Superbrains will not come out of a test tube


Stephen Hsu thinks super intelligent humans are coming. He thinks this because he's very impressed with genetic engineering (he's a physicist), and believes that the way to make people more intelligent is to adjust their genes, and therefore, more gene tweaking will lead to more intelligent people, inevitably. And not just intelligent, but super-intelligent, with IQs about 1000, even though he has no idea what that means, or for that matter, even though no one really knows what an IQ of 100 means. We're going to figure out all the genes that are involved in intelligence, and then we'll just turn the knob on each one of them up to their maximum, and boom, super-humans.

Good god, what a load of crap. Lots of people seem to think it's brilliant, though. It isn't.

Let's set aside one concern: is intelligence, and more of it, really that good for human beings? We're dependent on a certain level of smarts -- we're a technologically specialized species -- but it's not clear that we necessarily gain much by getting higher IQs. Fewer Donald Trump supporters, you might argue, but there intelligent, successful people supporting Trump, so there are more complex factors than just IQ behind that phenomenon. There is an argument to be made here against intelligence as a panacea, but as I say, let's pretend for a bit that it is nothing but good. We want more intelligent people. Society as a whole and every individual would benefit from being more intelligent, just for the sake of argument.

Then the question becomes one of whether such an increase is possible, and whether genetic engineering is a practical way to achieve it. My answer to the first is that it's unlikely, and the second is a flat no.

Our evolutionary history suggests that there was a period when Homo's brain was undergoing a long period of gradual enhancement. It wasn't Homo sapiens though; it was Homo erectus. Fossils of that species over its 2 million year history show a pattern of slow enlargement of the cranium -- they were getting larger brains. Their tools show some pattern of refinement, so there's some evidence they were using those bigger brains in more sophisticated ways. But taking a million years to figure out how to put a sharper edge on a stone hand axe isn't exactly a rapid development cycle.

Modern humans emerged out of Africa between 100,000 and 200,000 years. They were slightly smaller (and smaller brained) than the robust humans living in Asia and Europe, but they did bring about some advances in technology and swept over the world…and were adept at learning new skills. Again, we'll say for the sake of argument, they represented a clear adaptive advantage to greater intelligence, even though there is no biological basis for assuming they were more intelligent, or that it was their intelligence that allowed them to displace other human groups. (I suspect that more complex social structures and language, which are obviously a product of the brain, are more responsible than IQ).

But here's the thing: those early modern humans were pretty much indistinguishable from us today. They were about the same size, looked about the same, had the same capabilities we do now. If we used a time machine to go back and kidnap a Cro Magnon baby, bring her to our time and raise her in an ordinary American home, she'd probably grow up to play video games, shop at the mall, get a college degree, and land a job at an investment bank, and do just fine. Most of the evolving humanity has done since seems to be focused on their immune system and adaptations to agriculture and urban living.

One has to wonder, if IQ is such a great boon to humanity, why hasn't the biological basis for it shown much improvement in the last 100,000 years? Evolution is far better at tinkering than humans are, and has been tweaking our species for a long, long time, but super-brains haven't emerged yet. Somehow, genetic engineering is going to find amazing new solutions to intelligence, a quality of the brain that we don't even understand yet, and cause a great leap upward? Unlikely.

Hsu's answer is convincing only to the naive. He's basically proposing that the problem is that intelligence involves a lot of genes, so all we need to do is find the optimal variant in each of the genes involved (evolution has already done our work for us!), and then combine them all into one individual. If a thousand genes contribute to intelligence, and there's a variant of each that gives +1 IQ point overall, then all we have to do is a massive genetic adjustment that brings them all together. Easy, right?

To achieve this maximal type would require direct editing of the human genome, ensuring the favorable genetic variant at each of 10,000 loci. Optimistically, this might someday be possible with gene editing technologies similar to the recently discovered CRISPR/Cas system that has led to a revolution in genetic engineering in just the past year or two. Harvard genomicist George Church has even suggested that CRISPR will allow the resurrection of mammoths through the selective editing of Asian elephant embryo genomes. Assuming Church is right, we should add super-geniuses to mammoths on the list of wonders to be produced in the new genomic age.

Note his estimate of the number of genes that contribute to IQ: 10,000. That's half the human genome! Hmmm. I wonder if any of those genes play a role in other processes in human physiology that might be affected by his plan?

Here's an analogy for you: let's say a novice car designer has decided that the one quality of an automobile that is most important is speed, raw speed. He doesn't know much about cars, so he asks more qualified engineers about what elements of the car contribute to acceleration and velocity, and they start off with the obvious…details of the engine, fuel mixes, etc. Then they're talking tires. Aerodynamics. Weight. Pretty soon they have to admit that just about everything in the car is going to affect the speed at which it travels.

So our blithe designer decides that making a fast car is simple: we just look at each component of the car one by one, and we pick an available option for it entirely on the basis of which option makes the car go faster. We'll easily be able to make a car that can rocket along at a thousand miles an hour, he thinks.

But we have to ask whether we would want a car where the seats and steering were optimized for speed, where safety options were discarded, where something like visibility or reliability were jettisoned for the sole virtue of going really fast.

That's what Hsu is proposing. It's absurd. Humans are even more multidimensional than cars, and he thinks he can flatten people out to a single linear parameter. I think we've gone beyond imagining spherical cows to imagining human beings as a point on a line.

But OK, let's do the experiment. Let's grab a random human ovum, and our impossibly flawless CRISPR/Cas tools, and go down a list (one we haven't compiled yet) of 10,000 genetic alleles that each individually make some positive contribution to IQ, and we'll go through and serially edit each of those genes to conform to our hypothetical optimum.

Every biologist in the world is looking at that paragraph and saying, "Wait, we can't do any of that, and we don't have that information, and it's technically the next best thing to impossible." But don't worry, the techno-optimists who have no practical experience at all in this kind of molecular genetics will assure us, someday you definitely can. While the biologists will mutter in reply, "Then where's my flying car and my jetpack and my hoverboard, guy?"

But here in fantasy land, where I've been pretending all the difficult questions can be waved away, let's pretend we can carry out this experiment, and that it somehow gets past an ethics review board.

Now you've got a genetically engineered human egg, where half the genes have been ripped out and replaced with Stephen Hsu's chosen alleles. Who's the mommy and daddy? Who's going to take responsibility for this radical experiment in creative genetic engineering? Assuming this amply poked and prodded embryo makes it past the blastocyst stage (odds are it won't) and actually comes to full term (in someone's uterus, unless Hsu is so far into fairyland he's imaging artificial wombs), you're at some point going to have this infant with half its genome the progeny of a computer and a bacterial molecular pattern matching system. Now what?

This is where Hsu's whole idea flops down and dies in a flood of prolonged ignorance. All he talks about is genes, genes, genes, as if these humans will just pop out of a vat to take over his physics job. You know where most of the variability in intelligence comes from: it's in education and opportunity, not genes, and especially not genes that you don't understand and can't measure.

I'm all for increasing investment in biology and molecular genetics, but if you really want to create super-intelligent humans, the best strategy is to invest in sociology and education and social services. If Hsu actually believes all those optimized alleles are out there, why does he allow them to languish in poverty and want? If greater intelligence is an unalloyed good and an unquestionable virtue, why chase after untested genes in unborn individuals rather than crusading to have better nutrition for all children, better preschool education, better schools with more uniform standards?

Those are practical, achievable goals. They're also more expensive and require broader support than a mad scientist with a tissue culture facility, an electroporator, and a freezer full of reagents. But unlike the mad scientist, feeding poor children and creating a network of well-supported educational institutions would actually directly and effectively increase the overall intelligence and knowledge of the population and create material advantages to any society that pursued that goal within a generation.

If you actually cared about accomplishing an increase in human intelligence, that is. I'm not so sure that the people who are seeking magical solutions in a test tube actually do.


More like this

Where do these folks come from? Kind of makes you wonder about that IQ thing...

By David Jones (not verified) on 05 Apr 2016 #permalink

it's interesting this post coincides with the publication of the largest GWAS to date, and to my surprise it appears these guys have actually found something now.

the sample was entirely "white british" and it used an array > 100k SNPs on > 100k genome/phenotype pairs. and its hits for "cognitive ability" measured in a variety of ways coincided with many more of those putative hits already published than i expected.

We present lookups for all
available SNPs, as not all SNPs from the current study were
available due to differences in imputation reference panels. We
find that, of the 1115 genome-wide significant SNPs associated
with educational attainment, 327 (general cognitive function), 326
(years of education), 326 (college degree) and 267 (childhood
intelligence) were available in the published GWAS. Of these 158,
240, 211 and 47, respectively, showed replication at P < 0.05, for
general cognitive function, years of education, college degree and
childhood cognitive function.

Deary et al 2016

By chartreuse (not verified) on 06 Apr 2016 #permalink

the paper's title:

Genome-wide association study of cognitive functions and educational attainment in UK Biobank (N = 112 151).

i think it also worthy of note that the counter-example hsu has given on his own blog infoproc results from breeding, and he may not understand that though additive genetics is a good local approximation to the genotype x phenotype surface, the actual course of breeding through g x p space with environment held constant, just as with evolution, may be quite circuitous, that is, not a straight line as he assumes.

btw, i am a volunteer for hsu's BGI project, and he's banned me from his blog.

By chartreuse (not verified) on 06 Apr 2016 #permalink

I agree that IQ may not be useful - but that doesn't mean we won't work to increase IQ (however that is defined - which will continue to be defined many ways).

We have a long history of predicting what technologies will be achievable in the future - without being correct very often. We have just as long of a history predicting what technologies will *not* be achievable in the future - with the same lack of success.

By Howard Brazee (not verified) on 06 Apr 2016 #permalink

I read Hsu's article and thought, "more Transhumanist horse stuff," so thanks PZ for taking this on and shredding it (along with all the other Transhumanist cult stuff). seems to flirt with that from time to time, and could use to hear from some working scientists who can spell out in excruciating detail exactly why Transhumanism etc. belong in the same dustbin as astrology and alchemy.


As for intelligence increase:

1) We're not taking full advantage of the genetic capacity we already have. Some specifics:

improving maternal, prenatal, infant, and child nutrition. Yes that means a large increase in "welfare" and suchlike including free breakfasts & lunches for all students in public schools. That plus much else can be paid for with progressive taxation including excess wealth taxes.

Teaching parents how to provide enriched environments for their kids. Hint: major cutback in "screen time" for children *and* parents. More emphasis on reading to kids and playing with kids in their physical environments. Emphasis on providing sensory-enriched physical environments, with colors, textures, objects to play with that encourage imagination, etc.

The types of toys and games that are useful here should also be subsidized by more "welfare." And, good parenting practices should be covered in core public school curriculum starting just before the mean average age of onset of puberty.

2) We're not taking advantage of the full capacity of the brains we are already getting based on the parenting & schooling & nutrition etc. we are presently providing.

Public school should include music (promotes development of math abilities), creative & expressive arts, phys ed & athletics that include lots of physical movement in conjunction with positive peer feedback, etc.

"Emotional intelligence" includes not only empathy (the ability to infer the feelings of others and respond appropriately) but also the ability to recognize where one's emotions are interfering with one's ability to make reasoned decisions. Mindfulness meditation is good for the latter: learning how to recognize that one is having an emotion, is the first step toward recognizing how emotions affect reasoning.

3) Public school curriculum needs to back off on memorizing enormous quantities of details, and instead put more emphasis on "how to think." That would mean courses that cover the systems of logic developed by various schools of philosophy, courses on scientific method and how to apply it in routine daily life, the nature of "truth" as that term is used in various fields (e.g. the law, journalism, the arts, as well as the sciences: they are not interchangeable).

4) Public school & college curricula need to allow more time for students to just "sit and think," rather than having to produce enormous quantities of schoolwork like manic robots. Undirected/unscheduled time appears to be critical in problem-solving, see also the "release-of-effort effect."

5) Sleep, sleep, and more sleep: children, students, and adults. Sleep, and dreaming in particular, appear to be important for consolidation of short-term memory into long-term memory, and for emotional state regulation. Yet we are a chronically sleep-deprived nation with no end in sight, and with more Transhumanist-inspired efforts to completely eliminate the need for sleep in humans (highly unlikely to succeed without major side-effects).

6) Stop using graduates as "interns" whose job is to use their newly-earned degrees to make coffee for their "betters." Put them in positions where they can put their brains to use. The military is way ahead of the civilian world in this regard: smart kids can enlist and quickly get put into positions where their intelligence is put to use in ways that carry high levels of responsibility. The civilian world needs to take a lesson from that. See also Bill Gates and the two Steves (Wozniak & Jobs).

OK, this comment is too long as it is. But the bottom line is, there is no need to go tinkering with genes to produce smart humans. If Mom & Dad can't afford decent food, and if the local school's version of the Three Rs are Rote, Repetition, and Regurgitation, and if all of one's spare time is spent staring at circuses on a palm-sized screen: then it doesn't matter if you have the genetics for IQ 1,000, or only IQ 100: you'll end up with an under-utilized brain that is trained to resign itself to that state of affairs.

Surely we can do better: we must.

Thank you! I am beyond tired of that kind of load of crap. A dose of reality is an excellent public service.

As to why Hsu isn't also advocating realistic brain-boosters like good childhood nutrition and protection from lead poisoning, my guess is that that sort of crap grows out of a deep-seated techno-cognitive elitism. The vision isn't smarter kids today, it's a far future where all of humanity is in some sense a clone of Hsu.

...a deep-seated techno-cognitive elitism.

Or just regular old business + technology boosterism. Every day we see breathless news stories about "Bioengineering! It's happening NOW!" that tell us that the writers and the boosters know almost nothing about biology except that all you have to do is put "engineering" in front of it and soon there will be a whole industry churning out PROFITS.


Intelligence is over-rated. I'd like to see a discussion about the adaptive value of intelligence to the human species. For example, if it's true that the highly-evolved human brain is partly due to its relationship with the complex human social environment, I see some disadvantages. On a simple level, having an unsustainably large population of humans whose intelligence is invested in something other than physical survival is a threat to their physical survival.

Lots of animals and plants are "intelligent" in ways that don't count for humans, because they don't exhibit the kinds of social structures we expect in human societies. Or because we figure that multi-generational adaptations happen too slowly for us to care about.

Birds suggest this question, because they've been around (in various forms) for a hundred million years longer than humans. Yet they're hugely adaptable and able to learn whole ranges of behaviours, including limited forms of culture. Maybe you could call this a "distributed intelligence", not expensive enough to require re-building the whole cranium to accommodate a big brain, but plastic enough to allow extensive local variation.

And plant "intelligence" (is there a better word?) is so foreign to our mammalian minds that we're only now trying to comprehend that this might be worth including in our definition of intelligence*. It's becoming clear that there is some, it just exists on scales and dimensions we have no experience with.

*I have a Botany degree because I used to believe in a bunch of vague nonsense about plants and their secret lives.

your most recent rebuttal to shoe is the sort of thing he's banned me for.

good on ya!

you're obviously a lot smarter than professor shoe.

i myself have used the secretariat example many times. the problem with it is that thoroughbreds do NOT in fact have much genetic diversity. when i used the example i didn't know this.…

but still, secretariat's belmont is the single greatest feat in all of sports, 43 years later. "like a tremendous machine". it is forgotten today that secretariat didn't when the first two races by much, though he still holds the record in ALL THREE.

---BGI volunteer

By chartreuse (not verified) on 09 Apr 2016 #permalink

Inventing new labels (like “cognitive ability”) does not increase understanding.

and good on ya for that too. i've ridiculed professor shoe for exactly the same.

i think shoe is a great example of how the american education system selects for obedience, striving, pushiness, and a-morality much more than that of other countries and how it selects for intelligence or subtlety much less.

---BGI volunteer (banned by professor shoe)

By chartreuse (not verified) on 09 Apr 2016 #permalink

Re: "You know where most of the variability in intelligence comes from: it’s in education and opportunity, not genes". In fact, heritability for adult IQ is around 0.75.

By Tim Tyler (not verified) on 10 Apr 2016 #permalink

This post reminds me of a genuine question I've had for years but have never seen any analysis of: Why has intelligent life (apparently) only arisen once in the Earth's history?

I'll elaborate: There are certain readily-filled niches in ecosystems. Even when the slate is wiped clean by some mass-extinction event, the evolutionary process means that animals from wholly different orders are plastic enough to re-fill those niches quickly.

Thus, if terrestrial ecosystems have generally had room enough to tolerate thriving populations of arboreal animals, flying animals, burrowing animals, fast moving carnivores who prey on large lumbering herbivores, carrion eaters in their wake, semi-aquatic animals, and so on, then we have found that such niches are invariably filled.

Velociraptors, Moas, Tigers and Marsupial Lions have occupied one such niche (predatory carnivores) by turns. Apatosaurs, Elephants, Diprotodons another (large herbivores). Pterodactyla, Archaeopteryx, Modern birds, and bats still another, and so on.

When a living can be had as an occupant in one of those niches, it seems applicants have always queued up, regardless of whether they have cold blood, feathers or pouches. These niches must represent enduring evolutionary "sweet spots", since they are filled over and over.

Obviously, intelligence confers a huge survival advantage. It enhances the ability for creatures to plan, and to act in concert through the use of language. Although many other animals are social species, an intelligent individual's ability to survive and reproduce is further multiplied through greater co-operation with the whole. The aggregation of learned survival strategies suddenly can be passed down the generations via a means better than mimicry or instinct. An animal can only mimic what it has seen, but language means the memes for, for example, an improved hunting method, or of rendering a food otherwise poisonous fit for consumption, can be passed across continents and down the centuries by stories, and eventually, writing. Intelligence means an unprecedented ability for a creature change its environment to suit itself, rather than need to continually adapt to suit the environment.

So if nature has continually repeated herself through the repetition of forms and characteristics advantageous to exploit a niche, and intelligence is manifestly such a characteristic, why is there no indication that intelligent life or civilisation has ever appeared before in the half-billion years that have elapsed since complex life arose?

By Nathan Zamprogno (not verified) on 14 Apr 2016 #permalink