Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Some counterevidence for human sociobiology

0 Post author: taw 29 August 2009 02:08AM

I love seeing counter-evidence for everything. I estimate that while most of my beliefs are true (otherwise I wouldn't believe them in the first place), a small percentage is almost certainly completely false - and I don't really have any reliable way of telling the two apart.

Indiscriminatingly looking for counter-evidence for all of them can be very rewarding - the ones that are true are much more likely to sustain the assault of it than the ones that aren't. Yes, I might ignore counter-evidence of something that's false, or accept it for something that's true, ending up worse off, but it seems plausible that on average it should improve quality of my beliefs.

For example some of the standard beliefs about human sociobiology that seemed to be extremely widely held here are:

  • Men have lower chances of having any kids than women
  • Richer people, especially men, are more likely to have kids, and have more kids

Charting Parenthood: Statistical Portrait of Fathers and Mothers in America disagrees with them.

  • It's true that young men are less likely to have children than young women, but it reverses at old age, and total chance of having children during lifetime is - for people over 45 - 84% for men, and 86% for women. As some of childless men might still have children between 45 and their death (quite a few according to data), but almost no woman will, the difference must get smaller by the time of death, or it might even reverse. This is pretty convincing evidence against a major gender difference in chance of having children, at least as far as modern America is concerned.
  • The chance of having children is highest for people between 100% and 200% of poverty line (poverty line, not median income, these are all poorer than average people). For women going either lower or higher reduces chances of having children considerably. For men getting poorer reduces chance of having children considerably, while getting richer reduces it but only slightly. However - younger people are much more likely to fall below poverty line, and men tend to reproduce later, so even that can easily be an artifact of age-income relationship. The data is fully compatible with both poverty and wealth being negatively correlated with chance of having children in both genders.

These are not direct tests of sociobiological claims, so what we have is not exactly what we would like to, but I find them to be quite convincing counter-evidence. My belief in these sociobiological claims is definitely lower than before, at least as far as they concern modern world, even though I can imagine more focused studies changing my mind back.

More counter-evidence for things we commonly believe here, sociobiological or otherwise, welcomed in comments.

Comments (28)

Comment author: PhilGoetz 29 August 2009 07:06:40PM *  7 points [-]

I do appreciate your post; see Yvain's comment below for my response. It prompted me to re-read the famous letter Against Sociobiology from 1975. Points to note:

  • Several of the lynchpins of this letter are based on dogmas that we now look back on as having been upheld for purely ideological reasons. The opening paragraphs criticizing ideas that behaviors could be genetically-biased as "absurd" are one example. The claim that animal behavior can't teach us about human behavior is another. The revisionist pre-history story they allude to which rejects the idea that hunter-gatherers had sexually-defined roles is another.

  • The motivation for the letter, and its main arguing point, is not about truth, but about EO Wilson's alleged reactionary motives.

  • They associated Wilson with the Nazis in the second paragraph.

  • Recall that this letter is the very best argument ever made against sociobiology, by the most prestigious biologists (including Stephen J. Gould and Richard Lewontin), which most later arguments cite as authoritative.

  • This letter was written by a group that met regularly for over a month to compose it, in a room very near E.O. Wilson's office. No one ever told him about it until after it was published. If the authors of the letter had any interest in truth, they would have walked down the hall, shown him the letter, and said, "What do you say to this?"

I have the impression that 1970 marked the onset of a new dark age in science, after which ideology played a much larger role in the selection of ideas.

Comment author: Douglas_Knight 29 August 2009 10:54:38PM 2 points [-]

Recall that this letter is the very best argument ever made against sociobiology, by the most prestigious biologists (including Stephen J. Gould and Richard Lewontin), which most later arguments cite as authoritative.

Are you endorsing it as the best argument made against (brand-name?) sociobiology? I imagine that it's better than most of the later arguments that endorse it, but that's a much weaker claim than denying the existence of better arguments. You can't rely on opponents of an idea to filter for the good arguments against it. In particular, if the best arguments against it are less sweeping, they may be ignored.

Comment author: PhilGoetz 31 August 2009 11:34:24PM 2 points [-]

Perhaps I should have said "most famous" or "most influential". I'm not qualified to judge whether it's the best.

Comment author: Douglas_Knight 01 September 2009 01:46:38AM *  0 points [-]

I interpreted your comment as meaning that ad hominem (etc) arguments should be seen as in contrast to good arguments. I see it as the opposite. We should expect the use of effective rhetorical techniques and prestigious authors to lead to fame. These are good reasons to expect not to hear of better critiques. [ETA: this is what I was groping towards in my earlier comment]

I read the letter after writing my comment and it is not sweeping, contrary to my claim. Its arguments are pretty reasonable and it doesn't explicitly misrepresent Wilson much. But it is very effective at producing false beliefs, such as my belief that it was sweeping, and TAW's beliefs elsewhere on this thread.

Comment author: taw 30 August 2009 12:41:30AM -1 points [-]

The letter is a pretty good example of problems of communicating over a paradigm difference, how convincing you find it seems highly based on which paradigm you accept.

And it seems to be mostly criticizing a version of sociobiology that attaches changes in post-Paleolithic to changes in genetic basis of humanity, and existing variety of behaviour to existing variety of genes in modern individuals.

As far as I can tell, they're perfectly right about this, they won, and nobody holds such beliefs any more. What's left today is far milder claim that all humans share pretty much the same range of highly flexible behaviors that was well adapted to ancestral environment, and cultural evolution is based on memes not genes. (and this mild version is what I was looking for counter-evidence for in my post)

The point that evidence for what Wilson was talking about was pretty much non-existent is as valid today as it was then. There are very few genes known to be linked with anything behavioral, and now that we know how few genes there are, it severely limits possibility of their existence. As they point out, there is also very little evidence that our ancestral environment was anything like what sociobiology tends to so happily assume, and variety and rapidness of change of human cultures is huge.

The point that scientific theories are often given much more credit than they're due when they're convenient for those benefiting from status quo, is a valid argument. This doesn't apply as much to mild sociobiology of today as to radical sociobiology they are criticizing, which basically says poor people have bad genes etc., you can easily imagine people in power happily accepting it even without proper scientific evidence. If there was overwhelming evidence for it, then well, tough luck, but as they point out, such theories were often proposed and widely accepted against the bulk of scientific evidence. As far as I can tell, this seems historically correct.

Comment author: Tyrrell_McAllister 30 August 2009 02:19:39AM 3 points [-]

Was Wilson claiming that genetic variation within modern humans accounted for behavior variation within modern humans, other than in the case of male-female differences? I haven't read Sociobiology, but I don't remember hearing that claim attributed to him.

Perhaps the Against Sociobiology letter made valid arguments against the version of Sociobiology that it described. But was that the version put forth by Wilson, as the letter claimed?

Comment author: Douglas_Knight 30 August 2009 04:39:15AM *  1 point [-]

And it seems to be mostly criticizing a version of sociobiology that attaches changes in post-Paleolithic to changes in genetic basis of humanity, and existing variety of behaviour to existing variety of genes in modern individuals.

Where are you getting this? You go on to mention their complaint #5 that they don't like Wilson's reconstruction of the ancestral environment, which makes it sound like they're saying that he believes in uniformity.

They do say that some people say "poor people have bad genes" and they fear that those people will turn to Wilson, but they do not say that Wilson claims that. They talk about two forms of "biological determinism," contrasting Davenport, Jenkins, and Shockley as eugenicists / people who believe in diversity against people who make uniform claims about humanity, based on evolution namely Wilson, Lorenz, Ardley, and ... Spencer. OK, where they chose to put Spencer (who subscribes to both) speaks volumes, but there's a reason they don't come out and say it.

There are some places that could be read as saying that Wilson believes in diversity, but none of those places are very clear. The clearest is the part about homosexuality: surely, no gene for homosexuality is fixed, at least not fixed on! But they follow that with "for Wilson, what exists is adaptive, what is adaptive is good, therefore what exists is good" which makes it sound like he's talking about uniformity, as with "conformer genes" at the beginning of the paragraph. Anyhow, Wilson responds that he explicitly warns against the naturalistic fallacy and that the whole thing is a gross misrepresentation.

Nor does the letter seem to portray Wilson like Spencer, a believer in rapidly changing uniformity.

Comment author: tut 29 August 2009 08:20:49AM *  7 points [-]

Voted up because I like the attitude you show in your first two paragraphs.

However I think that you misunderstand the claims that you argue against.

Men have lower chances of having any kids than women

The obeservation underlying this is that that the average person has three times as many female as male ancestors. But every child has one male and one female parent. So the average male ancestor of now living people must have three times as many decendants as the average female ditto.

This might be because genocides were carried out by killing all men in a population, so that the women were forced to marry the conquerors. Or because two thirds of males that were born in some era were killed as children. Or because there was a tradition were some men killed other men's children, so that the surviving children had fewer fathers than the children that were born.

But in any case there was something going on that is not caught by your statistics. And the simplest (and nicest) explanation is probably that women got to choose who they had sex with, and some men were more attractive than others.

So why does your statistics say differently? I see two explanations off the top of my head. 1) Marriage, or in other words rationing of partners. Today each man marries one woman, and then other women are expected to marry another man and have at least one child with their husband. 2) Contraception, people have fewer children when they don't want children nowadays.

Also, I expect that if that study would talk about biological fathers rather than social fathers you would get different results.

Richer people, especially men, are more likely to have kids, and have more kids

  • Rich and rational people have as many children as they want, not as many as they would have had in the past.
  • See above about marriage. When many women could have the same man without being shamed more attractive men would have more children relative to others. Rich men are ceteris paribus more attractive than others, and that difference was probably bigger when normalcy for both men and women were on the limit of starvation
  • The "poverty line" is not really relevant in an evolutionary context. Being "above the poverty line" in America today means being inconceivably wealthy in the terms that were normal 200 years or more ago. Genetically we have not changed much in a couple of centuries.
  • Wealth in the modern sense is not really relevant either. The kind of property that would count in your survey has not been around long enough to make any significant changes to the human genome.
Comment author: pjeby 29 August 2009 05:30:51PM 3 points [-]

Also, I expect that if that study would talk about biological fathers rather than social fathers you would get different results.

That's probably an important point, since some studies have claimed that a pretty enormous number of social fathers are not their kids' biological fathers.

Comment author: knb 29 August 2009 06:56:49AM *  5 points [-]

Richer people, especially men, are more likely to have kids, and have more kids

Is this really a widely held belief? It seems to me that the actual claim made here is simply that rich men have higher sexual market value on average.

Number of children is not a good proxy for sexual market value, especially since reproduction and sex are now only weakly related.

Comment author: Yvain 29 August 2009 12:41:31PM 4 points [-]

If I wanted to test sociobiology as a theory with these statistics, I'd be much more interested in the numbers for hunter-gatherer tribes. In particular, I expect the invention of birth control has driven these numbers well away from what sociobiology would predict.

I guess these are a test of whether sociobiology can be applied willy-nilly to make predictions about every facet of our own society, but I hope people don't think it's that easy.

Comment author: PhilGoetz 29 August 2009 06:51:29PM *  5 points [-]

Agreed. You can't say "Sociobiology predicts X for every human society".

It's even a bit nonsensical to talk about evidence "against sociobiology". You can talk about evidence against particular sociobiological theories. But the idea that sociobiology itself is bankrupt, and should not be a field of study - that genetics and evolutionary psychology have no influence on human society - is absurd.

Comment author: teageegeepea 29 August 2009 06:18:20PM *  2 points [-]

There seems to be a boost in children among the richest.

I thought the higher death rate of men would show up in the data.

It should be noted that human beings are an unusually infertile species.

Comment author: Douglas_Knight 30 August 2009 08:23:20PM 0 points [-]

I thought the higher death rate of men would show up in the data.

Where? these are just percentages.

One of the deleted comments pointed out that married men live longer and that the numbers of old parents are really only old surviving parents. So if you took death into account, the gap would widen. I think this effect is small, though, since "old" is defined as 45+, a wide range.

Comment author: NancyLebovitz 29 August 2009 03:29:59PM 2 points [-]

One thing that seems to get left out of sociobiological discussions is arranged marriages. Some fraction of our history/genes isn't about people choosing their own partners.

Comment author: Alicorn 29 August 2009 03:59:37PM 3 points [-]

No, it's about people choosing their own children-in-law and the parents of their grandchildren - probably at least as efficient a system for picking out certain good genes.

Comment author: NancyLebovitz 30 August 2009 01:07:41AM 2 points [-]

It's efficient, but it presumably gives a somewhat different set of advantages.

Most of the sociobiology I see (casual discussions online) seems to assume that reproduction starts with attraction between strangers.

I don't know what the proportions are, but arranged marriages wouldn't give nearly as much of an advantage to "bad boys".

I suspect it gives a much larger advantage to prudence, and it might give a bigger advantage to skill at negotiation.

There's a third category: choosing people who are from the same smallish social system-- reputation might matter.

Comment author: Alicorn 30 August 2009 01:38:50AM 3 points [-]

Arranged marriages are also a way to unite families, whether they were particularly disharmonious before or not, and therefore were probably often engineered for resources more than the really obvious genetic factors. Wealth (and indirectly, the ability to obtain wealth) running in one's family netted the ability to secure a similarly wealthy/wealth-obtainment-advantaged spouse. Irresponsible "bad" people probably had (have) all of their children via non-marital encounters in societies where arranged marriage was (is) common.

Comment author: Theist 29 August 2009 03:40:38AM 2 points [-]

An important consideration is that our society has a very strong taboo against polygyny.

Comment author: Simon_Jester 29 August 2009 10:23:11AM *  1 point [-]

I wouldn't have assigned much of a prior probability to either of those common sociobiological beliefs, myself. It would hardly surprise me if they were both complete nonsense.

So what do you mean when you say that these beliefs are "standard" or "widely held?" Obviously, I am not a representative sample of the population, so I may have no opinion on a widely held belief. But I'm not aware of strong evidence that these beliefs are widely held, or at any rate are more widely held than the evidence would warrant.

Or, with tongue firmly in cheek, I claim that I'm presenting counterevidence for the common belief that [insert proposition here] is a common belief...

Comment author: Larks 30 August 2009 08:51:51AM 0 points [-]

I estimate that while most of my beliefs are true (otherwise I wouldn't believe them in the first place),

The former doesn't follow from the latter: it could be the case that each of your beliefs are your best guess, but you still give them a probability less than .5 if they contain more than a single bit of information.

Comment author: Douglas_Knight 29 August 2009 11:14:37PM 0 points [-]

Could you link to more specific things than the whole report? I think the first bullet point is supposed to link here.

Comment author: CronoDAS 29 August 2009 07:45:57PM *  0 points [-]

I posted this earlier. Relatively trivial interventions can skew results away from what evolutionary psychology might predict...

Comment deleted 29 August 2009 02:59:52AM *  [-]
Comment author: taw 29 August 2009 03:42:39AM 0 points [-]

People often talk as if these rules were universal. In fact we have very little evidence for their universality. We don't even have convincing evidence that they held more often than not in ancestral environment, even though it sounds plausible.

In particular is the second paragraph of your comment referring to the modern world, or a typical ancestral environment, or both, or do you claim it's universal?

Right now, male and female lifespans don't differ that much until long after fertility, so different life expectancies don't matter (and historically either of them could be longer, for example female infants might be getting less care than male infants - hard evidence is needed to claim typicality).

Calling industrial accidents and violence high risk strategies sounds very weird. It's like saying living in poor part of the city, or being black is a "high risk strategy" - it's not strategy, it's society abusing some of its members, and them not being able to do much about it.

And as for the association between high status and length of fertile life - that's exactly my point, it's one of those things that are commonly claimed as if they were widely proven, but there's very little evidence for them, except maybe for extremely low status people.

Comment deleted 29 August 2009 04:21:18AM [-]
Comment author: taw 29 August 2009 05:01:26AM -1 points [-]

There's an obvious problem with this argument as the risky jobs are not the high paying ones. Correlation between risk and payoff seems to be negative, not positive. And neither does violent crime pay much.

As an particularly extreme example of both, drug dealing in States is extremely risky (death far more likely than in military in Iraq, then risk of less than lethal violence, and imprisonment) and extremely unprofitable (wages far lower than minimum wage) behaviour. [ famously described in http://en.wikipedia.org/wiki/Freakonomics ]

Comment author: knb 29 August 2009 06:46:26AM *  3 points [-]

There is a real risk premium for jobs. Yes, high income jobs are less physically risky, but that isn't the relevant comparison. The relevant comparison is to other low skill jobs. So, for example, lumberjacks get paid more than doormen.

Comment author: Bo102010 30 August 2009 02:38:45PM 1 point [-]

I think Freakonomics supports the opposite view about risk and why low level drug dealers take it. It's the small chance of a huge payoff that motivates them, even though most of them end up living with their mothers.

Comment author: timtyler 31 August 2009 01:40:05PM 0 points [-]

"Even in their forties, men were less likely than women to have a child and more likely to have fewer children. "At age 50, almost 20% of men are childless. The estimate for women is about 13%.""