this is not a double post
December 19, 2011 4:28 PM   Subscribe

How can we better understand the interplay of nature and nurture in determining our personalities, behavior, and vulnerability to disease? Perhaps we should be looking at identical twins. (National Geographic January 2012 cover story)

Related
* Photo Gallery One
* Photo Gallery Two
* Interview with psychologist Nancy Segal, Director of the Twin Studies Center at California State University, Fullerton.
* Behind the Scenes Video of the cover shoot

In September 2010, the magazine also published a special feature of five different photo galleries, from the Twinsburg, OH Twins Day Festival mentioned in the above article:

* Festival Photos
* Oldest
* Youngest
* Most Alike
* Least Alike
* Match Game

Also see: A Town Full of Twins, (video) for a look at a town in Brazil where the rate of twin births is nearly a 1000% above the global average.

Previous coverage of the Ohio Twins Day Festival on MeFi: 2005, 2007
posted by zarq (89 comments total) 8 users marked this as a favorite
 
As a twin myself, I second this post!
posted by BigHeartedGuy at 4:59 PM on December 19, 2011 [3 favorites]


Nice post.

Unfortunately, though, the study of twins has more often confused the study of nature and nurture than clarified it.
posted by Tell Me No Lies at 5:10 PM on December 19, 2011


I know that NatGeo is a serious publication, but I really wish they stuck a poster of Arnold and Danny into the Least Alike gallery.
posted by filthy light thief at 5:11 PM on December 19, 2011 [1 favorite]


Are both the photo gallery links supposed to point to the same page? Not sure if you're being clever or if it was a mistake...
posted by polymath at 5:20 PM on December 19, 2011


My identical twin and I both work at the same company, although we came to it from very different directions. I can't convince him to consider Twinsburg or any similar festival. In fact, if we meet up on the weekends and he's wearing a similar shirt, he'll go change before we go out.

We both participate in the SRI twin studies registry. I encourage any other twins to register as well.
posted by blob at 5:25 PM on December 19, 2011 [2 favorites]


My mother, who taught school, always said that twins were a little odd, not quite right, which may be true but they're still kinda cool.
posted by shoesietart at 5:45 PM on December 19, 2011


After Cyril Burt, "twin study" == "scientific fraud."
posted by localroger at 5:53 PM on December 19, 2011


Unfortunately, though, the study of twins has more often confused the study of nature and nurture than clarified it.

Can you elaborate? How does the increase in data and analysis make things less clear? I don't understand what you're getting at.
posted by binturong at 5:59 PM on December 19, 2011


the study of twins has more often confused the study of nature and nurture than clarified it

I am currently working at the UK Adult Twin Registry, and this is nonsense. Twin studies are one of the most important tools for assessing the heritability of traits, and they remain one of the pillars of genetic epidemiology. It is true though that the current boom in genome-wide association studies has reduced the importance of twin studies as a route to discovering the genetic causes of disease. It's also broadly true that most complex diseases are about 50% heritable, meaning that for a large majority of diseases both nature and nurture have an important role.
posted by roofus at 6:00 PM on December 19, 2011 [8 favorites]


polymath: "4Are both the photo gallery links supposed to point to the same page? Not sure if you're being clever or if it was a mistake..."

Sorry, no. It was a mistake. The gallery links should be One. Two. I've flagged the post. Hopefully a mod will kindly fix it.
posted by zarq at 6:02 PM on December 19, 2011


It seems that the evidence of the genetic lottery upsets some peoples' notion of free will, human uniqueness, equality and whatever other ideological notions they hold. Witness this from the article: "Using data from four different tests, they came up with a heritability score of 0.75 for intelligence, suggesting the strong influence of heredity. This ran counter to the prevailing belief of behaviorists that our brains were blank slates waiting to be inscribed by experience. ..."The far-left groups on campus were trying to get me fired," Bouchard says."

Follow the evidence and a plague on all ideologies of left, right or centre. Scientists are not value-free but once you start embracing your conclusions before you start your study you are no longer engaged in scientific research.
posted by binturong at 6:10 PM on December 19, 2011 [2 favorites]


blob: " We both participate in the SRI twin studies registry. I encourage any other twins to register as well."

What they're doing is pretty neat, thanks for posting a link to them. My kids (3 yrs) are mixed gender twins and I'd register them if I lived in the area. But I did just email them to sign up for their newsletter.

From the site, for anyone who might be interested:
For more information on twins research and twin registries, watch these videos about "Twins: Unlocking the Secrets of Nature-Nurture Interactions": Background on twins research and an overview of Twin Registries from Dr. Gary Swan's talk at Cafe Scientifique Silicon Valley.

posted by zarq at 6:13 PM on December 19, 2011


It seems that the evidence of the genetic lottery upsets some peoples' notion of free will

Oh no, it's not that. It's the decades of forced sterilization, children taken from their parents, and systemic denial of opportunity based on data so vague that they are meaningless that kind of piss us off. Could there be a statistical correlation with intelligence? Sure. There probably is.

As a statstician you probably know that correlation does not imply causality. Low income, poor nutrition, lack of educational opportunities, and discrimination could easily account for your correlation, couldn't they?

For the first 50 years every single twin study ever done was fraudulent. This is known, and I'm not going to look up links because it's so well known that merely mentioning the name Cyril Burt should make the point. Why would any sane person even want to go there?
posted by localroger at 6:16 PM on December 19, 2011


And to add to the genetic lottery feeling, we have trisomic rescue.
posted by francesca too at 6:33 PM on December 19, 2011


I hear you, localroger. But you support my point. So people with an agenda will always distort data, misrepresent evidence, and abuse science to further their social policies. But to shut down studies of genetic influences and chant the mantra of "all people are created equal" is not a good way to go. Do you want to understand the human animal or do you want to pretend that genetic variability does not have a major role in every human being's life (as indeed it does in all species)? These twin studies suggest otherwise.
posted by binturong at 6:38 PM on December 19, 2011 [1 favorite]


binurong, I want to understand the human animal probably more than you do because I am in a certain sense malformed myself. But twin studies are so easy to abuse and so desireable to abuse by people with lots of money to fund them that they cannot be trusted. Even if your research is perfect people like me who are aware of the history will simply assume it's fraudulent. The well has been poisoned. There might be knowledge there, but the fascists and eugenecists have cloaked it.

Given the relatively small number of humans compared to, for example, bacteria, and our relative complexity, I doubt this is a problem that can be solved. There may be secrets waiting to be revealed by the differences between twins. But at our current level we have no chance of seeing those secrets past the noise of politics and prejudice.
posted by localroger at 6:46 PM on December 19, 2011


Can you elaborate? How does the increase in data and analysis make things less clear? I don't understand what you're getting at.

Identical twins aren't ideal for declarations about nature/nurture. To say, "both twins have this, therefore it must be genetic" is a fallacy.

1. Identical twins' upbringing is completely atypical, precisely because they have an identical twin.

2. One twin's upbringing, much like his/her genes, is basically the same as the other twin's, so...?

3. Any trait that is more likely in twins than in singletons -- whether because of nature or nurture -- is more likely in both twins, because each twin is a twin.
posted by Sys Rq at 7:00 PM on December 19, 2011


roofs wrote...
>the study of twins has more often confused the study of nature and nurture than clarified it
I am currently working at the UK Adult Twin Registry, and this is nonsense.


Sorry, I was over broad there. I was speaking from a psychology point of view. The study of twins has of course been tremendously useful in genetics.

binturong wrote...
Can you elaborate? How does the increase in data and analysis make things less clear?

An increase in data and analysis makes things less clear when a large amount of seemingly contradictory evidence is produced.

While cases of twins showing extraordinary similarities get all the press (see the Jim Twins) the actually situation gets very confused when you start looking at larger samples.

If you start poking around the Minnasota Center For Twin and Family Research results you start to run into all sorts of really specific correlations. For example there is a genetic correlation in how events affect twins, but only events that affect them individually and that they had a hand in making. There is no correlation for family events even if they affect the twin and none in any case for events that the twin had no hand in.

Great! You say. With information that specific you can really start digging into a model for human behavior.

But then you read the next study, and the next study, and the next study, and pretty soon you realize that all of these specific data points aren't really making any sense together. In many ways they're too sparse but in any case there's no way to take, say, three of them and make a useful predictive statement with them. So, more data and more confusion.

In any case I won't claim to be an expert on this. I'm just channeling my brother's frustration from his first pass at a psych thesis.
posted by Tell Me No Lies at 7:14 PM on December 19, 2011 [1 favorite]


One twin's upbringing, much like his/her genes, is basically the same as the other twin's, so...?

If you read the article, you will see that they allow for this by comparing identical twins brought up together and those brought up apart, as well as twins brought up not knowing they had a twin and those brought up apart but also knowing they had a twin.

To say, "both twins have this, therefore it must be genetic" is a fallacy.

And your statement is a strawperson. Why the resistance to accepting that many human qualities and characteristics are genetically determined? That would be a completely non-controversial statement applied to any other species.

Identical twins aren't ideal for declarations about nature/nurture.

Not ideal maybe but pretty damn good for biological studies of whole organisms.
posted by binturong at 7:15 PM on December 19, 2011 [2 favorites]


No one studies just identical twins.

Instead, researchers compare the correlation of a trait between same-gender fraternal twins, with the correlation of the same trait between identical twins.

Presumably there is no real difference in how the world treats two fraternal twins of the same gender vs. how it treats two identical twins. But the fraternal twins share 50% of their genes, while the identical twins share 100%. It's a perfect natural experiment. From that data, you can do statistics to figure out how much of the trait is heritable in the general population.

The article glosses over this idea, but it's the central idea behind twins research. It also solves a lot of the problems that people have brought up in this thread.
posted by miyabo at 7:25 PM on December 19, 2011 [2 favorites]


An increase in data and analysis makes things less clear when a large amount of seemingly contradictory evidence is produced.

Ha ha, yes, humans are a very difficult species to study! A lot of the psychological elements also affect non-twin siblings -- birth order, gender, age differences etc. all mean that children can experience different environmental influences growing up even when they are raised in the same family household. But still, it is fascinating to discover the importance of genetics and it always puzzles me how strongly some people kick back against the role of genes. As miyabo says, twin studies tease out the genetic contribution to traits from all the other variables.
posted by binturong at 7:32 PM on December 19, 2011 [1 favorite]


I think a major sticking point here is that psychologists define intelligence very differently from normal people.

I'm not a psychologist. When I say someone is "intelligent," I mean that they can clearly express their ideas; they have achieved career or educational success; they have worked hard to accomplish difficult goals. They have discipline, strong relationships, financial security, etc.

When a psychologist says someone is "intelligent," they mean they can solve an array of logical, visual, and verbal problems quickly. This is interesting to study because a person's speed at solving these problems is pretty easy to measure and remains just about constant throughout a person's adulthood. But it doesn't correlate well with what we call intelligence in normal conversation. In fact, after an IQ of 120, there is zero correlation between IQ and career or educational success. Other traits are just more important.

If you've ever had the (dis)pleasure of talking to people at a MENSA meeting, you know there are an awful lot of people who have very high IQs and yet are completely miserable.

There are certain things you do need a high IQ to do. The average IQ of a theoretical physicist is 140 -- that's the 99.997th %ile. But you also need to be extremely strong to be an Olympic weightlifter, or extremely good-looking to be a supermodel. I didn't win the genetic lottery that would enable me to do any of these things, but it's not something I lose sleep over.
posted by miyabo at 8:12 PM on December 19, 2011 [4 favorites]


I've never heard of Cyril Burt. Wikipedia has this to say about his fraudulent research in the larger context of research on twins and genetics:

Psychologists Arthur Jensen and J. Philippe Rushton have pointed out that the controversial correlations reported by Burt are in line with the correlations found in subsequent twin studies. Rushton (1997) wrote that five different studies on twins reared apart by independent researchers corroborated Cyril Burt's findings and had given almost the same heritability estimate (average estimate 0.75 vs. 0.77 by Burt).

Noting that later studies of the heritability of IQ have produced results very similar to those of Burt's, Hunt argues that Burt did not harm science in the narrow sense of misleading scientists with false results, but that in the broader sense science in general and behavior genetics in particular were profoundly harmed by the Burt Affair, leading to an unjustified general rejection of genetic studies of intelligence and a drying up of funding for such studies.


I guess you're disputing claims like that, localroger?

Is that phrase "twins reared apart by independent researchers" refer to something actually as horrifying as it sounds?
posted by straight at 8:15 PM on December 19, 2011


they allow for this by comparing identical twins brought up together and those brought up apart, as well as twins brought up not knowing they had a twin and those brought up apart but also knowing they had a twin.

It breaks my heart that there are enough twins fitting those descriptions for scientists to do that kind of research.
posted by straight at 8:19 PM on December 19, 2011


There are only about 100 known pairs of identical twins separated at birth in the world, so...yay?
posted by miyabo at 8:21 PM on December 19, 2011


For the first 50 years every single twin study ever done was fraudulent. This is known, and I'm not going to look up links because it's so well known that merely mentioning the name Cyril Burt should make the point. Why would any sane person even want to go there?

Um, in order to do the studies correctly?

Gregor Mendel appears to have somewhat falsified/idealized his research (in a manner curiously similar to the accusations against Burt)--which is the foundation of all modern research into genetic heritability. That hardly means that genetic research is a waste of time.
posted by yoink at 9:07 PM on December 19, 2011 [1 favorite]


Mod note: "Photo Gallery Two" link fixed
posted by taz (staff) at 10:07 PM on December 19, 2011 [1 favorite]


When a psychologist says someone is "intelligent," they mean they can solve an array of logical, visual, and verbal problems quickly

I doubt you'll find a paper from the last thirty years that lists simply "intelligence" as a measurable item. The whole I.Q. thing had its last gasp in the 1970s.

This has not stopped MENSA from treating it as a thing, but please don't confuse MENSA for credibility.
posted by Tell Me No Lies at 10:18 PM on December 19, 2011


Here's a list of six points from Cycles of Contingency by Susan Oyama et al, summarising what they call "developmental systems theory". It sums up nicely why the nature-nurture dichotomy is as problematic as ever, and why loads of new data will not help, because it's not about the data, it's about the underlying premises.


1. Joint determination by multiple causes—every trait is produced by the interaction of many developmental resources. The gene/environment dichotomy is only one of many ways to divide up these interactants.

2. Context sensitivity and contingency—the significance of any one cause is contingent upon the state of the rest of the system.

3. Extended inheritance—an organism inherits a wide range of resources that interact to construct that organism’s life cycle.

4. Development as construction—neither traits nor representations of traits are transmitted to offspring. Instead, traits are made—reconstructed—in development.

5. Distributed control—no one type of interactant controls development.

6. Evolution as construction—evolution is not a matter of organisms or populations being molded by their environments, but of organism-environment systems changing over time.
posted by Pyrogenesis at 12:05 AM on December 20, 2011 [3 favorites]


I know - what we need are more studies of separately reared siblings by identical researchers.
posted by hat_eater at 2:57 AM on December 20, 2011 [1 favorite]


I guess you're disputing claims like that, localroger?

Cyril Burt was one of the most brazen and influential frauds in the history of science. Some of his real "separated at birth" twins were separated at such advanced ages as 9 and 13. In his later work most of his twins were simply invented -- they never existed. He also made up a nonexistent research assistant whose decidedly non made-up salary Burt pocketed for many years.

Burt was extremely politically influential, and his "research" directly led to many policies such as forced sterilization and denial of educational opportunities based on ethnicity. He was so powerful that during his lifetime no one dared question him openly; it was only after he was safely dead that those people who suspected the credibility of his work began digging, and even his most vocal critics were shocked at what they found.

In the wake of these revelations a broad survey revealed that every twin study ever done up to that point had either been similarly fraudulent or so methodologically poor as to be useless.

The world is full of rich people who like being told they are rich because they really are better than everyone else and deserve it, so there will always be money to fund "studies" like Burt's. There is really only one possible interesting result that could come from a twin study, and the fact that people keep doing them suggests that that is the result they want to see; this creates a very strong attraction, even among people who think they are being honest, to get that result. And it's such a mushy field that it's not hard to get that result no matter what your actual data say.

Doing a twin study if you don't believe you'll get surprising correlations would be like stabbing yourself in the abdomen to prove it causes peritonitis. Nobody in their right mind would bother.
posted by localroger at 5:48 AM on December 20, 2011


Not getting into the contentious discussion. Just want to say: in the same house? I smell sitcom premise!
posted by psoas at 6:17 AM on December 20, 2011 [1 favorite]


taz: "[ "Photo Gallery Two" link fixed]"

Thank you!
posted by zarq at 6:55 AM on December 20, 2011


Come to think of it, I'm not even sure what the measure in percentages between nature and nurture even means, even ignoring the points I posted above. It's not that it's blatantly obvious what is being partitioned in this way, and what precisely does the number indicate. Even the link provided by roofus states that such a measurement is "sloppy". If I find out that my car drives at 100mph, with 60% contributed by fuel and 40% by the engine, then what have I found out exactly? It drives at 60mph when I remove the engine?

And all that is in addition to, and ignoring the points I posted above.
posted by Pyrogenesis at 7:35 AM on December 20, 2011


> After Cyril Burt, "twin study" == "scientific fraud."

If this is a scientific fraud it is the slickest scientific fraud I have ever seen. Application of Occam's razor suggests it ain't a scientific fraud.
posted by bukvich at 8:34 AM on December 20, 2011


I said: "My kids (3 yrs) are mixed gender twins and I'd register them if I lived in the area. But I did just email them to sign up for their newsletter. "

Just a followup on this... I just heard back from someone at SIRI via email and they do take registrations for twins outside the Bay area.
posted by zarq at 9:09 AM on December 20, 2011


SIRI? SRI. My kingdom for an edit window.
posted by zarq at 9:09 AM on December 20, 2011 [1 favorite]


bukvich, Cyril Burt was slick enough to get himself knighted. See my later comment -- Occam's Razor suggests that anyone doing a twin study expects a certain kind of result, and one should be extrememly suspicious of their methods and motivations.
posted by localroger at 10:00 AM on December 20, 2011


If I find out that my car drives at 100mph, with 60% contributed by fuel and 40% by the engine, then what have I found out exactly?

You've found out that cars with small fuel tanks shouldn't have access to higher education, obviously.
posted by localroger at 10:01 AM on December 20, 2011 [1 favorite]


Oh, and it is because the Burt affair and its consequences are such a devastating argument against the whole idea of twin studies that every once in awhile some brave soul launches an effort to rehabilitate the asshole's reputation so that they can reclaim the field. I will confidently label such people racists until proven otherwise, at which time I will confidently label them clueless naifs.
posted by localroger at 10:05 AM on December 20, 2011


Doing a twin study if you don't believe you'll get surprising correlations would be like stabbing yourself in the abdomen to prove it causes peritonitis. Nobody in their right mind would bother.

On the contrary, a lot of people assume that a whole range of things are caused more by genetics than environment. So someone with the opposite agenda might be attracted to twin studies in an attempt to demonstrate that genetics has a smaller role than is generally assumed.

But agendas aside, no one doubts that genetics and environment both play significant roles in how people develop. The value of twin studies seems to be trying to tease out the contributions of each. Even if you're just looking at mental development, it would be interesting to know that, say, language skills are much more highly correlated with genetics than math skills, or vice versa. It seems like that kind of information could be useful for more than giving aid and comfort to racists.
posted by straight at 11:30 AM on December 20, 2011 [1 favorite]


Come to think of it, I'm not even sure what the measure in percentages between nature and nurture even means, even ignoring the points I posted above.


Heritability is often defined as 'the proportion of trait variance within the population explainable by variance of genes.'

Razib Khan describes it as:
To say that a trait is .95 heritable does not mean that it is caused 95% by genes, that’s not even wrong. Rather, it is to say that 95% of the variance within the population can be accounted for by the variance of genes within the population. But heritable traits are also usually affected by environment; if you starve someone they will be short, but retain five fingers. The number of fingers you have on your hand is not heritable, because there’s no real variance within the population of the trait. It’s genetically specified, but not heritable.
That means that a trait like height will be less heritable in situations where a large percentage of children are undernourished. In a population of clones, none of the variation in a trait is due to variation in birth genes (thus heritability would be 0%) because clones (and identical twins) should be born with the same genes.

If you have an honest interest in biology, you can read about estimating trait heritability here.
posted by Human Flesh at 12:21 PM on December 20, 2011 [2 favorites]


Thanks, Human Flesh. I've actually had a "honest interest" in biology for nearly a decade on the academic level, just not in this topic. Khan's description is of course eminently sensible.
posted by Pyrogenesis at 3:06 PM on December 20, 2011


...a lot of people assume that a whole range of things are caused more by genetics than environment. So someone with the opposite agenda might be attracted to twin studies in an attempt to demonstrate that genetics has a smaller role than is generally assumed.

Except that, as poor as twin studies are as a tool for demonstrating heritability, the small sample sizes and poor controls make them really crappy tools for trying to disprove something.

Here is an argument of my own construction, which I find completely convincing, and does not depend on twin studies or statistical methods at all:

Consider the problem of how the genome builds the brain. The genome contains a very, very small amount of information compared to the brain. So where are the blueprints? (Note that the exact mechanism by which the brain is built is totally irrelevant and unnecessary; I am arguing from information theory that you can't put 100 gigabytes of information on a 360K floppy disk and expect to get it back.)

As it happens we have one -- EXACTLY one -- model for how this might be done, as a fractal expansion. There are many specific features of brain structure which also seem to suggest a fractal expansion, such as multiple levels of N-tuples of 100 in neural organization and growth. Also, lots of other living structures are more visibly obviously pulling the same trick (trees, for example).

Now, if this is correct (and we have no other mathematical model which makes any sense) then it means there is no such thing as a small change. The smallest change genetics could possibly make would be something like trisomy 21. The major difference between us and other animals is probably such a tiny change which added some layers to our prefrontal cortex.

There's another factor; the cortex is folded, and folding is another chaotic emergent property not subject to small changes. And everybody's cortex folds the same. This is very strong evidence that the neural growth pattern that forms it is exactly the same for everybody.

Now, there could be secondary factors involving hormones, neurotransmitter sensitivity, and whatnot which affect how many or how strong connections are made, or the strength of certain emotions; these may be subject to finer changes and probably explain phenomena like sociopathy. You might have the brain area for affection but never hook it up. But while that could be genetic, it could also be because your guardians abused you.

Here we are getting into an area where environmental influences are much more powerful. We know the effect that poor diet can have on development. We also know that songbirds exposed to birdsong grow heavier brains. These are huge differences which are almost impossible to control for when picking twins.

So in our twin study, if we see a pair of twins that gravitated toward the same career, do you honestly think that could be because of hormonal similarity? Or maybe they really weren't raised all that differently? Or maybe it's random chance? The latter is much more likely than twin study authors seem to like to think, but truly random data will regularly throw up combinations that look like they have to be causal or deliberate. It's a well known perceptual defect which keeps the casino industry in business.

So I see these coincidences and I think, first, that it is mathematically impossible for this phenomenon to be due to genetic influences; second it is very likely to be a pure coincidence or the result of poor methodology and scant data.

And if I wanted to disprove genetic influences I would say first, it's unnecessary because it's mathematically impossible, and second the scant data I am able to collect will be plagued with errors and coincidences. There aren't enough separated twins in the world to get into any meaningful long run.

In fact, the result I would expect is exactly what Tell Me No Lies describes upthread; a series of results each of which might seem to promise as useful correlation, but those really being coincidence no two studies would show the same correlations. And because what I would expect is also what has been regularly reported in the past I would not waste my time doing the study.
posted by localroger at 3:47 PM on December 20, 2011


Classic article on twins by Lawrence Wright in The New Yorker, 1995 (behind paywall, unfortunately).
posted by neuron at 7:27 PM on December 20, 2011


The genome contains a very, very small amount of information compared to the brain. So where are the blueprints? (Note that the exact mechanism by which the brain is built is totally irrelevant and unnecessary; I am arguing from information theory that you can't put 100 gigabytes of information on a 360K floppy disk and expect to get it back.)

...that it is mathematically impossible for this phenomenon to be due to genetic influences...



Are you willing to support that statement, or do you expect readers to just have faith in the words that you write? If the information in the genome is insufficient, then how much information is lacking (i. e. what is the minimum number of nucleotides needed to code for a nervous system)?
posted by Human Flesh at 3:05 AM on December 21, 2011 [1 favorite]


The human genome contains 2.9 billion base pairs which represent less than a gigabyte of data at 2 bits each. You could put your entire genome on a cheap USB flash drive.

The
human brain has about 100 billion neurons. That is about 30 whole neurons for every base pair in the entire genome. Furthermore, each neuron has many interconnections; the most common estimate I've seen is that the average is about 1,000 with other neurons.

It's a bit tricky to tell how much information those interconnections represent but it is a metric hell of a lot. Worst case, in a group of 100 billion neurons, it would take about 36 bits to target one individual connection to one individual neuron. Best case, even assuming most connections go to a growth-limited local population, you'd still need around half that many bits to target one single connection.

100 billion neurons (10^11) x 1000 connections = about 10^14 connections, times 30 bits divided by 8 for bytes not making much of a dent in the magnitude. The structure of the brain encodes somewhere around 100,000 times as much information as the entire genome does.

This is not biology; it is math. You can't get around it. There is no way the genome wires the brain without a hell of a lot of compression tricks. And it's well known how that kind of compression limits the algorithms that do the decompression. Most minor changes are simply catastrophic. The fact that evolution finds minor changes that actually improve things can only be put down to having tried quadrillions of such small changes, most of which simply crippled or killed their victims, in the course of letting the few lucky winners propagate their improvements.

This is why I find the idea that genetics have much at all to do with the kind of phenomena twin studies look at to be, to be as kind as possible about it, extremely misguided.
posted by localroger at 6:20 AM on December 21, 2011


The human brain has about 100 billion neurons. That is about 30 whole neurons for every base pair in the entire genome.

Most of our cells are not neurons. Would you say that DNA lacks the requisite information needed to code for a liver? I'm asking because I'd like to know if you're skeptical of DNA's role in anabolism as a whole, or if you're just skeptical of DNA's role in behavioural genetics?

A recipe for baking a cake doesn't need to account for every molecule in a cake. The science of genetics existed before anyone understood the molecular substrates of genetic information. Genetics is not merely the study of DNA. The principles of inheritance exist regardless of how much information you assume DNA codes for. We don't need to sequence DNA in order to estimate the probability that a trait will be expressed.
posted by Human Flesh at 3:34 AM on December 22, 2011


There aren't enough separated twins in the world to get into any meaningful long run.

As miyabo stated earlier, twins don't need to be separated at birth. Information can be gleaned by comparing monozygotic to dizygotic twins.
posted by Human Flesh at 4:00 AM on December 22, 2011


A recipe for baking a cake doesn't need to account for every molecule in a cake.

But the recipe for a computer most certainly does have to account for every interconnection in the computer, and the recipe for a computer program -- the instructions that tell the computer what to do -- most certainly does have to account for every bit. In essence, it is a recipe itself and leaving out a step will botch everything.

The fact that the brain isn't exactly like a computer doesn't matter; information theory applies in the same way. The argument that the brain starts off as a relatively unorganized mass waiting for programming is actually my argument. Just as a new unprogrammed computer consists mostly of similar elements repeated, the brain consists mostly of connections waiting to be made based on living experience.

But when you say that a trait is heritable, you are saying that the brain is built more like the Apollo AGC, that it is meticulously hand-wired to do certain things by instructions that exist when you are nothing but a fertilized egg. And the information to do that isn't there. That is exactly why I think these fine character traits are not heritable.

If you want to argue that the brain can be pre-programmed for things like sexual orientation or aptitude for math, then you have to account for the connections that form that programming. And when you consider what that entails -- please go hang out with the AGI guys some time if you want a few clues -- there are a hell of a lot of bits of data involved with even a relatively simple and primitive thing like recognizing the thing you are sexually attracted to.

The principles of inheritance exist regardless of how much information you assume DNA codes for.

This is absolutely, positively wrong. Traits which are not encoded in the genome (or some ancilliary source, like mitochondrial DNA) are flatly not heritable. As I said before, this is not a matter of biology, it is math. The trait has to be encoded in DNA to be heritable, and there are only so many base pairs. If a trait is heritable, eventually we should be able to point at a certain set of base pairs and say "there's the code."

At our current level of the art we can't do that, but we certainly can see that we are trying to pour 100,000 pounds of coffee into a 2 pound bag.

Information can be gleaned by comparing monozygotic to dizygotic twins.

Actually, given the prominent influence of early experience in doing the programming some misguided people think is done by the genome, both early separation and being raised in different environments are critically important, or you can't make make meaningful statements about heritability at all.

I know a little more than most people about what random results look like, a friend's card counting team having helped pay off my house. Random data have long runs which look, to human eyes, like they are really meaningful. I've personally seen roulette scoreboards lit up all in one color on at least five occasions, and I was only hanging out in casinoland for ten or twelve hours a week. When I say there aren't enough separated twins, I know what I am talking about. You will see these weird apparent associations because that's what random data does.

But when you do enough studies those "finds" won't hold up; you get exactly what Tell Me No Lies observed, runs of coincidence that individually seem solid enough to bet on but which don't hold up under scrutiny.
posted by localroger at 6:20 AM on December 22, 2011


...accidentally hit post. On monozygotic vs. dizygotic, I was going to add that the noise overwhelms the signal in even worse ways than separated twin studies, so even if you have more samples it's still not enough. It's similar to the problem of trying to use a twin study to disprove heritability.
posted by localroger at 6:25 AM on December 22, 2011


As I said before, this is not a matter of biology, it is math.

What do you think geneticists studied before DNA's role in reproduction was understood?
posted by Human Flesh at 9:22 AM on December 22, 2011


At our current level of the art we can't do that, but we certainly can see that we are trying to pour 100,000 pounds of coffee into a 2 pound bag.

Are you claiming that our genome lacks the requisite number of nucleotides needed to code for the 900 billion cells in our body that aren't neurons?
posted by Human Flesh at 9:39 AM on December 22, 2011


On monozygotic vs. dizygotic, I was going to add that the noise overwhelms the signal in even worse ways than separated twin studies, so even if you have more samples it's still not enough.

Why? How many monozygotic twins and dizygotic twins do you need?
posted by Human Flesh at 10:12 AM on December 22, 2011


localroger, wouldn't that same sort of mathematical analysis "prove" that it's impossible for identical twins to (physically) resemble each other as closely as most of them in fact do?
posted by straight at 12:06 PM on December 22, 2011


straight: No, it's completely different. You and Human Flesh need to draw a distinction between the brain, which is an organ and which is certainly built on similar principles to the rest of the organs of the body, and the personality which encodes all those traits twin studies waste their time studying.

Look at this way: Your computer is a general purpose device. Yours probably wasn't much different from mine when it left the factory. It might even be the same model. But once mine is set to XP Classic mode and loaded up with programming and PC board design tools and BDSM porn, it acts very different from the someone else's which was upgraded to Windows 7 and loaded up with music synthesizers, jazz MP3's, and Star Trek fanfic. What twin studies are trying to do, in effect, is comb the Dell factory to figure out why my computer acts differently than yours. Was Sharon on the assembly line both days? Did they switch suppliers for the LCD inverter? That is how absurd this whole thing is.

It's like studying 1,000 computers and announcing that Sharon was on the assembly line 68% of the days when units shipped that turned up with BDSM porn installed. You could seriously do that study and find all kinds of similar bogus correlations, which would fall down when you try to verify them with further study because they're really meaningless random noise. This is exactly what TMNL described.

The genome obviously contains enough information to build a brain; it has to, because brains get built. That's Dell's job. What it doesn't have enough information to do is program a brain. Not even close to enough. Dell doesn't even make the operating system, much less the programming IDE and RealPlayer. And all that stuff twin studies study -- that's programming. It might be expressed as dendrite and synapse growth instead of charges on teeny capacitors, but Claude Shannon proved that they are the same. Shannon came a little after Mendel, BTW.

To get back to Human Flesh:

What do you think geneticists studied before DNA's role in reproduction was understood?

They worked in igorance and did their best. Sometimes they figured things out the hard way; sometimes they were wrong. It was not really understood before Shannon's work that information can be measured and has limits. In some ways (current discussion being exhibit 1) this rather fundamental thing still hasn't penetrated everywhere it should.

How many monozygotic twins and dizygotic twins do you need?

A lot. I would really have to consult my friend who did all the bet sizing calculations for the card counting team, but the basic problem is the difference between looking for correlations and anti-correlations. As noted above, you can treat heritability as a percentage; this is mathematically equivalent to your expected return on a bet. What you are seeking to do is experimentally determine your EV, or what that percentage is, by looking at a series of actual results.

There are several formulas used by advantage play gamblers to do this. Considering that my friend and his team made close to USD$2 million doing this, and I was watching their spreadsheet data along the way, I have great faith in this math.

In order to confidently establish a percentage return, you need more samples the lower the EV / weaker the correlation is. When you are looking for correlations, by definition you are looking for strength; if your EV is 20% (similar to a correlation of 60%) it will show up reliably night after night, in relatively short order, with few reversals. You might only need a thousand bets or samples to feel pretty confident about it. Bear in mind that a thousand samples is an extremely low number for this kind of math.

But if you're trying to disprove a correlation, you have to address weaker correlations. At 1% EV, typical for card counters, and an important level because people get rich off of it, you need thousands of times as many samples as you do at the higher EV to establish whether your edge really exists. When evaluating a counting system counters would run computer simulations over millions of hands before feeling confident in the result. In practice, when deciding whether a game was poisoned by a cheating dealer or other factor, it might take months of actual play to confidently detect an anomaly. Every single hand played by every team member during that month is another set of twins you need.

It's been a long time since I did this stuff and I'm not going to run the numbers for you, but I am going to say that if what I just said doesn't seem perfectly clear then you have no business doing any kind of statistical study. Unfortunately, I've read several vocal complaints by actual statisticians that a lot of scientists don't understand the math I just described and publish bogus results all the time. I would confidently say that every twin study ever done falls in that category simply because I have another much more robust reason for suspecting their results.
posted by localroger at 1:49 PM on December 22, 2011


But, to use your analogy, localroger, if I try installing Windows XP both on machines with an x86 chip and a PowerPC chip, the difference in results is going to be very much determined by the hardware, even though the resulting functionality is vastly underdetermined by the hardware.

It seems very possible that a person's genetic inheritance could play a significant role in how environmental factors end up programing the brain (including complex, higher-order "personality" traits) even if the genetic inheritance by itself is not sufficient to create that programming.
posted by straight at 2:16 PM on December 22, 2011


If you try installing Windows XP to a PowerPC chip, you will get the result of a typical genetic mutation in the biological situation -- catastrophic failure. One flaw in my analogy is that the "installation process" for living things is, well, life. There is no chance for such a compatibility issue.

It seems very possible that a person's genetic inheritance could play a significant role in how environmental factors end up programing the brain

Yes, this is usually where this argument goes. The problem for me is that when you start talking about hormone levels, neurotransmitter activity, and so on the environment is a freaking huge influence. Trying to tease the genetic component out from that would be looking for a 0.01% EV in your game.

Here's a good example: Sociopathy. It's well described for over 200 years and even if the DSM-IV messed it up, there's lots of literature establishing that it's a real thing. Sociopathy seems to be caused by what might be some kind of neurotransmitter defect, perhaps a failure to properly respond to oxytocin. If you want to find a heritable personality trait, there's something I can actually believe in. You've got a pretty strong signal because it's around 1% to 3% of the population. You shouldn't even need twin studies; longitudinal studies should be able to pin it down. Getting millions of samples should be doable.

The kick in the pants is that even sociopathy doesn't seem to be heritable. At least I've never seen anyone even try to present evidence that it is -- and I'd believe that parsecs before I'd believe that something like math ability is heritable.
posted by localroger at 2:37 PM on December 22, 2011


You could seriously do that study and find all kinds of similar bogus correlations, which would fall down when you try to verify them with further study because they're really meaningless random noise.

If researchers found a strong correlation between a particular employee working at the factory and computers expressing a given trait, then that is a good reason to examine that employees work. Yes, it could be an irrelevant coincidence. Scientists understand that. That's why they publish p-values.
posted by Human Flesh at 2:12 AM on December 23, 2011


The genome obviously contains enough information to build a brain; it has to, because brains get built.

Do you retract what you wrote earlier regarding 'Consider the problem of how the genome builds the brain. The genome contains a very, very small amount of information compared to the brain. So where are the blueprints? (Note that the exact mechanism by which the brain is built is totally irrelevant and unnecessary; I am arguing from information theory that you can't put 100 gigabytes of information on a 360K floppy disk and expect to get it back.)'?

Are you skeptical of genetics as a whole, or are you just skeptical of behavioural genetics?
posted by Human Flesh at 2:30 AM on December 23, 2011


The kick in the pants is that even sociopathy doesn't seem to be heritable. At least I've never seen anyone even try to present evidence that it is

Viding et al. (J. Child Psychology and Psychiatry 46:592-597) found a heritability of 0.68 for psychopathic traits.

There is a lot of research on the ontology of anti-social behaviour.
posted by Human Flesh at 5:31 AM on December 23, 2011


If researchers found a strong correlation between a particular employee working at the factory and computers expressing a given trait, then that is a good reason to examine that employees work.

I have read a lot of really ignorant stuff since this whole internet thing came about, but this statement has achieved a new altitude record.

Do you retract what you wrote earlier regarding...

Did you read what I wrote earlier about the difference between hardware and programming?

a heritability of 0.68 for psychopathic traits.

Interesting. Do you have a link to the paper or a description of the methodology? If the sample size is less than five figures a correlation of 0.7 or less is basically meaningless. Which, getting back to the original topic, is the whole problem with twin studies.

Here's a good description of why.

All in all, statistical methods are a very difficult method of learning anything. You need far more samples than most people, even educated people who have taken statistics classes, really understand. That's why it's critically important not to rely on statistical methods when fundamentals are available. Gregor Mendel didn't have the work of Turing, Watson and Crick, Shannon, or Mandelbrot to help him limit his scope. Today we do.
posted by localroger at 6:09 AM on December 23, 2011


Do you have a link to the paper or a description of the methodology?

No. Do you think any research on behavioral genetics would change your mind?
posted by Human Flesh at 6:19 AM on December 23, 2011


Are you skeptical of genetics as a whole, or are you just skeptical of behavioural genetics?

To answer this question for about the fifth time in this discussion, of course I believe in genetics. The genome obviously codes for the 100 or so areas of cortex, the interareal connections, and the thickness of the cortex all of which are important for making us human. But I don't believe the genome can possibly code for behavior. Genetically, the difference between a cat brain and a human brain is probably just a few base pairs. Grow 100 areas instead of 60, grow the cortex thicker, things like that. Fine control is not within the scope of genetics. The neurons grow according to genetic instructions, but they form the connections that make us capable of navigating our environment by navigating our environment.

Most biological structures, including the brain, consist of repeated units in a fractal expansion. You don't need a lot of information to guide such construction, just as you only need a couple of lines of code to guide the construction of the Mandelbrot Set. But personality is software; it is completely different, and it involves a hell of a lot of information that can't be compressed that way.

Do you think any research on behavioral genetics would change your mind?

I would honestly be willing to be persuaded about sociopathy; it's always been the edge case for me because it's obviously a transmitter defect, and that sort of thing can be coded genetically. It wouldn't change the rest of my argument very much, though, because I don't think you can point to a neurotransmitter or hormone that codes for sexual orientation or math ability.
posted by localroger at 6:23 AM on December 23, 2011


Did you read what I wrote earlier about the difference between hardware and programming?

Yes. You wrote 'Consider the problem of how the genome builds the brain.'
posted by Human Flesh at 6:26 AM on December 23, 2011


But I don't believe the genome can possibly code for behavior.

Do you think that the herding behaviour of border collies and the web patterns that spiders make are attributable to purely environmental factors, or are you only skeptical behavioural genetics when it applies to humans?
posted by Human Flesh at 6:40 AM on December 23, 2011


If the sample size is less than five figures a correlation of 0.7 or less is basically meaningless.

Does this apply to all genetic studies, or just research on the genetics of behaviour. Also, wouldn't you expect the heritability figures to be low for traits that are not especially heritable?
posted by Human Flesh at 7:06 AM on December 23, 2011


You wrote 'Consider the problem of how the genome builds the brain.'

After that.

herding behavior: Probably hormonal.

web patterns: Pretty much the dictionary picture of a fractal emergent property.

Genetic and very old and not subject to much change, changes being catastrophic: The R-complex and limbic system, which make us mammals. These determine basic mechanisms of learning and certain very old and simple reflexes like the Moro and walking reflex. These are very similar not just among all humans, but among all similar animals.

Genetic and heritable: Number of areas, size of areas, thickness of cortex, broad interareal wiring, which make up the difference between our brains and other animals. All of which are the same for all humans, changes being mostly catastrophic, and making us a new and different animal if not.

Individually heritable: hormone and neurotransmitter sensitivity affecting broad behavior patterns (and those mostly involving emotion). I am quite open to including sociopathy here, I just hadn't seen any studies personally (and I did look, but it was 25 years ago in a library with paper catalogs and indeces). Any such changes should ultimately be verifiable through non-statistical methods, and treated with great skepticism until they are.

Not heritable: Specific complex behaviors, recognized patterns, and patterns of behavior. People have no idea how complex such a primitive thing as recognizing the thing you are sexually attracted to is. Considering that we know for a fact that simpler organisms like geese use non-genetic imprinting to program these complex functions, it's kind of stupid to think humans code them genetically. And nearly everything twin studies study seems to fall in this last category.

Does this apply to all genetic studies, or just research on the genetics of behaviour.

The article I linked was about a clinical drug trial. Does that answer your question?
posted by localroger at 7:12 AM on December 23, 2011


P.S. I did overspeak when I said the sociopathy study needed 10,000 samples to be meaningful at p=0.68. High three figures would make the result worth following up.
posted by localroger at 7:24 AM on December 23, 2011


P.P.S. after some digging, herding instinct does indeed seem to be one of those deep limbic or R-complex functions like the walking reflex, which probably evolved very early to encourage the protection of large groups of young. That's a good example of the outer range of what genetically coded behavior could look like. And it's an interesting point that while it requires that the animal recognize objects to herd, it doesn't seem to care what those objects are, indicating that the pattern recognition part of the instinct is extremely primitive.
posted by localroger at 7:35 AM on December 23, 2011


I'm not sure how I'd measure the complexity of a behaviour. Are ant and bee colonies complex or not complex?
posted by Human Flesh at 7:43 AM on December 23, 2011


This requires a bit of parsing on the word "complex." There are two kinds of complexity, emergent and what for want of another label what I'll call "arbitrary."

Fractals are emergently complex. They contain a great deal of detail generated from a relatively small set of instructions. This is math that was only discovered and understood for the first time in the 1980's, and it's very important because it answers the question "Why does the universe look the way it does if God didn't build it all deliberately" better than anything that ever came before.

Now you can get a complex emergent property from a simple system, but what you can't do is make minor changes to it. You want to build the Mandelbrot set but tweak one of the curlicues? Can't do it. If you make any change at all to the simple generative algorithm you get an entirely different pattern.

Arbitrary complexity, by contrast, can take any form. If you paint a painting, you can put the paint anywhere you want. There are no limits on how it can look.

More importantly, if you want to transmit the painting, you have to have enough memory to store every pixel of the image; there are no shortcuts. Because it is arbitrary you can't infer from one corner what the rest of it looks like.

But if you want to transmit the fractal, all you have to send is the generative algorithm and the instructions for how to use it. The fractal can emerge in all its glorious detail once it's expressed in a compatible system.

Ant and bee behavior is emergent. So is bird flocking. We even know some of the simple micro-behaviors that add up to the overall complexity in sufficient groups. Those micro-behaviors are sufficiently simple to be hard coded.

The gross structure of the brain -- how many neurons it has, how the major areas are organized and the broad interareal cabling -- is also emergent, and obviously genetic. It's also not subject much to genetic change except in a catastrophic sense.

The human brain, though, also has those 10^14 or so specific interconnections. We have lots of reasons to believe those interconnections are what define our experience, behavior, and personality. I know those aren't made genetically because they can't be; the math says so. This is a much more powerful observation than any statistical study could ever make.

If those interconnections were made emergently, we would all be the same, just as our cortexes all fold the same. And just as we see this broad sameness in, for example, bees where the honey dance is hard coded and every bee does it the same. Or in race horses, where there is a range of gaits but the gait can't be learned or changed, and a horse which doesn't instinctively run with the most efficient gait will never win the Triple Crown.

But humans aren't like this. We're all different, and our behaviors can change in radical ways as we live. And that means the cabling has to be arbitrary. And that means there is no room for it in the genome. The genome has enough base pairs to transmit the fractal algorithm for building a body, but it doesn't have even a fraction of what would be needed to set all the little local connections.

let's go back to one of the favorite hobby horses of some twin study fans, sexual orientation. Sexual orientation is not genetic. I will say that flatly and with great confidence because the patterns we imprint on are arbitrary. In some cases we even know where they come from; Robert Anton Wilson credited his lifelong affinity for redheads with an early babysitter. And where is the gene for a leather fetish? For shoes? For latex rubber? When I was younger I had a pet goose. Was Flicky's lifelong fascination with blue tennis shoes genetic, or maybe more likely due to the fact that I was always wearing blue tennis shoes when she imprinted?

Humans even have but don't use some of the instincts we know other animals have; like the racehorse whose gait can never be relearned, humans have a walking reflex. But we don't use it. It has to be suppressed so that we can learn to walk the human way -- from our environment. And that is why humans can not only walk, but also dance.
posted by localroger at 8:33 AM on December 23, 2011


Sexual orientation is not genetic.

Jeffrey Hall and Yoshiki Hotta discovered a gene that codes for homosexual behaviour in Drosophila melanogaster.
posted by Human Flesh at 8:17 PM on December 23, 2011


We are not flies. I would expect fly behavior, like bee and ant behavior, to be mostly hard coded. We don't use our walking reflex either.
posted by localroger at 6:17 AM on December 24, 2011


And a deserved elaboration: Insects recognize sexual contacts mostly by pheremones because they don't have our elaborate image recognition capability. That's certainly amenable to point genetic mutation.

We don't do it that way. We recognize sexual contacts mostly by perceptual cues. We don't even seem to have the potentially hard coded cues that some birds, mammals, and apes use; we can imprint on things as diverse as cars and furry animals of other species.

Those are images, and recognizing them requires both an arbitrary coding and a very complex recognition capability. The recognition algorithm is itself an emergent property and it's a prominent component of the larger algorithm we call "consicousness." There is a clear trend of evolution moving from using genetically coded behavior to using arbitrary pattern recognition from fish to amphibians to reptiles, birds, and mammals, with humans being human mainly because we hardly use hard-coded behaviors for anything.

Now while the ability to recognize complex patterns is a genetic legacy, the patterns themselves are not. They can't be. Any defect in the recognition algorithm itself is catastrophic because it's used throughout the brain for many other things. And as I keep saying, there is no room in the genome for all those arbitrary patterns.

Simply recognizing that an image represents a human being is a non-trivial trick -- Microsoft just spent a metric ton of money teaching the Kinect to do that, and the Kinect doesn't even know whether you're male or female.

And homosexuality is just the tip of the iceberg for human sexual orientation. We are vastly more complicated than the straight/gay divide. Where is the gene that makes Furries? Where is the gene that programmed this one dude I recall from a scholarly journal who was so in love with his car that he would masturbate into the tail pipe? Where is the gene of particular interest to me for BDSM? Where is the gene for shiny black rubber? (Actually, that one might be semi hard coded. At least it's not too complex and variable to represent with a small amount of data.)

No, we acquire our sexual cues through a series of imprinting events post-birth. And if there is any legacy of Drosophilia lurking within our genes, it's about probably buried even deeper than our genetic walking gait.
posted by localroger at 6:44 AM on December 24, 2011


More: There's an article making the rounds this morning about the surprising intelligence of birds.

This isn't really new news; my wife keeps birds and it's been known for awhile that birds like parrots, crows, and jays are much more intelligent than mammals with the same sized brains. What's not commonly understood is the rather obvious reason: Birds have a lot of genetically determined behaviors. The cortex available for them to do arbitrary processing can be devoted to higher order functions.

See, when you have an actual genetically coded behavior, there isn't any goddamn doubt about it. Birds respond to primitive reproductive cues, observe breeding seasons, migrate -- all genetically coded. P=1.0. It's a problem in preserving some endangered species; you put them in cages, and migration season comes, and BAM they beat themselves up trying to fly through the wire. Their chicks, who never even ever saw the other end of the route, will do exactly the same thing. They can't help it.

We don't do that.

But this is expensive. It makes childbirth perilous, childhood long, and development full of risks. It means we use 20% of the calories we consume keeping our brain alive. These are all harsh negatives for our survival as a species compared to other animals.

But what did it get us? Humans made our way to 6 of 7 continents, lived in frozen arctic wastelands and burning deserts and tropical islands and on high mountain plateaus and we thrived in all those environments long before we acquired modern technology. It put us at the top of the food chain even though in every environment where we lived there were animals stronger, faster, and better adapted than ourselves.

Not being genetically programmed is the genius of our species. It is the answer to the question "what makes humans special."

That anyone would prefer to think otherwise simply mystifies me.
posted by localroger at 7:48 AM on December 24, 2011 [1 favorite]


actually, given the prominent influence of early experience in doing the programming some misguided people think is done by the genome, both early separation and being raised in different environments are critically important, or you can't make make meaningful statements about heritability at all.

Why is that? That doesn't seem obvious at all.

On monozygotic vs. dizygotic, I was going to add that the noise overwhelms the signal in even worse ways than separated twin studies, so even if you have more samples it's still not enough. It's similar to the problem of trying to use a twin study to disprove heritability.

Why?

The article I linked was about a clinical drug trial. Does that answer your question?

Yes. I ask because it sounds like your criticisms apply to all research, not just research that compares monozygotic twins to dizygotic twins.
posted by Human Flesh at 11:44 AM on December 24, 2011


See, when you have an actual genetically coded behavior, there isn't any goddamn doubt about it. Birds respond to primitive reproductive cues, observe breeding seasons, migrate -- all genetically coded. P=1.0.


I think your distinction between 'hard-coded' and 'complex' traits is specious. Coronary heart disease has a number of environmental and genetic influences. Is coronary heart disease 'hard-coded' or 'complex'?
posted by Human Flesh at 12:58 PM on December 24, 2011


Heart disease is hard coded and simple.
posted by localroger at 6:25 PM on December 24, 2011


I would answer at more length but, merry xmas, my computer just got hosed by a worm and I'm using my wife's to post this.
posted by localroger at 6:26 PM on December 24, 2011


Twin studies show that monozygotic twins more closely resemble each other in their rate of coronary artery disease than dizygotic pairs resemble each other. One study on coronary artery disease occurring before age 60 showed a concordance rate of 0.83 for monozygotic pairs and 0.22 for dizygotic pairs.

Berg, K. (1983). In Progress in Medical Genetics, vol. V, pp. 36-90

If a disease that's an artefact of a number of environmental and genetic variables can be considered 'hard coded,' then I have no idea what you mean by hard coded. It's a good thing that geneticists don't use terms like 'hard coded.'
posted by Human Flesh at 3:19 AM on December 25, 2011


Your interpretation of gene expression might be wonky. If cytosine has two bits, how many bits does uracil have?
posted by Human Flesh at 3:26 AM on December 25, 2011


The smallest change genetics could possibly make would be something like trisomy 21.

Trisomy 21 is not a small change—it's a whole extra chromosome!
posted by Human Flesh at 3:36 AM on December 25, 2011


It is the answer to the question "what makes humans special."

That anyone would prefer to think otherwise simply mystifies me.



I question the tendency to call humans special because it seems a little too flattering. It doesn't jibe with my observation that we have a lot in common with our cousins in the animal kingdom. Biologists largely abandoned their essentialist intuitions because taking a population view of organisms helps us better understand genetics and evolution.
posted by Human Flesh at 3:55 AM on December 25, 2011


Well I'm mostly baaaack. Merry effing Christmas from hoseville.

If a disease that's an artefact of a number of environmental and genetic variables can be considered 'hard coded,' then I have no idea what you mean by hard coded.

The factors that result in heart disease are not emergent, they are direct. I am calling direct things "hard coded" by analogy to how software comes to exist, because that's kind of a natural vocabulary for me. By "direct" I mean that a a small change in the coding can make a small change in the result, such as an affinity for cholesterol or weakness in the artery walls. A similar small change in the coding that builds the brain results in a catastrophic fuckup.

Trisomy 21 is not a small change—it's a whole extra chromosome!

But the presence of that extra chromosome does produce what amounts to a small change in the generative algorithm that builds the brain. Point mutations would have a similar effect but they're much more rare.

OK, off to find more software to reinstall. Haven't got openoffice yet.
posted by localroger at 1:18 PM on December 25, 2011


But the presence of that extra chromosome does produce what amounts to a small change in the generative algorithm that builds the brain.

Please stop making things up. You don't know how nervous tissue is synthesised.


Point mutations would have a similar effect but they're much more rare.

That's not true. Mutations occur in each new generation.
posted by Human Flesh at 8:00 AM on December 26, 2011


Please stop making things up. You don't know how nervous tissue is synthesised.

OK, with this comment I am going to stop because we are talking past one another.

I have explained, several different ways, that the expansion of the brain is constrained by certain types of math regarding both chaos and information theory, both of which are relatively new but very important fields compared to genetics and anatomy. Both the gross structure of the brain and the fine programming growth represent information. That information has to come from somewhere. You consistently ignore this.

The gross structure of the brain is clearly a fractal expansion in n-tuples of around 100. Biologists have even directly observed the first and last stages of this growth pattern. The stem cell which will become the brain divides about 100 times and the daughters migrate outward, where it's generally believed (quite reasonably) that they become the basis for each of the brain's separate areas. The daughters then multiply 100 times each, their daughters each multiply about 100 times (there are striated patterns in cortical neural activity on this scale), and the final generation mutliplies about 100 times to establish the thickness of the cerebral cortex. 100 x 100 x 100 x 100 = 100,000,000 which is the most common estimate of how many neurons there are.

The first and last stages of this have been observed. The middle factor of 10,000 hasn't been directly, but I've seen lots of reasons to imply it based on observed patterns of neural activity at that scale. And it explains how such similar genomes can build such vastly different brains; you just change one of those 100's. (It's not necessary for every generation to be the same scale; there 's obviously also a final generation or it would grow forever.)

Do you really find any of this hard to believe? Does it sound like I know absolutely nothing about how the brain is assembled? Because it sounds to me like you know nothing about the information theory that must guide any such process, whether it's biological or mechanical. This is not just one of many theories that explains how the brain gets to look the way it does, it's the only theory we have that possibly could.

Now, the reason I said T21 presents as a point change in the algorithm is that mathematically, what it does is it changes one of those 100's. It's very likely that that "whole extra chromosome" is mostly just duplicating functions (after all, it's a duplicate not a whole different chromosome from an alien dimension) but the overfunctioning messes up some of those growth iteration parameters. Manwhile a point change to the algorithm that actually guides the overall construction is catastrophic. In such a brain there are no 100's. While such anencephaly isn't unknown, I think describing it as rare is fair.

Anyway, have a nice day. One day these things will be known without the use of statistical methods that are so easily twisted to unpleasant purposes.
posted by localroger at 10:38 AM on December 26, 2011


You still didn't support what you wrote earlier about:


actually, given the prominent influence of early experience in doing the programming some misguided people think is done by the genome, both early separation and being raised in different environments are critically important, or you can't make make meaningful statements about heritability at all.


On monozygotic vs. dizygotic, I was going to add that the noise overwhelms the signal in even worse ways than separated twin studies, so even if you have more samples it's still not enough. It's similar to the problem of trying to use a twin study to disprove heritability.

posted by Human Flesh at 1:50 PM on December 26, 2011


Here are some heritability figures.
posted by Human Flesh at 9:53 PM on December 30, 2011


« Older Deep Time   |   Cressus stole the etrog, and other stories. Newer »


This thread has been archived and is closed to new comments