Statistical overdrive?

In the 250 page book at reference 1, near thirty years old now, Mithen covers a lot of ground in an accessible way. Which is all well and good, but how careful has he been? Has the book stood the test of time?

With this in mind, I have taken a look at one of his assertions, built around a slightly adapted version of the figure above, taken from reference 2, with his version to be found on his page 12, near the beginning of chapter 1. He reads into this figure two spurts of brain enlargement: the first 2.0-1.5mya and the second 0.5-0.2mya, the two gaps in the record notwithstanding. Two spurts which I assume he will go on to build into his story about the evolution of the human mind. A figure which looks more like two gaps than two spurts to me.

Lots of other people have wondered about why exactly the brain got bigger and about how exactly that bigger brain was able to do new stuff, the new stuff which set us apart from other apes, not to mention all the other animals. Then why did this happen to humans but not to the others?

Putting it another way, it is clear that over the last couple of million years or so, humans have evolved to dominate the world, with intellectual powers of a quite different order from those of all other animals. Then over the same period, our brains have got a lot bigger. So what exactly are the selective pressures which drove this evolution? How exactly does bigger brain translate into more intellectual powers? Then was this evolution a smooth progression or did it happen in spurts?

With the result that when I start to dig, Bing rapidly turns up lots of stuff, including references 3, 4 and 5.

Reference 3, from the Smithsonian Institute offers an accessible introduction to the big brains of humans. It also offers a figure very similar to that offered by Mithen – but with a quite different spin. It puts the rapid acceleration of brain growth 0.8-0.2mya down to the need to adapt to the rapidly fluctuating Pleistocene climate. Part of this may have been improved social functions coming with these bigger brains: these newer humans could get along in bigger groups, the better to face the challenges of a changing world. Indeed, some workers see a strong correlation between the size of brains and the size of groups. And others see large-brain-enabled language as a better way to do social, to do groups, than grooming – something of which the other apes do a lot and which is very expensive in time.

Reference 4 is a bit anatomical for present purposes, but it does remind us that brain size is a popular proxy for progress because it is often available from the fossil record. We can measure how big a brain case is, even if we have little idea otherwise of what the brain that was once inside got up to. Some workers even look for lumps and bumps inside the brain case, rather in the way of phrenologists of old.

Reference 5 covers some of the same ground, but in a more accessible way. And it reminds us that of the 20 or so hominid species which turned up in the last 6 million years or so, only our species, Homo sapiens, has survived. With the Neanderthals hanging on until around 30,000 years ago – with their eventual failure to adapt being a point of particular interest. The suggestion that the diminutive Homo floresienses survived until quite recently appears to be a combination of typo and error: the story now seems to be that they died out well before the Neanderthals died out. As Kaas puts it, the evolution of big brained species did not always work out. 

So having acquired some background, it was time to turn back to reference 2, the source of Mithen’s brain sizes. Where I read that a lot of what we are can be explained as a consequence of our move out of the trees and into bipedalism. Which also resulted in bigger brains and bigger groups along the way. With the emphasis on bigger groups taking me back to Freud, his essay noticed at reference 9 and the need to balance the needs of the individual with those of the group.

Bigger brains and bipedalism were also mixed in with a lot of changes to the pelvis, which led me to references 6 and 7. Changes to the pelvis which affected the size and shape of the birth canal – which makes the birth of humans difficult and makes early birth desirable: a human needs to be born before the brain gets too big, which means in turn that humans, compared with many other animals, are born in a rather immature condition. They need a lot of care for a long time. But maybe the brain is all the better for this period of development in the outside world; it is learning to use stuff which was not available inside. All of which interacts with the size of the brain and the development of the systems supported by the brain. Bearing in mind that being born too immature is also bad, leading to all manner of problems. Design compromises everywhere.

Cranial capacity and body mass

One way to attack all this complexity was to look into how much of brain size was explained by body size with reference 11 being one such look, from near fifty years ago, and reference 12 a digression therefrom. Reference 11 rapidly gets too deep for me, but it starts off with the notion that brain size can be usefully modelled as a power of body size – yielding a straight line relationship – and linear regression – if you take logarithms.

A much more recent look at this is to be found at reference 8. A paper which attacks a small database of human remains with a great deal of computing and statistical machinery. The key variables are brain size, body size, species and age of find. Most of what follows is taken from this paper.

The small database

This database contains records for 615 finds of hominins, aka early humans. For 285 of them we have estimates of cranial capacity and for 431 of them we have estimates of body mass. For 101 we have both. But we are not offered much analysis of this database and various items are missing which I would have found helpful.

For example, more information about individual finds would have been helpful: are we talking about a more or less complete skeleton, a more or less complete skull (say less the lower jaw), a bit of femur or a tooth? The weight of the find would have been a helpful proxy for all this. Then there is age and sex, presumably mostly not available.

Denisovan finds appear to have been omitted from this version of the database, possibly because neither body weight nor cranial capacity can be estimated from the modest finds that we have. Denisovans were spread across Asia – while their DNA spread rather further – and they were roughly contemporary with the Neanderthals.

Chimpanzees and gorillas are also omitted.

All that apart, the variables of present interest are age of find (mya), species, cranial capacity (cc) and body mass (kg). The table above analyses the 615 finds by species, with indication of whether cranial capacity is present or not and average age. This last is the average value of the mean of upper and lower limits of age for each find – so just an indicator of age rather than something more concrete.

But it is clear that the finds are heavily weighted towards Homo sapiens and Homo neanderthalensis and that there are large gaps in the record. Meaning that one needs to be careful about drawing conclusions about what has happened over the past few million years.

The table above analyses the 615 finds by age in 500my bands. Column 2 counts finds, column 3 finds with cranial capacity and column 4 finds with body mass.

This database claims to be an exhaustive record of early human fossil finds but I have made no attempt to check that claim.

Taxonomic trees

The version of the tree given in the supplementary information. We get gorillas and chimpanzees but not bonobos. I don’t know why bonobos, our other near relative, are missing and the word ‘bonobo’ appears in neither text nor supporting text. Maybe because not many of them survive: there are a lot more chimpanzees about.

This from the Wikipedia article on Hominidae. The bonobos are small chimpanzees, otherwise Pan paniscus; sometimes eaten by the big chimpanzees, Pan troglodytes.

A zoom into an upper right hand part of the first tree. I believe that the horizontal black lines show the estimated interval in which the species existed, that the vertical black lines show the mean estimate of the date of the lineage splitting event and the coloured horizontal lines say something about the quality of the estimates of something or other. So at node 17, the ergaster lineage split off. And at node 18, a few hundred thousand years later, the other part of that lineage split into the naledi and floresiensis lineages. All subsequently extinct, but maybe not before there was a spot of out-breeding.

A picture of punctuated equilibrium: a lineage rolls along for tens of thousands of years, then quite quickly splits into two branches. I have not attempted to check how fair a picture this is.

The version of the tree given in the paper proper. Denisovan find omitted as there is no cranial capacity. The gorillas and chimpanzees also omitted. Two charts which, to my mind, both try to do too much.

The right hand panel, however, does serve to illustrate the patchiness of the record, with, roughly speaking, three clusters at two million year intervals. With a fair bit of filling in between the last two clusters. 

While the left hand panel arranges our twenty odd species in a taxonomist’s tree, derived using statistical methods which I do no pretend to understand – beyond thinking that it is a lot of statistics for a small amount of data.

Apart from being interesting in its own right, this tree is used in generating the 1,000 datasets derived from the base dataset, derived by applying probability distributions to uncertain data items. But again, I do not pretend to understand more.

Regressions

As far as I can make out, the two panels from Figure 3 snapped above summarise the main results of the paper.

The left hand panel tells us that while brain size overall has not increased over the period, there are big differences between species and brains do grow within species, at what appears to be an accelerating rate.

I account for this counterintuitive result by saying that while average brain size rose through the period, the mean difference between species did not.

While the right hand panel tells us that while body mass and cranial capacity overall are related by a simple power law (the grey band), within species they are only weakly related, with much more variation in body mass than in cranial capacity.

Which seems to be saying that, for selective pressures driving up brain size, one should be looking at particular species rather than at the hominin lineage as a whole. Which I do not really understand, but we do seem to be going back to the sort of things which interested Russell Lande, near fifty years ago, at reference 11.

Other matters

Reference 13 was a late entrant, brought to me by EAORC. But a bit too fierce at $35.95, so a non-starter – although I dare say it will leak out from behind its paywall before too long. What I can already see suggests looking at the cerebellum separately from the brain, with the cerebellum, in the case of mammals, particularly humans, containing a lot more neurons than the cerebral cortex. Contrariwise, some intelligent birds are the other way around. I have not checked, but I would have thought that estimates of brain volume from skulls would likely include the cerebellums.

Checking might be tiresome, given that the word ‘cerebellum’ appears in neither text nor supporting text of reference 8, from where most of the foregoing material is derived.

Reference 14 is another late entrant, in the sense that I only took a look towards the end of all this, inter alia, a reminder that big textbooks are indeed useful. At the very end of this one there is talk of ‘three major suites of adaptive shifts’ which took us away from the other apes – language which recognises the complexity of evolution. The three being the shift from dense forest to more open savannah, the shift to a higher energy regime and the shift towards bigger brains. With more reproduction being mentioned in connection with more energy, clearly a good thing in terms of selection.

With one complication being that the dental apparatus got bigger when we came down from the trees and relied less on fruit – and then smaller when we started on meat, particularly when it was cooked.

Lieberman suggests that brain sizes only really started to increase from around 2mya and then, relative to body size, only when there was plenty of energy available, say from the move to a more meaty diet around 1mya. Which seems to be broadly consistent with the story at reference 8.

All in all, a useful reminder that there is a lot more to the evolution of the hominin head, never mind the body, than the evolution of the brain.

Not yet got a grip on how the level of analysis affects all this, with the snap above attempting to summarize my present confusion.

I have read at reference 11, for example, that the slope of the allometric curve relating brain size to body size very much depends on taxonomic level. Very roughly speaking, the higher the level the steeper the slope.

But how does this play into the punctuated equilibrium mentioned above?

The Denisova Cave, in the Altai mountains, where the first reported Denisovans were found. Lifted from Wikipedia. To be found at gmaps reference 51.395697, 84.6770401.

Conclusions

I have not found anything which gives much support to the Mithen interpretation of the patchy record on brain size as showing two spurts of growth. But there is support for his focus on social skills and group size. Otherwise, I have found a great deal more complexity than I can cope with.

And I wonder whether statistical and computational wizardry has got a bit out of hand, a bit out of proportion to the data. I also worry that, for example, mixed teams of archaeologists are relying on the application of techniques which they only dimly understand. Most of them are going to have to take a lot on trust.

From where I associate to the impressive results of large language models – like Copilot (Microsoft) and Gemini (Google) – where one might like what one gets, but without having much understanding of how it was all done, how it all works. And to the workings of particle accelerators, where one infers the existence of new particles from odd blips on obscure graphs.

Work which may progress.

References

Reference 1: The prehistory of the mind: a search for the origins of art, science and religion – Mithen, S. J – 1996. 

Reference 2: Terrestriality, bipedalism and the origin of language – Aiello L – 1996. 

Reference 3: https://humanorigins.si.edu/human-characteristics/brains. 

Reference 4: https://en.wikipedia.org/wiki/Evolution_of_the_brain.

Reference 5: The Evolution of Brains from Early Mammals to Humans – Jon H Kaas – 2014. 

Reference 6: The runaway brain – Wills C – 1995. Might help with the pelvic evolution that Aiello talks about. 

Reference 7: The evolution of the human pelvis:  changing adaptations to bipedalism, obstetrics and thermoregulation – Laura Tobias Gruss, Daniel Schmitt – 2014. 

Reference 8: Hominin brain size increase has emerged from within-species encephalization – Thomas A. Püschel, Samuel L. Nicholson, Joanna Baker, Robert A. Barton, and Chris Venditti – 2024. 

Reference 9: https://psmv5.blogspot.com/2025/03/interim-report-no1.html. 

Reference 10: Evolution of early Homo: An integrated biological perspective - Susan C. Antón, Richard Potts, Leslie C. Aiello – 2014. 

Reference 11: Quantitative genetic analysis of multivariate evolution, applied to brain: Body size allometry – Russell Lande – 1979. May help with clarifying intra and inter specific regressions. Reference 46 at reference 8 above. 

Reference 12: https://psmv5.blogspot.com/2025/04/regularities.html. 

Reference 13: How do big brains evolve? - Cristián Gutiérrez-Ibáñez, Pavel Němec, Martin Paré, Douglas R. Wylie, Louis Lefebvre – 2025. Presently inaccessible.

Reference 14: The evolution of the human head – Daniel Lieberman – 2011. A book of more than 600 pages of text, more than 700 pages overall.