The Mismeasure of Dog

When you see a study that claims to predict behavior variation with cranial variation your anthropology radar might start pinging. If your anthropology radar's especially sensitive, it will ping regardless of the organism. And when that organism is the domestic dog, you might be tempted to ready the torpedoes before you've even read the abstract.

That's because, for one, people often like to compare human races to dog breeds. Conversations rooted in eugenics or racism often involve judgments of blood purity, much as American Kennel Club members and Westminster Dog Show contestants owe their worthiness to their pure bred ancestors. This isn't a perspective that many anthropologists abide.

And, secondly, there's a long history of measuring human heads, divvying them up by race, showing that they're distinct by race, and then explaining differences in behavior at the race-level in the general or specific context of those race-level differences in the head. Another practice that few anthropologists abide.

We're going to cover some ground on human heads before getting to the dogs'. So please... Sit... Stay... And if you do, we'll get to the dogs in just a second.

Brain size, or cranial capacity, has long been a favorite measure by folks interested in human variation. The cephalic index (CI) has too. These have been used in the name of science by racists, racialists, and folks who are neither. CI might be a favorite because not only does it appear to separate race categories and even populations within them, but it's easily, cheaply, and noninvasively obtained from live humans, while it also avoids phrenological subjectivity.

How phrenology-free is CI? It's not completely clear because, to my knowledge, no one has linked CI to behavior any better than they can with a lumpy left parietal. Remember, however, that many CI uses have not, and are not, for explaining behavior. CI's often measured to study human variation between or within populations, and maybe relatedness, and maybe change over time (evolution).

In anthropology, CI involves the ratio of the breadth to the length of the cranium.

Superior view of a human cranium. Red lines show the two parts of the cephalic index (CI = breadth/length x 100). (photo source)

People even created terms for different CIs: wide, broad skulls are brachycephalic and narrow, long ones are dolicocephalic. CI is influenced greatly by genes (accumulated in ancestry through any number of evolutionary mechanisms). The trait is complex and polygenic, sharing genes that affect other traits. For example, stature is linked to skull length: the taller a person the more dolicocephalic they might be. Environment also contributes to a person's CI which might explain apparent changes at the population-level seen between parents and offspring:

Depicting Franz Boas's CI data collected from 13,000 European-born immigrants and their American-born children in the New York area. (Figure 10.12 from Human Variation: Races, Types, and Ethnic Groups, 6th ed. by Stephen Molnar)

Notice how the changes in the American born children weren't all the same sort of change. That's probably one of the reasons that Corey Sparks and Richard Jantz wrote, "A reassessment of human cranial plasticity: Boas revisited." Boas's big classic dataset had long been used to demonstrate the strength of environmental effects on cranial variation. But with Sparks and Jantz's new consideration of the data, it looks like foreign born and American born people with similar ancestry have more similar head shapes to one another than foreign born compared to American born, regardless of ancestry.

Figure 1 from Sparks and Jantz (2002)

Sparks and Jantz make this point by extending the comparison:

"In America, both Blacks and Whites have experienced significant change in cranial morphology over the past 150 years but have not converged to a common morphology as might be expected if  environmental plasticity plays a major role (29).

We send out our genes wherever our genes may go and that includes, apparently, how our offsprings' adult heads will be shaped, roughly, regardless of where their heads live.

In conclusion, Sparks and Jantz discuss the context a hundred years ago:

"We also must consider the attitude of Boas toward the scientific racism of the day. Evidence of Boas’ disdain for the often typological and racist ideas in anthropology have been reviewed previously (45) and are evident also in his later publications (46–48). Boas’ motives for the immigrant study could have been entwined in his view that the racist and typological nature of early anthropology should end, and his argument for dramatic changes in head form would provide evidence sufficient to cull the typological thinking. We make no claim that Boas made deceptive or ill-contrived conclusions. In
Fig. 1 it is evident that there are differences between American and European-born samples. What we do claim is that when his data are subjected to a modern analysis, they do not support his statements about environmental influence on cranial form."

Acknowledging the strong possibility that genes are more powerful determinants of head shape than environment isn't offensive to many of us today, even those of us who are sensitive about these sorts of data because of their past uses or their potential for future abuses.

But maybe the environment affects different head shape genotypes differently? Why assume that similar environment will result in similar phenotypes if the underlying genotypes are different to start? Isn't the change in phenotype enough to show environment matters enough to consider it important? Perhaps Boas's data aren't sufficient for answering these questions.

And maybe humans just don't have very varied head shapes in the grand scheme of things, in the big picture of whatever "variation" is, and our perception of such typological differences is just what we're good at doing with things like human heads.

But humans and our variation and how we explain and perceive it aren't the main reason we're here today! 

I've introduced human CI, human cranial shape variation in general, and the history of typological race studies linking human head shape and human behavior, because...


One of the main findings of a paper recently published in PLOS ONE is that CI predicts some dog behavior:

link to article

[Now do you see what I meant by my opening paragraph? If not, there's a whole literature out there. Brace's Race is a Four-Letter Word is where you might start.]

For brevity, we're going to ignore the height and weight stuff in this paper and stick to the skull shape parts. But I recommend reading the paper for yourself.

The authors cite some papers to establish that skull shape might be linked to behavior.

One "noted that the morphology of working dogs' heads clustered according to their breed's original purpose. This observation was later supported by a series of studies focused on cephalic index."

"CI is correlated with a tendency for retinal ganglion cells to be concentrated in a form of an area centralis rather than a visual streak. This feature of short-skulled dogs means that they have more visual acuity in the centre of their visual field but less in the periphery."

And this is hypothesized to be linked to their being,

"more likely to follow a human pointing gesture, suggesting that the arrangement of retinal cell may link with aspects of canine social cognition."

Also, they cite a study of dog brain MRIs, linking skull shapes (particularly CI) to,

"progressive pitching of the brains, as well as with a downwards shift in the position of the olfactory lobe." 

This established their premise that differences in head shape reflect differences in brain organization and that,

"CI may be associated with changes in the way dogs perceive stimuli and possibly process information."

It's hard to write this through my insane jealousy, but they traveled to dog shows to collect their metric data. Photographs of the superior view of dogs' heads were taken and then used to measure CI.

They stuck to pure bred animals. Then they looked at a dog behavior survey called C-BARQ to see if CI predicted anything in the same breeds there. The survey's owner-reported, but of course, the author's explicitly acknowledge this source of bias that they can't do anything about (except not use it for science).

And lo and behold, CI predicts a few things:

Self-grooming, chasing, dog-directed aggression, allo-grooming, stranger-directed fear, persistent barking, compulsive staring, stealing food. 

All aren't exactly the kinds of behaviors that have been used to describe human races and populations, but some are.

Anyway, that's not what I really want to talk about. I want to talk about CI and its use for predicting behavior.

Here's how they measured it in the dogs:

Dog CI: "The length was measured from the fingertip to the tip of the nose, and the width was measured from each zygomatic arch, which was displayed by the tape placed around the widest part of the dog's head." (Figure 1 from McGreevy et al., 2014 with my blue lines added)

That's fine as a measure of something like the head. What're you gonna do otherwise at a dog show?

But how's that getting at brain shape? Remember, if the behavior's at all going to be about inbred biology as the premise of the paper sets us up to ask...

That is, if the behavior's not due to conditioning based on how the animal looks to humans and other environmental factors influencing behavioral development which are rightly acknowledged in the paper...

And if changes in behavior during life don't matter (which I don't believe are acknowledged in the paper but is obvious to anyone who's lived with a dog)...

Then we've got to be getting somehow at genes or brain shape or something with CI in order to use it to predict behavior.

But look at what their CI measures:

Boxer cranium with anthropology's CI (left; compare to human at top of post) and with McGreevy et al.'s CI (right). photo source 

Anthropology's CI gets at brain shape much better than McGreevy et al.'s. In fact, their measure isn't getting at brain shape any more than it's getting at meat helmet thickness (all that space between the zygomatic arches and the braincase) and snout length.

With this metric, the paper could be entirely rewritten to discuss not brain shape but meat helmet size as a predictor of behaviors instead.

There are many ways to take this discussion beyond merely pointing out that this measure of brain shape isn't. And I'll wrap-up with a few of those here.

First of all. I'm not an expert on the dog literature but it did stick out that they didn't cite a nice paper in American Naturalist from just a couple years ago--"Large-Scale Diversification of Skull Shape in Domestic Dogs: Disparity and Modularity" by Drake and  Klingenberg--that lends some support to their approach to dog head variation, by breed, and whether it's linked to behavioral variation by breed:

"The amount of shape variation among domestic dogs far exceeds that in wild species, and it is comparable to the disparity throughout the Carnivora. Moreover, domestic dogs occupy a range of novel shapes outside the domain of wild carnivorans."

Look at all the morphospace that dogs' heads are taking up! (from Drake and Klingenberg, 2010)

But Drake and Klingenberg's study also discovers and raises some really important issues that were not considered by McGreevy et al in PLOS ONE:

"The disparity among companion dogs substantially exceeds that of other classes of breeds, suggesting that relaxed functional demands facilitated diversification. Much of the diversity of dog skull shapes stems from variation between short and elongate skulls and from modularity of the face versus that of the neurocranium. These patterns of integration and modularity apply to variation among individuals and breeds, but they also apply to fluctuating asymmetry, indicating they have a shared developmental basis. These patterns of variation are also found for the wolf and across the Carnivora, suggesting that they existed before the domestication of dogs and are not a result of selective breeding."

So we're left wondering why this paper didn't even attempt to deal with genetics and development, that is, the phylogeny of dog breeds. If head shapes vary according to phylogeny, but within those clusters we see variation in behavior, then there's not such a strong link.

I think that phylogeny's ignored because it's not yet worked out. I just went back into the paper and found they do at least bring it up: "A cluster-based analysis of full genomes of these different breeeds may prove helpful in this domain."

It doesn't look like we're very far in dog breed phylogenetics according to (one of my oft go-to pieces for teaching examples of "simple" genetics) Parker, Shearin and Ostrander's review of dog genetics.

"An unsupervised cluster analysis of dogs and wolves. Using clustering algorithms with more than 43,000 single-nucleotide polymorphisms (SNPs), 85 dogs, representing 85 different breeds, along with 43 wolves from Europe and Asia, were assigned to 2–5 populations (inner circle to outer circle, respectively) based solely on genomic content. Each column represents a single individual divided into colors representing genomic populations. Blue indicates a wolf-specific signature, and red indicates a dog-specific signature. Note that the majority of crossover lies between ancient dog breeds and Chinese or Middle Eastern wolves. Figure originally published in Nature (158)." (Figure 1 from Parker et al., 2010)

That's not resolved well enough to be useful for McGreevy et al. (But I could be behind the times.). And granted, the figure I pasted there is aimed at seeing which dogs are most primitive or most like wolves and, maybe then, most like the earliest dogs. But domestic dog breed relatedness must be better worked out than this, given the things Wisdom Panel (using Ostrander's dog breed markers!) can do to pinpoint the breed ancestry of mutts like my pals Elroy and Murphy. (And here's where I learned from the horse's mouth what this test does.)

It does make one wonder, though, whether the dogs that have more wolf-like heads behave more like wolves. That kind of evidence would go a long way to support interest in linking domestic breed heads to their non-wolf-like behavior. But it's not in McGreevy's paper as far as I can tell.

But again, if you've got sensitive anthropoogy radar, it's pinging. First of all, isn't asking about which dogs are most like wolves the same as asking which humans are the most like chimps? Uh, no. But, yeah, kind of. And I'm in no state of mind to expound on that one here today given the trolled-up comment thread that would surely ensue.

But, second, it's troublesome that papers with easy measures and hypothetical (at best) or false (at worst) correlations between anatomy and behavior are published so far in advance of the best sort of data to answer their questions (e.g. a fully resolved phylogeny, let alone mapped genes for the brain shape or the genes for the behavior, etc etc...).

A deterministic and scientifically impatient (to be kind) approach to variation is what still haunts us about the history of physical anthropology. And to see it potentially playing out in other organisms like dogs need not raise any of our politically correct hackles, but I hope it arouses our scientifically critical ones.