> For example, a professional tennis player can follow the trajectory of a tennis ball after it is served at a speed as high as 160 miles per hour, move to the optimal spot on the court, position his or her arm, and swing the racket to return the ball in the opponent’s court, all within a few hundred milliseconds.
I suspect if you made a perfectly invisible ball, you'd find people would manage to hit it back quite often. There's a lot of signal from the opponent's body motion.
> For example, a professional tennis player can follow the trajectory of a tennis ball after it is served at a speed as high as 160 miles per hour, move to the optimal spot on the court, position his or her arm, and swing the racket to return the ball in the opponent’s court, all within a few hundred milliseconds.
This doesn't seem to be specific to the human brain. My cat is probably quicker.
Sure. The point is, biological brains are quite good at this compared to machines, despite it being a composition of trivial mechanics and kinematics problems, and a vision problem that is similar to ones that computers are pretty good at. We seem to do better at it and in a fraction of the power budget of computing.
Let's say you're building a robot tennis player. Ignoring the physical engineering for a second, let's assume it can move around and swing a racket with roughly the same speed as a human.
It doesn't have anything other than a camera to track the ball.
So in terms of it's control, we have two options:
1: track the ball in 3d, work out the velocity, trajectory, ballistics, etc and then calculate the optimal position and swing to return.
2: use some form of ML that recognises patterns and applies patterned responses.
I'm willing to bet that 2 will significantly outperform 1, for a fraction of the compute power (once the training data is done).
Because that's what humans do. No one is working out equations in real time.
Human brains aren't bad at this, but I'd bet large sums of money that machines will out perform us at this within a decade.
> Human brains aren't bad at this, but I'd bet large sums of money that machines will out perform us at this within a decade.
I'm not so sure. Many other spaces where computers outperform us come nowhere close to doing it in the same power budget, and the amount of power it takes isn't reducing quickly.
> Because that's what humans do. No one is working out equations in real time.
I believe we do something in-between-- like have a kinematics and tracking engine that is fuzzy and gets tuned/updated/supplemental training for particular problems we face a lot, but isn't very tennis-motion specific.
I’m not sure human brains don’t do equations in real time. Sure, not in the form a computer would do it, nor in a way we could use up intermediate calculations, but humans have great spacial/visual skills which are not solely akin to ML algorithms. I’m a layperson, but what I mean here is rotating a 3D object in your head is not the same mechanism as your general flow of thought. It can perhaps be thought of as “graphic acceleration”.
I think both brains and ML in practice "solve equations". The main thrust of machine learning is using linear algebra as a universal function fitter. Sure, the state space may be transformed in interesting ways and the overall process may be relatively opaque, but you could easily end up with the quadratic formula in there.
The equations may be very strange from the point of view of the underlying physics/kinematics problem. There's a lot of representations of the problem that fit well and may map better to what our perceptions see and the level of precision that our muscles need.
And it's likely very messy--- we don't have some clean, nested set of controllers here, but strange ways that many kinds of feedback are combined.
I honestly doubt ML will outperform a classical approach. A tennisbal is relatively easily recognized using basic filters. The kinematic equations to solve are cheap.
I think that's the answer to the question TFA poses: because your cat's enormous ancestors also had efficient, fast brains. In order to not get got, we needed to be similarly equipped, reaction-wise, and optimized for fuzzy analysis of high-speed, blurry objects.
This, of course, shunts the question one level of abstraction away, but the standard evolutionary arms-race handwaving suffices here. So long as one side does it, by random chance or a side effect of sexual selection or whatever, the other side eventually does, too.
What follows is in addition, not in contradiction.
The complexity of human musculoskeletal coordination is astonishing. Particularly in sport, but also in daily life. Consider: how many ways are there to raise a cup to your mouth? Basically: infinite. And yet you do it without a care, almost reflexively, barely paying attention. In sub-second sporting tasks, the role of the conscious brain is likely to be zero, but the role of anything above the brain stem is also often unclear.
The spine - home to reflex action - doesn't have the processing power to determine how to return a tennis serve, yet this sort of action is far quicker than any deliberate, calculated, conscious action.
When I last spent any time reading around the subject, the literature was full of stuff like how the central nervous system updates its expectations based on feedback when trying to achieve a goal; and hints about how musculoskeletal movement might involve "orchestration" of composable units of coordination.
It's a fascinating area of research.
Overlaps with robotics and ML are obvious, although I suspect those approaches are "inventing" rather than "discovering". Nothing inherently wrong with that, of course.
I don't think that returning a tennis serve is a literal reflex as you're claiming.
From my understanding a reflexive move requires some sort of stimulus, and there's no way that your spine can have stimulus of an incoming tennis ball unless the eyes and therefore the brain are involved.
My post isn't clear, in that case. I'm not claiming it's a literal reflex :)
A top-flight tennis player is responding extremely quickly, and extremely accurately. Deliberate, planned action is not happening. It's something else. It's not a spinal reflex, it's something that involves parts of the brain to very rapidly 'orchestrate' a highly complex musculoskeletal dance.
The raw processing necessary to coordinate that dance cannot (in my understanding) be achieved by familiar neurological pathways -- observing, planning, deciding, coordinating. The planning and deciding aspects are bypassed. Only a tiny fraction of the observing aspect is used, such as the moment the other player strikes the ball (because the ball then moves too fast to react to). Some form of prediction happens. Then a number of 'movement primitives' seem to be selected and orchestrated. Ideally, the whole system is tweaked in real-time based on feedback (visual, proprioceptive) regarding whether the goal was achieved -- i.e. did you return the serve well.
It's wild.
It's also possible I'm mis-remembering and mis-interpreting some of this stuff :)
This is most obvious in soccer and basketball where you’ll see two similarly skilled players almost get stuck in place trying to predict the other’s next move.
There was an experiment done on training for basketball players where a group was told to train for a couple of weeks 3 point shoots.
One group trained normally, as in, they would position themselves and throw the ball.
Another group trained the same way, except they had no ball. So they had to position themselves and throw an imaginary ball.
A final group was obviously told not to train at all.
The results of this study showed similar progression between the ball group and no ball group and no progression for the no training group. And the study continued on showing that the exercise of repeating the pattern even without the ball strengthened the neural pathways that are usually activated when doing the action. From brain to arm and that a big part of "learning" is not actually adjusting but reinforcing.
So I imagine the players in tennis would definitely be able to hit the ball if they trained without it. At least to a certain point (adjusting is important as well obviously).
Modern humans in the developed world have access to insane quantities of calories. Too bad that evolution is still scared of us starving to death so we store the calories as fat. I think over evolutionary time, our brains could do some unbelievable things using all those calories.
In Alastair Reynold's https://en.wikipedia.org/wiki/Revelation_Space_universe there is a genetically and cybernetically engineered subspecies of humans called Conjoiners who do just that. They take it to such an extreme that they eventually need to engineer their skulls to grow heat-dissipating fins, because otherwise their brains would cook from the waste heat.
I think the brain might be thermally limited so adding more fuel to the fire could only do so much. Maybe future humans will have to take up a lot more copper to deposit in the cooling veins.
If you beat the heart a lot faster you probably get pretty good cooling. Vent that in the extremities and start sweating. There is obviously still a limit but in day-to-day life we don't even really come close to the brain's cooling capacity.
Well your comment is typical to the bias “all calories are equals”
First, all calories are not equals, therefore you can be starving and pushed to store energy, in an abundant calories intake.
What matters here is bio availability or nutrient in the calories intake. If for instance your protein intake is too low, and available vitamins is poor, whatever the amount of “energy” ingested this will trigger starvation mode and energy saving.
Have you ever read scientific littérature about how hybernation is dietetary induced in hibernating mammals? As shift of diet to high sugar, high seed oil (from grain) greatly available during summer (weight increase period for hibernating mammals) by lowering metabolism rate, promoting fat storage and torpor.
In early 19th, average calories intake of French and Switzerland was around 4000 to 4500 kcal a day, without obesity crisis while in the same time in American states adopting seed oils emerging products while the daily calories intake was half those amount, overweight appeared and started to spread.
So no, calories in calories out, and all calories are equal, are not true and therefore cannot be used at any prémisses of rational development.
Probably not by natural pressures, though. We're more than intelligent enough to outcompete any species that might try to fill the same ecological niche as humanity. No benefit to even more brain power. It would more likely take intentional selective breeding if we wanted to go that route.
Most evolutionary progress comes from outcompeting inferior members of your own species, not others. But it's also true that in modern society, intelligence doesn't seem to confer reproductive fitness.
That's an interesting idea. Do we know what is different about the brains of geniuses that could be replicated by changing how the brain grows or consumes fuel in a normal pereson? I know descriptions of "high-IQ" and "normal" are extreme simplifications of a complex, multidimensional concepts, but I have to imagine someone like John Von Neumann who apparently never forgot anything he read, has had to have had something physically different in his brain.
For example, do we know if high IQ/abilities are associated with more neural connections in the same brain volume?
Not really answering your question, but many Nobel-prize winner people were of Ashkenazi origin, who have some genetic changes that can cause serious neural problems, but at the same time these genes may be responsible in part to their genius.
Evolution got around that by birthing what are essentially still fetuses compared to most other non-marsupial mammalian newborns and then continuing the base development process of the brain after birth. See how a lot of other newborn animals are already walking around just minutes after being born.
Yeah. If there's an evolutionary aspect to the increase, we should expect to see c-sections as a generational phenomena, and correlated with cranium size. I wonder if that study has been done.
I admittedly havent spent much time thinking about the problem, but it seems to me that the issue of getting one human out of another human is quite a difficult problem to solve.
Not how I read it. Thought they were saying that our brains could evolve to use more calories since they are abundant now, instead of storing the calories as fat as our bodies currently do. Basically what if you could make your brain 50% more powerful at the expense of needing more calories, it would seem like a pretty good trade off today since many of us eat more calories than we need anyway.
Energy budget of brain compared to rest of the body is already very substantial I wonder if we would face some issues if it was higher. Like maybe overheating.
A good point. Incidentally i need to eat like a whale despite sitting on a chair all day long. Brainpower is a hungry business but i dont get it my body needs to store some of that food as fat. I eat all day long! Perhaps we are feeding it the wrong stuff and what we are used to eating is no longer suitable for how we use our bodies in the modern day? But what about the super developed countries of east asia where people appear to be skinny? Is there something wrong with our diet in the european and american world?
Other than genetics I'm pretty sure there's a lot of things that make people less hungry.
I've heard sugar makes people want to eat more. I'm pretty sedentary, so when I cut back on sugar, I can easily just kind of forget to eat.
Evolution doesn't know about fridges and thinks people still need to store energy, like as if we are going to walk 10 miles after a day long fast at any moment. So many constraints no longer exists, seems like we should definitely be able to eat better than before.
Who knows how many things no longer apply to most people. Maybe some foods enhanced fertility and that preference was selected for, but most people don't need them now. Maybe some things helped people heal from specific blunt trauma injuries that are now rare, but at the cost of causing some extra heart disease.
Almost every overweight person could kick my ass just by falling on me. They can probably lift more than me. Many would do 200% better in the wild than me, but statistically they are at risk for heart problems and many already have assorted aches and pains.
The whole concept of health in popular culture is way too tied to strength and survival and naturalness. Why do we seem to pay more attention to who can climb a mountain and lift 400lbs than we do to which populations live the longest and have the least illnesses?
Although this sounds like a good idea for an individual, there’s really no point, because a group of well educated humans focused on a specific goal can easily have far more brain power than a single individual, and with far great redundancy. Therefore making one single individual far more intelligent and powerful at thinking has little use to society, unless they are focused on tasks that inherently can only be done effectively by a single individual, and which are becoming fewer as our tools for collaboration grow. It’s cool to have someone be like an Einstein figure, but Einstein alone could never match the combined brainpower and output of NASA for instance.
Hmm, I don't agree because I don't think intelligence scales out like that. Certainly a group of people can do more than a single person but I personally think things like increasing a persons short term memory could have pretty profound impact on our ability to have bigger ideas simply because you could hold more things in your head at once for example.
I think you're reading it wrong: GP is saying if the brain could use as many calories as were available, then people with access to fewer calories would be at a handicap; but such is not the case.
To be blunt, yes. People with poor nutrition, particularly in childhood, experience both physical and mental stunting, some of which is irreversible. Certain diseases which have been curtailed or eradicated in developed countries inflict a similar toll.
I'll give a concrete example that combines both effects. Hookworm is an intestinal parasite that causes nutritional deficiencies in the host. Children with hookworm are impaired across the board, mentally and physically. They can't run as fast or read as well as their hookworm-free peers. If we compare two otherwise comparable populations, one with hookworm and one without, we would expect the hookworm-infected population to be less intelligent on average.
I'm not saying this to look down on anyone. Human populations always show enormous variation, anyway, so it says nothing much about any individual. But that burden, of physical and mental impairment and chronic illness, due to poor nutrition and infection, is a significant barrier to development, and a major part of why parts of the less developed world remain less developed. And it's why I believe childhood vaccination, disease eradication and nutrition programs in poorer countries are some of the best things we could spend our resources on, in terms of furthering human development.
>To be blunt, yes. People with poor nutrition, particularly in childhood, experience both physical and mental stunting, some of which is irreversible.
I don't disagree, but I was trying to substantiate this point once and I couldn't find any one good source that would confirm such a statement. Would you mind sharing a reference to some good source material approving the conjecture that poor nutrition in the childhood causes mental and physical stunting?
This study [1] deals specifically with brain imaging in the malnourished, but it starts off with a pretty good literature review in section 1 and 2 that may offer you some pointers. This [2] is a review of the literature on the question of childhood nutrition and brain development, see particularly the section "long-term consequences of undernutrition in early life":
> Many studies have compared school-age children who had suffered from an episode of severe acute malnutrition in the first few years of life to matched controls or siblings who had not. These studies generally showed that those who had suffered from early malnutrition had poorer IQ levels, cognitive function, and school achievement, as well as greater behavioral problems. [...]
> Chronic malnutrition, as measured by physical growth that is far below average for a child's age, is also associated with reduced cognitive and motor development. From the first year of life through school age, children who are short for their age (stunted) or underweight for their age score lower than their normal-sized peers (on average) in cognitive and motor tasks and in school achievement. Longitudinal studies that have followed children from infancy throughout childhood have also consistently shown that children who became stunted (height for age < −2 SD below norm values) before 2 years of age continued to show deficits in cognition and school achievement from the age of 5 years to adolescence.
This is just one of the many ways of "proving" that people in developing nations are somehow inferior. The same people that grew up with poor nutrition perform quite well when relocated to other countries. While in developed regions there are people with access to food yet dumb like a kite. I think we should be careful with the conclusions we draw. Certainly nutrition, let alone disease or parasites, can lead to reduced mental performance, but deriving the fact that the developing world is somehow suffering from reduced mental power, because of food, as a whole is wrong and in my view dangerous.
A nice thing about the hookworm example is it applies within the United States. It was common historically in the South, but not the North. This may well account for some of the stereotypes about the lazy, stupid Southerner, as well as the gap in economic development and educational attainment. [1]
> How much credit, if any, hookworms can take for those lingering economic challenges and misconceptions, however, is nearly impossible to measure, although some have tried. Hoyt Bleakley, an associate professor of economics at the University of Michigan, used early to mid-20 th century census data and records from the Rockefeller Sanitary Commission to compare educational and economic gains in places where hookworm eradication did and did not take place. He found an increase in school attendance and literacy in relation to hookworm reduction and also discovered that those effects seemed to extend into adulthood, with better-educated children growing up to be higher-earning adults. This suggests, Bleakley writes , “that hookworm played a major role in the South’s lagging behind the rest of the country.”
> “If you compare places in the South with the worst versus the least hookworm problem, you’re talking differences in income of maybe 25%,” he says. “There are lots of reasons why the South had a different developmental path than the rest of the country, and while disease is not the whole story, it was certainly part of it.”
Generalising about the south in the us is as wrong as generalising about the rest of the world.
Incidentally in most european countries you find jokes about the south of the same countries as in the us. Its just something we do with people far and different we look at their ways and call them dumb.
The article tho has made an attempt at proving it with science. Correlation does not imply causation.
You cant simply draw the conclusion that a mass of people are dumb and then make up the science to prove it. Sure there individual and small localised groups affected by it but not whole nations or even massive areas of a country.
Edit: even here on this forum, there is anecdata from people that grew up in poverty with little to eat or poor nutrition yet they perform well given the opportunity. Some from the west, some from asia.
> There are disabled people who climbed Mount Everest, but that does not mean that disability is irrelevant for your chances.
More accurately, there are disabled people who have had sherpas drag them up Mount Everest. To be fair, that's what most so-called mountaineers do these days.
It's interesting, because in Germany the southerners are observably much more technically accomplished than the northerners. But there's still that you don't speak High German prejudice.
> While in developed regions there are people with access to food yet dumb like a kite.
But he explicitly said:
> > I'm not saying this to look down on anyone. Human populations always show enormous variation, anyway, so it says nothing much about any individual.
I'm curious for the other folks reading this post and interacting with this thread- do you believe that the logical pathway for expanding human intelligence is with A) Brain-Computer interfaces, B) Through AGI, C) Playing with our genes (e.g. designer babies), D) Something else (and I'd love to hear thoughts)
I'll go and play along with my own definitions, even though I think the question is hard to answer because we may not agree what it means at all.
Improving the development and deployment of existing human intelligence is the strongest logical pathway for expanding human intelligence.
1. Most humans don't develop to anywhere near their full potential.
2. It's unclear whether even the "best" humans are anywhere near actual limitations.
3. Most humans spend large amounts of time doing drudge, unrewarding work below their full developed potential that could be better accomplished in other ways.
Things like efforts to expand AI potentially chip away at 3.
It's pretty dated regarding the roles of women and men, but the ideas are very interesting nevertheless. One of the big insights is that many people might not be ready to operate at full potential - they would succumb to various forms of insanity.
Look how the Internet's 'network effect' amplified the voices of conspiracy theorists all over the world. Now imagine an enhancement of that kind of paranoia: this is what the super-rationalist protagonists had to deal with in Brainwave...
I kinda depends if they can agree on what needs to be done I suppose, just like in the real world. But TBH it sounds like it would quickly devolve into NIH, a bit like a team with too many seniors.
Oh I don't disagree that we should allow as many humans as possible to do self-actualize. It is clearly the right thing to do just from a moral standpoint. I just think that most of the people who fully "self-actualize to their full potential" as, say, a lobbyist or a conspiracy theorist podcaster would not be of as much benefit to the world as some of the enlightenment era scientists.
That was what I was getting at in my previous post: having everyone at peak potential is great when everyone is aligned. If there is internal strife, peak potential just means bigger conflicts. Taken to the extreme, that means bigger wars and billions of dead people. Until we can figure out the coordination problem, human limitations are not the biggest problem IMO.
My counterpoint is that even without people being aligned, we're far outperforming the past and what people hundreds of years ago thought we might do with increased privilege. Indeed, we even have less strife overall, even if we have many more axes of disagreement and fewer shared values.
> > 3. Most humans spend large amounts of time doing drudge, unrewarding work below their full developed potential that could be better accomplished in other ways.
Most of us have more interesting work that more fully uses our intellectual potential than typical people had a few hundred years ago.
And most of us also get more time in leisure, now, than people a few hundred years ago.
I think there's progress, but it's uneven. Look at what step changes of using a little more human creativity and intelligence-- the Renaissance, Enlightenment, Industrial and Information revolutions-- completely upending our society each time.
The main obstacle to a happy life for most of Humanity is not lack of intelligence but irrational or selfish impulses. That would be my D. Make a more liveable world and no, we won't get there with more emotions/drama. More like live and let live.
B won't expand human intelligence.
C is paved with unintended consequences. Wait until it's not playing.
C) you don’t even have to know which genes boost intelligence to know that kids from smart parents are more likely to be intelligent. Akademgorodok was a concentration of very smart Russian scientists and turned into an (accidental?) eugenics program producing a new generation of smart kids. Some of the smartest people I’ve ever met were born there. But once IQ boosting genes are known they can be selected for by sequencing embryos, I.e. pick the smartest 2 embryos from a candidate pool of 100. An average couple with average variance could still produce 130 IQ kids.
Genes don't even have to boost intelligence for kids from smart parents to be more likely to be intelligent.
Just think of a kid adopted out of the nursery by two smart parents. Is that kid: A) more likely to score higher on tests, B) more likely to score lower on tests, or C) exactly as likely to score high/low on tests no matter who his adoptive parents are?
> Just think of a kid adopted out of the nursery by two smart parents. Is that kid: A) more likely to score higher on tests, B) more likely to score lower on tests, or C) exactly as likely to score high/low on tests no matter who his adoptive parents are?
It is quite possible that the standards for adoption are such that only couples able to raise children in a nurturing environment capable of fostering intellectual growth are typically allowed to adopt and raise children. They are not an accurate cross-section of all potential parents by any means.
I am imagining if any couple could adopt; taking the same kid (which of course we cannot do) and having them raised by a couple of college professors vs. a couple of meth addicts living in a drug den - I would be absolutely stunned if there were no difference in what kind of test scores the kid would get growing up.
> taking the same kid (which of course we cannot do)
There were a bunch of studies on identical twins who were adopted in different families, so it can be tested. I think they showed the same thing (kids parents more important than adoptive parents) but I'm not certain and can't quickly find the studies to verify.
Twin studies are a good way to get another angle on this.
Here, unfortunately that doesn't get around the confounding factor that only parents deemed suitable candidates by adoption agencies are able to adopt and raise children, so they are not an accurate cross section of all possible parents. I don't dispute that there may be a threshhold at which some innate traits take over (or fail to do so) from the benefits of a nurturing environment (which may have diminishing returns past a certain point).
Terrible neglect results in low IQ. No one is arguing that. Obviously, whacking the kid on the head with a hammer, giving her terribly insufficient food, or never exposing her to language will depress IQ. These are all things that parents sometimes do.
Not only do we have evidence that the heritability of intelligence is high, from twin studies, but also: adopted siblings' intelligence is only slightly correlated-- barely better than strangers. Biological sibling intelligence raised in the same house are very closely correlated.
This seems to imply once parenting is "good enough", these types of environmental effects don't affect IQ in adulthood to a significant degree.
Perhaps the most interesting finding: parenting does seem to affect IQ scores in early childhood more than genetics, but by adulthood the strengths of these relationships have completely reversed.
I appreciate your well reasoned and thoughtful reply and I would like to add that I find your assessment completely plausible and I myself tend to agree with it. Although, I still question the validity of using IQ as a universal objective measurement of intelligence. At best I can see there being a fairly wide pareto front of optimal intelligence for most situations. But who can ultimately judge?
Into adulthood I find it very difficult to say whose intelligence is "better" in anything close to an objective way. While the results of genetic influence may become evident in some kinds of tests, it is just as easy to conceive tests at which the same candidate would fail yet someone else with a lower IQ might do very well. The possibilities are endless and I really don't think outside of cases of abuse or neglect creating specific developmental deficiencies as you mentioned that we can meaningfully quantify intelligence generally. If we could better do so I think the influence of parenting might still become more evident than what we see in IQ tests.
> No one is arguing that.
Well, my initial comment was simply an attempt to make the point that parenting can have an effect whether or not genetics do and the reply I got was "look it up".
> Into adulthood I find it very difficult to say whose intelligence is "better" in anything close to an objective way.
Intelligence is difficult to quantify and define. BUT: peoples abilities on a wide variety of tasks are positively correlated. This strongly implies there is some kind of underlying factor that makes you better at "all things", and this is what IQ tests attempt to measure.
Of course, the correlation coefficients are not 1; so of course someone can have a lower IQ than you but far outperform you in one of these tasks.
The results from IQ tests seem to be very strong predictors of many outcome measures later in life-- indeed, more than parenting seems to predict these things.
edit: from the wikipedia article on g factor:
" Research indicates that tests of g are the best single predictors of job performance, with an average validity coefficient of .55 across several meta-analyses of studies based on supervisor ratings and job samples. The average meta-analytic validity coefficient for performance in job training is .63.[76] The validity of g in the highest complexity jobs (professional, scientific, and upper management jobs) has been found to be greater than in the lowest complexity jobs, but g has predictive validity even for the simplest jobs. Research also shows that specific aptitude tests tailored for each job provide little or no increase in predictive validity over tests of general intelligence."
>having them raised by a couple of college professors vs. a couple of meth addicts living in a drug den
The usual analogy is that you can always lower the IQ of a child by hitting them on the head with a hammer or by lacing their food with lead powder. But no intervention has been found that raises IQ.
> The usual analogy is that you can always lower the IQ of a child by hitting them on the head with a hammer or by lacing their food with lead powder. But no intervention has been found that raises IQ.
Since almost no one, without deliberate intervention, experiences an environment completely free of known factors that are detrimental to IQ compared to the optimum environment, it's purely a semantic game to say that there are no known interventions that increase IQ.
I read recently of two identical twins separated at birth with on growing up in South Korea and the other in the United States, the twin who grew up in SK had an IQ 15 points higher than her identical sister. The connections in the brain either degrade or strength depending on how it's used and I can absolutely see a person's IQ being increased by engaging in subjects/tasks adjacent to what most IQ tests measure.
Yep. Note the US twin was literally whacked on the head, and that's probably why this MZA pair is so very unusual (and popularized, unlike other MZA reports like the authors' prior one, incidentally) and breaks away from the usual MZA correlation of ~.8.
Unfortunately intelligence is mostly genetic, look it up...
Edit: This is my opinion on what I consider a fact based on a fair bit of research and first hand observation. People are free to disagree with this but if they've made it to this point still holding that opinion then I think it's unlikely anything I could say would change their minds.
> Edit: This is my opinion on what I consider a fact based on a fair bit of research and first hand observation. People are free to disagree with this
You replied to my comment of "genes don't even have to boost intelligence for kids from smart parents to be more likely to be intelligent" with "Unfortunately intelligence is mostly genetic, look it up..." I'm not sure why you used the word "unfortunately" or what you want me to look up since you didn't even appear to be responding to what I wrote.
> but if they've made it to this point still holding that opinion[...]
I'm confused as to what this is intended to mean.
> [...]then I think it's unlikely anything I could say would change their minds.
How can you know to what extent anyone reading this has made up their mind about anything?
Are you suggesting, then, that the answer is C? I would hypothesize A is more likely, even assuming intelligence is genetic. Thus, my only point was there's a confounding variable if we try to use the relative tested levels of intelligence between parents an children as a basis to say whether intelligence is genetic. That's not to say there aren't potentially ways around that by designing studies carefully.
Given the difficulty of defining an objective measure for intelligence and the confounding factors of upbringing, I think it is almost impossible to find a good source on this sort of thing. Why should we go on a snipe hunt to prove your point?
* Identical twins have a much higher correlation coefficient of intelligence than fraternal twins (table 5).
* Identical twins raised apart have a correlation coefficient for intelligence of around .73 (MZA studies).
* Correlation coefficients of IQs of biologically unrelated siblings raised in the same adoptive environment are approximately 0.
There are obvious confounds the paper mentions:
* Fraternal twins may not have identical environments.
* Adoptive environments may not represent typical parenting environments.
* Some of these studies have had poor controls and practices.
* Distribution of results from studies looks bimodal which indicates something weird is going on.
Still, even if you take the most critical view I think you have to accept that there is still substantial heritability of intelligence. These results cannot be convincingly explained any other way.
This isn't really true. The heritability of IQ is pretty high. We know this from twin studies. Beyond this, adopted siblings, have a very low correlation coefficient of IQ by adulthood, but genetic siblings have a very high one.
All the other ones will happen to the best of our ability, but only for specific groups, and people(Or at least people like me) could wond up seeing them as almost cults, for a long while until it is totally proven safe and beneficial.
The rest of us, at least for now without major cultural change, will just do more and more of our "thinking" outside our heads.
Unless you can prove, via decades long studies, that your enhancements are totally safe and side effect free, with not even some headaches or 1 in 10000 complications, I'll just... use google.
Even if you did have the studies I'd still want to personally meet enhanced people and see for myself if it altered their personality or caused some random aches and pains, or had some other regrettable thing they didn't bother to do a study for.
Even after all that, there could still be social discrimination. If cyborgs were disliked my most people, I'll probably leave it to the people who are passionate about enhanced intelligence.
Sure, maybe I could get a better job. Maybe I would even be able to understand 3D space well enough to do things like driving and not occasionally walking directly into door jambs.
But by the time you have the enhancements cars will be self driving and computers will be able to do even more of the math I'd want to do.
I don't think I have any close friends who want a Neuralink.
CAD programs are essentially already a primitive form of neural prosthesis. We can work with shapes we couldn't work with mentally, and workflows outside of the traditional "Imagine, understand, write it down" are possible.
Even if ALL the problems are solved, it probably will only be useful to people working in specific cutting edge sectors, and for committed philosophers(Professional or amateur) who belong to a school that is not against enhancements in the first place.
I think in the medium term, probably genetics, because that's partly a trial-and-error approach. I feel like brain-computer interfaces, and other technology that relies on actually _understanding_ details of how the brain works, are quite a ways off.
But... in the short term, I think we could still make massive gains purely by improving education and child-rearing. If our schools and families actually raised and educated kids _properly_ (whatever 'properly' is), we'd see huge improvements in average IQ.
Real improvements that may be culturally infeasible, and also could be very damaging if they are done wrong.
My understanding is that removing lead from a variety of infrastructure has done wonders for people's intelligence, as has enabling better family planning.
I consider those to be in the category of not lowering people's IQ. Making someone dumber is fairly easy: just drop them on their heads as a child, malnourish them, and expose them to lead and lots of parasites. And make sure they don't get enough iodine. But in rich countries we've already fixed those things, and there doesn't seem to be much else we can do.
IQ is something very narrow. The fact that you can do some symbolic inference on paper does not mean that you will be more successful in life
From my school mates I have seen many of them who had on paper low IQ to have great careers as entrepreneurs, salesmen, athletes, craftsmen. I have also cases with top students who did not go anywhere.
So I am not sure whether an average increase of IQ would help the society.
Literacy for sure helps, but I am not convinced about IQ.
This is false: "The fact that you can do some symbolic inference on paper does not mean that you will be more successful in life"
IQ, more than any other psychometric measure, predicts a wide variety of positive life outcomes including income, health, longevity, not being incarcerated, etc.
Here [0] is a simplistic article about it, but there are endless studies and discussions about this, all with the same clear and uncontroversial result.
IQ correlates quite heavily with practice time on multiple choice tests though.
Who spends their time doing lots of tests? Rich kids. Being wealthy also predicts a wide variety of positive life out ones including income, health, longevity, and not being incarcerated
No, this issue of test practice being a confounder has been studied exhaustively for decades. There are many ways studies have been designed to determine its impact.
For example: Examining the predictive value of IQ tests between siblings in the same household (who generally have the same access to resources).
Answer: IQ remains very powerfully predictive even within a household.
A century of study and attempts at intervention have all pointed to an overwhelmingly clear result: It's not wealth that makes someone have high IQ. It's high IQ that makes people wealthy.
Children do inherit high IQ from their parents who had high IQ, but the mechanism of this isn't through wealth, it's genetic. About 80% of the variance in IQ is by genetic parentage.
Example study: Examining adoptees' IQ scores in relation to their genetic parents versus the parents who raised them. Answer: They inherit their measured IQ from their genetic parents; the adoptive parents have essentially no effect (the last 20% of variation is basically random).
C is the most practical in the short term, because we don't have to understand the brain all that well to find out how to tell an embryo to make it bigger and denser. It will involve a lot of failures.
I'm doing a comp-bio PhD right now, and the more I learn about the process of playing with genes, the further I think we are from expanding intelligence with them.
In the current state of the field, we simply don't know how cells work (in enough detail). We can't predict what cells do when genes are mutated, we definitely can't manufacture new genes that slot into the cell's existing machinery, and, for a lot of the proteins related to neurons in particular, we don't even know what they do.
If there is a breakthrough, it'll take luck at this point, and probably be something stupid, like realizing we can double production of some cell type without hurting the child *too* much. We're really not close to engineering anything, which might be a requirement for producing more intelligence without some very clumsy side effects.
Given the level of trial and error required for a lucky breakthrough, and the human cost of that error, we might just give up on this until the singularity anyways. That said, there's so much undiscovered now that there could be a perfect mechanism that just isn't known yet. And, indirectly, by just making sure people don't have allergies or a pre-disposition to heart disease... we might increase intelligence just by making people healthier.
Heritability of intelligence is high, and there's a few genes already existing in the human population that are well correlated with intelligence.
You could just try turning a few known knobs and see what happens. The efficacy is probably moderate-- maybe 7 IQ points or so with the currently known genes. But a change of mean IQ by 7 is actually a pretty substantial advantage-- it's half a standard deviation, and predicts about $4k more income per year.
Yes, it's unethical and there's probably substantial human cost.
Or, there's even a shortcut of just testing for people with high IQs that also possess the best known genotypes (i.e. test both genotype and phenotype) and forcing them to breed.
I would say probably A. Computers complement humans well, they are good at the things we suck at, and bad at the things we're good at. Lots of progress in computers nowadays is made by hanging problem-specific accelerators off the side of conventional CPUs, so either we can be "creativity accelerators" hanging off the computers, or they can be "computational accelerators" hanging off us.
It should also improve human-to-human communication speed. The whole trick of humans is that we can put our heads together and work as a super-brain. As a side-effect, the normal-humans have distill their work-product into 'ideas' that fit into the heads of other normal humans. It seems unlikely that we'll ever genetically engineer the whole human race, so continuing to produce normal-human-compatible ideas seems like a big win.
If we have genetically engineered super-humans going around, they might have some ideas that are just too big to fit in our normal brains. Breaking compatibility like that seems risky, don't want people out there whose word we just fundamentally have to take on some things.
If you eliminate the "G" in AGI to include any mechanical, electrical, or otherwise automated non-human computational processes we can use to augment our own brain's limitations, i.e. calculators, slide rulers, all the way up to Wikipedia, then AI is clearly the answer so far. Even just books and language, augmenting human intelligence by parallelizing thinking across many humans and not losing knowledge upon death, are huge wins.
But for the far future, all of the above seems like a good answer. At least one "something else" is the possibility of nutritional, electrical, or biochemical interventions that can increase brain function without needing to change our genes. Caffeine and beta blockers seem like the best things we've got so far, along with not being in a state of persistent starvation, but there are presumably ways we can do better.
If you didn't suffer from the presence of a moral code, gene editing babies for intelligence is something you could do today, or if not, with a month or two of preparations and waiting for suppliers.
I would bet that it is possible to expand actual consciousness into connected hardware (silicon or engineered nervous tissue), and have that hardware expand a metric you would call intelligence.
The most reasonable thing in a third direction would be improving brain development through specific training in childhood and training as an adult. A lot of intelligence comes down to how the brain organizes itself in response to inputs, and a scientific discipline to optimize this could make a lot of progress. Not that work isn't being done along these lines but I think there is a lot of room to grow here.
Are you asking me for a normative judgment (e.g. which one is morally correct?) or some kind of positive one (e.g. which one will improve outcome measure X most?)
Why are you assuming that it's just one? These choices don't seem mutually exclusive.
Human intelligence = the standard intelligence we exhibit today as humans vs. machine intelligence from something like ML
Not asking for a normative judgement or a positive one, I asked about what you thought would be the logical progression (under the assumption that humanity would like to continue to evolve and become more prosperous and becoming more intelligent or building more intelligent tools seems and obvious way to help enable that)
Could be more than one, was just trying to give some large buckets, that's why I added D and asked for people to add their thoughts.
Just to expand on the part of “ill defined question”. “Human intelligence” might not be a scale that goes up and down, so it’s a bit hard to say for sure. There is probably at least a tradeoff of flexibility vs specialization. There is also physical constraint where intelligence might be limited due to the whole brain/ machine becoming a distributed system (and you can’t have proper single point consciousness, for example). Though we don’t even know if consciousness is related to intelligence.
> Human intelligence = the standard intelligence we exhibit today as humans vs. machine intelligence from something like ML
Yah, but what is that?
Are you talking about an IQ score? Ability to do the tasks that ML is not currently good at isn't a great measure (a few years ago, Go or creating high quality paintings would have qualified).
E.g. one example of "D" is "rigorously studying and improving human education". But there's infinite clarification on what the problem you're trying to solve before we agree what improvement is:
* Mean, median, or peak outcomes?
* Performance on a given curriculum, or on a given abstract class of problems? Does creativity count? Is an education system that improves various kinds of quantitative metrics by 20% but results in a 5% less creative and flexible population a "win" or not?
Pretty much all the same arguments apply no matter which approach you consider.
> Not asking for a normative judgement or a positive one,
Well, normative considerations are important. If experimenting with brain-computer interfaces for quick advancement harms humans, is it a "good path"? How do we compare these paths? Quickness, magnitude of results, "best" resulting society, least harms, some kind of cost measure?
Why should this discussion play out again and again in this exact form is beyond me. It's always the same troupe of beaten up arguments trying to pass while being expressed in academia-flavored rhetorics, is it the canon critical thinking essay tone they teach in the US? Insofar as the grandparent post contained some useful inquiry this should amount to uncharitable derail.
So often this blows up when anything related to IQ is brought up, a profound generational trauma of american intellectualism gets a small scene to replay its discourse in suffocating detail.
Maybe this is as close to the root cause as we can publicly get? Our insane tendency to avoid productive discussion and stifle true creative activity, all in favor of hollow wordplay (quasi-statistical if the situation requires, not less hollow). Our education - a glorified daycare with rote-learning for line-workers, of which we somehow expect intelligence gains if only we could spend more time instilling it (going against all solid evidence to the contrary, because the doctrine of never having enough education has to be preached).
I don't like the way you are talking to me, and I especially don't like that you may not have read carefully enough to understand that parts may be agreeing with you.
> a glorified daycare with rote-learning for line-workers, of which we somehow expect intelligence gains if only we could spend more time instilling it
Drilling tons of math in rigorous ways for excessive amounts of time clearly improves various kinds of outcome measures. It also tends to cause other harms.
If we want to improve intelligence, first we need to agree what intelligence is. Then we can talk about what measures are appropriate or ethical to use to try and improve it.
Even if we just take a somewhat stupid measure, and say "all we care about is a hybrid of IQ and SAT-I scores at age 16"-- there's still the question of whether we care more about moving the top performance upwards or the entire distribution.
If we don't agree what intelligence is, we can't answer questions like this. If we use a traditional measure like IQ as a measure of G factor-- I don't think brain-computer interfaces are going to swing things much and they may even make it worse. (But they could also allow explosions in human productivity or quality of life that are not measured on that scale).
e.g. Smartphones have not made us smarter, but they may sometimes allow us to do smarter and better things.
Psychologist Robert Sternberg defined intelligence as "mental activity directed toward purposive adaptation to, selection, and shaping of real-world environments relevant to one's life."
An IQ score is a measure of intelligence, it's not intelligence itself. Just as the temperature you read off a thermometer is only a measure of the heat in the room. The thermometer tells you very little about the position and momentum of the gas particles in the room, only that higher temperatures correlate with more momentum.
The value of IQ scores for measuring human intelligence is that they correlate with all sorts of life outcomes. The hypothesis behind the research is that both the life outcomes and the IQ scores are caused by a single factor, known as g, for general intelligence.
Obviously, if we train some model on IQ tests we can probably get a machine to ridiculously high IQ scores. Does that mean we have created intelligence? No. The machine is not going to show any of the life outcomes we expect from a human with high IQ. It's not going to do anything! That's why none of these machines are intelligent.
Yes, I understand all that (and have talked about and mentioned g in other comments in this thread).
The point is, what's "improving intelligence"? It's something that we have a hard time quantifying on a linear, agreed axis. As you point out, IQ scores are a somewhat flawed measure of g, which itself is a somewhat handwavy observation of a underlying factor that is a strong correlate to human performance in a broad set of areas.
And even if you could put that human performance in a linear measure, there's still a question of how you value different changes in that distribution. Which is better-- raising everyone by .1 standard deviations, or boosting a few people near the top by 1 standard deviation?
I think baby steps: just do an IQ test with and without access to Google or Alexa, see where we are now with brain augmentation. There's a paper in that.
> Another salient property of the brain ... is that the connection strengths between neurons can be modified in response to activity and experience—a process that is widely believed by neuroscientists to be the basis for learning and memory. Repetitive training enables the neuronal circuits to become better configured for the tasks being performed, resulting in greatly improved speed and precision.
This seems like a clear description of a bayesian process. The new connections are a posterior updated from the likelihood informed from input sensory data. But the fact that they don't describe it as such makes me wonder if there's other reasonable ways to model this learning behaviour.
I am familiar with Karl Friston's work (as a layperson) of animal behaviour and learning as a form of bayesian update via variational inference. Does anyone know if his ideas, or the general idea of learning as a bayesian process well accepted in the neuroscience industry?
The general idea of a Bayesian brain is fairly well-accepted. Not everyone agrees that oh, this must be the way the brain works, but there are definitely experiments demonstrating set-ups in which you can get a good approximation of a Bayesian posterior distribution out of subjects.
The question is how the brain might do Bayesian inference in a very broad class of generative models (not just affine-Gaussian!), and what else it might be doing inference in service to.
That makes sense. Since you seem to have expertise in this domain, may I ask what is the current state-of-the-art for generative models for modeling learning/behaviour, in the neuroscience community? If any...
The neuroscience community doesn't really keep "benchmarks" the way the machine learning community does. The trouble this causes for us right now is that many experimental tasks now used in neuroscience are simpler than the corresponding tasks used in ML, and so could be "solved" by a significantly simpler model.
> This latter property is particularly useful for enhancing the precision of information processing. For example, information represented by an individual neuron may be noisy (say, with a precision of 1 in 100). By taking the average of input from 100 neurons carrying the same information, the common downstream partner neuron can represent the information with much higher precision (about 1 in 1,000 in this case).
What does precision mean here, mathematically? Precision as in precision/recall doesn't seem to work, as precision in that context means TP / (TP + FP), and going from 1/100 to 1/1000 would mean worse precision.
Does it mean something like floating point precision, where increasing bits can increase information represented, i.e. 2 bits represent max of 11=3, 3 bits represent maximum of 111=7 and so on? I can't quite this to work...
Let's say you want to estimate some quantity, so you put a Gaussian prior over it. You then have a bunch of different sensors subject to white (Gaussian) noise, and some link-function connecting the latent value to the mean of the sensor reading with fixed precision. Gaussian with Gaussian is a conjugate prior-likelihood pair, so you can calculate the posterior parameters in closed form.
Your posterior mean is going to be a precision-weighted average of your sample mean and your prior mean, while your posterior precision (signal-to-noise) is going to be proportional to the number of sensors you used.
Thank you, that's a very clever explaination. Even though I can't grok the mathematics off the top of my head (I would have to look up the conjugate gaussian-gaussian formula and work this out) I think converting this to a posterior probability calculation is much more intuitive, then the article's explaination.
So, my crude understanding is that therefore the signal and noise is the mean and variance of the Gaussian, respectively. The posterior uncertainty in this is case is thus a function of a weighted average of the sample data, weighted by the priors, normalized by the data, which results in a higher signal-to-noise ratio, or lower uncertainty.
>So, my crude understanding is that therefore the signal and noise is the mean and variance of the Gaussian, respectively. The posterior uncertainty in this is case is thus a function of a weighted average of the sample data, weighted by the priors, normalized by the data, which results in a higher signal-to-noise ratio, or lower uncertainty.
The variance of the Gaussian is proportional to a noise-to-signal ratio, so you have to turn it upside-down to get the signal-to-noise we usually work with.
BCI. We still interact with keyboards, which is a very cumbersome and roundabout way to talk to each other and to the machine. BCI will connect our intelligence in unprecedented ways, and hopefully will allow 'uploading' so that there's continuity even after death.
Genetics is not the right way to go about it - carbon and protein chemistry was a very slow 'minecraft' way to build intelligence , and each of our senses and motors can be replaced with much much simpler electronic devices.
I don't know what AGI means in this context. Surely a BCI can upload itself to a brain-simulator and live there. No need for 'artificial' persons, although it is in principle possible (why not after all).
> Surely a BCI can upload itself to a brain-simulator and live there.
I'm not sure. I have a strong suspicion that we can make a useful brain-computer interface, without any idea how the brain really works. And I kind of expect that's how it will go. I think this because, in a technical sense, we already have brain-computer interfaces. Cochlear implants. They are used to restore hearing. The auditory nerve and cochlea are, anatomically speaking, actually part of the brain.
A long, extremely thin wire probe is inserted into the cochlea and auditory nerve. The probe has electrical contacts which are driven by the cochlear implant speech processor. The probes themselves are interesting in their own right; early ones only had a few electrical contacts, while modern ones have dozens or hundreds of electrodes spaced out along the length, each of which can be individually controlled, providing an almost infinite variety of possible inputs which the brain can distinguish.
So here's the wild bit. We have only a vague idea how auditory signals are encoded by the cochlea and sent down the auditory nerve. This was even more true when cochlear implants were first devised, in the 1970s.
It turns out that doesn't matter. If you start applying varying voltages at different electrodes along the probe, that correspond with activation of various frequency bands at varying intensities, in the vaguest mimicry of auditory nerve signalling, the brain pretty quickly figures it out.
A person who previously had natural hearing and lost it can learn to understand speech this way sometimes in a matter of days. People who were born profoundly deaf and were not implanted in infancy often regret the decision to be implanted and usually do not learn to decode sound in a useful way. Children who are implanted in infancy usually do. As an aside, the accounts given by people who have had this generally unpleasant experience, suddenly given an extra sense but with no neural structures developed life-long to process it, are both fascinating and harrowing.
As you pointed out, the brain is very plastic and it will learn the mapping of its input, like it learns to read those squiggles we call letters.
We don't need to know how the brain works, just how its neurons work and its anatomy. Hopefully those will be enough to build the simulator, and the rest of the work of wiring itself up, will be done by the simulator itself.
How deep do we have to simulate before we capture what the brain computes? If we have to model particle interactions then there's not going to be much hope of accurately "uploading" a person.
we don't necessarily need to "copy" the existing brain. I guess by communicating with a "brain extension" it will gradually learn everything i know , enough to be a fascimile of my person. I don't need to know how it did it, just like i dont need to know all the weights of an artificial neural network.
Just think, some day instead of choosing between religions you'll get to decide between the alphabet afterlife or the meta heaven. Hopefully you don't commit some cardinal sin in life and get yourself shadowbanned first before you die.
One will eventually come to believe that the other's afterlife is a crime against the dignity of the soul, and attempt to destroy it. In this way we will fulfill the prophecy we created ourselves by waging war in heaven.
Frankly, that is quite improbable. Unless you can work over time to replace the brain piecewise so it integrates with non biological parts, transfer of consciousness is not happening
Because it's had millions of years to converge on a near-optimal solution for the environment that it's in. In fact, this is true for all organs within all extant species.
Nature is far more advanced than technology, and I don't believe that we'll ever quite be able to one-up her. Most efforts to do that have seriously harmed us.
We think we know everything, but this dangerous hubris can destroy all planetary life in the blink of an eye.
I'll take the third side of that bet: we've already so fundamentally altered "nature" to our whims, often in very destructive and short-sighted ways, that we're well past the point of trying to pretend like we're preserving any "natural" state of affairs ( https://xkcd.com/1338/ !). The best we can hope for at this point, barring a severe and abrupt population decline, is to continue engineering our way out of this mess as best we can, with the long-term interests of humans at the forefront.
#/media/File:Chabris2007a.svg){kind=link}
I suspect if you made a perfectly invisible ball, you'd find people would manage to hit it back quite often. There's a lot of signal from the opponent's body motion.