Category Archives: science

the third reviewer. always, the third reviewer…


an aside

A commenter over at Bergh points out an interesting feature of Nick Bostrom’s cv: he lists financial support from CSN under ‘honours and grants.’ CSN is the Swedish national student aid authority. Generally, students are supported with a combination of loans and grants. These grants are means tested, not based on merit; if you get a loan, you get a grant. In other words, it’s not exactly a feather in one’s cap to be supported by CSN.

OK, so he has this on his cv. No real harm done. But it is fascinating that someone who also has three seven-figure grants, publications in places like Nature, Ethics and the Journal of Philosophy, books at OUP, not to mention being the director of an Oxford Institute, feels compelled to pad his cv such. Still, at least he probably knows the difference between a letter to the editor, an op-ed and an article. Unlike some prominent Swedish ‘producers of ideas.’

Update: the link is now fixed.

mind the gap… in sensibilities

How much is explained by IQ? I suppose making arguments about the importance of IQ in explanations of economic growth, longevity and so on is about as controversial as research gets. Anyone brave enough to tread into this minefield is sure to be hit from all sides. Bryan Caplan is of the opinion that IQ matters for all kinds of things and that to ignore it in, say, a model of the effect of education on earnings will mean overestimating the effect of schooling.

Thus, IQ is highly policy-relevant after all. The left-wing ideologues who damn anyone who even thinks the letters “IQ” are actually on to something: IQ research does turn out to be a rationale for “right-wing” laissez-faire policies. The more IQ matters, the more likely it becomes that existing government policies are a waste of money – and that you would get a bigger payoff by doing less – or maybe nothing at all.

This is controversial enough for many people. Now add race (implicitly, if not explicitly) to the mix. Speaking of mix, another one who has partaken of the mixed grill, Satoshi Kanazawa, has a paper in the current issue of the British Journal of Health Psychology that is sure to stir the pot. Here’s the abstract:

Wilkinson contends that economic inequality reduces the health and life expectancy of the whole population but his argument does not make sense within its own evolutionary framework. Recent evolutionary psychological theory suggests that the human brain, adapted to the ancestral environment, has difficulty comprehending and dealing with entities and situations that did not exist in the ancestral environment and that general intelligence evolved as a domain-specific adaptation to solve evolutionarily novel problems. Since most dangers to health in the contemporary society are evolutionarily novel, it follows that more intelligent individuals are better able to recognize and deal with such dangers and live longer. Consistent with the theory, the macro-level analyses show that income inequality and economic development have no effect on life expectancy at birth, infant mortality and age-specific mortality net of average intelligence quotient (IQ) in 126 countries. They also show that an average IQ has a very large and significant effect on population health but not in the evolutionarily familiar sub-Saharan Africa. At the micro level, the General Social Survey data show that, while both income and intelligence have independent positive effects on self-reported health, intelligence has a stronger effect than income. The data collectively suggest that individuals in wealthier and more egalitarian societies live longer and stay healthier, not because they are wealthier or more egalitarian but because they are more intelligent.

I haven’t read the paper so I can’t really comment on the data or analysis. No one can accuse him of studying boring stuff though.

what if life is just some hard equation, on a chalkboard in a science class for ghosts?

I recently read about how Spore is going to change the face of gaming, nay the world. Brought to us by the creator of The Sims, the game ‘draws on the theory of natural selection. It seeks to replicate algorithmically the conditions by which evolution works, and render the process as a game.’

The article brought to mind Nick Bostrom‘s simulation argument. Here’s the abstract:

This paper argues that at least one of the following propositions is true: (1) the human species is very likely to go extinct before reaching a “posthuman” stage; (2) any posthuman civilization is extremely unlikely to run a significant number of simulations of their evolutionary history (or variations thereof); (3) we are almost certainly living in a computer simulation. It follows that the belief that there is a significant chance that we will one day become posthumans who run ancestor-simulations is false, unless we are currently living in a simulation. A number of other consequences of this result are also discussed.

I’m not sure of the necessity of posthumanity to the result (or even what posthumanity is; Bostrom gives a definition here). That aside, it looks like the ancestor-simulations are on the way, though not perhaps imminent.

One thing about the simulation argument strikes me: if we are (almost certainly) living in someone’s simulation, the same or similar logic could be applied to the author of the simulation: they are (almost certainly) living in someone else’s simulation. I just checked, and Bostrom acknowledges this in the original paper (p9). As for whether it is ‘necessary for the hierarchy to bottom out at some stage’, ‘the metaphysical status of this claim is somewhat obscure’. Could it all go around in a circle? Are we living in a simulation created by someone living in a simulation created by someone who will eventually live in a simulation we create? Is existence just a multiple equilibrium story? Jeez, this is more fun than economics.

[Aside: Bostrom too has had the mixed grill.]

P.S. This is the first in an occasional series of posts with titles by the Silver Jews.

from here to paternity

Oooh, he’s got his father’s eyes/hands/vacant expression. Such are the comments every new parent has to tolerate. Never mind that three-week-old ricardito looks like your standard generic infant: he apparently has my nose (`nicely rounded’ says the mother-in-law; `bulbous’ sighs the wife).

The degree to which newborn babies resemble their fathers is, of course, interesting from an evolutionary perspective. Males don’t want to invest resources rearing other males’ children, and so looking like your father can be adaptive. The catch is that it’s only adaptive if your father is in fact the male proposing to rear you. It might equally suit philandering bandits (and their female partners) to have children that at most resemble their mother. Empirically the matter seems not to be settled, as a glance here indicates, but it does appear that infants’ likeness to their fathers is not great.

[Aside: I’m not sure how, or even if, such empirical studies account for the fact that any lack of paternal resemblance in the study group can arise for two reasons: first, that kids don’t look like their fathers all that much; second, that their fathers might not be their fathers. ]

Sticking to theory, what range of `strategies’ might we expect to see in macro equilibrium once we take the above forces into account? If we lump mating behaviour in there too, my hunch is that we should observe equilibria where both {adultery, no paternal resemblance} and {fidelity, paternal resemblance} strategy combinations are played. From the male viewpoint, resemblance is only adaptive if one is in danger of being cuckolded by philanderous bandits playing the first strategy pair, while banditry only pays if there are potential cuckolds out there.

This at first seems to predict an `arms race’ along the resemblance dimension. One problem with this idea, though, is that there’s an inherent asymmetry. The philanderer can’t make his offspring resemble the cuckold, since there’s no gene for that; at best, the offspring will look like the mother.

In these circumstances, why wouldn’t a faithful male (also `playing’ the paternal resemblance strategy) leave if there is no such resemblance and his mate’s child resembles her alone? This paper (discussed here) introduces another strategy level, that of deceit and gullibility. It is obviously in the mother’s interest always to tell her mate that the baby looks like him, regardless of the actual paternity. A main result of the paper is that it may be in the mate’s interest to believe her, as long as the probability of deceit is low enough that the expected benefit from sticking around exceeds the expected opportunity cost.

The problem with this story is that `low probability of deceit’ equals ‘low proportion of bandits’. And a low proportion of bandits is just not consistent with stable equilibrium if the non-bandits are gullible and babies do not overly resemble their fathers. Philandery will be highly adaptive, bandits will flourish and gullibility will cease to be adaptive.

Something is missing here. I’ll have to think.

r&r for peer review

The quality of scientific research is maintained by the peer review process. Or at least that’s the idea of peer review. But that process is expensive, time consuming and anything but fail-safe. It puts a considerable strain on academics’ time. Some that I know, claim to get multiple requests to review papers every day—though I’m not sure I believe them. In any case, it takes time to review papers properly and you are expected to do it as part of your service to the scientific community. In other words, you do it for free. There are serious doubts about the utility of this system of self policing both in the natural and social sciences. There are plenty of cases of good or even great papers that get rejected and mediocre papers that get published in elite journals. It’s a capricious process.

Now Nature, as part of an ongoing debate on peer review, has launched an interesting experiment. Authors can choose to participate in a trial online peer review where they post papers to a website and academics in the field can provide comments and review the manuscripts. Submitted papers will still go through the traditional review process, but the editors will be able to take both sets of comments into account when making publication decisions.

This seems like a pretty good idea to me. It happens all the time that papers get rejected/published because editors don’t have enough information, or rely too heavily on the judgement of a bad reviewer or ignore that of a good reviewer. Hopefully the Nature trial will work in such a way that bad reviews will stand out from the crowd as just that.

This and this are also kind of interesting.