Hacker News Reader

Comment by romaaeterna | Hacker News Reader

fMRI results show that children from lower-SES backgrounds are less stimulated by monetary rewards.

"The findings suggest that lower SES circumstances may prompt the brain to adapt to the environment..."

Wow! How did they control for the possibility that a heritable reward-focus in the parents leads to the SES results? Or that other factors, such as higher engagement or game-competence, might have led to this result? The story doesn't say, but if someone could post the relevant discussion from the journal article, I'd be interested in seeing it.

I'm sure it's there.

View on HN

They obviously didn't, and didn't need to - that sentence is not making any assertion, it is suggesting a hypothesis that can be tested in future work by them or others. The sole assertion this study is making is about the existence of the effect, and it is only speculating (and not hiding that it's speculating) about a possible mechanism causing that effect.

And obviously further details relating to this observation aren't already tested and published, it would take at least a few months for that to happen - if someone had already done and published a follow-up analysis, this article wouldn't be news but be outdated already.

And, indeed, makes for some rather interesting follow up questions...

* Is this genuinely just a different way of processing rewards?

* Do low SES children not understand that money is a reward?

* Is there a capability gap at playing games? Or different responses to games themselves?

I don't think it makes intuitive sense that limited rewards leads to a lower response, so my guess is low SES children probably have a different relationship with money. My observation has been, in a weird but important way, that only a subset of high SES families seem to treat money as a desirable thing. Most people seem to have a what I see as a complex love-hate-contempt relationship with money. There isn't conclusive evidence in the article though.

> Do low SES children not understand that money is a reward?

Maybe they understand that the money will be soon taken away by the adults.

Either literally taken away, or simply the next time the kids need something, they will be told "now that you have money, you can buy it for yourself" (effectively taxing the income at ~ 100% rate).

"* Do low SES children not understand that money is a reward?"

As in they're stupid? How would they not understand it's a reward.

"* Is there a capability gap at playing games? Or different responses to games themselves?"

From the article - "Unbeknownst to the participants, the game was set up to control whether the guess would be correct or incorrect."

So no... that's not a possibility

> As in they're stupid? How would they not understand it's a reward.

A lot of people, if you give them a dollar they'll try to spend it on something. And that, counterintuitively, means they don't see the dollar as a reward. Someone who sees the dollar as a reward will put the dollar on their pile of dollars and then feel good because they have a dollar. The dollar, in itself, is the reward.

If someone is dirt poor, I assume they don't have a habit of seeing money as the end in itself.

"A lot of people, if you give them a dollar they'll try to spend it on something. And that, counterintuitively, means they don't see the dollar as a reward."

I think Homer already deconstructed your argument, no not that Homer: https://www.youtube.com/watch?v=A81DYZh6KaQ

But the question is, how can you attribute that behavior to the poor kids only if there is no mention of it in the study or article? Undoubtably some number of both the poor and rich kids are going to want to buy something with the money. Buried in your assumption is the assumption that poor kids are less likely to try to save money or that they don't see financial stability as rewarding.

And, in my opinion, the fact that the reward center of all the kids brains lit up fundamentally contradicts your central point. The poor kids did see the money as a reward for correctly guessing the answer in the game.

> But the question is, how can you ...

I can't. That was the hidden meaning behind the "interesting follow up questions" and "There isn't conclusive evidence in the article though." parts of my original comment.

And yes, that Homer clip is exactly the sort of flow I'm thinking about. The idea obviously isn't a new one. Compare it to a similar style of scene but with Mr Burns - he'd see the money and start comically salivating or something. Homer has a 10 second delay before even realising it is a good thing, just because he doesn't see money in itself as a goal.

If the dollar bill itself is the reward, it is a worthless reward. I may as well give you a piece of paper that says "reward" on it. The dollar is a reward because it can be exchanged for something.

I agree that that sentence didn't, but further down in the article stronger claims are made, sometimes obliquely referencing prior work. So it would be interesting to see the original paper to see what they cite as supporting evidence for this hypothesis.

Obviously this is something that can be controlled somewhat with the appropriate selection of subjects (there are a variety of families that have experienced generational rags-to-riches or riches-to-rags, for example). Though the N in those cases is likely to be smaller.

I think a problem is that this press release is mixing content from the scientific study with an interview of experts talking about it, and the colloquial assertions made in that conversation happen to have a bunch of extra speculation that isn't part of the study they discuss; for half of the press release the "standard of evidence" is like in science and for the other half it is like in this HN discussion thread, where people do rise many interesting (and possibly true) hypotheses where we don't have any evidence (yet).

I agree.

A nod's as good as a wink to a blind bat, say no more, say no more!

Most of an individual's intelligence is inherited from their parents, and that's been proven multiple times with large-scale studies of adopted twins. It wouldn't surprise me if reward response was similar.

But it's unlikely we'll make any advancements in our understanding of genes and intelligence or behavior, because this kind of research is out-of-favor and almost impossible to fund now.

Unfortunately the field of observational statistics called "psychometrics" which has "proved" this using "twin studies" is largely bunk.

The premise of those studies is linear additive effects from each causal factor, which is almost certainly false; and in general, there is no way of establishing causation without interventional experiments. If the effects were plausibly non-linear, we would need 1 trillion+ data points even to establish statistically significant differences.

These fields are all cargo-cult stats, putting data into GUIs and printing pro-forma reports. The people involved are in way over their heads, to the degree that there isnt anything to "reproduce" let alone try to reproduce the experiments: they dont have any.

The whole thing is little more than principle component analysis on astrological star data.

> The whole thing is little more than principle component analysis on astrological star data.

When and where we are born definitely has an effect on personality, irrespective of family, in most cases. Astrological star data is a proxy for the when and where, and while thus would not be causative, would often be correlative. So it makes sense to do PCA on it as you'll identify a lot of good correlations.

You are correct that star positions are proxies. The problem in the case of psychometrics is that the claim is that the stars (here: genes) arent proxies, but bonafide causes.

So in the case of astrological analysis we would, if we wished to change population personalities change, say where people were born -- in the case of psychometrics we're told that we have no control over the causes whatsoever (genes).

In the vast vast majority of observational genetic studies genes are just proxies, in the the same way the star signs are. In other cases, they're mere correlations.

In the case of IQ (and basically all social genetic studies) it is highly likely that any correlated genes are merely proxies for geographic-historical distributions of peoples over time and it was the long-time-scale geographic and historical effects (culture, education, wealth, famine, poverty, war...) that we see in attainment.

(Setting aside that IQ tests aren't valid above average IQ and have extremely bad properties both at the individual and population level for most of the measured range. )

We do know that genes are a cause of IQ (we can see this comparing our nearest non-Homo relatives. But yes, we are far away from saying which genes do cause it, much less how they cause it, and how they interact with the environment and other genes to cause it.

IQ is a test, it isnt intelligence and isnt even a measure of intelligence. It is valid only for establishing certain sorts of mental retardation in a human population.

To say that "genes cause IQ" is several levels of confusion. The claim has nothing to do with whether genes determine the physiological body structure, the claim is whether relevant aspects of human intelligence are a product of the at-birth or developmental structure -- or of that acquired from the environemnt, ie., neural plasticity / brain structure / etc.

Almost all evidence we have is that relevant aspects of human intelligence are environmental, ie., acquired. There is little evidence of the alternative because the experiments needed to establish it are large impossible or unethical. So it might be true, but we have no evidence of it.

Insofar as any claims are made about genes and IQ they amount to associating markers of geographical-historical migration (ie., breeding across time) with a very bad proxy measure of abstract reasoning which, at best, only detects deficits with any reliabiliy.

So we have proxies on top of correlations on top of terrible measures of anything with zero reliability in the range of interest. This whole thing is operating nearly at fraudulent levels of "research".

There's very very little this field can say scientifically, since there arent any experiments. It's all pseudoscientific uses of stats.

To simplify matters, suppose that we know for sure that all above-average differences in IQ are 100% environmental/acquired-during-lifetime. Now, so what do twin studies show, on this assumption?

They show that genes must be correlating with these environmental measures, and twin studies give us an 'effect size' for geographical-historical patterns of reporduction.

So twin studies, even if they were valid (and their assumption of linearity renders them invalid), do not decide the matter. You need indpednent evidence of causal mechanisms from experiments, which doesnt exist.

> IQ is a test, it isnt intelligence and isnt even a measure of intelligence.

Are you sure about that? If I find someone that we all agree is “pretty damn smart” and they regularly score 115+ and we find someone that we all agree is “not very smart at all” and they regularly score < 90. And we repeat that dozens of times, it seems like the IQ test is a measure of intelligence.

It may be imperfect, and pretty clearly does not have three significant digits of precision, but it seems it’s still a measure of the thing we’re trying to estimate, same as a set of mixed stones and a playground seesaw can provide a comparative measure of weight.

Er, sure, if you can define some objective measures of performance that are taken to be symptomatic of intelligence and correlate those with IQ test results -- and then show test-retest validity across a wide population (etc. etc.), then you'd have a case.

But all attempts to do this, i'm aware of, show in fact IQ tests don't behave this way. That all correlation with objective measures of performs, with reliable measures, end up random >100 -- and weakly non-random <100.

So what we, in fact find, is that IQ tests only reliably correlate in the 'mental retardation' region of performance.

There is, possibly, a sort of non-objective culturally feeling to the kinds of people who might do well on IQ tests. Call these personalities, "puzzler bureaucrats" it isnt clear that even for people in this population we're percing any sort of objective intelligence-related performance.

Eg., suppose we had an objective measure of intelligence based on a deep causal model of the brain and how it interacts with body/environment. Suppose we found out that a population who score 100-110 on IQ tests were vastly more intelligent than another scoring 120-130.

Now, could we find independent measures of performance to substantiate that? Yes, trivially. Take, eg., a group of CEOs and a group of maths undergrads. The former, could by supposition, have a much higher performance on a variety of objective measures.

> So what we, in fact find, is that IQ tests only reliably correlate in the 'mental retardation' region of performance.

Even assuming we agree that that’s the only usefulness of an IQ test, in what way does that make them not a measure of intelligence?

A ruler is still a measure of length even if it can’t measure the size of an atom or the distance to the moon.

It's more like a binary classifier: retarded, not-retarded.

I don't understand why you're presupposing "performance" as a measure of intelligence, and what you mean by it. No one is saying that intelligence alone (above a threshold) equates to demonstrated, or even potential, life outcomes.

then your definitions are circular, "intelligence is whatever the IQ test measures, assuming it measures anything"

this is no way to establish validity or reliability. you need an independent theory of intelligence and a variety of indepenent objective measures -- otherwise the whole enterprise is circular

Again, I wrote that I don't understand what you mean by "performance", so you can't claim a circular definition yet. In the context of intelligence or IQ the word "performance" means to me either demonstration of something (and I don't understand what you expect CEOs to be demonstrating other than attaining a particular social position), or the "performance score" of the Wechler IQ test.

> you need an independent theory of intelligence

Sure. We have some of these. Despite it's flaws and limitations I'm partial to portions of Guilford's Structure of Intellect.

> and a variety of indepenent objective measures

The various sub-tests in an IQ test are nominally independent, yes? Going back to your "CEOs", I don't see how a social position can be assumed to be objective when it is subjectively contingent on external social processes.

> Almost all evidence we have is that relevant aspects of human intelligence are environmental, ie., acquired. There is little evidence of the alternative because the experiments needed to establish it are large impossible or unethical.

The experiment raising an infant chimpanzee alongside a human infant has been done. To presuppose that human intelligent is categorically distinct from chimpanzee intelligence such that this experiment does not lend evidence to a genetic basis of variations in human intelligence is further than I would go.

I presume that some amounts of variation in intelligence, as well as IQ scores, is due to factors that are neither directly genetic nor environmental, but are more random (such as cell division and post-fertilization DNA replication and rearrangements).

> Unfortunately the field of observational statistics called "psychometrics" which has "proved" this using "twin studies" is largely bunk

Do you have a credible source for this claim? The criticism I'm seeing is largely constrained [1].

[1] https://en.wikipedia.org/wiki/Twin_study#Criticism

> The premise of those studies is linear additive effects from each causal factor

I don't understand this. Why would you need linear additive effects to think that if identical twins are more similar than fraternal twins and half siblings are less similar than full siblings along some axis then there is probably a genetic component?

Well, I think the issue isn't the research exactly, it's the motivations behind doing the research.

Like, if it was determined tomorrow, with a huge article in nature or another journal that's suitably popular/prestigious that intelligence (and rewards for the sake of this thread) is totally inherited or even 80% inherited, what material changes would you want/expect to see in society in relation to that news.

> if it was determined tomorrow...that intelligence (and rewards for the sake of this thread) is totally inherited or even 80% inherited

IQ is 50 to 80% inherited [1]. (We can only predict about 10% of it with genetics [2].)

[1] https://en.wikipedia.org/wiki/Heritability_of_IQ

[2] https://www.nature.com/articles/nrg.2017.104

For whatever it's worth, since you cited Plomin, if you Google him (or that study) you'll get a lot of interesting writing about how not-especially-useful that result turns out to be. A good shot/chaser on polygenetic behavior indicators is Shalizi's "g" writing, which is mostly about the distinction between exploratory and confirmatory data analysis, which seems pertinent.

Because you end up with bad science, like (potentially) the study posted here. If you refuse to consider the role of genetics, then everything is going to look like environmental impacts.

This has a real impact on policy and investment, because we waste time on money on programs that have no chance of success.

Trying to remain ignorant isn't going to help us here, even if we have good intentions.

> because we waste time on money on programs that have no chance of success.

And presumably learn from it. While wasting a lot of money and effort (and people) on programs that won't work isn't good, neither is not trying because the effect is presumed, pre hoc, to be small or non-existent.

We live in a world where most people aren't enabled to achieve their potential. Figuring out that extrinsic rewards don't work to motivate behavioral changes or achievement for some people of a certain category, regardless of why, will hopefully result in the testing of other non-extrinsic-reward interventions that may work to motivate.

> And presumably learn from it.

While it might be easier than for regular people, it's still extremely hard for a politician/academic or political faction/academic faction to admit error. It's almost always possible to come up with a reason for why something didn't work that doesn't require giving up on a centrally important idea.

As far as I can tell, the reason why science is continually trying to prove black people inferior (let's not bury the lede) is in order to put the current condition of black people in America as a result of nature and not of hundreds of years of slavery and slave breeding, and of the current condition of black people in Africa to their inferior brains instead of the violent takeovers of their countries, mass enslavement and massacres, and the installation of friendly dictators to ship their resource wealth overseas.

Normally you look for deficits in people in order to account for them with special help, but this is a science just looking to shed blame. It's like testing for dyslexia to find kids that you shouldn't bother trying to teach to read. Or even worse, retroactively justifying your decision not to teach redheaded kids how to read by doing a survey of redheaded kids who were never taught to read, over a dozen generations, and finding them stupid.

That's the problem. The field is hopelessly tainted, and being an area that would be difficult under any circumstances (due to mass social effects rather than mass genetic effects), meaningful designs would be hard to find even if the researchers were motivated by something other than justifying the past.

In my experience, the people most obsessed with arguing and proving the usefulness of IQ are somewhere between 110-120. The people obsessed with the heritability of IQ and the supposed dominance of nature over nurture are often people who have raised terrible children.

aside: one thing I'm curious about and haven't found is that there are a number of researchers who show differences between American black achievement (slave descendants) and black immigrants from Africa and the Caribbean (who started arriving in 1965.) I haven't seen a lot of race science speculating about that: did slavers (either Western or the Africans who sold them) have some sort of filter that only put the stupid blacks into slavery? Were all the smart ones culled, and/or their intelligence intentionally bred out of them over generations? Did the admixture of the genetics of so many white slavers into their slave stock lower slave intelligence, or raise their propensity to violence? Or does that entire line of inquiry (or intense study of slave breeding in general) hold no interest for people trying to justify 18th century race science?

The Caribbean has a history of slavery not unlike the U.S., so the fact that Caribbean immigrants are better off is genuinely interesting. Of course, institutional factors are the most likely difference; among other things, the British got rid of slavery a generation or so before the U.S. did, and there was no real equivalent there to the U.S. Civil War and failed Reconstruction. There are also cultural differences that will of course matter, but they're causally downstream of that basic stylized fact.

Thank you for the comment. Everyone else is talking around the issue but you cut right to it, and did so very cleanly.

> what material changes would you want/expect to see in society in relation to that news.

Governments offering free embryo selection plausibly.

> Most of an individual's intelligence is inherited from their parents

Note that heritability numbers are specific to a particular environment. E.g. if you have a more random environment with some kids exposed to lead poisoning, some to malnourishment and some to parasites you will get lower heritability figures.

Most studies on this are conducted in countries like Sweden and the USA on middle class populations.

This was my thought too, ofc. it's not a popular idea these days that genetics and heredity might have some bearing on success, but it does seem worthwhile looking into..

Anecdote time, I identified the reward-seeking behaviour in my peers in early childhood and didn't understand it, I felt they were acting like dogs, being overtly manipulated to do what the grownups wanted, and I felt disgusted with the idea. (I was not an easy child, but then again, I grew up to be a rather disagreeable adult)

I come from a fairly normal lower middle-class family, but was consistently rewarded, except my response to praise and reward was mostly negative, and to some degree still is. I'm pretty sure my father was the same way, he'd do stuff for their own sake and his own sake, not because someone rewarded him for it. Unsurprisingly, taking into account how much potential I've been told I have, I'm a relatively low achiever, I'm simply not attracted to reward or winning very much, except maybe winning for its own sake, which just seems dumb.

Looking into? Sure! Claiming that it's already established science when it absolutely isn't however…

We know for a fact that genetics has a large influence on intelligence.

We know for a fact that genetics has a large influence on personality.

We know for a fact that (at scale*) intelligence is the single largest influence on success.

IIRC, we're pretty sure that (at scale*) personality is the second largest influence on success.

I.e., if you measure all the possible things that might influence success and run a principal component analysis, intelligence is the first component and personality the second.

* Nepotism and such exist, but extreme cases are actually not that common.

It should not be controversial to acknowledge both part A and part B at the same time.

No we don't know for a fact any of these things.

Perhaps you don't, but dismissing an entire field of science...

Well, if you can make a strong statement without any supporting argument, so can I.

>the possibility that a heritable reward-focus in the parents leads to the SES results

This would imply that the problem with poor people lies within themselves rather than the society, which is an unacceptable proposition.

> which is an unacceptable proposition

Why?

Well first of all its circular logic.

They're dumb because they're poor <-> They're poor because they're dumb.

Setting aside my moral issues with the idea of Economic Phrenology I think this is an acceptable checkpoint to stop at when trying to reason out an answer. Feel free to stop, look around stretch out your legs. And then get back on that train of thought and take it to the end. Why are they poor?

There's the circular logic argument of "well cuz they're dumb" but that ignores generational wealth, the gender wage gap, systemic inequality, luck, economics, and just a general lack of equity of opportunity.

Some people smart people choose low paying jobs because it brings them a fulfillment money cant. Some dumb people get rich off the lottery.

And like... We've all seen rich peoples kids. They're not all geniuses.

It's just not a coherent belief. It is demonstrably not true.

"Dumb" is just a projection. It takes a lot of checked boxes to become successful. Yes, one of them just may be a hereditary propensity to value reward more. It's not an incoherent belief any more than that some people could never play in NBA, it just takes another set of boxes, some of them hereditary.

I remember high school, coach organized a wrestling match played out over a week. You had 32 matches. The winners competed in 16 matches. The winners of that competed in 8 matches. Then 4 matches, then 2 matches then the final match.

When you've built an economic system that distributes wealth that way you can't cast moral aspersions on the losers. Because that's what it does, creates losers by design.

There's repechage, and there's plenty of repechage in life. At the same time, yes life is not fair, and that's not really a property of the economic system anyone built.

Reply on news.ycombinator.com