The brain of the hobbit

News story on MSNBC

News story on BBC

Michael Balter story in Science


The original post is listed below. A reader has kindly forwarded the text of the Falk et al. (2005) article, so I add a few comments to the original. The paper is a very nice description of the endocast, with pretty much every detail covered whether in the affirmative or negative. If a similar paper could be published about each of the skeletal elements, they would constitute a very decent monograph. My comments focus on the pathology issue and the analyses of shape.

As Maciej Henneberg has pointed out online, the study includes a specimen with primary microcephaly as a comparison. Henneberg and others have suggested the specimen may have secondary microcephaly. According to Falk et al. (2005:1):

The only criterion for secondary microcephaly is an occipitofrontal circumference below -2 standard deviations for age and sex, but these data are unavailable for LB1's population. Unless a Homo erectus-like endocast shape is characteristic of an unrecognized form of secondary microcephaly, we reject the hypothesis that LB1 was a pathological microcephalic.

And in the Supplementary material:

Since LB1 lacks the diagnostic head shape associated with [primary microcephaly] and lacks the gyral morphology associated with [microcephaly with simplified gyral pattern], its interpretation as a microcephalic can only be made by claiming that it is a secondary microcephalic. This amounts to saying LB1 is small-headed (literally microcephalic) because it is small-headed, which does not lend itself to hypothesis testing.

It seems to me this passage pointedly glosses over the troublesome parts of the pathology hypothesis. As defined here, secondary microcephaly does not result in the gross abnormalities in brain shape that can result from primary microcephaly. Instead, it is defined by brain size. Clearly the brain size of LB1 would qualify as microcephalic under this definition -- even if it were a member of a Homo erectus-like population.

So what response is there to this? Firstly, we don't know what brain size the rest of the LB1 population had. If they were all small like LB1, then clearly that specimen can't be microcephalic. And secondly, LB1's endocast is abnormally-shaped; it is shaped just like a Homo erectus endocast. If secondary microcephaly doesn't result in abnormal brain shape, then it certainly can't produce a Homo erectus-shaped brain.

I don't know about anyone else, but that argument doesn't satisfy me. Basically, what I wrote below still holds, and I would suggest a working relationship between a pathologist specializing in brain disorders and a paleopathologist who can give some context on the ways that Stone Age lifestyles can alter the characteristic symptoms of developmental abnormalities.

Endocast shape

Falk et al. (2005) provide a series of metric comparisons to give an idea of the similarities in endocast shape between LB1 and other fossil hominids and extant hominoids. The major method of comparison is by principal components analysis, performed upon endocast ratios and/or measurements. In a first analysis, the gross measurements of size (length, breadth, height, and frontal breadth) were tabulated into six bivariate ratios, to which principal components analysis was applied. The second principal components analysis was performed on six measurements of the endocast base.

The reason for doing a PC analysis of ratios escapes me entirely, since there is no information in the four ratios that is not already in the measurements. One possible justification is to discard information, with the notion that ratios will be scale-free. But several of these ratios appear to be quite strongly correlated with cranial size. And there are only three important PCs for a four variable analysis. At any rate, the second PC in this analysis is almost entirely determined by the microcephalic, so the inclusion of that specimen really dominates the overall variation in the samples. Meanwhile, within-population variability is essentially assumed away by the use of population means for humans and chimpanzees instead of real data. In the end, the analysis appears to show that LB1 is most similar to ... the Black Skull! And next most similar to ... chimpanzees! In my opinion, this analysis isn't telling us about anything other than cranial size -- WT 17000 comes out as similar because it is the most similar in size (410 cc).

The second analysis is the one that places LB1 near humans. Again, this is a PC plot based on ratios, in this case including some values that express the difference between two linear measurements as a ratio with a third linear measurement. LB1 is similar to Homo because of the third PC, which is loaded heavily on the distance to the olfactory bulb from the rostral end of the endocast, and the distance between the rostral and caudal ends of the olfactory bulbs. The reason for this similarity is not obvious to me, but it may be important. In any event, it would help to have ellipses of real data instead of just species means, so the variability in the values could be evaluated.

Cognitive capabilities

The lede of the story is the idea that the convolutions on the endocast of LB 1 show that its brain may have had relatively advanced cognitive abilities. In particular, the temporal lobes are relatively large, and the frontal lobe is fairly strongly convoluted. Dean Falk says that the greatest convolutions correspond to Brodmann's area 10, "which is expanded in modern humans and is involved in undertaking initiatives and planning future actions -- key components of higher cognition" (Balter 2005:1386).

According to the story:

This enlarged area suggests that the little Flores people may well have been capable of creating the stone tools that were found near them, which are more typical of those made by prehistoric modern humans than earlier hominids including Homo erectus. "The real take-home message here is that advanced behaviors, like making sophisticated stone tools, do not necessarily require a large, modern, humanlike brain," says [Fred] Spoor. "It can be done by reorganizing a small brain, with convolutions and rewiring, and this goes to the heart of our understanding of human evolution." (Balter 2005:1386)

The idea that there are huge frontal convolutions here that are absent in H. erectus certainly is not evidence of cognitive function. That is to say, if they are new in this putative species compared to its ancestor, then they cannot be said to be homologous with modern human frontal convolutions, which means we can conclude nothing about their function. If, and that's a big if, the brain actually shrank from a larger brain in an ancestor species, then to maintain the same function it might well have increased in convolutions, thereby retaining a more similar surface area (and thereby neurons) with respect to volume. These convolutions would not attest to advanced cognitive skills, except in the sense that they might be beyond those expected for a 380 mL brain.

For me, the key facts here are the ones that are missing. There is no suggestion that Broca's area was evident. Considering that KNM-ER 1470 has a pronounced Broca's area, this is potentially important. Endocasts have little to say about brain function in the first place, but this endocast in particular has little to set it apart from any early hominid.

UPDATE: Falk et al. (2005:2) say this about Broca's area:

[W]e decline to identify rami that border the human pars triangularis (part of Broca's area) on the left, although the general morphology in this region would be consistent with their existence.

No smoking gun, in other words.

Of course the key evidence that the Flores hominids had advanced behavior is the archaeological record of that behavior. But there really is no reason to think the archaeology at Liang Bua was made by the hominids in the cave (even if they all represent the same population). The tools really are more typical of those made by modern humans. And there is every reason to think that by 18,000 years ago (and quite probably by 70,000 years ago or earlier) there were modern humans on Flores. There were modern humans in Australia by 50,000 years ago, and modern humans in island Melanesia by 30,000 years ago -- both places reached by oceangoing boats. In this context, it is extraordinary to suggest that the modern humans weren't responsible for the modern human-looking tools. So the archaeology is far from evidence of cognition for the LB 1 individual or its putative population.


The paper provides a graphic comparison of the LB 1 endocast with the virtual endocast of a modern human microcephalic. They are not the same shape, but they are approximately the same size. The paper argues that this shows that the skull did not have microcephaly.

Personally, I think this interpretation is premature. Pathologies may affect modern humans in roughly consistent ways, but there is no reason to think they would affect an ancient human skull in exactly the same way. There are just too many potential variables when we turn to humans who did not live a modern lifestyle. We find this with every other kind of paleopathology, and I would expect microcephaly to be no different. Every case may be sufficiently ideosyncratic to defy generalization.

But we don't even have data about the distribution of brain shapes in modern microcephalics. Microcephaly is not only a disorder of the brain; it affects many aspects of cranial shape. If these are developmentally different to any substantial extent, the brain shape may be radically different. The same is true when we consider artificial cranial deformation -- the gross shape of the brain says little about what most of the genes would have it look like, and a few mechanical factors can greatly alter it. Again, the external shape of the brain just doesn't say that much about its wiring, and we really know too little to say anything with confidence about its development in this case.

The Balter article quotes Fred Spoor as saying, "Colleagues advocating that [the Flores hominid] is a modern human microcephalic should start publishing hard evidence in peer-reviewed journals to underpin their claims" (1387). I can see this point -- the "put up or shut up" priniciple -- but I just don't think we are there yet. Has anyone involved any experts in human brain pathologies in studying the skull yet? Is there even a single expert on microcephaly who has said anything about the skull? I know enough not to pretend that I can say this is or isn't pathological. I suspect that any qualified clinician would say the same; there is just no telling how microcephaly would manifest in a Stone Age specimen.

Australopithecines again

It's my hypothesis, and it won't go away. Here's what Balter writes (1388):

But the new paper urges consideration of an alternative hypothesis, that a small-brained, pre-erectus hominid managed to get to Flores in the distant past, and then, in a case of parallel evolution with modern humans, evolved a relatively advanced brain on its own. "Some of [the hominid's] traits indicate that the ancestral population may predate Homo erectus," says Morwood.

I find little to criticise here, except the idea that these hominids were necessarily cognitively advanced. That comes from the assumption that they made the tools, which I cannot justify.

The main problem with the australopithecine hypothesis is the small teeth, which did have to evolve in parallel with modern humans. But then, they also had to evolve in parallel with modern humans in the case that LB 1 descended from H. erectus, since LB 1 has smaller teeth than any other archaic hominid. Otherwise, it's a good australopithecine -- complete with a broad pelvis, long arms, and small brain and body sizes.

Now, if somebody would just give me credit for it. I really did publish it first, on this site.

More on the Flores hominids