err, disagree? free will is logical uncertainty about your own choice. it’s not written until you write it, which you were always going to do. then you find out you wrote it.
I’m curious—is this your attempt to “do your homework”, according to Eliezer? I ask because it doesn’t seem like you’re addressing the question he actually asked. I would really try to do that before opening this spoiler. The question is: what is the belief in free will? How does it work? What does it actually do?
I see people converge onto this kind of view all the time, but I have to admit (a strong personal failing!) that I still don’t understand how it happens. When you say you feel like you have free will, what do you actually mean by that? It doesn’t seem the case to me—certainly, I don’t think I feel that way—that people have a self-evident sense that their decision-making ability exists as an Ultimate Sourceless Cause. People say things like “I feel like I can choose between options, and I feel like I have agency, I can do things, I can direct my actions”. Why is it self-evident that these are claims that someone can break physics to so many people? Do these words just have an “air of magic” to them? It just seems straightforwardly clear to me that this is no more incompatible with physics than the idea that there is a chair over there; certainly a lossy abstraction, but a lossy abstraction that is an attempt to give a rough description of a real underlying thing. There is a real entity (an assemblage of matter configured in a certain way) implementing all “your” decision-processes, and that entity is you. Ergo, you make the choices. A billiard ball’s being pushed by another ball doesn’t make its pushing yet another billiard ball any less pushy.
I wasn’t attempting to do Eliezer’s homework. I wasn’t even aware it existed when I wrote this post. What happened is I wrote this post and then, while tagging it, I discovered that Eliezer had written about the subject and there was a prominent puzzle that I might be spoiling (part of) the answer to. I added the spoiler just because I didn’t want to contribute any hints.
I honestly never understood the debate around free will. The brain is basically a computer that makes decisions about actions. You are the thing that makes the decision (depending on how you define The Self, of course.) It would be silly to say of humans “there is no will”, and I don’t really care whether that will is “free” according to some irrelevant definition.
Does resolving an irrelevant semantic puzzle help us solve alignment?
EDIT: Hopefully I’m not coming off as snarky or argumentative. This is a topic where I’m honestly confused about why rationalists (non-dualists) would see any mystery or puzzle. Free will is only a puzzle if you are committed to both scientific materialism and dualism, two views which are obviously incompatible anyway, although many moderns try to keep a foot in both camps. Since I doubt there are very many dualists on this forum, I just don’t see why anyone here would be mystified.
I think there are some important aspects to it; if you want to develop your moral thinking about personal responsibility and blame, you might want to reflect a lot on what “freedom” means to you. You might also not care about it at all due to some other a priori considerations, but in my experience this is the sort of thing that can push people one way or the other in ethics.
Meditators who find their feeling of free will and self “dissolved’ aren’t noted to act any differently by their family/coworkers afterwards, a fact that many find very confusing and upsetting. So if there isn’t a difference in behavior (other than talking about meditation a lot), why would there be a selection advantage?
First of all, talking about meditation a lot probably is selected against (at least slightly). Even slight selection effects add up rapidly over evolutionary timescales.
But I don’t think that’s the crux of the issue. In general, meditators who achieve enlightenment or who “pull up hatred by the roots” also don’t show sudden major changes in longterm behavior.
Suppose instead of “free will” that we were discussing dukkha (roughly “suffering”).
Dukkha serves an obvious evolutionary purpose. Dukkha (just like free will) is a useful component to motivating intentional behavior. But advanced meditators frequently transcend dukkha. When they do, they tend to keep doing basically what they have always been doing (with some flamboyant exceptions that are not relevant to this discussion). Why? Because they don’t suddenly forget all the good habits and mental models they built up over many years of living.
I think the illusion of free will, just like dukkha, is most important to brains without strong meta-cognition. Especially young ones.
OK, I looked at some of the research for enlightenment again, and it looks like the biggest problem associated with enlightenment is severe memory issues? That would explain why it would be selected against—memory is pretty important!
And I think I understand what you’re saying. Sarkic pain/pleasure is required to teach new minds behaviors and patterns. Children born without pain reception tend to do things like chew off their own limbs for fun. Removing the grounding for pain/pleasure doesn’t affect the learned behaviors much once they’re fully formed.
Some have found negative effects, which look to me like what I would expect of someone who had actually expunged their sense of self. The subjects sound like burnt-out potheads.
Curiously, people are willing to impute thought, choice, decision-making, understanding, purpose, action to achieve purposes, and so on, to machines like the GPT family, yet are often diffident or hostile to attributing these things to people. I find this very strange.
I left “free will” out of that list, because it is not clear to me what people are asserting, when they say they have “free will”, or what they are denying, when they say they do not.
We can all agree that there is no such thing as a unicorn. We can do that because we all know what the word means: a supposed horse-like creature with a single horn in its forehead, and having magical attributes variously told of. Because we know what we mean, we would recognise one if we found one. But we know it is only a legend. Explorers have never found any such creature, not even something close enough to have inspired the legend. There is no such thing as a unicorn.
But when speaking of free will, whether to assert it or deny it, people describe it only in phrases that seem to point at the same thing in different words, whatever that thing is. “Choice”, “the possibility of having done differently”, and so on. Contrast “unicorn”, which is described in terms that make no reference to unicorns.
Thus “free will” is the name given to a certain subjective experience that people are unable to articulate any further, the experience of acting in the world. When they try to articulate it they confabulate non-answers, answers that no more answer the question “what is free will?” than “gravity” answers “why do things fall?”, or “energy” answers “how does a clockwork toy work?”, or “fluency” answers “how do you speak a foreign language well?”[1] These non-answers are easily shown to be empty or absurd, but this does not negate the existence of the phenomena that they fail to elucidate. Things fall, clockwork toys move, and mastery of a foreign language is possible.
People clearly do think, choose, make decisions, understand things, have purposes, act to achieve them, and so on. Given that, what is left to talk about on the subject of “free will”? Free will is the quale of doing such things, and can be no more articulated than the sensation of hearing a trumpet.
I don’t understand the counterfactual you’re using to declare the false belief “useful”. Are you saying a p-zombie would somehow behave differently, or that it’s impossible, or something else?
I mostly bite the bullet that it’s an illusion, but I suspect it’s a side-effect of the efficient behavior encoding in a human brain, not a separable thing that can be evaluated as to utility. I do not know if other active deterministic structures have or need such an illusion.
One mind knows that the universe is purely deterministic
Some thoughts:
The word “knows” does a lot of work here. What does it mean for a brain to know that it is deterministic? I guess you appeal to an intuitive understanding, but that makes the argument weaker.
I also don’t think that the conclusion follows. Many organisms get a lot of things done without knowing it.
Evolution cannot work on concepts in the neocortex, and thus your argument doesn’t work unless you assume some mechanism for how the brain stem brings about the concepts in the neocortex. In this case, the argument is really about that mechanism and not about our high-level summary “knows itself”.
If your argument works, though, that would indicate a new (?) alignment strategy: Sufficiently strongly ensuring that the mind doesn’t produce self-awareness could reduce agency to zero.
The human brain is too big and too warm for quantum indeterminacy to matter
What you read was: the human brain is to big and too warm for large scale superpositions to form. Any single particle experiment amplifies quantum indeterminacy to the macroscopic level.
ETA
Most people feel like they have free will. But nobody actually does
What makes that true? Just the claim about quantum.mechanics?
err, disagree? free will is logical uncertainty about your own choice. it’s not written until you write it, which you were always going to do. then you find out you wrote it.
I’m curious—is this your attempt to “do your homework”, according to Eliezer? I ask because it doesn’t seem like you’re addressing the question he actually asked. I would really try to do that before opening this spoiler. The question is: what is the belief in free will? How does it work? What does it actually do?
I see people converge onto this kind of view all the time, but I have to admit (a strong personal failing!) that I still don’t understand how it happens. When you say you feel like you have free will, what do you actually mean by that? It doesn’t seem the case to me—certainly, I don’t think I feel that way—that people have a self-evident sense that their decision-making ability exists as an Ultimate Sourceless Cause. People say things like “I feel like I can choose between options, and I feel like I have agency, I can do things, I can direct my actions”. Why is it self-evident that these are claims that someone can break physics to so many people? Do these words just have an “air of magic” to them? It just seems straightforwardly clear to me that this is no more incompatible with physics than the idea that there is a chair over there; certainly a lossy abstraction, but a lossy abstraction that is an attempt to give a rough description of a real underlying thing. There is a real entity (an assemblage of matter configured in a certain way) implementing all “your” decision-processes, and that entity is you. Ergo, you make the choices. A billiard ball’s being pushed by another ball doesn’t make its pushing yet another billiard ball any less pushy.
I wasn’t attempting to do Eliezer’s homework. I wasn’t even aware it existed when I wrote this post. What happened is I wrote this post and then, while tagging it, I discovered that Eliezer had written about the subject and there was a prominent puzzle that I might be spoiling (part of) the answer to. I added the spoiler just because I didn’t want to contribute any hints.
Understandable! I recommend the “solution to free will” sequence, since you’ve gotten this far. The core stuff is here and here if you’re in a hurry.
I honestly never understood the debate around free will. The brain is basically a computer that makes decisions about actions. You are the thing that makes the decision (depending on how you define The Self, of course.) It would be silly to say of humans “there is no will”, and I don’t really care whether that will is “free” according to some irrelevant definition.
Freedom is the freedom to avoid things
The future where we call get killed by ASI is inevitable, if everything is inevitable. But some people care a lot about avoiding it.
I don’t see what that has to do with free will.
It’s a potential reason why it’s worth caring about.
But if you don’t care about anything, you don’t care about anything.
Does resolving an irrelevant semantic puzzle help us solve alignment?
EDIT: Hopefully I’m not coming off as snarky or argumentative. This is a topic where I’m honestly confused about why rationalists (non-dualists) would see any mystery or puzzle. Free will is only a puzzle if you are committed to both scientific materialism and dualism, two views which are obviously incompatible anyway, although many moderns try to keep a foot in both camps. Since I doubt there are very many dualists on this forum, I just don’t see why anyone here would be mystified.
Dualism is only relevant to free will under certain definitions,so the semantic issue is relevant.
I think there are some important aspects to it; if you want to develop your moral thinking about personal responsibility and blame, you might want to reflect a lot on what “freedom” means to you. You might also not care about it at all due to some other a priori considerations, but in my experience this is the sort of thing that can push people one way or the other in ethics.
Meditators who find their feeling of free will and self “dissolved’ aren’t noted to act any differently by their family/coworkers afterwards, a fact that many find very confusing and upsetting. So if there isn’t a difference in behavior (other than talking about meditation a lot), why would there be a selection advantage?
Good question.
First of all, talking about meditation a lot probably is selected against (at least slightly). Even slight selection effects add up rapidly over evolutionary timescales.
But I don’t think that’s the crux of the issue. In general, meditators who achieve enlightenment or who “pull up hatred by the roots” also don’t show sudden major changes in longterm behavior.
Suppose instead of “free will” that we were discussing dukkha (roughly “suffering”).
Dukkha serves an obvious evolutionary purpose. Dukkha (just like free will) is a useful component to motivating intentional behavior. But advanced meditators frequently transcend dukkha. When they do, they tend to keep doing basically what they have always been doing (with some flamboyant exceptions that are not relevant to this discussion). Why? Because they don’t suddenly forget all the good habits and mental models they built up over many years of living.
I think the illusion of free will, just like dukkha, is most important to brains without strong meta-cognition. Especially young ones.
OK, I looked at some of the research for enlightenment again, and it looks like the biggest problem associated with enlightenment is severe memory issues? That would explain why it would be selected against—memory is pretty important!
And I think I understand what you’re saying. Sarkic pain/pleasure is required to teach new minds behaviors and patterns. Children born without pain reception tend to do things like chew off their own limbs for fun. Removing the grounding for pain/pleasure doesn’t affect the learned behaviors much once they’re fully formed.
Some have found negative effects, which look to me like what I would expect of someone who had actually expunged their sense of self. The subjects sound like burnt-out potheads.
Curiously, people are willing to impute thought, choice, decision-making, understanding, purpose, action to achieve purposes, and so on, to machines like the GPT family, yet are often diffident or hostile to attributing these things to people. I find this very strange.
I left “free will” out of that list, because it is not clear to me what people are asserting, when they say they have “free will”, or what they are denying, when they say they do not.
We can all agree that there is no such thing as a unicorn. We can do that because we all know what the word means: a supposed horse-like creature with a single horn in its forehead, and having magical attributes variously told of. Because we know what we mean, we would recognise one if we found one. But we know it is only a legend. Explorers have never found any such creature, not even something close enough to have inspired the legend. There is no such thing as a unicorn.
But when speaking of free will, whether to assert it or deny it, people describe it only in phrases that seem to point at the same thing in different words, whatever that thing is. “Choice”, “the possibility of having done differently”, and so on. Contrast “unicorn”, which is described in terms that make no reference to unicorns.
Thus “free will” is the name given to a certain subjective experience that people are unable to articulate any further, the experience of acting in the world. When they try to articulate it they confabulate non-answers, answers that no more answer the question “what is free will?” than “gravity” answers “why do things fall?”, or “energy” answers “how does a clockwork toy work?”, or “fluency” answers “how do you speak a foreign language well?”[1] These non-answers are easily shown to be empty or absurd, but this does not negate the existence of the phenomena that they fail to elucidate. Things fall, clockwork toys move, and mastery of a foreign language is possible.
People clearly do think, choose, make decisions, understand things, have purposes, act to achieve them, and so on. Given that, what is left to talk about on the subject of “free will”? Free will is the quale of doing such things, and can be no more articulated than the sensation of hearing a trumpet.
The first two examples are from Feynman, and the last from my late mother, who often “explained” things in that manner.
I don’t understand the counterfactual you’re using to declare the false belief “useful”. Are you saying a p-zombie would somehow behave differently, or that it’s impossible, or something else?
I mostly bite the bullet that it’s an illusion, but I suspect it’s a side-effect of the efficient behavior encoding in a human brain, not a separable thing that can be evaluated as to utility. I do not know if other active deterministic structures have or need such an illusion.
lsusr, if it was proven that the human brain actually does work on quantum principles, how would that change your view on free will?
It would add technical complications to the logic but wouldn’t change the conclusion.
Some thoughts:
The word “knows” does a lot of work here. What does it mean for a brain to know that it is deterministic? I guess you appeal to an intuitive understanding, but that makes the argument weaker.
I also don’t think that the conclusion follows. Many organisms get a lot of things done without knowing it.
Evolution cannot work on concepts in the neocortex, and thus your argument doesn’t work unless you assume some mechanism for how the brain stem brings about the concepts in the neocortex. In this case, the argument is really about that mechanism and not about our high-level summary “knows itself”.
If your argument works, though, that would indicate a new (?) alignment strategy: Sufficiently strongly ensuring that the mind doesn’t produce self-awareness could reduce agency to zero.
What you read was: the human brain is to big and too warm for large scale superpositions to form. Any single particle experiment amplifies quantum indeterminacy to the macroscopic level.
ETA
What makes that true? Just the claim about quantum.mechanics?