Incredible finding. It does seem to push human speech, and humans as we more or less recognise them back pretty early? Like 1.5 mya maybe 2.5 MILLION years ago? That's shocking honestly. Alan Rogers is going to win "found earliest divergence" award at this rate. That's well into the oldowan era.
I get the issue of the presence of other hominid fossils and the paucity in key areas of what might be more likely our own predecessors, but the thought experiment I work with is
If naledi was the oldest we'd find we'd assume it was ancestral to us. Until we found jebel irhoud
Likewise with dmanisi, until we've now found that denisovanesque looking Yunxian skull from 1.76 mya. I get that's almost rabbit in the Cambrian era (as a friend put it) but it at least gives us another strand of possibilities. Humans as we more or less recognise them, going way back.
And as you're highlighting here we apparently haven't even scratched the surface in terms of the fossil sites we could be digging.
Kind of speechless here we are in the relative golden age of palaeoanthropology. Utterly disgraceful funding is getting cut I'm hoping its just reversed in two years. Thank you for these constant updates Professor Hawks.
There are no simple answers to questions like these of course. Pushing the common ancestors of today's humans further back does have to open our minds to the possibility that traits like language are quite a lot deeper than we used to think. It's certainly no big news for Neanderthals to be talking when we accept that modern people have come from lines that separated 1.5 million years ago, long before Neanderthals ever existed.
At the same time, when we accept evolution, we're accepting that humans transformed from something else. We all evolved from ancestors that did not talk. Evolution doesn't automatically give us a reason to think that language evolved before that last common ancestor, when it could just as well have evolved afterward. We underwent so many changes in that span of time, why not change in our system of communication?
But I do think that we have underestimated the cultural capacities of most hominins. That shouldn't be controversial, it's manifestly clear that we underestimated other nonhuman primates, and social mammals broadly, in cognition and behavior. So if some Homo erectus branched off 2.5 million years ago and headed into Eurasia, and two million years later their descendants connected with some early Denisovans, that's not very different from chimpanzees and bonobos interbreeding, and a lot closer than the ghost population that contributed to bonobos. We just have to accept that the ability to learn from each other and interact was broadly present across early humans.
As you state, as an evolutionary process, speech can develop after. I imagine in this model perhaps the "capacity" or "reservoir" so to speak exists in both strands , and its like equids having alien like abilities to diverge and rejoin without too much of an issue, so speech getting "encountered" independently by the two different strands. Although that seems to raise some degree of difficulty(what on earth did that look like, what intermediate stages existed if it developed in steps and what would it look like for a group that speaks to mate with a group that doesn't ), without denying its scientific validity. After all (if im not mistaken) nerves developed independently as did communication in the general sense. With our absence of erectus DNA, at least for now we're left mapping scientific possibilities, possibilities we are tracing and rapidly constraining but possibilities nonetheless.
If I'm getting them right, researchers seem to say Turkana boy could not have spoken as we do-while the software may have been there in his brain, he seems to have lacked the hardware and not as a generic defect.
If we entertain the the thought that *didn't happen* like, as we seem to have found with Neanderthals, they simply were like us (were us really), spoke as we do while sounding a bit different, bit odder looking at a thanksgiving dinner table but us more or less seen on a train or bus, we do avoid some difficulties while raising some others (why nearly no fossils sans that Yunxian one?).
The question is, what seems to raise fewer difficulties, a reservoir of language capacity allowing for independent development or humans as we know them just being around far further back than we ever conceived of. I can't wait for us to explore that! These recent and I'm guessing extremely unexpected findings(who conceived of something like this even twenty years ago?) make those questions far more important today than they ever were in the past.
I'll follow up on the Turkana Boy story sometime. This idea comes from the early analysis of the skeleton led by Alan Walker, and there's actually an interesting set of twists and turns on how people have looked at the skeleton since then.
Re: Turkana boy's physical capacity for speech, would that limitation apply to click-based languages? Click-based languages are rare today, but I think it's plausible that they were much more widespread in the past.
I could see a scenario where language originally developed with click sounds and new sounds gradually as physically abilities improved.
Was there not an epigenetic study recently that showed major differences in gene expression for the voice box between Neanderthals and modern humans as well?
The first paragraph is a breathtaking example of scientific curiosity in action. Wow. Nobody is as good at opening minds to fresh ways of thinking than John Hawks. I found it impossible to read further without first taking time to consider the implications of his words.
Hawks writes so clearly that even though I lack a background in genetics, I was fascinated by how generously he incorporates and credits the work of others to suggest a new approach to understanding the evolutionary history of Homo sapiens. I love that Hawks concludes his piece by asking, “What do you think?” I’m not sure what I think. I only know that I’d love to get these researchers together in one room and ask them all kinds of questions!
I think the evidence of the continental separation can be found in the modern population structure as well. If one looks at African-only populations in the PCA space, it consist of three main clusters (not counting much smaller ones for Koisan and pygmies). One is clearly Eurasian-derived, I would call it "Egyptian" although any NE Africa will do. The second is "Bantu-Yoruban" (B) and the third is Nilo-Saharan or "Dinka" (D). Everything else is connections (clines) between these clusters. B and D are very far from E, so we have a kind of triangle EBD with EB and ED sides very long (EDis a bit shorter), but BD is much shorter. Small clusters for Khoisans, Mbuti and Biaka are closer to B and very far from E, Hadza is a bit closer to D.
So the whole African diversity is mostly a cline (or rather two clines), connecting Eurasian-derived ancestry with very deep African ones, with separation between them corresponding to at least many hundreds of thousands years of mostly separate development. There is simply not an African cluster to be seen that may serve as the source of "recent exit" into Eurasia.
I think the "Recent Out-of-Africa" dogma blindsided paleoanthropology for more than 30 years and needs a major revision, not just some patches. It looks like human diversity is far greater than thought before with a lot of separate developments in different regions and separate deeply archaic admixtures in each regions.
It is remarkable that in the last few years several different computational models converge on similar answers for very ancient history of humanity, but generate new puzzles.
After reading the "branches A/B" paper I thought there must be geographical structure in these branches, it is very unlikely that they spread evenly everywhere. And such structure can be found indeed, at least in the model values, though it is buried in the supplement and I don't remember ever mentioned in discussions.
The timings of split and merge were separate for different populations, with the most contrast between African and Eurasian ones. For Eurasian populations the split/merge timings had max probability ~1800 and 380 Kya respectively, while for African populations - ~1000 and 190 Kya.
I haven't seen separate population estimates for A/B fractions (maybe they are buried among other charts) but I think there is very little chance that it doesn't have a strong geographical structure while othe parameters do.
This in my view makes an overwhelming case for the continental separation of A and B branches, with A being mostly a Eurasian one and B - African one. The timings parameters may also a scenario with 3 populations, including an intermediate one between A and B, such as "Southwest African" (B), "Northeast African" (AB) and "Eurasian" (A). A exited from Africa soon after 2 mln years with early Acheulean technology and lived in Eurasia ever since. AB and A kept in contact until ~1 mln years ago but then also separated due to demographic slump. Then A and AB reconnected ~300 Kya probably with Levallois technology, but the contact with B was reestablished much later.
This would fit the geographical structure reasonably well.
There is a hint of such structure in the "2 stems" model as well.
The question of geographic structure in the two components is a central one. The data are not there yet to answer it.
The Ragsdale model enabled heterogeneous gene flow from the two stems into ancestral groups after 110,000 years ago in Africa, and their model is based on a fit that may partially include "ghost" contributions previously found by other researchers. The Cousins model and Rogers model basically assume that A and B were fully mixed prior to differentiation of later African groups, and don't resolve whether B may have persisted as a ghost.
I've seen Chris Stringer write he supports a late mixture for A and B, my instinct is more for the early mixture with possible additional ghosts, but I wouldn't say the data speak strongly to this.
All of these models are basically island models in form, and what probably is closer to reality is some kind of isolation-by-distance model with dynamics over time. So there's a good bit of oversimplification in thinking about what 80-20 mixture actually means. If there is structure that we could detect in the ratio of the A and B stems in present-day Africans that would lead to a test of the biogeography of mixing.
Is it not possible to get uni-parental genetic markers(Y and mtDNA) from the many fossil records of Denisovans, Neanderthals and other archaic species like Homo Floriensis / Luzonensis. Maybe we can infer something by looking at the diversity of those markers and trace their evolution/migration just as we do for humans. There can also be a search focused on finding the Indian/Australian fossil records of early hominids. Also if possible can modern scientific methods not extract DNA from Jebel Irhoud just as we did with the Denisovan fossils and also establish correct dates for the remains. There are fossils in Europe/China/Australia which had shown earlier dates than what Out of Africa had permitted(Liujiang man comes to mind as do certain fossils in Europe).
Also for a definitive answer to human origins, migration and dispersal we would need to look at the Y DNA and mtDNA. I have repeated this on other places but I always find the distribution of certain Y DNA clades is way too weird(eg Onge and Japanese D clade). While reading up on the origin of Y and mtDNA clades, I found there was a proposal that Y DNA diversification of most non-Africans might have either taken place in the Arabian Peninsula ie South West Asia or in South East Asia. Also there was also an idea that the very basal mtDNA clade of L originated in South East Asia rather than East Africa. For instance, the initial findings on Mungo Man LM3 reported a very diverged clade of mtDNA (possibly indicating that his mtDNA was more archaic than what Out of Africa could explain?). Later that finding was overturned due to "possible contamination" but did they conclusively rule out the earlier findings and determine the mtDNA clade.
Also from Wikipedia, I learnt that the JC virus tracking places us in Asia rather than Africa. Is this not a cause of concern for people in the palaeo-anthropologic field as to why or how this disparity is occurring?
Finally, someone(Mondal et al) in 2019 used AI and determined that there was an evidence of another ghost population which contributed to human evolution and origins. With more sophisticated hardware, algorithms can we not revisit that possibility.
TL;DR: Greater scrutiny on genetic rather than morphological records, especially parental markers would probably solve human origins.
Incredible finding. It does seem to push human speech, and humans as we more or less recognise them back pretty early? Like 1.5 mya maybe 2.5 MILLION years ago? That's shocking honestly. Alan Rogers is going to win "found earliest divergence" award at this rate. That's well into the oldowan era.
I get the issue of the presence of other hominid fossils and the paucity in key areas of what might be more likely our own predecessors, but the thought experiment I work with is
If naledi was the oldest we'd find we'd assume it was ancestral to us. Until we found jebel irhoud
Likewise with dmanisi, until we've now found that denisovanesque looking Yunxian skull from 1.76 mya. I get that's almost rabbit in the Cambrian era (as a friend put it) but it at least gives us another strand of possibilities. Humans as we more or less recognise them, going way back.
And as you're highlighting here we apparently haven't even scratched the surface in terms of the fossil sites we could be digging.
Kind of speechless here we are in the relative golden age of palaeoanthropology. Utterly disgraceful funding is getting cut I'm hoping its just reversed in two years. Thank you for these constant updates Professor Hawks.
Outstanding comment, thanks for this!
There are no simple answers to questions like these of course. Pushing the common ancestors of today's humans further back does have to open our minds to the possibility that traits like language are quite a lot deeper than we used to think. It's certainly no big news for Neanderthals to be talking when we accept that modern people have come from lines that separated 1.5 million years ago, long before Neanderthals ever existed.
At the same time, when we accept evolution, we're accepting that humans transformed from something else. We all evolved from ancestors that did not talk. Evolution doesn't automatically give us a reason to think that language evolved before that last common ancestor, when it could just as well have evolved afterward. We underwent so many changes in that span of time, why not change in our system of communication?
But I do think that we have underestimated the cultural capacities of most hominins. That shouldn't be controversial, it's manifestly clear that we underestimated other nonhuman primates, and social mammals broadly, in cognition and behavior. So if some Homo erectus branched off 2.5 million years ago and headed into Eurasia, and two million years later their descendants connected with some early Denisovans, that's not very different from chimpanzees and bonobos interbreeding, and a lot closer than the ghost population that contributed to bonobos. We just have to accept that the ability to learn from each other and interact was broadly present across early humans.
As you state, as an evolutionary process, speech can develop after. I imagine in this model perhaps the "capacity" or "reservoir" so to speak exists in both strands , and its like equids having alien like abilities to diverge and rejoin without too much of an issue, so speech getting "encountered" independently by the two different strands. Although that seems to raise some degree of difficulty(what on earth did that look like, what intermediate stages existed if it developed in steps and what would it look like for a group that speaks to mate with a group that doesn't ), without denying its scientific validity. After all (if im not mistaken) nerves developed independently as did communication in the general sense. With our absence of erectus DNA, at least for now we're left mapping scientific possibilities, possibilities we are tracing and rapidly constraining but possibilities nonetheless.
If I'm getting them right, researchers seem to say Turkana boy could not have spoken as we do-while the software may have been there in his brain, he seems to have lacked the hardware and not as a generic defect.
If we entertain the the thought that *didn't happen* like, as we seem to have found with Neanderthals, they simply were like us (were us really), spoke as we do while sounding a bit different, bit odder looking at a thanksgiving dinner table but us more or less seen on a train or bus, we do avoid some difficulties while raising some others (why nearly no fossils sans that Yunxian one?).
The question is, what seems to raise fewer difficulties, a reservoir of language capacity allowing for independent development or humans as we know them just being around far further back than we ever conceived of. I can't wait for us to explore that! These recent and I'm guessing extremely unexpected findings(who conceived of something like this even twenty years ago?) make those questions far more important today than they ever were in the past.
I'll follow up on the Turkana Boy story sometime. This idea comes from the early analysis of the skeleton led by Alan Walker, and there's actually an interesting set of twists and turns on how people have looked at the skeleton since then.
Veeery interested in this.....can't wait
Re: Turkana boy's physical capacity for speech, would that limitation apply to click-based languages? Click-based languages are rare today, but I think it's plausible that they were much more widespread in the past.
I could see a scenario where language originally developed with click sounds and new sounds gradually as physically abilities improved.
Was there not an epigenetic study recently that showed major differences in gene expression for the voice box between Neanderthals and modern humans as well?
The first paragraph is a breathtaking example of scientific curiosity in action. Wow. Nobody is as good at opening minds to fresh ways of thinking than John Hawks. I found it impossible to read further without first taking time to consider the implications of his words.
Hawks writes so clearly that even though I lack a background in genetics, I was fascinated by how generously he incorporates and credits the work of others to suggest a new approach to understanding the evolutionary history of Homo sapiens. I love that Hawks concludes his piece by asking, “What do you think?” I’m not sure what I think. I only know that I’d love to get these researchers together in one room and ask them all kinds of questions!
Such a kind comment, I appreciate so much!
I think the evidence of the continental separation can be found in the modern population structure as well. If one looks at African-only populations in the PCA space, it consist of three main clusters (not counting much smaller ones for Koisan and pygmies). One is clearly Eurasian-derived, I would call it "Egyptian" although any NE Africa will do. The second is "Bantu-Yoruban" (B) and the third is Nilo-Saharan or "Dinka" (D). Everything else is connections (clines) between these clusters. B and D are very far from E, so we have a kind of triangle EBD with EB and ED sides very long (EDis a bit shorter), but BD is much shorter. Small clusters for Khoisans, Mbuti and Biaka are closer to B and very far from E, Hadza is a bit closer to D.
So the whole African diversity is mostly a cline (or rather two clines), connecting Eurasian-derived ancestry with very deep African ones, with separation between them corresponding to at least many hundreds of thousands years of mostly separate development. There is simply not an African cluster to be seen that may serve as the source of "recent exit" into Eurasia.
I think the "Recent Out-of-Africa" dogma blindsided paleoanthropology for more than 30 years and needs a major revision, not just some patches. It looks like human diversity is far greater than thought before with a lot of separate developments in different regions and separate deeply archaic admixtures in each regions.
This is a very profound, thought-provoking read!
It is remarkable that in the last few years several different computational models converge on similar answers for very ancient history of humanity, but generate new puzzles.
After reading the "branches A/B" paper I thought there must be geographical structure in these branches, it is very unlikely that they spread evenly everywhere. And such structure can be found indeed, at least in the model values, though it is buried in the supplement and I don't remember ever mentioned in discussions.
The timings of split and merge were separate for different populations, with the most contrast between African and Eurasian ones. For Eurasian populations the split/merge timings had max probability ~1800 and 380 Kya respectively, while for African populations - ~1000 and 190 Kya.
I haven't seen separate population estimates for A/B fractions (maybe they are buried among other charts) but I think there is very little chance that it doesn't have a strong geographical structure while othe parameters do.
This in my view makes an overwhelming case for the continental separation of A and B branches, with A being mostly a Eurasian one and B - African one. The timings parameters may also a scenario with 3 populations, including an intermediate one between A and B, such as "Southwest African" (B), "Northeast African" (AB) and "Eurasian" (A). A exited from Africa soon after 2 mln years with early Acheulean technology and lived in Eurasia ever since. AB and A kept in contact until ~1 mln years ago but then also separated due to demographic slump. Then A and AB reconnected ~300 Kya probably with Levallois technology, but the contact with B was reestablished much later.
This would fit the geographical structure reasonably well.
There is a hint of such structure in the "2 stems" model as well.
The question of geographic structure in the two components is a central one. The data are not there yet to answer it.
The Ragsdale model enabled heterogeneous gene flow from the two stems into ancestral groups after 110,000 years ago in Africa, and their model is based on a fit that may partially include "ghost" contributions previously found by other researchers. The Cousins model and Rogers model basically assume that A and B were fully mixed prior to differentiation of later African groups, and don't resolve whether B may have persisted as a ghost.
I've seen Chris Stringer write he supports a late mixture for A and B, my instinct is more for the early mixture with possible additional ghosts, but I wouldn't say the data speak strongly to this.
All of these models are basically island models in form, and what probably is closer to reality is some kind of isolation-by-distance model with dynamics over time. So there's a good bit of oversimplification in thinking about what 80-20 mixture actually means. If there is structure that we could detect in the ratio of the A and B stems in present-day Africans that would lead to a test of the biogeography of mixing.
Is it not possible to get uni-parental genetic markers(Y and mtDNA) from the many fossil records of Denisovans, Neanderthals and other archaic species like Homo Floriensis / Luzonensis. Maybe we can infer something by looking at the diversity of those markers and trace their evolution/migration just as we do for humans. There can also be a search focused on finding the Indian/Australian fossil records of early hominids. Also if possible can modern scientific methods not extract DNA from Jebel Irhoud just as we did with the Denisovan fossils and also establish correct dates for the remains. There are fossils in Europe/China/Australia which had shown earlier dates than what Out of Africa had permitted(Liujiang man comes to mind as do certain fossils in Europe).
Also for a definitive answer to human origins, migration and dispersal we would need to look at the Y DNA and mtDNA. I have repeated this on other places but I always find the distribution of certain Y DNA clades is way too weird(eg Onge and Japanese D clade). While reading up on the origin of Y and mtDNA clades, I found there was a proposal that Y DNA diversification of most non-Africans might have either taken place in the Arabian Peninsula ie South West Asia or in South East Asia. Also there was also an idea that the very basal mtDNA clade of L originated in South East Asia rather than East Africa. For instance, the initial findings on Mungo Man LM3 reported a very diverged clade of mtDNA (possibly indicating that his mtDNA was more archaic than what Out of Africa could explain?). Later that finding was overturned due to "possible contamination" but did they conclusively rule out the earlier findings and determine the mtDNA clade.
Also from Wikipedia, I learnt that the JC virus tracking places us in Asia rather than Africa. Is this not a cause of concern for people in the palaeo-anthropologic field as to why or how this disparity is occurring?
Finally, someone(Mondal et al) in 2019 used AI and determined that there was an evidence of another ghost population which contributed to human evolution and origins. With more sophisticated hardware, algorithms can we not revisit that possibility.
TL;DR: Greater scrutiny on genetic rather than morphological records, especially parental markers would probably solve human origins.