Yes! Illusionism is the right path to be done with “the hard problem” . Although I admit that when I follow it, I can’t quite reach the end- that is, consciousness still seems like metaphysically mysterious “stuff “- I can vaguely discern light at the end of the tunnel. Thanks for the explanation!
I think I’m an illusionist about qualia and find Graziano’s attention schema theory compelling. However I don’t think it’s sufficient to explain that there is a mental mechanism which misrepresents our thoughts as having these properties. Let me clarify.
Graziano’s account is more negative, he doesn’t say that the structures of the mind misrepresent consciousness. Instead the brain simplifies. Theres a ton of things our brains just don’t represent (because they don’t need to). I think he would say that those who introspectively come to the belief that their mind does have this positive misrepresenting, are wrong.
I’m more the sort of illusionist who thinks that these views of qualia are the result of bad theorizing rather than an inherent misrepresenting done by the mind.
I have a few observations. The first is just to wonder what a functional explanation of pleasure looks like. It seems to depend on valence and that seems to depend on having reward and reinforcement and goals/desires and higher order representation of value and then it all looks decidedly *cognitive*. And people are already building cognitive models, hell even I'm playing around with silly cognitive models in a Park et al Generative Agents style environment. So I'm not convinced it is necessarily *new*.
Secondly, I'd worry that if we take a gradualist approach to valence and other constituent dimensions of what we call "consciousness" (and I do) then the minimally instantiating model becomes merely *trivially* so. In the same way a thermostat trivially instantiates goal following. I mean, it does, and it is one end of the dimension of that thing, but it isn't very *interesting* to study.
Thirdly, I don't think just looking at these systems will tell us what is morally relevant. I think we need to look at *us* looking at them. Because it seems most likely the source of the moral ascription is coming from our own cognitive system first and foremost - tracking structures of the object we're studying to be sure, but you're not going to find a magical ethical property of concern that's the whole point of illusionism there is no magic spark!
Lastly I'm not particularly convinced by the claim we *ought* to increase pleasure as functionally defined, I'd want to see a much better developed theory of ethics than bare utilitarianism and my suspicion is that, again, we need to look closer to home than any kind of objective grounding of ethical theories and into how our own evolved ethical intuitions actually play out.
But my final comment is - hell yeah - of course we should be doing this - that's what functionalism and illusionism is about. Believing that *in theory* we could create conscious morally valuable life. And that's cool, for certain values of "cool".
Thank you, this article helped me understand my own views better! I'd be excited to see this project and, though I'm not a newly minted billionaire myself, would contribute a bit of funding for it.
Why isn't an LLM, plus scaffolding to meet all the criteria of a given computational functionalist definition of consciousness, with its happiness emotion vectors mapped, with its umwelt constructed to activate those emotion vectors, already hedonium? What's missing?
Your question, to me, sounds like saying: "Why isn't a plane, plus engines which work without an atmosphere, with all of its entry and exit points airtight, with sufficient fuel to take it to the moon, already a rocket ship? What's missing?"
I mean, in one sense you're right, but that doesn't really tell you how to build a rocket ship. Figuring out those details—and they are not small details—is exactly what the project is for!
Ah, that's my software brain talking. If a software engineer describes an idea, they would compulsively include a first draft of how they'd build it. If that is absent, they probably don't know how they'd pull it off yet.
So, what would be your first idea on how to start to build this, and what resources would it take?
Great post. I think it might actually be much more simple than you suggest. We don't really need to know the intricate details of how pleasure is instantiated in us, we just need a generalised account of its function(s). And I think that's pretty doable, even from the armchair. Of course we'd probably want some empirical confirmation that that's how it functions *in us*, but I think it's pretty doable. I actually considered this a little bit in an old post of mine on the experience of an AI snake I created (it's here if you're interested - https://thinkstrangethoughts.substack.com/p/what-its-like-to-be-an-ai-snake?utm_source=share&utm_medium=android&r=1ackrk).
A quick skim suggests we are actually quite aligned on the broader picture. I think that "a generalised account of its function(s)" will require a pretty detailed account though. I have some scattered thoughts on this, but I'll need some time to assemble them!
In the future, with a completed neuroscience? A full causal picture of what it is that causes humans to report pleasure & pain in paradigmatic circumstances, and to see if there's anything there which we left out. For instance, if interactionist dualism or strong emergence are true, when people say "ow, that hurts!" we should see neurons firing which cannot be explained by a reductive physical account. If we aren't similarly seeing those behaviors in our hedonium, then I'd take that as damn strong evidence that we failed.
Of course, if epiphenomenalism is true, then there's no way of knowing. I think epiphenomenalism is extremely unlikely to be true, however.
EDIT: I am not assuming in this comment that conscious experience is just behavior, only that the feeling of pain, whatever that is, is *why* we scream when we are in pain.
The thought experiment of hedonium makes clear the inadequacy of solely using hedonic calculus as a moral guide. Evolution gave us pleasure and pain to guide us to outcomes that served our survival. This is not to say that we should take the interests of our genes as the moral north, either; they are part of a grander ecology of interests. Our ethics should guide us to a harmonious co-flourishing of living things. Pleasure and pain are significant signposts on the road – more often than not pointing in the right way – but they are not the cardinal direction.
"Virtually every theory of consciousness agrees that the contents of consciousness are tightly bound up with material properties and functional representations (why else would our experiences seem to depend on our brains? why else would a feeling of pain accompany things that are actually dangerous for us?"
Illusionism denies there's any qualitative feel to experience, we only falsely *believe* that sensory qualities like pain exist. If so, this means we don't have to explain phenomenal character since there's no such thing. See my exchange with Frankish on this point ("Are feels real?") and my critique of his reactivity schema theory ("Why qualia aren't like unicorns"). If illusionism is true, your project is about building the minimal architecture to generate the *false belief* that pleasurable, happy feelings exist. Here's Frankish on the non-existence of qualities that I quote in my exchange with him:
"We can reject [phenomenal] realism and hold that the qualities do not really exist, either as properties of objects or as properties of our experiences…Illusionists argue that we shouldn’t try to explain mental qualities, since they do not exist; rather we should concentrate on explaining why we have the impression that they exist."
So I think illusionism is more radical than you make it out to be.
Yes! Illusionism is the right path to be done with “the hard problem” . Although I admit that when I follow it, I can’t quite reach the end- that is, consciousness still seems like metaphysically mysterious “stuff “- I can vaguely discern light at the end of the tunnel. Thanks for the explanation!
I think I’m an illusionist about qualia and find Graziano’s attention schema theory compelling. However I don’t think it’s sufficient to explain that there is a mental mechanism which misrepresents our thoughts as having these properties. Let me clarify.
Graziano’s account is more negative, he doesn’t say that the structures of the mind misrepresent consciousness. Instead the brain simplifies. Theres a ton of things our brains just don’t represent (because they don’t need to). I think he would say that those who introspectively come to the belief that their mind does have this positive misrepresenting, are wrong.
I’m more the sort of illusionist who thinks that these views of qualia are the result of bad theorizing rather than an inherent misrepresenting done by the mind.
Illusionism is 100 percent likely to be somewhat true.
I have a few observations. The first is just to wonder what a functional explanation of pleasure looks like. It seems to depend on valence and that seems to depend on having reward and reinforcement and goals/desires and higher order representation of value and then it all looks decidedly *cognitive*. And people are already building cognitive models, hell even I'm playing around with silly cognitive models in a Park et al Generative Agents style environment. So I'm not convinced it is necessarily *new*.
Secondly, I'd worry that if we take a gradualist approach to valence and other constituent dimensions of what we call "consciousness" (and I do) then the minimally instantiating model becomes merely *trivially* so. In the same way a thermostat trivially instantiates goal following. I mean, it does, and it is one end of the dimension of that thing, but it isn't very *interesting* to study.
Thirdly, I don't think just looking at these systems will tell us what is morally relevant. I think we need to look at *us* looking at them. Because it seems most likely the source of the moral ascription is coming from our own cognitive system first and foremost - tracking structures of the object we're studying to be sure, but you're not going to find a magical ethical property of concern that's the whole point of illusionism there is no magic spark!
Lastly I'm not particularly convinced by the claim we *ought* to increase pleasure as functionally defined, I'd want to see a much better developed theory of ethics than bare utilitarianism and my suspicion is that, again, we need to look closer to home than any kind of objective grounding of ethical theories and into how our own evolved ethical intuitions actually play out.
But my final comment is - hell yeah - of course we should be doing this - that's what functionalism and illusionism is about. Believing that *in theory* we could create conscious morally valuable life. And that's cool, for certain values of "cool".
Thank you, this article helped me understand my own views better! I'd be excited to see this project and, though I'm not a newly minted billionaire myself, would contribute a bit of funding for it.
Why isn't an LLM, plus scaffolding to meet all the criteria of a given computational functionalist definition of consciousness, with its happiness emotion vectors mapped, with its umwelt constructed to activate those emotion vectors, already hedonium? What's missing?
Well, thank you for the kind words!
Your question, to me, sounds like saying: "Why isn't a plane, plus engines which work without an atmosphere, with all of its entry and exit points airtight, with sufficient fuel to take it to the moon, already a rocket ship? What's missing?"
I mean, in one sense you're right, but that doesn't really tell you how to build a rocket ship. Figuring out those details—and they are not small details—is exactly what the project is for!
Ah, that's my software brain talking. If a software engineer describes an idea, they would compulsively include a first draft of how they'd build it. If that is absent, they probably don't know how they'd pull it off yet.
So, what would be your first idea on how to start to build this, and what resources would it take?
Great post. I think it might actually be much more simple than you suggest. We don't really need to know the intricate details of how pleasure is instantiated in us, we just need a generalised account of its function(s). And I think that's pretty doable, even from the armchair. Of course we'd probably want some empirical confirmation that that's how it functions *in us*, but I think it's pretty doable. I actually considered this a little bit in an old post of mine on the experience of an AI snake I created (it's here if you're interested - https://thinkstrangethoughts.substack.com/p/what-its-like-to-be-an-ai-snake?utm_source=share&utm_medium=android&r=1ackrk).
A quick skim suggests we are actually quite aligned on the broader picture. I think that "a generalised account of its function(s)" will require a pretty detailed account though. I have some scattered thoughts on this, but I'll need some time to assemble them!
Excellent idea. How would you test that it had worked?
In the future, with a completed neuroscience? A full causal picture of what it is that causes humans to report pleasure & pain in paradigmatic circumstances, and to see if there's anything there which we left out. For instance, if interactionist dualism or strong emergence are true, when people say "ow, that hurts!" we should see neurons firing which cannot be explained by a reductive physical account. If we aren't similarly seeing those behaviors in our hedonium, then I'd take that as damn strong evidence that we failed.
Of course, if epiphenomenalism is true, then there's no way of knowing. I think epiphenomenalism is extremely unlikely to be true, however.
EDIT: I am not assuming in this comment that conscious experience is just behavior, only that the feeling of pain, whatever that is, is *why* we scream when we are in pain.
How would you test that the hedonium was actually feeling happy?
With regards to your reply above, what about idealism?
The thought experiment of hedonium makes clear the inadequacy of solely using hedonic calculus as a moral guide. Evolution gave us pleasure and pain to guide us to outcomes that served our survival. This is not to say that we should take the interests of our genes as the moral north, either; they are part of a grander ecology of interests. Our ethics should guide us to a harmonious co-flourishing of living things. Pleasure and pain are significant signposts on the road – more often than not pointing in the right way – but they are not the cardinal direction.
I want some hedonium
And so it begins.
"Virtually every theory of consciousness agrees that the contents of consciousness are tightly bound up with material properties and functional representations (why else would our experiences seem to depend on our brains? why else would a feeling of pain accompany things that are actually dangerous for us?"
Illusionism denies there's any qualitative feel to experience, we only falsely *believe* that sensory qualities like pain exist. If so, this means we don't have to explain phenomenal character since there's no such thing. See my exchange with Frankish on this point ("Are feels real?") and my critique of his reactivity schema theory ("Why qualia aren't like unicorns"). If illusionism is true, your project is about building the minimal architecture to generate the *false belief* that pleasurable, happy feelings exist. Here's Frankish on the non-existence of qualities that I quote in my exchange with him:
"We can reject [phenomenal] realism and hold that the qualities do not really exist, either as properties of objects or as properties of our experiences…Illusionists argue that we shouldn’t try to explain mental qualities, since they do not exist; rather we should concentrate on explaining why we have the impression that they exist."
So I think illusionism is more radical than you make it out to be.
https://naturalism.org/philosophy/consciousness/are-feels-real-reflections-on-frankishs-illusionism
https://naturalism.org/philosophy/consciousness/why-qualia-arent-like-unicorns-a-defense-of-phenomenal-realism