james_davis_nicoll | Two questions about Myrmecophaga tridactyla

Sigh. I say the sample size is probably low because I check the references. Take a look at that table. It gives its sources on brain size as Eisenberg and Wilson on bats (nope), Eisenberg and Wilson on possums (nope), Mace on small mammals (nope), and by the process of elimination, Eisenberg in The Mammalian Radiations.

Except it's a trick. I happen to know those data aren't actually in The Mammalian Radiations. It's got some awesome tables, but for brain size, Eisenberg only refers to where he found the data sets.

Eisenberg gives four general references for Mammalia, plus some others to fill in phylogenetic holes so he can weight his regressions more "authentically". There's Crile and Quiring's 1940 paper, "A record of the body weight and certain organ and gland weights of 3690 animals," where the modal sample size is one (1) per species. There's Von Bonin's 1937 compilation, "Brain-weight and body-weight of mammals". Von Bonin usually got his specimens from zoos: small sample sizes. (He nearly freaked out when a TB epidemic hit the baboons at the Paris Zoo right before WWII, and he got to work with what he considered an extremely large sample size of baboon brains: n = 42.)

Eisenberg's third reference is to Jerison (1973), which McNab and Eisenberg also refer to originally. For direct brain measurements, Jerison mainly uses C&Q, VB, and Count's 1947 compilation, "Brain and body weight in man: their antecedents in growth and evolution," where again, the modal sample size is one (1). Count seems to be the source for the capybara brain and body mass numbers. n = 1 for the capybara in Count.

I haven't read Eisenberg's fourth source, Wirz (1950), "Studien über die Cerebralisation: zur quantitativen Bestimmung der Rangordnung bei Säugetieren," because my German is just not up to snuff, but given the difficulties of conducting research in immediate postwar Germany, I would be astonished if sample sizes weren't also small.

(Maybe you can tell this is not the first time I've investigated this topic. I would think saying "I happen to have Marshall McLuhan right here," would be a giveaway.)

Anyway. Given that now you know how the data were collected, how heavy an evidential weight would you put on those specific numbers?

I bow to your more extensive library; The Mammalian Radiations is something I don't have and couldn't check, so I will take your word for it.

If it is indeed on a sample of one, then yes, there is an issue; although having said that, even on a sample of one I would point out that the point is ratio of brain to body size, and that this proportion tends to remain the same for different absolute values of body mass within a species. Even given that body mass can very widely for both capybaras and hyenas, one would expect the ratio of brain to body mass to remain the same ratio; so a comparison of different species in this way can still have value. Yes, of course with a sample size n=1 there is a possibility that you have an aberration rather than a typical specimen of any given species; how much of an issue is this? I don't know.

Given the amount of work that has been done with brain evolution and brain:body mass proportions in mammals, there are surely other sources. I'm going to go out on a limb here and make a prediction: that these sources also support the idea that brains tend to be more massive in predators than in prey, and in challenging rather than relatively peaceful environments with abundant food, and that hyenas have significantly bigger brains relative to body size and more functional intelligence than capybaras. Do you have anything in your library which would refute that, or if you want to be fair about it, support it?

Um. Allometry is used to determine what scaling (power) law, if any, is most appropriate to relate biological variables. A simple ratio is nice, but it implies a scaling law of one-to-one. And a single point is less than helpful to determine a scaling law, since a scaling law has two parameters.

In mass comparisons like McNab and Eisenberg, one hopes that the variations in data collection cancel each other out in the interspecific regression. (Eisenberg recognizes this difficulty in his early calculations of EQ, making comments like "? obese individual" or "seasonal variation" in his data sets.) But for that reason, you can't compare two species data points and conclude a "just-so" story from them. It's data exploration at best, and data dredging at worst.

There's an additional difficulty in comparing allometric variables across lineages. Many evolutionary changes are developmental changes, which directly affect scaling laws. Thus, different lineages will tend to have somewhat different scaling laws for body variables. A small change in exponent can have large effects, especially at higher body masses.

However, lineage is also strongly correlated with lifestyle. Upshot: one can't make the simple claim that a predatory member of Carnivora will have a larger brain at the same body mass as a herbivore of Rodentia because it's a predator. It might have a larger brain because it's a member of Carnivora.

I haven't seen anything that tries to break down these effects across groups, although I know studies within groups (from order to genus level) support your idea. It's probably time for me to review the literature again.

Finally, I trust I don't have to go into why brain size is only a loose correlate for functional intelligence, as I fight off the elephant/orca invasion of Brooklyn. Hopefully, as techniques for non-invasive neuroanatomy become more common, we'll move past the "weigh on a scale" approach and get to the nitty-gritty. Jerison used skull castings to estimate cortex proportions, which to my mind is a more interesting variable than brain weight, but that method has fallen out of favor.

Anyway. I don't doubt there's a lot more data for body and brain size out there for hyenas and capybaras and maybe even anteaters. I'd prefer more behavioral studies myself, but that's a personal taste.

Ok, I will absolutely not, under any circumstances, argue that it isn't more complicted than what we've been squabbling over. No, you are absolutely correct that it isn't necessarily a simple ratio. Yes, I was deliberately arguing it on a simple level, though. But I have to admit, your arguments about scaling laws: I'd be stupid to try to dispute all that. You make good points. Although there is something nagging at the back of my mind -- I'll have to let it stew in its own juices for a bit before the thought gels, though.

I think, in the first instance, I was simply having a knee-jerk reaction to the notion that the hyena and capybara are pretty much on par with each other-- they aren't, and I will stick by that.

Of course, when you consider relative neocortex size, cortical folding (which I personally have an interest in, although I haven't had the opportunity to explore that interest to any detailed degree) and all the other fun modular aspects of brain evolution, the arguments can get complicated indeed. (Any idea how much cortical folding a capybara has? Frankly, if they have much more than their cousin the cavy I would be vastly surprised.) But anyway, when you say "brain size is only a loose correlate for functional intelligence", yeah, no argument. It's a base measure. It's a starting point. There are obvious points where the general correlation breaks down (for instance, I would suggest that marsupials fall off the small end because so much of their gestation ends up being ex utero, given that most placental mammal brain growth takes place in utero; on the other end, I don't have the reference to hand but I believe there is a "minimum mammal brain size" so that very tiny mammals, like the smallest species of mice, have a lot more brain for their mass than you would expect). ...But it is a rough measure, against which you would expect relatively bigger-brained critters to generally be functionally more intelligent than smaller brained ones. (Enough caveats?)

(Of course, once you move outside mammals you do run into MAJOR problems with that; look at all the recent papers on the cognitive abilities of corvids! I'm just talking mammals, above.)

...lineage is also strongly correlated with lifestyle. Upshot: one can't make the simple claim that a predatory member of Carnivora will have a larger brain at the same body mass as a herbivore of Rodentia because it's a predator. It might have a larger brain because it's a member of Carnivora.

I don't know if you were paying any attention to my argument with Doug M., below, but that was precisely my basis for asserting that "social complexity" is a valid driver of brain size even though it is not a perfect correlate with brain size in orang-utans -- given that orangs are primates (and arboreal, come to that), and primates hit the top end of the scale in general. And that a relatively nonsocial species with a big brain does not invalidate the hypothesis of social complexity as a driver for brain size, given that there are strong arguments that social living is an ancestral primate trait which was partially lost in orangs, but this didn't automatically correlate to a loss in inherited brain size. (Not to mention all the confounding factors: arboreal vs. terrestrial, diet, etc.)

Anyway, I think we have a better idea of where we each stand, anyway. I don't think I disagree with you that much. ;-)

Edited 2009-05-28 19:05 (UTC)

Actually, I think mainly we use the word "comparable" differently. :-)

I think allometry shows some very useful empirical relationships. But there are a bunch of reasons I wouldn't put strong weight on allometric hypotheses that aren't also strongly grounded in some other aspect of biology. Some are mathematical, some are methodological. The arboreal correlation strikes me as something that might be better explained as a lineage effect, for instance. When you look at the earlier literature, it's not far removed from playing with log-log graph paper.

As for cortexes, the capybara is gyrencephalic, just like the anteater and the hyena (but unlike the guinea pig or even the porcupine, which are smooth-brained). I don't have data on the degree though.

I'd love to see more behavioral studies on all three species. I've never read a modern (i.e. post-Skinnerian) study that showed an animal was duller than researchers thought, though of course there's publication bias.

I'm going to have to look up details of capybara brains now.

But...why are you so eager to discount the arboreal environment as a driver of brain size in its own right? Think about it; although a terrestrial animal has to be aware of what is going on in three dimensions to some degree, it is largely moving in two, with activities like jumping or climbing being only occasional -- and it is negotiating what is for the most part a single, continuous surface.

Arboreal animals, on the other hand, are fully mobile in three dimensions, and rather than negotiating a single large surface they are negotiating a series of narrowly restricted horizontal and vertical surfaces, many of which interact in highly complex ways and are mobile in their own right. And a misstep 40 feet above the ground is frequently going to have more dire consequences than a misstep at ground level --Gravity Is Not Always Your Friend. I would have thought that arboreal environments are fundamentally more challenging to any animal which is attempting to move at speed (so leaving aside sloths, then) in terms of perception, physical coordination and necessary speed of reaction. Why would that not be a spur for a more developed set of sensory and motor cortices?

Not eager per se, but a) the explanation, while plausible sounding, is very hand-wavey, and b) the literature already reports them to be strongly clustered in phylogenetic groups.

E.g., from McNab and Eisenberg, "Many of the species with the largest brains relative to the mass standard are arboreal, most notably xenarthrans, carnivores, primates, and tree squirrels."

The primates and the carnivores are already known to have their own allometric series which might be confounding the arboreal effect. Tree squirrels, I don't know. But arboreal xenarthrans -- they're sloths! Like you said, it's really hard to argue that they have a larger brain than predicted because of their enhanced three-dimensional mobility.

So you have two orders where developmental allometric differences might be confounding matters, one order where the standard explanation makes little sense, and tree squirrels.

(Eisenberg did look within smaller phylogenetic groupings for evidence of the arboreal-brain size connection -- it's in his possum paper referred to in McNab and Eisenberg -- and while he found a relationship, it was hard to disentangle from other factors.)

So it's an Occam's razor sort of thing.

But you have to start somewhere, and some hand-wavy just-so stories make a more logical starting hypothesis than others.

Here's a question for you: were tree sloths always so slow?

Oh, as for "I've never read a modern (i.e. post-Skinnerian) study that showed an animal was duller than researchers thought" -- might that not be because up until very recently, let's face it, the baseline was low. The debate over whether or not non-human animals could even truly feel pain wasn't that long ago.

Wasn't there some debate over whether infants felt pain?

God help us, yes.

Doug M.

In the department of weird-ass coincidences, the paper "Brain-size evolution and sociality in Carnivora," Finarelli and Flynn (2009), was put up on PNAS exactly while we were having our discussion. It basically does cross-regressions within Carnivora, using brain volume instead of mass in order to include fossil members.

This'll take a little time to digest. But, as they put it, "Valid statistical analysis of comparative data in biological systems always requires information on the phylogenetic relationships of the organisms being analyzed to account for the statistical nonindependence of observed character values for closely related taxa."

Off to read that one!

Two questions about Myrmecophaga tridactyla

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject