I think most of us have a real difficulty appreciating the possibile events that humans can experience over 100,000 years. For any population, there were abundant chances of "perfect storms" of climate change, conflict with other populations, and -- indeed -- perfect storms. Even very rare events can occur many times over several hundred thousand years.
Both American continents were populated in less than 20,000 years. The Pacific islands were populated in much less time. There would have been many failures in trial-and-error that do not leave any traces.
This is a long way to say that the actual story is much more complicated than was described a few decades ago. Time and space were both very big, with lots of possibilities.
On the other hand, it also is difficult to understand why there were so few technological changes in the physical record over the millennia. Of course, we know little about wooden and fabric technology, and of the people who were not living in caves.
We look forward to the ongoing research and the rich complexity of our ancestors.
Thanks for this comment! A way of looking at the odds is that success is a long tail distribution with a few favorable events and circumstances making a lot of difference on the timescale of millennia.
The flip side of this is the most of the individuals and circumstances that archaeologists happen to find are unlikely to be these massive successes. Most of our sites come from people who succeeded in modest terms, who would have been entirely typical of their time, but who were ultimately doomed.
Their strategies were locally optimized, certainly to the extent they could plan them. Yet some few groups, maybe far away, were climbing higher adaptive peaks that would propel their descendants ahead of the average, exponential growth doing its inevitable thing.
Genetics traces the latter events, while archaeology mostly traces the former, and as you note this is a challenge of timescales.
One idea I like (sorry, no attribution) is that in Asia the use of bamboo might have driven cultural development differently from that in bamboo-deficient Europe. Perhaps microscopic analysis of Denisovan tools could reveal bamboo-specific use.
I've read the Reich's preprint and this post again, and it is an interesting indirect exchange of views.
Reich contrasts his model with the idea of a "single episode" of encounter between early HS and N some 300 Kya, and I think he is spot on here - it was likely a long and gradual introgression of HS genes into early N territory. He is puzzled by the fact (as he mentioned many times) of a complete replacement of N mtDNA and Y-DNA by early HS but only modest autosomal input, and I think his simulation model hadn't fully answered that - the actual data is reproduced in the model only with an (unrealistically) strong survival advantage of HS mtDNA relative to N mtDNA. Otherwise there would be much more of HS autosomal admixture into N that would accompany the sweep.
Your view on the matter is that a sweep by uniparental lines can happen at a lower level of admixture and relatively fast. I think it is a valid point and it is known that uniparental time depth is much lower than the whole genomic one. Yet it is not clear if a sweep of both uniparental lines can occur at such a small level of autosomal admixture. Maybe we underestimate the level of admixture from early HS into N.
What I strongly agree with Reich is the idea of the connection of the spread of Levallois technology and other MP innovations with the demographic expansion of some early HS population (during MIS 11-9 time frame) that produced genetic connections both with N and some African archaics (and the ideas of "80-20 stems"). I don't think the lack of HS admixture in contemporary Denisovans that also used Levallois tools disprove it - they could get it from N later.
I actually expressed fairly similar ideas (though without a simulation model) in my post
Regarding the uniparental systems, no disagreement here. In the end this is testable and I expect we'll know once we have enough aDNA from later Middle Pleistocene contexts. There's enough remaining uncertainty about alignment of mtDNA, YDNA and autosomal chronologies for modern human expansions that it's unlikely we have the power to know for sure how different their coalescence times are.
For the Levallois/prepared core problem, the disagreement I have with the Reich idea falls along two lines of my thinking.
The widespread presence of these reduction strategies in Denisovans and possibly other hominins is one of them. There is a widespread assumption that technologies and hominin populations (species in many people's view) can be reliably associated. In this way of thinking, Acheulean tools are markers of Homo heidelbergensis and Homo erectus, Mousterian a marker of Neanderthals, and so on. Reich's idea would break Neanderthals in half: the "real" Neanderthals after Levallois arrives and the "pre" Neanderthals before that. This follows from the assumption that technology is a reliable marker of population.
I have a deep unease with this assumption. The data on association are really not very good. The widespread presence of Levallois and prepared core strategies more broadly in East and South Asia is an easy thing to point out. The absolute lack of information about which hominin populations made most of the MSA weighs strongly on me, because I'm busy studying African Middle Pleistocene hominins that probably made some of these early MSA tools.
The other line of my thinking is about the evolution of technology. Consider "Levallois" or "Acheulean": named technostrategies or tool traditions. John Shea and other archaeologists have written and thought critically about how these named tool traditions are actually misleading people about what they are. Levallois is not particularly special. It's not an idea that occurred once and was spread from a single source. It is a limited way of describing one method for making large flakes that depends on a hierarchical organization of flaking. It and other hierarchical strategies emerge repeatedly in Early Pleistocene and Middle Pleistocene contexts with large cutting tools like handaxes. The reason for this repeated emergence in biface-dominated contexts is clear: If you take biface reduction to a certain point, the only way to proceed is with removal of large thinning flakes. Large thinning flakes themselves are incredibly useful, and they're tricky to make. Making them on a smaller core than a biface core is a pretty useful thing to be able to do, and hominins solved this problem many times, in a few different ways.
The East Asian contexts are really interesting because they expand this even further. There you sometimes get discoidal cores and Levallois or Levallois-like reduction combined with an otherwise Oldowan-looking background. There's a pathway here that mostly skips the bifaces altogether.
In other words, we know many circumstances in which independent invention is very likely to happen, and we have abundant evidence that independent invention did happen again and again. As a population marker, this is not helpful.
Now, that does not mean that understanding the technology is useless for building hypotheses about the hominins. Repeated independent invention during the time period between 400,000 and 250,000 years ago might be a marker for an acceleration of invention, which might implicate social or cognitive changes that did evolve. Gene flow might be involved in this process of evolution. Greater connectivity of populations and dispersal of one or more from Africa may be the pathway of this gene flow.
However I wonder whether the archaeological data are good enough to actually test whether the period between 400,000 and 250,000 is actually a different pattern, say, when compared to the equal-length period between 550,000 and 400,000. These datasets are really poor. That's a subject for a longer post.
This is extremely interesting. Most of the recent genetic models show a major population expansion around 350-250 Kya from a low level during the 1000-500 Kya period. After 200 Kya there was a gradual decline, and then a major decline during MIS 4 (or post-Toba). Real picture surely had more ups and downs (e.g. there likely was an expansion during the Eemian interglacial), but models lack sufficient resolution. Archaeology also shows a major increase of finds after 400 Kya. So it looks like something very important occurred during 400-350 Kya and a new technology (Levallois) was a big part of that. What is very intriguing is that it started to appear all over the places in this period – India, Armenia, Europe, East Africa. It looks like the demographic expansion led to rapid spread of the new technology and genetic connections to previously isolated populations, such as early Neanderthals and African archaics. Remarkably, the recent “80-20 two branches” model shows a merger of these branches around this time. As you wrote, this must be more than just “Pan-African”, this historically crucial process surely involved huge regions both in Eurasia and Africa. I think the development in this research area is incredible and we are starting to approach much more realistic history of early humanity.
I think most of us have a real difficulty appreciating the possibile events that humans can experience over 100,000 years. For any population, there were abundant chances of "perfect storms" of climate change, conflict with other populations, and -- indeed -- perfect storms. Even very rare events can occur many times over several hundred thousand years.
Both American continents were populated in less than 20,000 years. The Pacific islands were populated in much less time. There would have been many failures in trial-and-error that do not leave any traces.
This is a long way to say that the actual story is much more complicated than was described a few decades ago. Time and space were both very big, with lots of possibilities.
On the other hand, it also is difficult to understand why there were so few technological changes in the physical record over the millennia. Of course, we know little about wooden and fabric technology, and of the people who were not living in caves.
We look forward to the ongoing research and the rich complexity of our ancestors.
Thanks for this comment! A way of looking at the odds is that success is a long tail distribution with a few favorable events and circumstances making a lot of difference on the timescale of millennia.
The flip side of this is the most of the individuals and circumstances that archaeologists happen to find are unlikely to be these massive successes. Most of our sites come from people who succeeded in modest terms, who would have been entirely typical of their time, but who were ultimately doomed.
Their strategies were locally optimized, certainly to the extent they could plan them. Yet some few groups, maybe far away, were climbing higher adaptive peaks that would propel their descendants ahead of the average, exponential growth doing its inevitable thing.
Genetics traces the latter events, while archaeology mostly traces the former, and as you note this is a challenge of timescales.
One idea I like (sorry, no attribution) is that in Asia the use of bamboo might have driven cultural development differently from that in bamboo-deficient Europe. Perhaps microscopic analysis of Denisovan tools could reveal bamboo-specific use.
I recently attended a great talk about the bamboo idea by a couple of good colleagues, and I am hoping they will publish their ideas.
I've read the Reich's preprint and this post again, and it is an interesting indirect exchange of views.
Reich contrasts his model with the idea of a "single episode" of encounter between early HS and N some 300 Kya, and I think he is spot on here - it was likely a long and gradual introgression of HS genes into early N territory. He is puzzled by the fact (as he mentioned many times) of a complete replacement of N mtDNA and Y-DNA by early HS but only modest autosomal input, and I think his simulation model hadn't fully answered that - the actual data is reproduced in the model only with an (unrealistically) strong survival advantage of HS mtDNA relative to N mtDNA. Otherwise there would be much more of HS autosomal admixture into N that would accompany the sweep.
Your view on the matter is that a sweep by uniparental lines can happen at a lower level of admixture and relatively fast. I think it is a valid point and it is known that uniparental time depth is much lower than the whole genomic one. Yet it is not clear if a sweep of both uniparental lines can occur at such a small level of autosomal admixture. Maybe we underestimate the level of admixture from early HS into N.
What I strongly agree with Reich is the idea of the connection of the spread of Levallois technology and other MP innovations with the demographic expansion of some early HS population (during MIS 11-9 time frame) that produced genetic connections both with N and some African archaics (and the ideas of "80-20 stems"). I don't think the lack of HS admixture in contemporary Denisovans that also used Levallois tools disprove it - they could get it from N later.
I actually expressed fairly similar ideas (though without a simulation model) in my post
https://kirillpankratov.substack.com/p/two-million-years-of-human-history
I think this period 400-250 Kya represents a critical juncture in human prehistory and our understanding of what happened will change in a major way.
Really outstanding comment, thanks!
Regarding the uniparental systems, no disagreement here. In the end this is testable and I expect we'll know once we have enough aDNA from later Middle Pleistocene contexts. There's enough remaining uncertainty about alignment of mtDNA, YDNA and autosomal chronologies for modern human expansions that it's unlikely we have the power to know for sure how different their coalescence times are.
For the Levallois/prepared core problem, the disagreement I have with the Reich idea falls along two lines of my thinking.
The widespread presence of these reduction strategies in Denisovans and possibly other hominins is one of them. There is a widespread assumption that technologies and hominin populations (species in many people's view) can be reliably associated. In this way of thinking, Acheulean tools are markers of Homo heidelbergensis and Homo erectus, Mousterian a marker of Neanderthals, and so on. Reich's idea would break Neanderthals in half: the "real" Neanderthals after Levallois arrives and the "pre" Neanderthals before that. This follows from the assumption that technology is a reliable marker of population.
I have a deep unease with this assumption. The data on association are really not very good. The widespread presence of Levallois and prepared core strategies more broadly in East and South Asia is an easy thing to point out. The absolute lack of information about which hominin populations made most of the MSA weighs strongly on me, because I'm busy studying African Middle Pleistocene hominins that probably made some of these early MSA tools.
The other line of my thinking is about the evolution of technology. Consider "Levallois" or "Acheulean": named technostrategies or tool traditions. John Shea and other archaeologists have written and thought critically about how these named tool traditions are actually misleading people about what they are. Levallois is not particularly special. It's not an idea that occurred once and was spread from a single source. It is a limited way of describing one method for making large flakes that depends on a hierarchical organization of flaking. It and other hierarchical strategies emerge repeatedly in Early Pleistocene and Middle Pleistocene contexts with large cutting tools like handaxes. The reason for this repeated emergence in biface-dominated contexts is clear: If you take biface reduction to a certain point, the only way to proceed is with removal of large thinning flakes. Large thinning flakes themselves are incredibly useful, and they're tricky to make. Making them on a smaller core than a biface core is a pretty useful thing to be able to do, and hominins solved this problem many times, in a few different ways.
The East Asian contexts are really interesting because they expand this even further. There you sometimes get discoidal cores and Levallois or Levallois-like reduction combined with an otherwise Oldowan-looking background. There's a pathway here that mostly skips the bifaces altogether.
In other words, we know many circumstances in which independent invention is very likely to happen, and we have abundant evidence that independent invention did happen again and again. As a population marker, this is not helpful.
Now, that does not mean that understanding the technology is useless for building hypotheses about the hominins. Repeated independent invention during the time period between 400,000 and 250,000 years ago might be a marker for an acceleration of invention, which might implicate social or cognitive changes that did evolve. Gene flow might be involved in this process of evolution. Greater connectivity of populations and dispersal of one or more from Africa may be the pathway of this gene flow.
However I wonder whether the archaeological data are good enough to actually test whether the period between 400,000 and 250,000 is actually a different pattern, say, when compared to the equal-length period between 550,000 and 400,000. These datasets are really poor. That's a subject for a longer post.
This is extremely interesting. Most of the recent genetic models show a major population expansion around 350-250 Kya from a low level during the 1000-500 Kya period. After 200 Kya there was a gradual decline, and then a major decline during MIS 4 (or post-Toba). Real picture surely had more ups and downs (e.g. there likely was an expansion during the Eemian interglacial), but models lack sufficient resolution. Archaeology also shows a major increase of finds after 400 Kya. So it looks like something very important occurred during 400-350 Kya and a new technology (Levallois) was a big part of that. What is very intriguing is that it started to appear all over the places in this period – India, Armenia, Europe, East Africa. It looks like the demographic expansion led to rapid spread of the new technology and genetic connections to previously isolated populations, such as early Neanderthals and African archaics. Remarkably, the recent “80-20 two branches” model shows a merger of these branches around this time. As you wrote, this must be more than just “Pan-African”, this historically crucial process surely involved huge regions both in Eurasia and Africa. I think the development in this research area is incredible and we are starting to approach much more realistic history of early humanity.