The point of dissolving the free will question was that it doesn’t matter what physics we run on. There is in fact no physics which could possibly cause me to believe I had “free will” in the sense of somehow determining my actions outside of physics, because any method for determining my actions is a physical process. In every possible consistent physics, where the laws of physics are more or less constant, I will believe I have “free will” in the sense that the output of my brain corresponds to the algorithm I feel like I’m implementing. It’s logically impossible for me to determine my own algorithm, because I’d need some starting algorithm to select one, recurring infinitely. Not to mention that when I say “me” I’m really just referring to the algorithm I’m currently running anyway. I accept (as I think everyone does intuitively) that I exist due to forces outside my control. My mental algorithm was shaped over time by forces I could not control. Even now my algorithm can be changed by events outside my control, although this usually happens only in very small degrees and/or with consent from my then-current algorithm. But when I actually make a decision, the output of my body consistently lines up to what feels like the output of my mental algorithm. That’s what free will means, that’s where the feeling comes from; when I act, my thoughts and actions are in sync. That’s how free will gets dissolved, physics be damned. In fact, randomness and unpredictability makes me feel less in control than otherwise. If I really have free will, I ought to be perfectly predictable! You could just simulate the algorithm I run on, to whatever degree that’s actually accurate to how my mental architecture actually operates. If my decisions are unpredictable, even to me, then I start to worry about getting some fundamental part of me randomly erased, or finding that my mental narrative is a lie and I don’t control my own actions. Since this doesn’t happen (much) I can fairly safely say that random factors are not strongly important in my mental processes, though it’s possible I could be mistaken (humans are good at making up consistent narratives).
There is in fact no physics which could possibly cause me to believe I had “free will” in the sense of somehow determining my actions outside of physics, because any method for determining my actions is a physical process.
I don’t think this is a valid argument without the premise that if my will is a physical cause, it must itself be subject to physical causality. In other words, why couldn’t my will be a ‘one-way’ physical cause, able to cause things in the physical world, but unable to be affected by the physical world? Are you certain that every possible physics excludes this kind of one-way causation? Because while the idea strikes me as wildly implausible, it doesn’t seem to be contradictory or anything.
I haven’t seen an argument anywhere in the sequences (or elsewhere) defending this premise. This has always bothered me, so I’d appreciate it if you could supply the argument or point me to wherever I may have missed it.
In other words, why couldn’t my will be a ‘one-way’ physical cause, able to cause things in the physical world, but unable to be affected by the physical world?
The simplest examples of one-way causes may be the laws of physics. They cause the physical universe to have its properties but the universe does not cause them to exist or affect their nature. Theoretically there could be other “laws of wills” governing our behavior in a similar way but I would hesitate calling them actual (or especially individual) wills because of their effective non-agency. Agents’ behavior is caused by interaction with the physical universe, whereas the nature of laws is apparently not caused by interaction with the physical universe. A one-way will would be completely sensory-deprived and thus lack effective agency.
I think this is a very interesting thought, one famously articulated by Kant: the CI is essentially a law in the style of natural law, only pertaining to the will. He agrees with you that the law can’t be identified with each individual will (for one thing, some of us are bad or irrational). This avoids the ‘sensory deprivation’ problem, but keeps the idea that insofar as we’re governed by the law of the will, we’re free. The result is that we’re free only to the extent that we’re good.
Actually, that premise isn’t needed. No matter what the causes are, or even if I am an ontologically basic mental entity, it still remains true that I did not cause myself. I may not have been caused by anything else, but I certainly didn’t determine which algorithm is making my decisions. Not to mention that the rest of the world has to affect my will somehow, or I couldn’t actually perceive it or act on it intentionally (it’s a simple inversion of the argument against epiphenominalism). One-way causation is easily possible; I could write a computer program that worked like that. But the very act of writing the computer program violates the “free will” in the strict knee jerk reaction sense of “determining my own actions”. Determining your own actions requires cyclic causality, and even then I would struggle to accept that I really was determining my actions, rather than just saying that they happened basically by chance (I cannot recall where, but I recently saw something by EY about circular causality and time-turners in Conway’s Game of Life in which he said that the only way to calculate it with a Turing machine is to iterate over all possible universes and rule out the ones where it doesn’t happen by chance).
The point of dissolving the free will question was that it doesn’t matter what physics we run on. There is in fact no physics which could possibly cause me to believe I had “free will” in the sense of somehow determining my actions outside of physics, because any method for determining my actions is a physical process. In every possible consistent physics, where the laws of physics are more or less constant, I will believe I have “free will” in the sense that the output of my brain corresponds to the algorithm I feel like I’m implementing. It’s logically impossible for me to determine my own algorithm, because I’d need some starting algorithm to select one, recurring infinitely. Not to mention that when I say “me” I’m really just referring to the algorithm I’m currently running anyway.
This only makes sense because you’ve heavily internalized how our actual universe seems to work. There’s no reason a priori that you couldn’t have a universe with ontologicaly irreducible mental entities which aren’t subject to any restrictions on their actions (this is not too distant from the suggested idea of “freebits”).
Even if I was an ontologically basic mental entity, it remains true that I did not determine my own algorithm, I merely execute it. I mean, “freedom” to do anything doesn’t actually feel like freedom unless it’s curiously structured so that I can do precisely those things I choose to do and no others. Added randomness makes me a lot less confident in my free will, because then my actions might not line up with what I would otherwise have chosen. There aren’t supposed to be any possible worlds in which I make a different decision (given identical perception), merely worlds in which a different “I” is there to make the decision. I can counter factually imagine making all sorts of decisions, but which one I will actually make ought to be determined by what algorithm I actually implement; and if I don’t implement any consistent algorithm then I ought not to feel like I have free will, since I will notice that my decisions are based on chance rather than on my own preferences and decision algorithm. “Freedom” to do anything is meaningless; either I actually do things at random, or I am constrained by my own thought processes (which ought to be consistent and which cannot have been originally caused by me).
Even if I was an ontologically basic mental entity, it remains true that I did not determine my own algorithm
Among other issues, this assumes that you have an algorithm. It looks like stuff in our universe works that way, and it is important to develop intuitions (in so far as intuitions are internalized experience) that actually reflect how our universe works, but that shouldn’t stop you from imagining very different universes.
I mean, if I don’t have a consistent algorithm (Turing-computable or not) then I won’t feel like I have free will, at least not in the sense that I do right now. Unpredictable is equivalent to random; and if I’m making random decisions then I won’t feel like my decisions match up to some internal narrative. The more I notice that my decisions are random, unpredictable in foresight, the more I will feel like I have no free will (more precisely, feel like I have no will rather than that it is not free). But I’m not sure it’s even coherent to have those sorts of feelings without implementing some kind of consistent algorithm in the background (I’m not sure it isn’t, either, but it certainly feels like there’s a potential problem there).
Not to mention, even if I do not implement any consistent algorithm, it does not follow that “I” (whatever on earth “I” can actually mean in such a case) am able to determine my own decisions. Unpredictable decisions do not in any way support the idea of me having free will, or that whatever is determining my decisions is itself a mental entity which has free will.
I suspect that you are using free will in a way that’s very different from how many people use the term or have intuitions about it. Many when discussing free will emphasize lack of predictability as a basic part. Have you read Scott’s piece? He discusses why that makes sense in some detail. Maybe I should ask what do you think free will would look/feel like?
I think that the initial knee jerk intuitions most people have about free will are incoherent. It wouldn’t look like anything if they were actually true, because they are logically incoherent; there is no way the world could be where a conscious entity decides its own actions, unless potentially cyclic causality is involved (a quick example of a problem, though perhaps not unsolvable: if I determine all of my own actions, and my state of mind determines my actions (this must be true in order for me to feel like I have free will, regardless of whether I do for some definition), then I must determine my own state of mind; and how did I determine the very first state I was in?). However, that’s a very different question to why people, including me, feel like we have free will. The feeling of free will is pretty well linked to our feeling of being able to conceive of counterfactuals in which we make all sorts of different decisions, and then decide between those decisions. Regardless of how over decided (read: predictable) our decision is, we feel like we could, if we just wanted to, make a different decision. The key is that we can’t actually make ourselves want to make a different decision from the one we do in fact want to make. We can imagine that we could want something different, because we don’t know yet what choice we will want to make, but at no point can we actually change which decision we want to make, in a way which does not regress to us wanting to do so.
I also hold, though this was not explored on Less Wrong that I remember, that the existence of a consistent internal narrative is key for free will. Without it we would feel like we were not at the least in complete control of our decisions; we would decide to do one thing but then do another, or remember doing things but not be able to understand why. To the extent these phenomena actually happen in real life, it seems that this holds (and it certainly seems to hold in fantasy, where mind control is often illustrated as feeling like this).
I should also note that while do not hold what I understand to be a standard Compatibilist conception of free will, Compatibilism certainly also holds that unpredictability is not a requirement for free will. Certainly this is not a new idea, and at least a part of my understanding does fall into standard Compatibilism as I understand it. My views are also derived from the free will subsequence here on LW. These ideas are debatable, but they are certainly not all that special. Perhaps I was assuming too little inferential difference; I didn’t attempt to derive the entire argument here, not did I give a link to the sequence which formed my beliefs; I think I assumed that many people would notice the connection, but perhaps not.
You’re straw manning. An entity controlling its own behaviour is so non contradictory that there is a branch of engineering dedicated to it: cybernetics.
Free will isn’t generally defined as necessarily operating outside of physics, so that is more an attack on a strawmman than a dissolution of the question.
Free will also doesn’t mean a feeling.
If you didn’t k ow the output of you’re algorithm, you might feel free, and if you had free will you might feel free too. EYs solution isn’t unique.
The point of dissolving the free will question was that it doesn’t matter what physics we run on. There is in fact no physics which could possibly cause me to believe I had “free will” in the sense of somehow determining my actions outside of physics, because any method for determining my actions is a physical process. In every possible consistent physics, where the laws of physics are more or less constant, I will believe I have “free will” in the sense that the output of my brain corresponds to the algorithm I feel like I’m implementing. It’s logically impossible for me to determine my own algorithm, because I’d need some starting algorithm to select one, recurring infinitely. Not to mention that when I say “me” I’m really just referring to the algorithm I’m currently running anyway. I accept (as I think everyone does intuitively) that I exist due to forces outside my control. My mental algorithm was shaped over time by forces I could not control. Even now my algorithm can be changed by events outside my control, although this usually happens only in very small degrees and/or with consent from my then-current algorithm. But when I actually make a decision, the output of my body consistently lines up to what feels like the output of my mental algorithm. That’s what free will means, that’s where the feeling comes from; when I act, my thoughts and actions are in sync. That’s how free will gets dissolved, physics be damned. In fact, randomness and unpredictability makes me feel less in control than otherwise. If I really have free will, I ought to be perfectly predictable! You could just simulate the algorithm I run on, to whatever degree that’s actually accurate to how my mental architecture actually operates. If my decisions are unpredictable, even to me, then I start to worry about getting some fundamental part of me randomly erased, or finding that my mental narrative is a lie and I don’t control my own actions. Since this doesn’t happen (much) I can fairly safely say that random factors are not strongly important in my mental processes, though it’s possible I could be mistaken (humans are good at making up consistent narratives).
I don’t think this is a valid argument without the premise that if my will is a physical cause, it must itself be subject to physical causality. In other words, why couldn’t my will be a ‘one-way’ physical cause, able to cause things in the physical world, but unable to be affected by the physical world? Are you certain that every possible physics excludes this kind of one-way causation? Because while the idea strikes me as wildly implausible, it doesn’t seem to be contradictory or anything.
I haven’t seen an argument anywhere in the sequences (or elsewhere) defending this premise. This has always bothered me, so I’d appreciate it if you could supply the argument or point me to wherever I may have missed it.
The simplest examples of one-way causes may be the laws of physics. They cause the physical universe to have its properties but the universe does not cause them to exist or affect their nature. Theoretically there could be other “laws of wills” governing our behavior in a similar way but I would hesitate calling them actual (or especially individual) wills because of their effective non-agency. Agents’ behavior is caused by interaction with the physical universe, whereas the nature of laws is apparently not caused by interaction with the physical universe. A one-way will would be completely sensory-deprived and thus lack effective agency.
I think this is a very interesting thought, one famously articulated by Kant: the CI is essentially a law in the style of natural law, only pertaining to the will. He agrees with you that the law can’t be identified with each individual will (for one thing, some of us are bad or irrational). This avoids the ‘sensory deprivation’ problem, but keeps the idea that insofar as we’re governed by the law of the will, we’re free. The result is that we’re free only to the extent that we’re good.
Actually, that premise isn’t needed. No matter what the causes are, or even if I am an ontologically basic mental entity, it still remains true that I did not cause myself. I may not have been caused by anything else, but I certainly didn’t determine which algorithm is making my decisions. Not to mention that the rest of the world has to affect my will somehow, or I couldn’t actually perceive it or act on it intentionally (it’s a simple inversion of the argument against epiphenominalism). One-way causation is easily possible; I could write a computer program that worked like that. But the very act of writing the computer program violates the “free will” in the strict knee jerk reaction sense of “determining my own actions”. Determining your own actions requires cyclic causality, and even then I would struggle to accept that I really was determining my actions, rather than just saying that they happened basically by chance (I cannot recall where, but I recently saw something by EY about circular causality and time-turners in Conway’s Game of Life in which he said that the only way to calculate it with a Turing machine is to iterate over all possible universes and rule out the ones where it doesn’t happen by chance).
It sounds like the premise is not just needed, but quite complicated!
This only makes sense because you’ve heavily internalized how our actual universe seems to work. There’s no reason a priori that you couldn’t have a universe with ontologicaly irreducible mental entities which aren’t subject to any restrictions on their actions (this is not too distant from the suggested idea of “freebits”).
Even if I was an ontologically basic mental entity, it remains true that I did not determine my own algorithm, I merely execute it. I mean, “freedom” to do anything doesn’t actually feel like freedom unless it’s curiously structured so that I can do precisely those things I choose to do and no others. Added randomness makes me a lot less confident in my free will, because then my actions might not line up with what I would otherwise have chosen. There aren’t supposed to be any possible worlds in which I make a different decision (given identical perception), merely worlds in which a different “I” is there to make the decision. I can counter factually imagine making all sorts of decisions, but which one I will actually make ought to be determined by what algorithm I actually implement; and if I don’t implement any consistent algorithm then I ought not to feel like I have free will, since I will notice that my decisions are based on chance rather than on my own preferences and decision algorithm. “Freedom” to do anything is meaningless; either I actually do things at random, or I am constrained by my own thought processes (which ought to be consistent and which cannot have been originally caused by me).
Among other issues, this assumes that you have an algorithm. It looks like stuff in our universe works that way, and it is important to develop intuitions (in so far as intuitions are internalized experience) that actually reflect how our universe works, but that shouldn’t stop you from imagining very different universes.
I mean, if I don’t have a consistent algorithm (Turing-computable or not) then I won’t feel like I have free will, at least not in the sense that I do right now. Unpredictable is equivalent to random; and if I’m making random decisions then I won’t feel like my decisions match up to some internal narrative. The more I notice that my decisions are random, unpredictable in foresight, the more I will feel like I have no free will (more precisely, feel like I have no will rather than that it is not free). But I’m not sure it’s even coherent to have those sorts of feelings without implementing some kind of consistent algorithm in the background (I’m not sure it isn’t, either, but it certainly feels like there’s a potential problem there).
Not to mention, even if I do not implement any consistent algorithm, it does not follow that “I” (whatever on earth “I” can actually mean in such a case) am able to determine my own decisions. Unpredictable decisions do not in any way support the idea of me having free will, or that whatever is determining my decisions is itself a mental entity which has free will.
I suspect that you are using free will in a way that’s very different from how many people use the term or have intuitions about it. Many when discussing free will emphasize lack of predictability as a basic part. Have you read Scott’s piece? He discusses why that makes sense in some detail. Maybe I should ask what do you think free will would look/feel like?
I think that the initial knee jerk intuitions most people have about free will are incoherent. It wouldn’t look like anything if they were actually true, because they are logically incoherent; there is no way the world could be where a conscious entity decides its own actions, unless potentially cyclic causality is involved (a quick example of a problem, though perhaps not unsolvable: if I determine all of my own actions, and my state of mind determines my actions (this must be true in order for me to feel like I have free will, regardless of whether I do for some definition), then I must determine my own state of mind; and how did I determine the very first state I was in?). However, that’s a very different question to why people, including me, feel like we have free will. The feeling of free will is pretty well linked to our feeling of being able to conceive of counterfactuals in which we make all sorts of different decisions, and then decide between those decisions. Regardless of how over decided (read: predictable) our decision is, we feel like we could, if we just wanted to, make a different decision. The key is that we can’t actually make ourselves want to make a different decision from the one we do in fact want to make. We can imagine that we could want something different, because we don’t know yet what choice we will want to make, but at no point can we actually change which decision we want to make, in a way which does not regress to us wanting to do so.
I also hold, though this was not explored on Less Wrong that I remember, that the existence of a consistent internal narrative is key for free will. Without it we would feel like we were not at the least in complete control of our decisions; we would decide to do one thing but then do another, or remember doing things but not be able to understand why. To the extent these phenomena actually happen in real life, it seems that this holds (and it certainly seems to hold in fantasy, where mind control is often illustrated as feeling like this).
I should also note that while do not hold what I understand to be a standard Compatibilist conception of free will, Compatibilism certainly also holds that unpredictability is not a requirement for free will. Certainly this is not a new idea, and at least a part of my understanding does fall into standard Compatibilism as I understand it. My views are also derived from the free will subsequence here on LW. These ideas are debatable, but they are certainly not all that special. Perhaps I was assuming too little inferential difference; I didn’t attempt to derive the entire argument here, not did I give a link to the sequence which formed my beliefs; I think I assumed that many people would notice the connection, but perhaps not.
You’re straw manning. An entity controlling its own behaviour is so non contradictory that there is a branch of engineering dedicated to it: cybernetics.
There’s plenty of evidence that people rationalise decisions after the event, so you would have a feeling of narrative under any circumstances.
If there are N things you might want to do, why not make a random choice between them?
You may be constrained, but that doesn’t imply constrained down to one.
Free will isn’t generally defined as necessarily operating outside of physics, so that is more an attack on a strawmman than a dissolution of the question.
Free will also doesn’t mean a feeling.
If you didn’t k ow the output of you’re algorithm, you might feel free, and if you had free will you might feel free too. EYs solution isn’t unique.