Text

Physicist, Startup Founder, Blogger, Dad

Saturday, August 31, 2013

Another species, an evolution beyond man




Readers might be interested in this interview I did, which is on the MIRI (Machine Intelligence Research Institute, in Berkeley) website. Some excerpts below.
... I think there is good evidence that existing genetic variants in the human population (i.e., alleles affecting intelligence that are found today in the collective world population, but not necessarily in a single person) can be combined to produce a phenotype which is far beyond anything yet seen in human history. This would not surprise an animal or plant breeder — experiments on corn, cows, chickens, drosophila, etc. have shifted population means by many standard deviations (e.g., +30 SD in the case of corn).

... I think we already have some hints in this direction. Take the case of John von Neumann, widely regarded as one of the greatest intellects in the 20th century, and a famous polymath. He made fundamental contributions in mathematics, physics, nuclear weapons research, computer architecture, game theory and automata theory.

In addition to his abstract reasoning ability, von Neumann had formidable powers of mental calculation and a photographic memory. In my opinion, genotypes exist that correspond to phenotypes as far beyond von Neumann as he was beyond a normal human.

I have known a great many intelligent people in my life. I knew Planck, von Laue and Heisenberg. Paul Dirac was my brother in law; Leo Szilard and Edward Teller have been among my closest friends; and Albert Einstein was a good friend, too. But none of them had a mind as quick and acute as Jansci [John] von Neumann. I have often remarked this in the presence of those men and no one ever disputed me. – Nobel Laureate Eugene Wigner

You know, Herb, how much faster I am in thinking than you are. That is how much faster von Neumann is compared to me. – Nobel Laureate Enrico Fermi to his former PhD student Herb Anderson.

One of his remarkable abilities was his power of absolute recall. As far as I could tell, von Neumann was able on once reading a book or article to quote it back verbatim; moreover, he could do it years later without hesitation. He could also translate it at no diminution in speed from its original language into English. On one occasion I tested his ability by asking him to tell me how The Tale of Two Cities started. Whereupon, without any pause, he immediately began to recite the first chapter and continued until asked to stop after about ten or fifteen minutes. – Herman Goldstine, mathematician and computer pioneer.

I always thought Von Neumann’s brain indicated that he was from another species, an evolution beyond man. – Nobel Laureate Hans A. Bethe.

The quantitative argument for why there are many SD's to be had from tuning genotypes is so simple that I'll summarize it here (see also, e.g., here or here).  Suppose variation in cognitive ability is

1. highly polygenic (i.e., controlled by N loci, where N is large; N is almost certainly more than 1k -- perhaps roughly 10k), and

2. approximately linear (note the additive heritability of g is larger than the non-additive part).

Then the population SD for the trait corresponds to an excess of roughly Sqrt(N) positive alleles. A genius like vN might be +6 SD, so would have roughly 6 Sqrt(N) more positive alleles than the average person (e.g., 200 extra positive alleles if N = 1000). But there are roughly +Sqrt(N) SDs in phenotype to be had by an individual who has essentially all of the N positive alleles. As long as Sqrt(N) >> 6, there is ample extant variation for selection to act on to produce a type superior to any that has existed before. (The probability of producing a "maximal type" through random breeding is ~ exp( - N), and for large N the historical human population is insufficient to have made this likely.)

This basic calculation underlies the work of animal and plant breeders, who have in many cases (corn, drosophila, cows, dogs) moved the "wild type" population by many SD through selection. See, e.g., this essay by famed geneticist James Crow of Wisconsin.

42 comments:

Kudzu_Bob said...

How many among us are qualified to teach (and above all properly socialize) Homo novus?

David said...

Not quite sure about this. I mean what would such a huge deviation in terms of IQ look like? I would say right now the smartest humans in the world are Terrence Tao, Grigori Perelman, Ed Witten, and Charles Fefferman. My guess is all of them have IQ in the 200-230 range. Are you saying that we could see people with IQ of say 500? Why haven't we seen such people yet? (at least that we know of). Super smart people have children all the time, and although intelligence is certainly genetic, I haven't seen evidence which suggests that IQ is a trait that increases from one generation to another in the same way certain physical traits are.

steve hsu said...

I suggest you look up terms like parental midpoint, regression to the mean, etc. IQ doesn't generally increase from generation to generation except due to selection pressure.

David Coughlin said...

We'll walk them to the edge of knowledge at a very young age, then push them over and hope they can fly.

JayMan said...

The X-Men's Homo sapiens superior? ;)

5371 said...

Yes, some remarkably intelligent strains of corn and breeds of cow have been bred.

steve hsu said...

As you know I was referring to different phenotypes for those species, although the math is basically the same.

In the case of dogs, it's clear that they are shifted in cognitive ability relative to their wild cousins (wolves). The average dog is a many SD outlier among wolves in ability to interact with humans.

http://blogs.scientificamerican.com/thoughtful-animal/2012/04/30/dogs-but-not-wolves-use-humans-as-tools/

LondonYoung said...

IIRC, Darwin always emphasized that within an animal species competition is among members of the same sex. Geoffrey Miller hypothesizes that all these intelligence genes are meant to help men stay sufficiently smart even when subjected to inadequate nutrition and/or disease - so that they stay attractive to the women. But, given plenty of food and medicine, the selection pressure is off since the random human is more than smart enough for what nature puts in his way ... Selection pressure may even be the other way now if these genes bring negative effects with them ...

Kudzu_Bob said...

I'm sure that they will love us for putting them in the same category as we do transgenic fish tomatoes, that is, as a resource to be exploited. And if not, well, what could these super-intellects possibly do about it?

Richard Seiter said...

Speaking of the other phenotypes, is there any reason to believe IQ scales with respect to the underlying physiology with the same exponent that those do? If not, that might make a difference in what the math implies. As an example, I wonder how IQ scales with nerve conduction velocity. For small changes this might not make much of a difference, but by 30 SDs it could.

Diogenes said...

What a joke. Only a nerd. Three Jews, one Chinaman. Pfff! And all mathletes.



All of these are idiots compared to the Pennsylvania Dutch and Afrikaner Elon Musk.

Diogenes said...

Yeah that explains the Flynn effect. Ha! Ha! Ha!


Dem cavemen was so clever dey'd make Einstein look dumb.


Typical Randroid.

Diogenes said...

Homo novus won't require socialization.



Typical. Nerds don't be understandin' regular people, Dey's got no "social skills". I gots dem. I's better than any nerds.

dxie48 said...

http://en.wikipedia.org/wiki/Mental_chronometry

Researchers have reported medium-sized correlations between reaction time and measures of intelligence:

dxie48 said...

http://en.wikipedia.org/wiki/Flynn_effect

For the Raven's Progressive Matrices test, subjects born over a 100-year period were compared in Des Moines, Iowa, and separately in Dumfries, Scotland.
Improvements were remarkably consistent across the whole period, in both countries.[1]
This effect of an apparent increase in IQ has also been observed in
various other parts of the world, though the rates of increase vary..

dxie48 said...

http://www.nature.com/nature/journal/v407/n6803/full/407470a0.html

Intelligence: Maze-solving by an amoeboid organism

...Here we show that this simple organism has the ability to find the minimum-length solution between two points in a labyrinth.

Richard Seiter said...

I knew about the reported correlation, but thanks for the link. It contains this which I hadn't known:

"Standard deviations of reaction times have been found to be more strongly correlated with measures of general intelligence (g) than mean reaction times. The reaction times of low-g individuals are more spread-out than those of high-g individuals.[26]"



The reference (26) is interesting. I think the right skew of reaction time (where IQ is close to normal) may be relevant to my point. It also mentions that inspection time is better correlated with g than reaction time.

5371 said...

Worth noticing that for all his gifts, Neumann mostly picked low-hanging fruit in his research career. Many men made more original contributions to mathematics and physics in his time.

Diogenes said...

Miracles are found for those the Church would make saints. Even IQ obsessed atheists need saints.

In the sesquicentennial of our Ford, as Brave New World would have it, living in a luxury he made possible, strangely, the cleverness of useless men has never been thought so great, and the esteem for men, epigones of our Ford, has neverbeen less.

eqn said...

Extremely smart people with questionable conscience/ethics exist, and are capable of (arguably, have already done) great harm. Should we just dial the intelligence gauge on some embrios to the max level and hope for the best? Is intelligence really the best "quantity" to increase, if our aim is to solve the problems of humanity?

tractal said...

This doesn't make sense to me. Do you have a link?

David Coughlin said...

We'll breed these people, our putative successors, so they will arrive faster than they would by natural selection. They were bred to work on problems, so we will give them problems to work on. Their genesis will, otherwise, put them at odds with the world. What would you do for someone who equals your adult capability by the time they are eight? How about if it takes them until they are forty to be cognitively mature, but they have four times your IQ when they get there? Who is best capable of discovering and defining the new style relationships these people may need? Strikes me as heady to think that we can plan for their socialization.

Diogenes said...

Ah, but why is intelligence or language like the Peaock's tail?

Much more likely explanation for man's supposed excess intelligence is what Gould called spandrels. That is, unintended consequences. That is, one can do useful things with intelligence, but he can also a lot of useless things.

And when were women attracted to men for anything other than their use-value? Isn't there a name for a charming and beautiful man? I hear
they smoke them in London.

Miller is a psych with degrees in psych. That's all one need know about him.

Kudzu_Bob said...

You haven't thought this through, although you are hardly alone in this. Homo novus will grow up in a world of people that to all intents and purposes are retarded. This is bound to be psychologically stressful for for them. How do we ensure that they won't regard us with disdain or even hatred?

Kudzu_Bob said...

Not planning for their socialization means that we might well end up with a large group of super-geniuses who don't like us. Either we should come up with ways to promote their happiness and mental health so that they get along with us retards, or else we had better implant miniature, remote-controlled bombs in their skulls at birth in case the experiment goes badly awry.

David Coughlin said...

I struggle to understand your ethics.

Kudzu_Bob said...

Our potential evolutionary successors might struggle to understand our ethics as well. This is why we should think about these things now, while there is time, rather than assuming that just because Homo novus is smart that everything will somehow turn out okay.

Diogenes said...

I agree. We might be used in their medical experiments just as we use chimps sometimes when rats or dogs won't do.


And we already have an example in a commenter here. (Sorry for all the metaphors. It's what came to mind.)



LondonYoung is a world-controller and inner party member, that is a partner at the vampire squid, and he believes that 1. the social equivalent of the damned are lazy and stupid and 2. that they deserve their damnation even though he also believes in predestination, that is, he is a social Darwinist; he believes that psychological traits are innate.

Diogenes said...

We will be hateful to them just as we are for Steve. Beyond Good and Evil there is taste and manners and class, as Nietzsche pointed out. If the Elizabeth visited an Eastern Kentucky holler...it would be like that.

Cornelius said...

Why does everyone think that we're all of a sudden going to have a bunch of children running around with 4000-level IQs.



How is genetic engineering likely to start? As embryo selection, and the overall process won't be cheap for some time.



So let's assume we find a way to grow eggs outside of the a woman's body. A couple decides to have a child so they go to their fertility doctor who produces 1000 embryos for them to select from. Even if we assume no regression to the mean, and if we assume that we know ALL the genes that lead to desirable and undesirable phenotypes, we've still got a long way to go before our children are fundamentally a different species. Why? Because parents are not going to select the embryo with the highest IQ, they're going to select the embryo with the best SET of characteristics, and that set will include, IQ, EQ, height, attractiveness, health, longevity, hand-eye coordination, athleticism etc. Of course different parents will have different sets of characteristics but we're still left in a world where our children are on average maybe 1sd smarter than us. Not exactly a recipe for disaster. In fact, many of us high IQ types grew up with parents who were 1sd lower than us in IQ and I don't think we have fundamentally different relationships with our parents than low IQ people do.

dxie48 said...

From the psychologists,

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0057989

Does “Science” Make You Moral? The Effects of Priming Science on Moral Judgments and Behavior



Conclusions/Significance

These studies demonstrated the morally normative effects of lay notions of
science. Thinking about science leads individuals to endorse more
stringent moral norms and exhibit more morally normative behavior. These
studies are the first of their kind to systematically and empirically
test the relationship between science and morality. The present findings
speak to this question and elucidate the value-laden outcomes of the
notion of science.

hardly said...

nietzsche thought the indian caste system was essentially a selective breeding program that funneled genes for intellect, aggression and social skill into different castes.

disqus_ZsHlaWpOsF said...

kudzu
Disdain, not; benevolent pity, yes. Till you hurt them.

Brian said...

You don't. I only have an IQ of about 135 and that's how I see the person of 90 IQ-- mentally retarded.

jinna said...

hi i just wanted to say thank you for your comment on my article on RoK, and to ask if you blog or so. It would be interesting to read what like-minded people think about feminism, traditional stuff etc.

MUltan said...

Standard deviations become meaningless well before you get out to even today's smartest people. The reliability of testing at that level just doesn't exist, and the assumption of a normal distribution isn't valid - maybe it's log-normal or Pearson type IV, but any rate the right tail looks increasingly fatter the further you go out. The real problem is that IQ is a rarity measure relative to an age group, not an ability measure.

Absolute (ratio scale) ability measures do exist, but are not widely known. They are called Rasch measures, and are used for example on the Stanford-Binet V as its Change Sensitive Scale (CSS). This scale lets us see how much human intelligence actually varies. Quoting from an old post of mine:In the SB5 norming samples, which included participants from ages 2 to 85 and an additional gifted norming group, the highest Rasch measure seen was 592, on a scale with the average for a 10 year old chosen to be 500. The score of 592 likely corresponds to an adult standard-deviation IQ of 150-160 but the documentation is unclear on this point. To give a sense of how absolute ability correlates with age, a 2.25 year-old would score 435, 4 year-old would score 460, a five year-old 470 and a 16 year-old about 509 or 510. You could say the high scorer is only about 15% smarter than the average person, but on the other hand, you could say that the difference in scores is greater between the high scorer and the average person than between the average person and a 2 year-old.

So John von Neumann with a CSS of 600-700 was likely less than 50% smarter on an absolute scale than an average toddler, and certainly less than twice as smart. An engineered posthuman super-intellecton the other hand might be more than twice as smart as that, CSS ~1250 if we figure an adult standard deviation as being about 25 CSS points.

ESRogs said...

I thought I understood this when I read it before, but coming back to it just now I realize I don't know why we expect the population SD to be ~sqrt(N) positive alleles. Is that a straightforward derivation from statistics, or is it based on an evolutionary / population genetics argument?

ESRogs said...

Some of my confusion is that, from reading other posts, such as [this](http://infoproc.blogspot.com/2012/10/deleterious-variants-affecting-traits.html), I thought your model was that there were approximately 10,000 total genes contributing to g, and of these, the average person has about 9000 good alleles and 1000 bad alleles. So is the N that we should be taking the square root of the 10k total loci or the 1k average # of bad alleles?

steve hsu said...

Toy model: 10k loci, prob 0.9 of + allele at each site. Then the *average* person has 9000 + alleles and 1000 bad ones. But the SD of this distribution is ~ sqrt( p (1-p) N ) ~ sqrt ( .1 N ) ~ 30 or so. Most people have 1000 plus or minus 30 bad alleles, but exceptional people might have 900 or 800 or 1200, with correspondingly high or low g scores.

Look up binomial distribution. http://en.wikipedia.org/wiki/Binomial_distribution

ESRogs said...

Oh, right. Now I remember calculating that out before. I was just confused because in the post you say Sqrt(N), rather than Sqrt(.1 N). Thanks for the clarification!

steve hsu said...

Trying to avoid details in the post discussion above. The "~" symbol hides some coefficients :-)

ESRogs said...

Fair enough :-)

Blog Archive

Labels