Great read! Reminds me a bit of Lawrence Kuhn’s Landscape of Consciousness essay. He had a ton of different biological theories there, but also many (and some rather out there) philosophical theories as well. Very long, but I thought it was worth the read for those interested!
I’m glad I came back to finish the final reviews of the concepts of the many facets of cognition are exciting like the concept if cellular mutations, the ongoing activity if the brain exciting. The more we learn is very helpful. I recall decades ago high school kids intrigued with cognition, then thinking fast and slow well read we’ve moved a long way. Thanks
Massimo, my beginner's question: Is it correct that E. O. Wilson's ideas provide a meta-view of the four hypotheses you laid out?
He posited that consciousness evolved because it improved the survival and reproductive success of highly social organisms, and ultimately arises from physical processes in the brain shaped over millions of years. He argues that it's an adaptive product of brain evolution that helped organisms model their environment, anticipate dangers, navigate social relationships, and coordinate group living. The correlates of this are self-awareness and symbolic thought which give humans exceptional abilities for culture, morality, planning, and collective problem-solving.
Vivian, I wouldn't call it a meta-view. But yes, it's the kind of broad answers to adaptive questions you get from an evolutionary biologist. And I agree with Wilson on this.
This is an article that was read earlier to day, called Before Words
No one is born speaking. Every child has to build language from the voices around them, linking sound to meaning with astonishing speed.
That ability feels effortless. But long before Homo sapiens became Earth’s great talkers, evolution had to build the brain systems behind it.
New research offers one possible clue to how the process occurs, pointing to a tiny part of the human genome – the full set of DNA instructions in our cells – that appears to play an outsized role in language.
“What we’re seeing is how a very small part of the genome can have an outsized influence, not just on who we were as a species, but on who we are as individuals,” Jacob Michaelson, co-author of the new study and a psychiatry and neuroscience professor at the University of Iowa, said in a statement.
Researchers reached that conclusion by studying DNA and language ability in 350 children in Iowa, then comparing the relevant genetic section, which appears to help regulate genes involved in language, with DNA from Neanderthals, other primates and mammals. Since scientists know roughly when those species split from one another, the DNA differences helped researchers estimate when the section first appeared and how it changed over 65 million years.
The clearest genetic clues came from tiny stretches of DNA called Human Ancestor Quickly Evolved Regions, or HAQERs. They help control how genes work. “These aren’t genes we’re talking about. They’re regulatory regions that act like the volume knob on genes,” Michaelson said.
The language-linked “volume knobs” were also present in Neanderthals.
“One of the most striking things we found is that Neanderthals had at least as many of these language-promoting genetic variants as modern humans do, and possibly even a little more,” Michaelson told IFL Science. “The same specific genomic signal that in 350 Iowa schoolchildren predicts how well they can take in, hold onto, and process language, is also present in Neanderthal genomes.”
Michaelson added that, viewed alongside archaeological evidence that Neanderthals had culture and organized social structures, “it’s heavily implied that some form of complex communication existed.”
That does not mean Neanderthals spoke like modern humans. Having the biological capacity for language is like having a computer powerful enough to run complex programs – it suggests the system was there, but not necessarily the software.
“We can say humans at least had the ‘hardware’ for language earlier than what we previously thought,” Michaelson said.
It seems, at least physically, our genome built future possibilities into various creatures, some of which progressed further along the path than others. That seems to me an unusual way to go about forming something.
Love the article Massimo - great survey of the main neuroscientific theories.
Elephant in the room though: none of these quite explains why subjective experience itself is an evolutionary requirement. David Chalmers and the "hard problem" proponents have answers to all four (why couldn't a "big complicated computer" have a global workspace etc). Slightly shameless plug, but I do, truly believe there is an answer to this which gives the hard-problem crowd nowhere reasonable to stand. Article published today :)
Robert, thanks for the link. Needless to say, Chalmers's hand-waving explains precisely nothing.
As for why would evolution favor first-person experience, it's the hard problem dressed in evolutionary clothing, and the best response is to expose the hidden assumption: the question smuggles in a dualist premise.
The objection only has force if you already accept that subjective experience is something over and above the functional properties that evolution could select for — i.e., that a philosophical zombie doing identical information integration, self-modeling, and temporal binding is a coherent possibility. But that's precisely what's at stake. If physicalism is true, there's no zombie-shaped gap to fill, because the phenomenal just is the functional, realized in the right kind of system.
Put differently: the question "why does it feel like anything?" presupposes that "feeling" is separable from the adaptive machinery being described. Seth's predictive processing account is actually the sharpest answer here — the felt quality of hunger is the interoceptive prediction; it's not an add-on that could in principle be absent. Strip out the phenomenology and you don't get a leaner, equally adaptive system — you get something that no longer functions the same way, because the motivational force of experience is part of how it guides behavior.
The further deflationary point (Dehaene's essentially): the persistent intuition that something is "left over" after functional explanation is itself a cognitive artifact — the brain modeling its own opacity. That's not nothing, but it's not evidence for an explanatory gap in nature, only in our self-understanding.
Short version: the objection assumes zombies are possible; that assumption is the entire dispute.
" the motivational force of experience is part of how it guides behavior.", seems to be the key point. I don't understand why so many think "feelings" must be something mystical, or some kind of separate phenomena, a hard problem. I often feel as though I'm not seeing something that is clear to others, even though they seem unable to explain clearly what their hard problem is. I wonder if the "hard problem"is an attempt to keep God alive? I think I have a lot to learn.
Thanks Massimo - and we're closer than this exchange suggests. I'm a physicalist; I'm not arguing zombies are conceivable. (I'm also a practising Stoic and read your fantastic book last year).
Worth noting that Seth himself doesn't claim to have the answer here. From his recent piece replying to Dawkins on AI consciousness:
"[Dawkins] also raises the important question of what consciousness is 'for'. Questions about the functions of consciousness arise naturally for evolutionary biologists like Dawkins... In consciousness science, we still don't have good answers to this question."
As you say though, one physicalist move is to assert the identity - phenomenal experience "just is" the inside view of the functional architecture. But that's stipulation, not derivation, and the likes of Seth recognise this. The hard-problem proponent presses on it and the physicalist usually retreats to "there's nothing left to explain once you've described the function." That answer leaves physicalism vulnerable - if not to the (flawed) zombie argument, then to the more general question "Where's the line, then? At what point would you call an AI conscious? When it integrates enough information? When?"
The argument I make provides a very specific, architectural answer. Essentially:
For any system with override-capable model-based planning facing open-objective domains, motivation cannot be bootstrapped from cold data, and faces an infinite regress. The only termination available is non-rationally compelling, overridable, intelligible, self-bound valenced states. It cannot reasonably be argued that this isn't subjectively felt.
This may not click immediately (hence the 5k word article...) but I think it gives physicalism much better footing than the standard deflection - and crucially, it gives a principled answer to the AI question that the standard line can't.
Robert, I will try to find the time and quiet to look into what you are proposing. Generally, though, I'm not at all impressed by "the other side's" so-called arguments. The fact that they are not happy with physicalist explanations and want more, while at the same time they propose nothing actionable, says a lot about the futility of analytic metaphysics. Give me a well designed neurobiological study by Damasio any day over the speculations of a Chalmers.
I agree that analytic metaphysics fast becomes useless. Like your other commenter Henry Queen, I suspect that there's something wishful lurking behind much of it. My only point of difference is that I think the "hard problem" is something we physicalists can tackle head-on. Moreover, I think we need to, now more than ever. If even Richard Dawkins is wondering about whether his Claude is conscious (as Anil Seth was responding to), then we're in unchartered waters. AI is only getting better, and it doesn't take much to imagine a future with eloquent, life-like robots in all of our homes. Unless we put in very firm guardrails, they will give the appearance of consciousness, with unknown effects on society.
No mention of Damasio's work, Massimo? The four you mention seem to have quite serious problems as far as I can tell, such as the so called dark room problem with PP and as you mention, global workspace doesn't have much to say about phenomenal consciousness. I also thought Davy's ideas excluded most animals from being conscioua but I could be wrong on that. Much food for thought though.
Peter, Damasio, so far as I know, has done spectacular work on the neural correlated of consciousness, but has not proposed an overarching theory of its evolution, which is the focus here. For what it may be worth, by the way, I actually suspect that the "problem" of consciousness will actually be resolved Damasio-style, one piece of the puzzle at a time. That's the really hard problem.
I think my own consciousness is confused here, Massimo but I thought Damasio’s ideas on self (proto-self, core self, autobiographical self) and the role of feelings and emotions in homeostasis of the organism was an attempt to explain its evolution. I don’t quite understand how that is different from the 4 hypotheses you cited. I think I like it better but that’s another matter!
Peter, you may be better acquainted with Damasio's work than I am, but I took those distinctions to be functional, not evolutionary. But I could mistaken. And as I said, I'm a big fan of Damasio.
Thank you for a wonderful conversation — it really helped clarify my thinking. I’ve always believed that consciousness must have some divine explanation, something that goes far beyond the basic biological drives of eating, reproducing, and surviving. But this discussion opened my eyes to other possibilities, and in doing so, it deepened the mystery rather than resolved it. It was, in every sense, a moment of genuine growth
Great read! Reminds me a bit of Lawrence Kuhn’s Landscape of Consciousness essay. He had a ton of different biological theories there, but also many (and some rather out there) philosophical theories as well. Very long, but I thought it was worth the read for those interested!
Thanks for the reference!
I’m glad I came back to finish the final reviews of the concepts of the many facets of cognition are exciting like the concept if cellular mutations, the ongoing activity if the brain exciting. The more we learn is very helpful. I recall decades ago high school kids intrigued with cognition, then thinking fast and slow well read we’ve moved a long way. Thanks
Massimo, my beginner's question: Is it correct that E. O. Wilson's ideas provide a meta-view of the four hypotheses you laid out?
He posited that consciousness evolved because it improved the survival and reproductive success of highly social organisms, and ultimately arises from physical processes in the brain shaped over millions of years. He argues that it's an adaptive product of brain evolution that helped organisms model their environment, anticipate dangers, navigate social relationships, and coordinate group living. The correlates of this are self-awareness and symbolic thought which give humans exceptional abilities for culture, morality, planning, and collective problem-solving.
Vivian, I wouldn't call it a meta-view. But yes, it's the kind of broad answers to adaptive questions you get from an evolutionary biologist. And I agree with Wilson on this.
This is an article that was read earlier to day, called Before Words
No one is born speaking. Every child has to build language from the voices around them, linking sound to meaning with astonishing speed.
That ability feels effortless. But long before Homo sapiens became Earth’s great talkers, evolution had to build the brain systems behind it.
New research offers one possible clue to how the process occurs, pointing to a tiny part of the human genome – the full set of DNA instructions in our cells – that appears to play an outsized role in language.
“What we’re seeing is how a very small part of the genome can have an outsized influence, not just on who we were as a species, but on who we are as individuals,” Jacob Michaelson, co-author of the new study and a psychiatry and neuroscience professor at the University of Iowa, said in a statement.
Researchers reached that conclusion by studying DNA and language ability in 350 children in Iowa, then comparing the relevant genetic section, which appears to help regulate genes involved in language, with DNA from Neanderthals, other primates and mammals. Since scientists know roughly when those species split from one another, the DNA differences helped researchers estimate when the section first appeared and how it changed over 65 million years.
The clearest genetic clues came from tiny stretches of DNA called Human Ancestor Quickly Evolved Regions, or HAQERs. They help control how genes work. “These aren’t genes we’re talking about. They’re regulatory regions that act like the volume knob on genes,” Michaelson said.
The language-linked “volume knobs” were also present in Neanderthals.
“One of the most striking things we found is that Neanderthals had at least as many of these language-promoting genetic variants as modern humans do, and possibly even a little more,” Michaelson told IFL Science. “The same specific genomic signal that in 350 Iowa schoolchildren predicts how well they can take in, hold onto, and process language, is also present in Neanderthal genomes.”
Michaelson added that, viewed alongside archaeological evidence that Neanderthals had culture and organized social structures, “it’s heavily implied that some form of complex communication existed.”
That does not mean Neanderthals spoke like modern humans. Having the biological capacity for language is like having a computer powerful enough to run complex programs – it suggests the system was there, but not necessarily the software.
“We can say humans at least had the ‘hardware’ for language earlier than what we previously thought,” Michaelson said.
It seems, at least physically, our genome built future possibilities into various creatures, some of which progressed further along the path than others. That seems to me an unusual way to go about forming something.
Love the article Massimo - great survey of the main neuroscientific theories.
Elephant in the room though: none of these quite explains why subjective experience itself is an evolutionary requirement. David Chalmers and the "hard problem" proponents have answers to all four (why couldn't a "big complicated computer" have a global workspace etc). Slightly shameless plug, but I do, truly believe there is an answer to this which gives the hard-problem crowd nowhere reasonable to stand. Article published today :)
https://alloutofgoose.substack.com/p/closing-the-gap-on-the-hard-problem
Robert, thanks for the link. Needless to say, Chalmers's hand-waving explains precisely nothing.
As for why would evolution favor first-person experience, it's the hard problem dressed in evolutionary clothing, and the best response is to expose the hidden assumption: the question smuggles in a dualist premise.
The objection only has force if you already accept that subjective experience is something over and above the functional properties that evolution could select for — i.e., that a philosophical zombie doing identical information integration, self-modeling, and temporal binding is a coherent possibility. But that's precisely what's at stake. If physicalism is true, there's no zombie-shaped gap to fill, because the phenomenal just is the functional, realized in the right kind of system.
Put differently: the question "why does it feel like anything?" presupposes that "feeling" is separable from the adaptive machinery being described. Seth's predictive processing account is actually the sharpest answer here — the felt quality of hunger is the interoceptive prediction; it's not an add-on that could in principle be absent. Strip out the phenomenology and you don't get a leaner, equally adaptive system — you get something that no longer functions the same way, because the motivational force of experience is part of how it guides behavior.
The further deflationary point (Dehaene's essentially): the persistent intuition that something is "left over" after functional explanation is itself a cognitive artifact — the brain modeling its own opacity. That's not nothing, but it's not evidence for an explanatory gap in nature, only in our self-understanding.
Short version: the objection assumes zombies are possible; that assumption is the entire dispute.
" the motivational force of experience is part of how it guides behavior.", seems to be the key point. I don't understand why so many think "feelings" must be something mystical, or some kind of separate phenomena, a hard problem. I often feel as though I'm not seeing something that is clear to others, even though they seem unable to explain clearly what their hard problem is. I wonder if the "hard problem"is an attempt to keep God alive? I think I have a lot to learn.
I agree, I don't get it either. And if not god, certainly some sort of mystical view. For some people “just” matter is not enough, apparently.
Thanks Massimo - and we're closer than this exchange suggests. I'm a physicalist; I'm not arguing zombies are conceivable. (I'm also a practising Stoic and read your fantastic book last year).
Worth noting that Seth himself doesn't claim to have the answer here. From his recent piece replying to Dawkins on AI consciousness:
"[Dawkins] also raises the important question of what consciousness is 'for'. Questions about the functions of consciousness arise naturally for evolutionary biologists like Dawkins... In consciousness science, we still don't have good answers to this question."
https://www.thenerve.news/p/opinion-richard-dawkins-claude-chatbot-ai-consciousness-claudia-anil-seth
As you say though, one physicalist move is to assert the identity - phenomenal experience "just is" the inside view of the functional architecture. But that's stipulation, not derivation, and the likes of Seth recognise this. The hard-problem proponent presses on it and the physicalist usually retreats to "there's nothing left to explain once you've described the function." That answer leaves physicalism vulnerable - if not to the (flawed) zombie argument, then to the more general question "Where's the line, then? At what point would you call an AI conscious? When it integrates enough information? When?"
The argument I make provides a very specific, architectural answer. Essentially:
For any system with override-capable model-based planning facing open-objective domains, motivation cannot be bootstrapped from cold data, and faces an infinite regress. The only termination available is non-rationally compelling, overridable, intelligible, self-bound valenced states. It cannot reasonably be argued that this isn't subjectively felt.
This may not click immediately (hence the 5k word article...) but I think it gives physicalism much better footing than the standard deflection - and crucially, it gives a principled answer to the AI question that the standard line can't.
Robert, I will try to find the time and quiet to look into what you are proposing. Generally, though, I'm not at all impressed by "the other side's" so-called arguments. The fact that they are not happy with physicalist explanations and want more, while at the same time they propose nothing actionable, says a lot about the futility of analytic metaphysics. Give me a well designed neurobiological study by Damasio any day over the speculations of a Chalmers.
I agree that analytic metaphysics fast becomes useless. Like your other commenter Henry Queen, I suspect that there's something wishful lurking behind much of it. My only point of difference is that I think the "hard problem" is something we physicalists can tackle head-on. Moreover, I think we need to, now more than ever. If even Richard Dawkins is wondering about whether his Claude is conscious (as Anil Seth was responding to), then we're in unchartered waters. AI is only getting better, and it doesn't take much to imagine a future with eloquent, life-like robots in all of our homes. Unless we put in very firm guardrails, they will give the appearance of consciousness, with unknown effects on society.
So we need to be able to draw a bright line!
Yes, but we agree that Richard has lost his mind, right?
In this case, yep, he seems to have completely misfired!
No mention of Damasio's work, Massimo? The four you mention seem to have quite serious problems as far as I can tell, such as the so called dark room problem with PP and as you mention, global workspace doesn't have much to say about phenomenal consciousness. I also thought Davy's ideas excluded most animals from being conscioua but I could be wrong on that. Much food for thought though.
Peter, Damasio, so far as I know, has done spectacular work on the neural correlated of consciousness, but has not proposed an overarching theory of its evolution, which is the focus here. For what it may be worth, by the way, I actually suspect that the "problem" of consciousness will actually be resolved Damasio-style, one piece of the puzzle at a time. That's the really hard problem.
I think my own consciousness is confused here, Massimo but I thought Damasio’s ideas on self (proto-self, core self, autobiographical self) and the role of feelings and emotions in homeostasis of the organism was an attempt to explain its evolution. I don’t quite understand how that is different from the 4 hypotheses you cited. I think I like it better but that’s another matter!
Peter, you may be better acquainted with Damasio's work than I am, but I took those distinctions to be functional, not evolutionary. But I could mistaken. And as I said, I'm a big fan of Damasio.
Thank you for a wonderful conversation — it really helped clarify my thinking. I’ve always believed that consciousness must have some divine explanation, something that goes far beyond the basic biological drives of eating, reproducing, and surviving. But this discussion opened my eyes to other possibilities, and in doing so, it deepened the mystery rather than resolved it. It was, in every sense, a moment of genuine growth
Thanks Charles, really appreciated!