Thanks for this post. I've very much enjoyed some of Yudkowsky's work, especially the brilliant _Harry Potter and the Methods of Rationality_. I was unaware of his very implausible views that you mentioned above.
Regarding the Zombie argument, it sounds as if he might be unaware of the meaning of "metaphysical possibility". Most people who haven't studied philosophy use "possible" in a sense closer to physical or nomological possibility. So he probably thinks that the Zombie argument uses the premise that Zombies are nomologically possible. He'd almost be correct to think that, if consciousness is in fact efficacious, then Zombies are *nomologically* impossible. I guess that's what's going on.
I have to disagree with you on the zombie argument. You say you can imagine a world where everything- down to atoms- is same, but there is no Consciousness. But this is impossible; it is the same as the case of married bachelors and square circle. I think if everything, down to atoms, is same, then consciousness automatically arises. It is a resultant property. I cannot see how this thought experiment disproves physicalism.
When we get to the zombie argument premise I got lost a bit. When you claim that you find the zombie argument convincing, you have for me done nothing to set that up. As you say:
> They sure seem possible. I can quite vividly imagine a version of me that continues through its daily goings-on but that lacks consciousness.
I cannot. I do struggle to argue against this point as it is a matter of imagination, but I feel (perhaps incorrectly?) that I can push it aside for that reason.
> It’s very plausible that if something is impossible, there should be some reason that it is impossible
Perhaps, but I don't think this compels or even moves me in the direction of accepting the premise above.
This is of course fine, but as it is the first detailed disagreement with EY, I thought it poignant that I got lost at exactly that point and wanted to note it. I'm not saying EYs position in the paragraphs to follow make sense to me either, I just fail to be moved by your arguments till this point as well.
Hope I'm not coming across as too negative, I'll update when I'm done reading!
(For further context about my beliefs: I personally lean towards physicalism, probably partially because of EY, but I find multiple non physicalist positions interesting and have not committed either way)
So the evidence for him being definitely and egregiously wrong is disagreeing with you on two points where it is not at all clear from the provided text that you understand his arguments.
Re: the zombie argument, maybe you’ve written more on this elsewhere, but what makes you confident that you can actually imagine zombies? I.e. you think you can imagine a world that is physically identical but in which no physical structures are conscious, but how do you know your imagination is actually possible?
I know this is tangential at best, but my new favorite illustration of the difference between causal and evidential decision theory is one I heard on Sean Carroll’s podcast a couple of weeks ago: Calvinist predestination. If your fate in the afterlife is fixed before you’re born, a causal decision theorist will think it’s pointless to do a bunch of pious stuff to try to please God, but an evidential decision theorist will disagree.
I think this can also be extended to correspondence theory, scientific realism, the normativity of classical logic, and maybe moral realism (although it's hard to tell if he really is one since he invents his own jargon and it's not always clear what he means).
If you're really interested I can write a critique of these positions.
> And Eliezer does often offer good advice. He is right that people often reason poorly, and there are ways people can improve their thinking. Humans are riddled by biases, and it’s worth reflecting on how that distorts our beliefs.
Hey, you don't get to dump those all other criticisms and then say : "but there is that ONE thing where he doesn't make the same mistakes that he keeps making everywhere else (and by the way, I am not as specialist in that field (?))" :p
If the brains capability to model itself as a requirement for consciousness is understood as the capability to self-referentialism, then it is most certainly correct. What makes it so incredibly different adjucate between different modes of consciousness is the absence of systems theory in the American discourse, and it is this absence that is to blame for 95% of those scientific debates that have been running endless circles around themselves for decades without ever bearing fruit. Read Luhmann, it's about as fruitful as it gets.
Unless a physicalist denied the existence of consciousness, they would not believe that if a block is moving right while thinking "I want to move right," it would describe all the physical facts to say the block is moving right. For physicalists who don't deny consciousness, the block's thinking "I want to move right" is also a physical fact, so that would need to be stated too. According to the block analogy as you've phrased it, physicalists are all consciousness deniers.
The descriptions you've attributed to the epiphenomenalist and interactionist are both descriptions the non-consciousness-denying physicalist would prefer over the description you've attributed to the physicalist.
I mostly agree with your last two points but I think the first isn't quite "egregiously" wrong.
If I understand correctly, your main point about zombies is that although Eliezer's argument is a good one against epiphenomenalism, it doesn't exclude all non-physicalist theories of consciousness. However:
- Of the two alternatives Chalmers lists, type-D interactionism (~dualism) and type-F monism, Eliezer argued at length against the former in other places. He treats dualism separately from epiphenomenalism in his essay: "Why not postulate the true stuff of consciousness which no amount of mere mechanical atoms can add up to, and then, having gone that far already, let this true stuff of consciousness have causal effects like making philosophers talk about consciousness?"
- So, it seems that the main problem is that he hasn't addressed type-F monism. He does bring this up in his reply to Chalmers:
>Type-F monism is a bit harder to grasp, but presumably, on this view, it is not possible for anything to be real at all without being made out of the stuff of consciousness, in which case the zombie world is structurally identical to our own but contains no consciousness by virtue of not being real, nothing to breathe fire into the equations. If you can subtract the monist consciousness of the electron and leave behind the electron's structure and have the structure still be real, then that is equivalent to property dualism or E. This gets us into a whole separate set of issues, really; but I wonder if this isn't isomorphic to what most materialists believe. After all, presumably the standard materialist theory says that there are computations that could exist, but don't exist, and therefore aren't conscious. Though this is an issue on which I confess to still being confused.
But now we've shifted from "The argument is completely egregiously wrong" to "The argument successfully challenges epiphenomenalism, but it assumes that Chalmers is a strong advocate of E when he actually prefers D and F, and it doesn't conclusively prove that physicalism is correct because he doesn't also address type-F monism (although he does argue against D elsewhere and is at least somewhat open to F)." These are still significant errors, especially the first, but they're not as catastrophic as the headline implies, IMO.
I wonder whether Yudkowsky got his false interpretation of the zombie-argument from Sean Carroll: Carroll has made the claim that the P-zombie argument actually SUPPORTS physicalism, because consciousness would have no explanatory role otherwise
On the Smoking Lesion case: Are you familiar with Arif Ahmed's tickle defence? The idea is that the fact that you WANT to smoke is already evidence that you have the gene, so your final decision to smoke doesn't give you any further "bad news". So, as the case is formulated, it's just false that EDT recommends refraining from smoking (there are ways to re-formulate the case to make it more tricky for EDT, but I think Ahmed has good responses to those as well)
Rebbetzin Devorah Fastag's recent responses to his 2005 writing can be found on my substack. You gotta respect a guy who barely writes anything, but whatever he writes becomes timeless and universally referred to.
Thanks for this post. I've very much enjoyed some of Yudkowsky's work, especially the brilliant _Harry Potter and the Methods of Rationality_. I was unaware of his very implausible views that you mentioned above.
Regarding the Zombie argument, it sounds as if he might be unaware of the meaning of "metaphysical possibility". Most people who haven't studied philosophy use "possible" in a sense closer to physical or nomological possibility. So he probably thinks that the Zombie argument uses the premise that Zombies are nomologically possible. He'd almost be correct to think that, if consciousness is in fact efficacious, then Zombies are *nomologically* impossible. I guess that's what's going on.
I have to disagree with you on the zombie argument. You say you can imagine a world where everything- down to atoms- is same, but there is no Consciousness. But this is impossible; it is the same as the case of married bachelors and square circle. I think if everything, down to atoms, is same, then consciousness automatically arises. It is a resultant property. I cannot see how this thought experiment disproves physicalism.
(midway comment, Haven't finished it yet)
When we get to the zombie argument premise I got lost a bit. When you claim that you find the zombie argument convincing, you have for me done nothing to set that up. As you say:
> They sure seem possible. I can quite vividly imagine a version of me that continues through its daily goings-on but that lacks consciousness.
I cannot. I do struggle to argue against this point as it is a matter of imagination, but I feel (perhaps incorrectly?) that I can push it aside for that reason.
> It’s very plausible that if something is impossible, there should be some reason that it is impossible
Perhaps, but I don't think this compels or even moves me in the direction of accepting the premise above.
This is of course fine, but as it is the first detailed disagreement with EY, I thought it poignant that I got lost at exactly that point and wanted to note it. I'm not saying EYs position in the paragraphs to follow make sense to me either, I just fail to be moved by your arguments till this point as well.
Hope I'm not coming across as too negative, I'll update when I'm done reading!
(For further context about my beliefs: I personally lean towards physicalism, probably partially because of EY, but I find multiple non physicalist positions interesting and have not committed either way)
So the evidence for him being definitely and egregiously wrong is disagreeing with you on two points where it is not at all clear from the provided text that you understand his arguments.
" I know nothing about quantum physics, and he sounds persuasive when talking about quantum physics. "
I know a lot about QM, and he's confused and over confident about that too. Also Solomonoff induction, Aumamns theorem, and Bayes
Re: the zombie argument, maybe you’ve written more on this elsewhere, but what makes you confident that you can actually imagine zombies? I.e. you think you can imagine a world that is physically identical but in which no physical structures are conscious, but how do you know your imagination is actually possible?
I know this is tangential at best, but my new favorite illustration of the difference between causal and evidential decision theory is one I heard on Sean Carroll’s podcast a couple of weeks ago: Calvinist predestination. If your fate in the afterlife is fixed before you’re born, a causal decision theorist will think it’s pointless to do a bunch of pious stuff to try to please God, but an evidential decision theorist will disagree.
TL;DR Calvinists are one-boxers.
It's so over for LessWrongcels
I agree, and you might want to downvote this post: https://forum.effectivealtruism.org/posts/2S3CHPwaJBE5h8umW/read-the-sequences?commentId=9J76kQkF8k86zcyWF
I think this can also be extended to correspondence theory, scientific realism, the normativity of classical logic, and maybe moral realism (although it's hard to tell if he really is one since he invents his own jargon and it's not always clear what he means).
If you're really interested I can write a critique of these positions.
> And Eliezer does often offer good advice. He is right that people often reason poorly, and there are ways people can improve their thinking. Humans are riddled by biases, and it’s worth reflecting on how that distorts our beliefs.
Hey, you don't get to dump those all other criticisms and then say : "but there is that ONE thing where he doesn't make the same mistakes that he keeps making everywhere else (and by the way, I am not as specialist in that field (?))" :p
Any zététitians around ?
P.S.: bonuses about Jordan Peterson :
https://web.archive.org/web/20181224093855/https://medium.com/s/story/peterson-historian-aide-m%C3%A9moire-9aa3b6b3de04
https://ndpr.nd.edu/reviews/french-theory-how-foucault-derrida-deleuze-co-transformed-the-intellectual-life-of-the-united-states/
https://medium.com/@Corax/argue-like-jordan-peterson-265e4c11b235
If the brains capability to model itself as a requirement for consciousness is understood as the capability to self-referentialism, then it is most certainly correct. What makes it so incredibly different adjucate between different modes of consciousness is the absence of systems theory in the American discourse, and it is this absence that is to blame for 95% of those scientific debates that have been running endless circles around themselves for decades without ever bearing fruit. Read Luhmann, it's about as fruitful as it gets.
Unless a physicalist denied the existence of consciousness, they would not believe that if a block is moving right while thinking "I want to move right," it would describe all the physical facts to say the block is moving right. For physicalists who don't deny consciousness, the block's thinking "I want to move right" is also a physical fact, so that would need to be stated too. According to the block analogy as you've phrased it, physicalists are all consciousness deniers.
The descriptions you've attributed to the epiphenomenalist and interactionist are both descriptions the non-consciousness-denying physicalist would prefer over the description you've attributed to the physicalist.
I mostly agree with your last two points but I think the first isn't quite "egregiously" wrong.
If I understand correctly, your main point about zombies is that although Eliezer's argument is a good one against epiphenomenalism, it doesn't exclude all non-physicalist theories of consciousness. However:
- Of the two alternatives Chalmers lists, type-D interactionism (~dualism) and type-F monism, Eliezer argued at length against the former in other places. He treats dualism separately from epiphenomenalism in his essay: "Why not postulate the true stuff of consciousness which no amount of mere mechanical atoms can add up to, and then, having gone that far already, let this true stuff of consciousness have causal effects like making philosophers talk about consciousness?"
- So, it seems that the main problem is that he hasn't addressed type-F monism. He does bring this up in his reply to Chalmers:
>Type-F monism is a bit harder to grasp, but presumably, on this view, it is not possible for anything to be real at all without being made out of the stuff of consciousness, in which case the zombie world is structurally identical to our own but contains no consciousness by virtue of not being real, nothing to breathe fire into the equations. If you can subtract the monist consciousness of the electron and leave behind the electron's structure and have the structure still be real, then that is equivalent to property dualism or E. This gets us into a whole separate set of issues, really; but I wonder if this isn't isomorphic to what most materialists believe. After all, presumably the standard materialist theory says that there are computations that could exist, but don't exist, and therefore aren't conscious. Though this is an issue on which I confess to still being confused.
But now we've shifted from "The argument is completely egregiously wrong" to "The argument successfully challenges epiphenomenalism, but it assumes that Chalmers is a strong advocate of E when he actually prefers D and F, and it doesn't conclusively prove that physicalism is correct because he doesn't also address type-F monism (although he does argue against D elsewhere and is at least somewhat open to F)." These are still significant errors, especially the first, but they're not as catastrophic as the headline implies, IMO.
I wonder whether Yudkowsky got his false interpretation of the zombie-argument from Sean Carroll: Carroll has made the claim that the P-zombie argument actually SUPPORTS physicalism, because consciousness would have no explanatory role otherwise
On the Smoking Lesion case: Are you familiar with Arif Ahmed's tickle defence? The idea is that the fact that you WANT to smoke is already evidence that you have the gene, so your final decision to smoke doesn't give you any further "bad news". So, as the case is formulated, it's just false that EDT recommends refraining from smoking (there are ways to re-formulate the case to make it more tricky for EDT, but I think Ahmed has good responses to those as well)
Rebbetzin Devorah Fastag's recent responses to his 2005 writing can be found on my substack. You gotta respect a guy who barely writes anything, but whatever he writes becomes timeless and universally referred to.