Thursday, December 06, 2012

Cavemen, quadrupeds and science, oh my!

So... there's a new paper in PLoS ONE about how 'cavemen' depicted four-legged animals better than 'modern' artists (Horvath et al. 2012). I usually try to refrain from paper bashing here, but there is such a high density of wrong (if not downright fail) in this one, that it's hard not to. Becky Farbstein agrees, and points out that:

1) anyone using the word 'cavemen' with a straight face in a scientific publication today cannot be taken to know anything about the time period in question;

2) the paper's conclusions are only surprising or noteworthy if you assume that cavemen (and by extension hunter-gatherers) are somehow less advanced at a fundamental level than 'modern' folks (again, whatever that is);

3) and - extremely importantly - that there is no reason to expect that an artist, archaic or modern, necessarily operates with the goal of depicting quadrupeds realistically; to assume that this is the case fundamentally misinterprets what art can be and usually is, i.e., not strictly about representing reality.

Basically, as a friend said on Facebook, "With their methodology, you could argue that gravity was invented in the Renaissance because figures in 15th-century paintings begin to be grounded, rather than seeming to float as in Medieval art!" The same friend also added that that's why everyone needs to study at least some art history, but that's another issue (and for those readers wondering, I have [way] more than one snarky friend on FB).

I agree with all of this. However, another extremely problematic aspect that I haven't yet seen discussed concerns the data the authors use to make their case. Even if you disregard the points above, issues with the composition of their 'caveman art' sample data alone should be enough to laugh these people out of town. I was immediately suspicious when I saw the elephant image that accompanies the press release on the paper - an elephant is not usually what one associates with 'cavemen'. So I dug into their data, in the off chance they simply had an extensive sample of prehistoric art. After all, they boast about their data base being 1000 images strong. As it turns out, only 35* of those are 'prehistoric'. That's less than 4% of their total sample (I wonder how often they could bootstrap a similar trend from their modern sample just by randomly selecting 35 representations at a time...). But surely, these images all come from the same site or a handful of sites that date to the same period, right? Sadly, no: we're looking at 11 Paleolithic drawings (9 from Lascaux, one from Niaux, and another from Altamira) that span several thousand years and two countries. At least with those, a tenuous link to the idea of cavemen paintings could be made. But the 24 other representations come from sites scattered across Libya (consistently misspelled as 'Libia' in the paper, for crying out loud!), Morocco, various parts of India, and South Africa, with little if any chronological control. The point is that these pictures don't actually form a coherent body of evidence by any stretch of the imagination. Oh, and these conflate paintings, incisions, and engravings, all of which impose their own specific constraints on how art is produced. Add to that there is no reason to think that many of these representations were necessarily meant to be seen only in two dimensions or outside of the panels on which they were created (and in relation to the other designs these panels comprise), and you've got some of the shoddiest sample selection I have ever seen in a scientific paper.

I can't really figure out what their 'modern' sample is, but the issues with the prehistoric sample alone are enough to damn the paper. I don't care if they think that they can simply disregard context to focus on how walking was represented - the fact is you can't dissociate art from its context. Furthermore, with such a tiny and far-flung (temporally and spatially) sample of 'caveman' art, we can basically have no confidence at all that we're actually looking at how 'prehistoric' people represented animals, let alone their gait.

*Edit (Dec. 6, 2012; 11:43PM: Table 1 in the paper states a prehistoric sample of 39 'prehistoric' representations, whereas the supplementary information only details 35. Given this, I went with 35 being the most reliable value, since it's impossible to figure out what the other four representations were (of and from). Ultimately, however, using one or the other figure does not affect my point here - it's still less than 4% of the total sample, and a very small number of representation given the staggering geographic and temporal scales over which they are spread.


Horvath G, Farkas E, Boncz I, Blaho M, Kriska G (2012) Cavemen Were Better at Depicting Quadruped Walking than Modern Artists: Erroneous Walking Illustrations in the Fine Arts from Prehistory to Today. PLoS ONE 7(12): e49786. doi:10.1371/journal.pone.0049786 (Link)


terryt said...

"an elephant is not what usually what one associates with 'cavemen'".

You don't think that elephants might have lived in caves in those days? Perhaps the people just got it to pose in the cave while they made a painting.

Julien Riel-Salvatore said...

Terry -
I'd say there was less than a 35/1000 chance of that ;)


terryt said...

I used to paint wildlife quite a lot when I was younger, and reasonably well if I do say so myself. It has always amazed me how well those 'cavemen' drew the animals deep in the caves. Presumably they did so from memory, something that I would have found difficult to do well at all. My guess is that they probably practiced on perishable material for some time before they managed to do such magnificent drawings so far underground. The paper presumably does not take this ability into account when comparing artistic ability across the ages either.

T. said...

Your post reminds me to this old info on possible paleolithic face drawings, does one know by now more on that?:

Julien Riel-Salvatore said...

Hi T.,
As concerns La Marche, I am mostly familiar with the controversy surrounding the engraved limestone slabs at the site. That said, if these engravings are authentic, then it certainly implies that a high level of realism could be represented during the Paleolithic. To answer your question, though, to my knowledge Rappenglueck has never published his results on La Marche, although I recall a recent paper (in French) that used microtomogrpahy to look at some of these engraved slabs.

Turning back to the overlap b/w the news story you linked to and this new paper, I just want to point out that my comment wasn't about saying that Paleolithic artists couldn't accurately represent the natural world. For what it's worth, I think that they likely were highly skilled artist. My point is simply that the paper by Horvath et al. is so fraught with conceptual and data problems that it cannot be used to address this question at all.


T. said...

Dear Julien,

thanks! Can you give me the bibl. infos (or, even better, a link) to that french artcle on La Marche?