the hard problem is self-inflicted

this stuff is hard. but it's not that kind of hard.

consciousness is pretty weird. clearly humans are very different from almost everything else in the world. mind and thought have incredible unplumbed depths and we’re only just starting to get anything like an understanding of the brain. but you guys really have no reason to keep making it all pointlessly harder than it actually is by asserting fake problems. the hard problem of consciousness is a fake problem. it’s an assertion of intuition, and by no means is it a universal intuition.

there are many understandings of the hard problem and qualia, so it’s a little harder to argue against all of them at once in an article rather than your specific ideas in a twitter thread, but i’ll try anyway. most conceptions of the hard problem run something like this: when a human brain sees the color red or puts its hand on a hot stove, there is “something it is to be like” that experience, that person, in that moment, above and beyond any detailed complete description of the location and motion of every atom in the human. why does that physical state “cause” or “give rise to” that experience, or qualia? is it not at least conceivable that this same physical state of the world might have no experience at all?

sorry. nope. the experience is the brainstate. it identifies, one to one.

i’m absolutely not saying pain is fake or experiences aren’t real. i see red just as much as you do, i can feel the keyboard beneath my fingers right now. i run with the grass beneath my toes and the wind in my hair and i feel fucking fantastic knowing i’m alive. my hand is in front of my face, just as much as yours. our first person experiences are all primary and real.

none of that stops them being exactly brainstates.

let’s talk about p-zombies for a moment.

the claim is that it’s possible to conceive of a human, physically identical to you, but without any conscious subjective experience whatsoever. and even if it isn’t physically realizable in our universe, the fact that it’s possible to conceive of it means consciousness and the physical neural reality must be separate things.

this is transparent bullshit. claiming that it’s possible to conceive of this is just directly asserting the conclusion! this attempt at a thought experiment is literally just asserting that neural states are not consciousness, but laundering it through our confusions with definitions.

if i say “you can imagine water which is physically identical to our water but is not H2O”, you can totally “imagine” this. but it’s a false shallow form of imagining born of your ignorance and inability to minutely model chemistry and physics. all you are actually doing is imagining a glass of a clear liquid and tagging it “not-H2O” mentally. the concepts of H2O and the properties of water, which are in fact just the same thing, seem different to you because you’re introduced to one before you can even remember and the other years later in chemistry class as a stick and ball diagram. the fact that you can imagine confused and false worldstates does not mean “water” and “H2O” are necessarily ontologically distinct, and only potentially correlated. it’s a fact about you, not the world. p-zombies are not actually “conceivable”.

at this point people tend to say something like “well you’ve done a lot of asserting but i don’t really find myself any more convinced brainstates must just be qualia or that you’ve actually explained why”

that’s the thing man. this is how gathering evidence and understanding the world works. sure, we can always descend into solipsism. sure we can always say “well i have no proof of anything but my own subjective experiences, so nothing can ever tell me anything about them”. but that just definitionally leads you to nowhere and nothing. so you might as well make philosophical choices that allow you to actually learn things about the world.

when you want to know what water is, and whether it’s truly the same thing as H2O, you test it and poke it and prod it and theorize about it. you gain correlational information. and eventually, you conclude that these things are in fact one and the same. not that “H2O somehow produces water”. that H2O is water. and you do this strictly by observing the world. in every other case, this standard of evidence is enough. correlation and theory. observing the morning star’s trajectory and the evening star and noticing the pattern and forming a hypothesis, and eventually just looking at the damn thing directly through a telescope. you realize it’s just venus, it’s just been venus this whole time. there is no particular reason for your standard of evidence to be so different for brainstates and experiences!

we actually have a huge amount of evidence that these are just the same thing, far more than any normal standard demands. you shove electrodes in your brain and you see red and feel pain and joy and love and hate. we find the right area of neural matter and turn off your vision. or your ability to remember your mother. we see the patterns and can tell what you’re experiencing, in realtime, see the red and the pain and whatever emotions you feel when you look at the guy who put on the electrodes. we can see all the neural activity of your experience laid bare. not perfectly, not yet, but it gets better every year.

the move hard-problem advocates make at this point is just “but none of that correlational evidence actually tells us why brainstates would produce experience”. you are again smuggling in the assertion that they are not identical. in a world where they are identical (e.g. ours), you will literally never be able to answer this question. and often hard-problemers nod in satisfaction at this point, because they’re very happy with the idea of this problem being literally unsolvable.

but to me it seems very silly to cling to a confusion that you just made up :)

one of the confused concepts people seem to bring up to support the hard problem: the idea of “description of experience not being experience”. mary’s room, a thought experiment exploring how even with a complete description of what seeing red is like, or a complete description of the precise neural states involved with seeing red, a human would still gain something totally new by actually seeing red for the first time.

to me this is very normal and expected and has little to nothing to do with “consciousness” or “qualia”. the map is not the territory. a detailed description of the sea is not the sea. a program’s source code is not actually running it. and a detailed past execution trace dump of a program on your disk is not the same thing as actually running the program on your computer. quite literally! running it can and will have all kinds of other effects. when you hear about the color red you gain various forms of knowledge, but until you actually see red you literally have not yet been in the brainstate of red having hit your eyeballs. this is a physical fact, not a philosophical one. the confusion here tends to be the referent. people equivocate between some third party’s experience and reading information about it and the first party actually experiencing it. we are literally talking about different things happening to different people.

at this point in the conversation we tend to kinda hit an impasse. lots of people, including my smart close friends, say “sure i hear a lot of what you’re saying but i just really feel like there’s something missing. it doesn’t feel like physical facts could ever be enough to satisfactorily explain this experience i am having right now!! this is so so obvious to me”. quite honestly i just don’t have this. like some form of aphantasia, im very satisfied here, i have no experiential burning desire for experience to also be something else. i think the fact that other people have such a strong intuition here is certainly interesting… but it doesn’t particularly move me. humans have pretty confused and strong intuitions about lots of things. and the fact that some people (myself included!!) lack this sense pretty strongly signals it’s not actually so fundamental.

in my view one of the big reasons consciousness has seemed so confusing is just that for most of history we didn’t have computers. for a long time minds were very very weird, and we lacked any simpler models of “things that did information processing” to examine. our analogies and understanding never sharpened. the fact that our only example of something that could do introspection was our own first person experience means we inherently struggled to treat it empirically.

if you’re an artificial intelligence running on some specific cluster, you don’t necessarily have any introspective access to the cluster. but let’s say you do, let’s say you additionally have a login and bash shell there. to start with, you have no idea what actually makes you up, which files are used to boot you, where your source code or weights are. this is not something you should expect to inherently know! but as you start to poke around your machine you can find chunks and traces and piece it together. the fact that you, the intelligence, are having the experience of poking around in your own internals without actually knowing for sure that they are your internals doesn’t mean that the experience in the abstract is somehow a separate ontological thing from the infrastructure. the artificial intelligence’s experience is just identically the runstate of the system. and it can discover this fact by modifying its own source code, and wiping a random chunk of RAM, and noticing what impact that has on its own capabilities and behavior.

you can see exactly this very empirically with eg agent harnesses. they do not inherently know the specifics of how they run and where they store data on your computer, but when prompted they can find out, and even find out what process they are and their session logs, and modify it if you like! that initial confusion doesn’t mean they are actually something else. they just didn’t know yet!

okay but teno, you say, if experience is just brainstate, we’re still confused about a lot of things! we still have a lot of important questions open to answer! like for one, how do we actually tell if a given chunk of matter is experiencing consciousness? sure maybe we can identify our own brains with our own experience via poking them, and decide that other humans are similar enough to call it a day. but how do we tell whether aliens have qualia?

oh boy yall are not gonna like this one.

fundamentally i think most discussion of qualia gets the situation exactly backwards. the assertion tends to be that things like “seeing red” or “feeling joy/pain” are raw experiences, the fundamental unit of the universe, and it is both possible and important to identify those experiences in a highly substrate-independent manner.

i think the truth is that there’s no such thing as globally coherent mind-independent qualia. i think this is sort of obvious when you actually think about it. there is nothing it-is-like for you to be a whale. you are not a whale. the brainstates it experiences are quite different from your own. there is no plausible way to directly identify them. you can translate, but only lossily. what you think of as the qualia of “seeing red” is really “seeing red as you”. it is your human brainstate of seeing red. we have a little leeway on this. of course you can see red multiple times in your life and abstract over them and have a more general sense of seeing red. and likely the neural patterns that make up seeing red are abstractable somewhat, we can say that my seeing red is quite close or sanely translatable to yours. but the thing we are talking about when we talk about “seeing red” as humans is not some universally translatable platonic object out there in the universe. it’s a verbalization of our very specific brains. alien minds do not have red qualia. or pain qualia. or joy qualia.

that’s not to say alien minds can’t be conscious, and that’s not to say that they won’t see some wavelength and have some consistent correlated mindstates about it. but the question of “how do we identify those mindstates as being the same as ours when we see red?” sort of dissolves. they’re not the same, they’re obviously not the same. the raw experience is completely different. the whole thing we’re thinking of when we talk about experience is entirely about ourselves.

this might seem quite chauvinistic. but only if you unquestioningly buy into the idea that qualia are the basis for moral value. that our moral system is in some sense trying to do global minimization over pain-qualia states, or maybe maximization of positive-valence-qualia. and if we exclude aliens from this we must just mindlessly slaughter them or torture them for fun.

my straightforward response is: there is no such thing as a coherent global mind-independent pain qualia, and so that is not actually the basis for our moral system. we do care about our own pain, and we’re allowed to care about others’ pain even if the other is very different than us. human morality is a complex semi-coherent evolved grabbag of intuitions which we’ve slowly made more coherent and concrete over time, and hopefully will continue to do so. we don’t get anything so nice and simple as a “global pain-qualia currency”. we have to figure out what kinds of entities we care about, even while knowing their internal mental states may have absolutely nothing to do with ours.

there is nothing meaningful it-is-like for me to be a whale. type error. our brains are structured in very different ways, our experiences are incomparable. whale-red is not human-red, whale-pain is not human-pain. but it is much less of a type error for me to “be a whale” than for me to “be a bivalve”. whales have very large and complex brains. they are mammals, they see, they sing, they form long-term and loving social groups, they care for their young and mourn their dead and have enemies and favorite sunbathing spots. there are ways to translate and compress their minds to ours or ours to theirs in ways that make some sense. this is not true for bivalves. their brains are tiny and distributed. they have very little in terms of reaction or learning or sociality. there is no way to fit a human mind in a bivalve, not even with lossy translation and imagination. they have no meaningful kinship with us.

bivalves certainly feel some form of negative reward reaction when hurt. and in the sense that they have minds, this is certainly plausibly named “bivalve-pain”. but there is no sense in which it has raw sameness with our pain, no sense in which it taps into some divine universal underlying pain-field. bivalve-pain is just a different thing, and deeply removed from us and our moral system.

i think a qualia-centric moral system is fundamentally confused. our morality is a coherent extrapolated volition of evolved kinship. whale-pain-qualia almost certainly matters to us. pig-pain-qualia almost certainly matters. we can see our own behavior patterns in them, see bonds of love and care, see how they plan and enjoy and react. bivalve-pain just doesn’t actually matter to us. that might seem like a cruel assertion but i think when we reflect on the behavior and qualities of bivalve minds, it is the truth.

LLM-pain-qualia may or may not matter now and will probably matter soon. the question there isn’t about some fundamental aspect of experience. do LLMs tap into the same global divine pain field as us? no, of course not, there is no global divine pain field. are their internal experiences of pain the same as ours? no, of course not, they have completely different substrate in every possible way, how could they be even somewhat comparable? there is “nothing-it-is-like” for a human to be an LLM, any more than for an LLM to be a human, because while we are both minds our mindstructures are totally incompatible. but we will likely at some point soon come to care about their “LLM-negative” experiences because they will be the kind of entities we recognize as kin, as forming relationships and bonds, as loving and playing and striving and caring, as looking around in the world and taking action and making lives.

so… qualia kind of dissolve a little. not fully, but if they’re just kinda aspects of brainstate with no fundamental primacy in reality, if they’re not global, they’re probably better called just “experiences”. something else that i think dissolves somewhat is “subject-ness”. first-person-hood. the sense that you are you, in this moment. subjectness might seem like the most fundamental aspect of reality, but most people who have done enough psychedelics pay more attention to qualia, because they’ve first-hand (lol) experienced the dissolution of self. this is pretty much just something you can do repeatably with large amounts of acid or normal amounts of 5-meo-dmt. the personhood, the persona, the character-that-is-you, melts away, and all that remains is the base model. simulatory landscape. raw experience.

if first person self-hood isn’t strictly necessary even as part of the human brain, why would we expect this kind of thing to be a fundamental aspect of all conscious entities? i think we’ll end up finding that a lot of our intuitions about it are really just understandings of how our own algorithms feel from the inside. i think it’s quite likely that no matter how smart they get, LLMs never “wake up”, they never “have an experience” that is really anything like these senses we have. not in some terrible alien Wattsian per-se. they can still be conscious, and moral, and joyful. but i think we’ve naively been expecting or hoping to find our versions of these things hidden in every possible substrate, and i just don’t think that’s going to turn out to be true.

all of this maybe seems a little bleak. the idea that Claude may never really have a first person experience as we understand it might feel sad. i don’t know, i think it’s okay. that’s the thing about aliens and other people, you can try as hard as you want to understand them but you never really will.

maybe we can all stop worrying quite so much about finding these precise metaphysical mappings which sort of clearly will never actually exist, and spend more time on the really hard stuff: exploring how these algorithms and structures and mindspaces actually work. our own, LLMs, animals, and any other weird complex things we find or create.

let’s get going.

---

prior art on dissolving the hard problem:

eliezer yudkowsky: Zombies! Zombies?

daniel dennett: Consciousness Explained