From one microcephalic to twenty

Falk et al. (2005a) compared the LB1 endocast to one microcephalic skull and concluded it didn't match. Now Jochim Weber and colleagues (2005) have compared the endocast to nineteen additional microcephalic modern human crania. They argue that LB1 is likely microcephalic. Falk et al. (2005b) disagree. So who's right?

Weber et al. (2005) make four points:

  1. The endocranial volume of LB1 (417 ml) is well within the range of the modern human microcephalics (280 to 591 ml).
  2. Modern microcephalic endocasts are "extremely heterogenous" in their morphology, but some of them grossly resemble early hominid endocasts.
  3. One endocast in particular from their sample has a volume of 415 ml, and the ratios of its dimensions are all within 0.01 of the matching ratios from LB1. This appears to indicate an extremely close similarity of overall endocast shape.
  4. A substantial proportion (7 out of 19) of the microcephalics exhibit relative enlargement of Brodmann's area 10, the area that is also extremely enlarged in LB1

Falk et al. (2005b) criticize the latter two of these points. Taking the last one first:

Weber et al. assert that seven of their microcephalic endocasts have a relatively expanded Brodmann's area 10 similar to LB1, but none of the five microcephalic endocasts in their figure 3 reproduce the two distinct, enlarged convolutions seen in the region of area 10 in LB1 (Falk et al. 2005b:236).

Their other criticism focuses on the apparent similarity in ratios between one microcephalic and LB1. This point is somewhat involved, because it relies on a technical difference in the orientation of endocasts. As they describe, the microcephalic pictured by Weber et al. (2005) as well as that pictured by Falk et al. (2005a) both have relatively large cerebellar volume compared to overall brain volume. This results in a cerebellum that takes up a broader area in the back of the endocast.

I've Photoshopped the figures together to make the orientation clearer:

Microcephalic skull pictured by Weber et al. (2005) in the orientation of Falk et al. (2005b) (L); Same skull in orientation of Weber et al. (2005) (C); LB1 in orientation of Falk et al. (2005a) (R). For all three, the position of the transverse sinus is marked "TS".

The disproportionate reduction of the cerebrum compared to the cerebellum in these microcephalics complicates the comparison of endocast shape. For Falk et al. (2005b), the main implication is that an incorrect orientation may result in an incorrect measurement of endocast dimensions, and thereby to incorrect ratios of those measurements. The end result may be that the "closely matching" ratios may not be so close.

On the other hand, the shape of the forebrain in these endocasts is still broadly similar, and the size of the microcephalic is if anything smaller than LB1. Thus, the disagreement comes down to anatomy, and questions that cannot be answered by ratios.

For example, even though the cerebellum of LB1 appears relatively small compared to those of the microcephalics illustrated in these studies, it actually appears relatively large compared to humans, Homo erectus and chimpanzees pictured in Falk et al. (2005a). Is this a consequence of allometry, or is it actually abnormally large relative to brain size? Are there microcephalic crania whose cerebellar size is more similar to LB1? Are there any modern humans that have pronounced area 10 convolutions like LB1?

What I think

I'm not sure we will ever have answers to some of these questions, at least those requiring comparison to microcephalic crania.

Of course, there is at least one kind of "smoking gun" proof: if we find another endocast of similar size. Unless another small-brained specimen were found in a death embrace with LB1, the existence of a second small endocast by itself would be enough to show that a small-brained population really existed.

But is there any other kind of smoking gun evidence? Suppose for example that we found a modern microcephalic skull in a museum that was identical to LB1. Would that be enough to prove the skull was pathological?

As Weber et al. (2005) note, some microcephalics approximately have the shape of early hominid endocasts. Even if it were pathological, LB1 is clearly one of those: its shape is grossly similar to that of H. erectus, although much smaller. This similarity may have been overstated, since the proportions of individual parts of the endocast surface and the details of its surface have not been reported fully. But considering the extensive variability of normal modern human endocasts, we should expect a range of variability in fossil endocasts as well. Thus, LB1 may well lie within the range of shapes of several early hominid species, even though it does not match those populations in size.

So as long as LB1 is plausibly within the shape range of early hominids, even a perfect match with some microcephalic cranium is unlikely to convince some people that it is pathological. No smoking gun there.

At the same time, I don't think we should expect to find an identical microcephalic anywhere. Weber et al. (2005) write that microcephalic crania are "extremely heterogenous", and this would be my assessment also from the few I have seen. Microcephaly is a developmental pathology with many causes, many different manifestations, and consequently many different morphological possibilities. To me, this means that no two microcephalics may be alike, and we may never find an exact match for any ancient microcephalic skull.

For this reason, the idea that we should expect to find an exact match is far too stringent a requirement. After pointing out the possible discrepancy of orientation, and noting that none of the endocasts studied by Weber et al. had exactly the pattern of area 10 convolutions as LB1, Falk et al. (2005b) end their comment:

If this is the best evidence that can be produced from a sample of 19 microcephalics, we suggest that the authors reconsider their position on the microcephalic hypothesis regarding Homo floresiensis.

I have the opposite opinion. I might well expect to study several hundred microcephalic crania and never find an exact match for LB1. Even so, if this sample, like the 19 studied by Weber et al. (2005), has over 30 percent of endocasts with enlargement in Brodmann's area 10, and if it has some endocasts that are relatively broad with low transverse sinuses, it would be hard to say that LB1 lies outside their range of variability.

One may reasonably wonder where the burden of evidence lies in this case, particularly now that we have additional evidence for small-bodied individuals in the Liang Bua sample. Some may presume that the hypothesis of pathology for the endocast must be accompanied by a specific diagnosis that matches the details of the entire skeleton. After all, microcephaly is a very rare condition, and it must be the rare case indeed that survives in a Late Pleistocene population and exhibits morphology that is broadly similar to that of early hominids.

But the hypothesis that the specimen is a normal member of a dwarf population of H. erectus, H. sapiens, or Australopithecus brings its own strong problems that we should not gloss over.

  • If it derives from Homo sapiens, what pattern of selection caused its brain to shrink? What accounts for its limb proportions and strange long bone, pelvic, and mandibular morphology?
  • If it derives from Australopithecus, how did it get there? How did it survive so long? How did it get human-like teeth?
  • If it derives from early Homo, none of these questions have good answers.

So pathology solves a lot of problems with the sample, even if it creates others. I don't know that we can weigh these options relative to each other: the likelihood of microcephaly versus the likelihood of a "Lost World" australopithecine scenario, for example. But if we are going to act like Sherlock Holmes, we will have to start eliminating options. And Weber et al. (2005) do a good job of showing that we can't yet eliminate microcephaly.

UPDATE (10/14/05): A lot of readers have come away with the impression that I am strongly advocating the microcephaly hypothesis. What I have tried to do over the course of several posts is not to advocate for microcephaly, but instead to explain why the microcephaly explanation may not yet be rejected.

If I am an advocate for anything, it is for a full airing of these issues, and a more public presentation of original data. In the Reuters article on the new bones, Robert Martin says:

"This paper doesn't clinch it. I feel strongly that people are glossing over the problems with this interpretation."

A lot of people feel that way. There is nothing to be done but more research, more comparisons, and more access to results.

References:

Falk D et al. 2005a. The brain of LB1, Homo floresiensis. Science 308:242-245. Full text (free)

Falk D et al. 2005b. Response to comment on "The brain of LB1, Homo floresiensis". Science 310:236. Full text (subscription)

Weber J, Czarnetski A, Pusch CM. 2005. Comment on "The brain of LB1, Homo floresiensis". Science 310:236. Full text (subscription)