What a Mess
A paper that purports to show that more social cetaceans are smarter actually proves the opposite, and my first analysis of it contained two important errors. My first mistake was due to difficulties with reading files on supporting data. I misidentified the points representing the bowhead whale as the sperm whale. Bowheads are marked as being one of the longest lived whales, having twice the known lifespan of any other cetacean, at over 200 years. Lifespan is one of the few metrics for which we have reason to believe could prove more tightly linked to intelligence than absolute brain size, and that brings me back to that paper, and how I missed how strongly its data support that opposite conclusion. Let’s start again…
We might have some considerable grasp of primate intelligence, but we are far off being able to create a similar scale for whale cognition. Because of the limited nature of direct testing among dolphins and whales, we may get our first decent overview via indirect genetic correlates. One of the best ways of doing this within primates is to gauge the evolutionary pressure put on brain structure genes, as they vary among groups and species. That process has begun with a paper published late in August.
Genetic Basis of Brain Size Evolution in Cetaceans
THIS paper attempts to gauge the selective pressures on cetacean brains (here, microcephaly genes). Their objective was to test which factors might be responsible for the sudden increase in encephalisation apparent in the fossil record of this group. The conclusion they came to was that its correlation with group size leant support for the Machiavellian Intelligence Hypothesis. Given that my previous post showed how strongly that hypothesis was discredited by all datasets restricted to direct measures of intelligence, that result would be surprising.
It might have been that cetaceans are a special and unusual exception to the rule in which the Machiavellian Intelligence Hypothesis actually applies. It might have been that the comprehensive data we have to date hides a bimodal distribution in which intelligence for social animals does tend to increase with group size, but that the solitary lifestyle presents the highest cognitive demands of all (at least for carnivora, solitary animals are more intelligent, p=0.02), and that there is a dearth of solitary animals among the cetaceans. However, what seems to have happened, is that their use of data lead them to inadvertently treat positive and negative selection, as a continuum, where both types are generated by the same selective pressures.
In this paper they use the metric ω = the ratio of nonsynonymous to synonymous exon substitutions, such that ω > 1 indicates positive selection, and ω < 1 indicates negative selection. Then they take a closer look at the four microcephalin genes that have shown signs of positive selection in some groups.
Also important here is the way encephalisation quotient is used in that paper. Although I cannot find any evidence to support this metric in primates as an indicator of intelligence (primates = the only group for which we have adequate empirical data), it contain importance in this: the ratio by which brain size exceeds that expected for a given body size should reflect the relative selective pressure that lineage experienced for increasing cognitive demands in general. However, even in that context, such a metric contains a flaw. There is an upper limit to brain weight, past which the organ becomes so vulnerable to linear and rotational waves that it is susceptible to injury. Already we are seeing this in humans with careers in high impact sport, but even for the largest, most buffered organisms, the increased susceptibility to shock waves causes the maximum to top out at a brain of around 10kg. Depending on how steeply dangers increase near this limit, many EQ values might have to be readjusted, or, at the other extreme, just two: those of killer whales and sperm whales.
In the paper, bowhead whales represent an extreme outlier. Despite having the 7th to 9th largest brain on Earth at 2.7 kg, its calculated log(EQ) = -1.06, makes it an extreme outlier, and thus a means by which p values may mislead, and make it more likely that EQ trends artificially, in the same direction as group size. I suspect that this further confused the authors, since they seem to use the trend of EQ against ω as conformation of their methodology.
Below is figure 3 from the paper. The bowhead is the only solitary animal in their survey. It is seen at the bottom with log(group size) = 0
Note that the bowhead whale is the only cetacean in the survey with WDR62 under positive selective pressure – and that it shows the highest positive selective pressure of any gene in any graph. Note also that both CDK5RAP2 and ASPM are under the highest purifying pressure in the case of the bowhead whale. So this data shows, as clearly as is possible, that the solitary animal has the highest demands of its microcephalin genes. Note how treating purifying selection and positive selection as part of the same spectrum, then drawing trend lines on that basis, has obfuscated the task. It is a procedure for which they make no justification in the paper. To broaden this finding, remove the bowhead datapoint and add the y axis from the zero point in your minds eye. That new vertical line represents minimised selective pressure. Note how the points for those cetaceans with large group sizes seem to cluster close to that axis, suggesting that this result generalises, and the less sociable the cetacean, the higher the demands on its intelligence tend to be.
I’m not sure if this finding is good news for this website as the group size for sperm whales are about the average for this order. If anyone can explain to me how the findings could still stand, in the direction indicated by the authors, I would be grateful, particularly as I kinda like the Machiavellian Intelligence Hypothesis. Unfortunately, after this paper, it looks dead even for the particular case of the cetacean order.
It may be useful for me to point out that, a potential confusion in reading the referenced paper, is that it uses ‘IQ’ exclusively to refer to an amino acid sequence where isoleucine (symbol, I) is followed by glutamine (symbol, Q).