk about them, as logical structures. For example, in the universe of Super Mario Bros., there may be certain states of the game that are unreachable no matter how hard the player tries to play (say, walking through a wall). But it still makes sense to talk about what the game would look like when such an impossible state is reached, e.g., if someone hacks the game and places Mario straight into a wall.