Hobbit cretin FAQ

It's all over the news this week: Australian researchers Peter Obendorf, Charles Oxnard, and Ben Kefford claim that the Homo floresiensis skeleton LB 1 belonged to an individiual suffering from congenital hypothyroidism, or cretinism. The disorder has a number of developmental and skeletal effects, including short stature, and they run through a list of LB 1's characters that appear to match the disorder.

Needless to say, others disagree. So here is my take:

Is LB 1 a cretin?

No.

The authors of this research provide a list of characters of the fossil that are consistent with the diagnosis of cretinism. Some of them are interesting, in that they have not previously been connected to any pathology. But the most important observation is simply wrong. With that falls the hypothesis.

Is the pituitary fossa of LB 1 large?

The diagnosis of cretinism would be most strongly supported by the authors' claim that LB 1 has a large pituitary fossa.

If you haven't learned this lesson yet, take note: If Ralph Holloway and Dean Falk agree on an anatomical observation, then it is correct.

The two have been interviewed in a number of the press articles. My favorite comes from Kate Wong, where both Holloway and Falk give long statements. These are really too long for me to block quote, so go read them. But I'll take one sentence from each:

Falk:

Now that we've reexamimined the CT images, we can tell
you that there is absolutely no way that the length of the pituitary fossa could be 12.9 mm.

Holloway:

...the pituitary fossa on my endocast is, to my mind, tiny, and I don't get much more than about 6 mm in dimensions, either [anterior-posteriorly] or in breadth, so I don't understand where they have data to make such a claim.

That's the end of this story.

How did the paper get that so wrong? I mean, didn't they just measure it?

They took the images of CT scans presented in the supplementary data to Falk and colleagues' 2005 paper on the LB1 endocast, blew them up, and attempted to measure the length of the pituitary that way. To understand where they went wrong, I did the same thing. Here's the picture, blown up:

LB1 endocast, basal view

Supplementary Figure 2e from Falk et al. 2005, focusing on LB1

An estimate of 12.9 mm is wrong on many levels. For one thing, how many significant digits do you think you could get out of that figure? The blown-up version is clearly very pixelated. That by itself might not be so bad -- after all, a medical CT begins with limited resolution anyway -- but in this case there is no clear way to identify the borders of the pituitary fossa. We might well do better with the endocast itself, or with the ability to rotate and relight the CT image, because we could explore the contours more thoroughly. Here all we have is a computerized rendering of the surface in which our recognition of the detail depends entirely on the simulated lighting.

This is where they went wrong. No replicable estimate is possible from that rendering, but they went with one anyway.

But everyone knows that you can never do any real research using photographs. You must examine the original specimens!

In my opinion, that would be exactly the wrong conclusion to draw from this case. The problem was not that they attempted a new measurement on a photograph; it was that this rendering is not a photo, and does not provide sufficient information for such a measurement.

In some cases, a well-resolved photograph can give better basis for a measurement or comparison. In almost all cases, a research article accompanied with original photos will allow experts to assess the accuracy of claims and replicate the observations. Making work as easily replicable as possible should be the goal of every good scientist.

It doesn't help to read things like this, from the Rex Dalton article in Nature:

[Peter] Brown is critical of the cretin theory. "I am the only person on the planet to have seen what's left of the pituitary fossa," he declares. "It is very poorly preserved and not capable of meaningful measurement."

It may be true, but that doesn't make it science. If nobody can see it, then nobody can replicate it. Which means we have no reason to believe it.

Happily, in this case independent experts have access to the scans and can tell us what they look like. On the other hand, if I were sitting on scans like these, with people publishing critical articles every couple of months, you can bet I would put them on an FTP site and let everybody have them. Three quarters of the problems would immediately vanish, because people could refute their own hypotheses before they went anywhere, and reviewers could work from the best information also.

The remaining quarter of the problems would at least be interesting!

Oh, now that can't be the most boorish of the comments in the press. I mean, after all, this is the hobbit!

Well, let's see....from The Australian:

The [cretinism] notion's been ignored for good reason, said evolutionary anatomist and paleoanthropologist William Jungers of Stony Brook University in New York state.
"The cretin and hobbit (body types) exhibit virtually no similarities except for short stature. That is, they're both short. End of story," said Professor Jungers who has studied the hobbit remains first-hand.
"The only merit to this paper is their correct dismissal of a competing 'pathology du jour' called Laron Syndrome (which causes skeletal deformities). The rest is a rather large and stinky pile of misinformation and wild speculation," he claimed.

Or, this one, from the same article:

A final kick came from biological anthropologist Colin Groves of the Canberra's Australian National University: "I recall spending an hour or so in the pub with Peter Obendorf about three years ago when he confided to me about this latest bee in his bonnet."
"As fast as he produced supposed similarities I put stumbling blocks in his way. I warned him that he would simply be laughed to scorn if he produce what is mainly idle speculation," Professor Groves claimed.

Ha, ha, ha! Here's Groves again, in The Guardian:

"I regret to say that this paper cannot be regarded as a contribution to our understanding of the Flores hominin," said Prof Colin Groves, a bioanthropologist at the Australian National University, Canberra. "Many of the claims lack evidence (ie they are sheer speculation), some even fly in the face of the evidence. I am very sorry indeed to see serious scientists involved in such a travesty."

Remind me never to sit in a pub with Colin Groves!

Then there's this piece of legal news:

"The Tolkien trustees do not file lawsuits lightly, and have tried unsuccessfully to resolve their claims out of court," Steven Maier, an attorney for the Tolkien estate based in Britain, said in a statement. "New Line has not paid the plaintiffs even one penny of its contractual share of gross receipts despite the billions of dollars of gross revenue generated by these wildly successful motion pictures."

Ooops....wrong hobbit....

Is there anything salvageable out of this?

Well, they wrap up a number of the skeletal features of LB 1 into their cretinism hypothesis. My personal opinion is that you have to start by explaining the brain size, and cretinism doesn't. None of their comparative sample of European cretin specimens has a brain size smaller than 1000 g. The paper claims that these would "scale with height" down to 700 g, but that is just a projection outside the data's range based on the regression.

Without an explanation for the brain size, and without any unique character to confirm the cretinism hypothesis (like the pituitary size), they have to resort to a list of the known problematic characters of LB 1, such as the Tomes root, the humeral torsion, the relatively large foot and broad diaphyseal breadths. But these match other pathological explanations as well -- most of them were cited in the Laron syndrome paper, for example.

To the extent that these features are "developmental abnormalities," they may be explained by any number of conditions. It's possible that they may have resulted from a unique evolutionary history, either as side effects of other adaptive changes in a small population, or as fixed deleterious variants.

The multivariate analysis of the skull measurements is not convincing. Of course, it is extraordinarily rare for me to find a multivariate comparison convincing about anything.

References:

Dalton R. 2008. Hobbit was 'a cretin.' Nature 452:12. doi:10.1038/news.2008.643

Culotta E. 2008. Were the Flores hobbits really cretins? ScienceNOW March 5, 2008. Full text

Obendorf PJ, Oxnard CE, Kefford BJ. 2008. Are the small human-like fossils found on Flores human endemic cretins? Proc Roy Soc Lond B (early) doi:10.1098/rspb.2007.1488a>