"Gaps between epigenetic age and chronological age have been linked to health problems in the past, but again it's too soon to draw those kind of conclusions: the scientists readily admit they don't yet know how this will affect the kids later in life."
And this is the crux of the matter.
I don't doubt that children who are being cuddles more could have their genetics changed. Whether it actually has an effect on their prospects later in life I highly doubt.
Historically, mothers were blamed for causing their kids to have autism by being "cold" towards them. "Refrigerator mom's" was one name for them.
With modern insight it seems more likely that moms somewhere on the autism spectrum themselves were being blamed because they seemed odd to neurotypical medics.
I have read about the refrigerator mom concept, but didn't make the link that these mothers could have potentially been autistic themselves. On this note, you are probably aware of the recent attention being paid to the diagnosis of women and girls with autism. Until recently the common conception was that males were more likely to have autism because most of the early research was done on autistic boys! And, in case you're curious to learn more and haven't heard, I've linked to an article from Scientific American (below) which reviews some newish research suggesting that females with autism have brains more closely resembling neurotypical males (and the interesting "protected sex" hypothesis, dealing with how testosterone influences the development of autism).
My dad is likely autistic, so there's the link for me. I don't know him at all, but genetics are a whacky little code system and I've magically taken a liking to electronics and computers with the same degree of passion that he supposedly applies to his work with ham radio! I was raised in a highly artistic community, and although I have developed a decent creative ability and do art as a hobby, my career path is in maths and computer science. Interesting!
> I don't doubt that children who are being cuddles more could have their genetics changed.
Looks like actual experts in the field disagree with you. [1]
Also, deny a child affection and you have a mentally ill adult. Sadly, there is now an entire generation of parents who have been advertised things like the "cry-out" method. [2]
Just because someone has done something for generations doesn't make it right. They also send children to kindergarten by the 3rd month, for economic reasons.
Having traveled a bit in Northern Africa, I noticed baby sleep problems are virtually non-existent there. They do what humans have done for eons: co-sleeping and breastfeeding. They do it so naturally, mothers are never fully awake while they breastfeed during the night (wife had to sleep in a room full of women and children on several occasions).
Biologically, this makes a lot more sense than giving the signal to the most immaturely born placental mammal that she can't rely on the mother's care.
Wait are we talking morals or are we talking genetic and statistics?
Denmark sends their kids to nursery school early too and because many women work full time there it means most children aren't with their parent's for a long duration of the day (used ot be much earlier).
What you think about it morally isn't interesting to me. Unless you can prove it's somehow making them worse of then my original position is intact.
Maybe I'm reading it wrong but i don't think parent is raising a morality issue. I think the point was parent's experience with cosleeping having positive effects re baby sleep. I'm not saying this is scientifically valid, btw.
However, saying "unless you can prove it" shows ignorance. Not everything is known. Lack of proof does not indicate a hypothesis' invalidity. and I'm not aware of solid proof against it either.
I think you might have misinterpreted my comment as an argument about moral.
Leaving babies to cry or separating them from their mothers for extended periods of time is primarily a health issue which I believe should concern society. [1][2][3] Unfortunately, it fits our economic model very well -- so there is that.
> Sadly, there is now an entire generation of parents who have been advertised things like the "cry-out" method.
For sleep training? It’s a somewhat rough 2-7 nights when they’re in the 8-12 week age range, or way, way worse later and a ton of bad nights until then. Probably some kids don’t take to it, but that’s true of anything you try. We’re 3-for-3 on it, with the worst taking three nights. A+++++, would not accustom my kids to someone coming in and rescuing them from learning how to chain REM cycles every time they wimper again. Only way I’ve seen it go wrong with ours or with friends’ kids is 1) inconsistency on the part of the parents, or 2) illness getting in the way.
Yeah but that's what I am doubting what even means to be honest. Premature kids behave very differently and many of them are void of contact in the beginning.
Bzzt. Nope. Epigenetics is heritable as are child-rearing practices and this is a cross-sectional design, not longitudinal. (Not even within-family comparisons!) For all they know, those epigenetic differences were present from birth, as is often the case. Yet another epigenetics study whose overhyped causal claims are dead on arrival.
Handily dismissed by a hypothesis conceived in a moment with zero effort.
Between the cell's nucleus, where complex enzymes transcribe to RNA in a chemical process, resulting in gene expression, that concludes with the construction of proteins, your claim is that high stress at early developmental moments, not only has no long term effect, but no effect at all?
It sounds like you've conducted a study. Is your sample size N=1 where 1 equals you?
This person is not claiming to know a cause: this person is claiming that there is another hypothesis that you can conceive "in a moment with zero effort" that this study design was incpaable of isolating.
There are multiple reasons that could have happened: one is that the scientists didn't think of it (which to me would undermine their credibility as it is really obvious as a possibility), a second is that this study was much cheaper and they weren't sure there would be a correlation at all (so now there is enough evidence to warrant a better study), and a third is that they didn't care about "why" and were only interested in "that".
To the scientist's credit, the article quotes them as saying that they don't know the cause, which is great, and they say "too early" so they might intend to keep going. To their discredit, they gave enough of an explanation based on stuff they weren't sure of to the reporter to guide it to the point where the headline is just guesswork, not real science.
On the other hand, you certainly... and apparently the scientists :/... are simply assuming your explanation and then using a study that is incapable of discerning the the difference between these two hypotheses to dismiss the idea, which is clearly not OK.
[I am sorry if someone saw a draft of this comment or if it appeared under the wrong parent for a moment: I was having issues replying on my iPhone today :/.]
> Handily dismissed by a hypothesis conceived in a moment with zero effort.
Not at all. What I am saying is a standard criticism of overhyped epigenetics studies which follows from the lazy design which does not control for very large known confounds (pretty much all effects are smaller in a within-family comparison) or actually demonstrate it's not reverse causation, which as in OP results in such wild overhyping of causal implications. This is as much 'zero effort' as noting, of the latest nutrition study claiming that fat (or carbs, or red meat, or protein, or milk, or eggs, or...) will definitely kill you, that the methods simply do not work for what they are claimed to prove. If you find the criticisms tiresomely cliche & predictable, that's an indictment of the researchers or funders, not the criticisms. (I know I find it tiresome to point out the same problems every time and respond to comments shocked that correlation!=causation, or that most human traits are heritable, or that p-values mean nothing interesting, estimates are inflated by publication bias, one shouldn't control for intermediate variables, gains fade out etc. I suppose I do it out of stubborn perversity. I keep resolving not to get into arguments about these things, and keep failing.)
It has been demonstrated many times that epigenetics are heritable, affected by gene-environment correlations, and are not themselves causal for many outcomes they have been claimed to be but passive indicators of environment; so OP is, at best, possible but not probable and the speculation that the epigenetics changes will have any causal effects later on to be taken with considerable skepticism... (as they are more likely to simply indicate the manifestation of the childrens' genes as inherited from their parents). Epigenetics is of tremendous biological interest as it is often how genetics exerts its effects, but unfortunately, that's not what people want epigenetics to be. You may remember the scientific reaction to Siddhartha Mukherjee's New Yorker article on epigenetics last year which similarly took likely-confounded & reverse causation epigenetics claims at face value.
My thought was that there are lots of variations and they just found a coincidence that will require a followup experiment to replicate and rule out chance. Your points are valid as well - if you're claiming a change, you gotta show the before and after.
While you raise valid limitations of the study, those limitations do not provide a valid cuse to dismiss the result outright. Random assignment to different levels of "cuddles" in humans is probably not ethical,and longitudinal studies are expensive. To get a grant funding a longitudinal study you usually have to have a) animal model based evidence from the literature b)have already demonstrated in humans the phenomenon cross-sectionally. As a side note, your comment would have been perfectly reasonable sounding if you had left off the first two words, and the last sentence. But you didn't.
None of that is true. We have a long history of epigenetics studies which let us known this is almost certainly yet another failure. Random assignment is not unethical under many designs (exercise for the reader: think of one. Hint: draw inspiration from economics), is not necessary to improve over what they did, and longitudinal designs at their simplest merely require two sampling periods. It is also untrue that you 'usually' have to have demonstrated it in animals or cross-sectionally because many longitudinal studies are convenience samples, record-based or population registries, or collect large numbers of measurements to see what comes out in later analyses of things they hadn't thought of before.
Clearly I was not claiming that random assignment is un ethical generally, but specifically in the case of assigning the degree of physical affection/cuddling to a baby is probably unethical.
As someone who has been involved in multiple longitudinal studies, although not in epigenetics, I can say that they are more expensive.
In terms of what I said about prior evidence, I was coming from a cognitive psych background where we do not typically deal with record-based data. We deal with participants directly. It is expensive and time consuming. Longitudinal studies usually follow cross-sectional evidence.
> but specifically in the case of assigning the degree of physical affection/cuddling to a baby is probably unethical.
Alright, since you refuse to think, I'll give it to you for free: do an intent-to-treat design with incentives for additional cuddling for either the parents or hired helpers like nannies, or passive controls; as no infant receives less physical affection than they would otherwise, no one is hurt by the randomization and it is not unethical, but you still get a causal effect. (Not all of them will use it, and the ones who do, the difference will be smaller than if you could remove cuddling, but that's merely a matter of power.) Such designs are often used in economics, and are applicable to many topics where randomization is considered infeasible for 'ethical' reasons: if you have a limited supply of X, it can be allocated by lottery; if you want to measure the impact of health insurance, surely this is unethical? How could you ever study it? But no, you have limited Medicare funds, so you simply allocate by lottery. Charter schools, lead poisoning, vaccines, basic incomes, daycare, enrichment programs, cuddle subsidies etc. The winners get the additional treatment while the losers serve as baseline controls, finessing the ethical problems. As it is usually the case that there's not enough money to deliver the claimed-beneficial intervention to all possible beneficiaries, it's quite a general strategy.
> As someone who has been involved in multiple longitudinal studies, although not in epigenetics, I can say that they are more expensive.
I didn't say that they weren't, just that they were only somewhat more expensive, and they can be done cheaply in a number of ways. There are considerable economies of scale in these things - look at the UK Biobank.
First, you may not be aware of yourself sufficiently, but you are being a real asshole. Second, you are correct that I didn't consider that design. This may be because it is a little stupid, or at least impractical without the right target population. For example, if I wanted to study the effects of malnutrition it would make sense to do random assignment to individuals to food and food+extrafood conditions, but only if my sample contained a number of malnourished kids. If I took my sample from middle class America, and I assigned them to food and food+plusmorefood conditions (via force feed I guess?), I'd be studying obesity more likely than malnutrition. Levels matter. I'd guess that the primary interest in the present research is the effect of not getting physical affection as an infant. The effect between some and more affection, is probably qualitatively different than none and some affection.
> First, you may not be aware of yourself sufficiently, but you are being a real asshole.
Uh huh.
> Second, you are correct that I didn't consider that design. This may be because it is a little stupid, or at least impractical without the right target population.
It's neither stupid nor impractical, and I gave several examples, many of which could or have been done in the US. Even healthcare can be randomized in the US this way - consider the Oregon Medicaid or the RAND healthcare experiment. (Both of them, BTW, deliver the profoundly surprising result that correlation!=causation and additional healthcare does little for health.) As it is a priori unlikely that all US children are above some hypothetical threshold for cuddling - in which case OP would be even more dead on arrival that it looks - the design would be perfectly adequate to investigate the causal effects of cuddling. One can screen for low cuddling levels if need be to increase power.
> but only if my sample contained a number of malnourished kids. If I took my sample from middle class America, and I assigned them to food and food+plusmorefood conditions (via force feed I guess?), I'd be studying obesity more likely than malnutrition.
This is a bad example as middle class American kids are not 100% micronutrient replete. Iron deficiency/anemia is common, and even iodine deficiency is actually quite common according to the urinary surveys. (If you look, you'll find at least two iodine RCTs claiming noticeable cognitive benefits in New Zealand college students, a country typically considered better off than the US on public health.) Vitamin D is another question mark - the official RDAs are very low but they're heavily criticized and if you accept some of the higher recommendations, then middle class American kids are quite likely to be deficient (all that time indoors).
> The effect between some and more affection, is probably qualitatively different than none and some affection.
That would be very unfortunate for OP, then, as I doubt any of the kids in their study had literally zero affection, as they were not recruiting from Cold War Romanian orphanages.
You have brought up some points I will ponder. Thank you.
That said, your approach to discourse seems to be: argue everything, concede nothing. Scientists of this sort (and spouses for that matter) are the most frustrating people to deal with. It would have been helpful to the discourse, for example, if you could have conceded the obvious fact that some hypotheses should be tested at the level where the effect of interest lay; that levels matter not just quantitatively, but qualitatively. By conceding something as minor as this you could go a long way to becoming a better communicator of your ideas.
We appreciate your many valuable contributions to HN, but you have to follow the civility rules here like anyone else, and have stepped badly out of line in this thread. Please don't do that.
I don't believe I was trying to make that case. My point was that no single design is perfect, and that research is an iterative endeavor, where initial (easy) evidence is collected prior to collecting expensive and time consuming but more definitive evidence.
Epigenetic tags are inheritable, yes, but when they're expressed and to what extent the epigenetic tags are expressed changes between individuals.
Since expression of these tags result from external influences, I don't see what the relevance is. Can you explain? How does inheritance of these epigenetic tags, unexpressed, contradict the article?
I have understood that epigenetics refers to heritable changes in gene function without changes in the genome. Change in gene expression even if it last the rest of the life of an individual not epigenetic change. But even university press release says otherwise.
You are missing nothing, its just that modern molecular biology has somewhat changed in what it regards as epigentics. Now, we usually refer to any somewhat lasting changes (a few cell cycles, sometimes just for a cell phase) to DNA as epigentic. This change in meaning has happend (as far as I understand it) because some of the somewhat traditional epigentic marks (like DNA methylation) have been found to play an even bigger role in relatively short time processes. Now almost all base modifications of DNA (and the bound nucleosomes) are regarded as epigenetic (in the sense of temporal genetic changes) even though the appearing changes are not really short term inheritance reasoned anymore but mere functional changes.
And now, the whole story is becoming even more confusing because those base modifications also happend in RNA (transcribed DNA) which lead to the term epitranscriptomics, which to be honstest, has nothing to do anymore with inheritance, it is just about how a RNA molecule interacts with other cell processes.
Wouldn't be surprised if it changes the genetics of the cuddling parents too, it's just that the researchers didn't think of it yet. All those hormones that are released are going to have an impact.
If by "good" you mean improved genetic fitness (likelihood of reproduction), then there is not going to be enough data to support such a claim, since this would require tracking and comparing the reproductive outcomes of hundreds of individuals both cuddled and not.
53 comments
[ 3.7 ms ] story [ 51.5 ms ] threadAnd this is the crux of the matter.
I don't doubt that children who are being cuddles more could have their genetics changed. Whether it actually has an effect on their prospects later in life I highly doubt.
¯\_(ツ)_/¯
With modern insight it seems more likely that moms somewhere on the autism spectrum themselves were being blamed because they seemed odd to neurotypical medics.
https://en.m.wikipedia.org/wiki/Refrigerator_mother_theory
My dad is likely autistic, so there's the link for me. I don't know him at all, but genetics are a whacky little code system and I've magically taken a liking to electronics and computers with the same degree of passion that he supposedly applies to his work with ham radio! I was raised in a highly artistic community, and although I have developed a decent creative ability and do art as a hobby, my career path is in maths and computer science. Interesting!
https://www.scientificamerican.com/article/autism-it-s-diffe...
"Early-Experiences-Can-Alter-Gene-Expression-and-Affect-Long-Term-Development.pdf"
http://developingchild.harvard.edu/wp-content/uploads/2010/0...
Looks like actual experts in the field disagree with you. [1]
Also, deny a child affection and you have a mentally ill adult. Sadly, there is now an entire generation of parents who have been advertised things like the "cry-out" method. [2]
[1] http://developingchild.harvard.edu/wp-content/uploads/2010/0...
[2] https://www.goodreads.com/book/show/192222.Solve_Your_Child_...
Having traveled a bit in Northern Africa, I noticed baby sleep problems are virtually non-existent there. They do what humans have done for eons: co-sleeping and breastfeeding. They do it so naturally, mothers are never fully awake while they breastfeed during the night (wife had to sleep in a room full of women and children on several occasions).
Biologically, this makes a lot more sense than giving the signal to the most immaturely born placental mammal that she can't rely on the mother's care.
Denmark sends their kids to nursery school early too and because many women work full time there it means most children aren't with their parent's for a long duration of the day (used ot be much earlier).
What you think about it morally isn't interesting to me. Unless you can prove it's somehow making them worse of then my original position is intact.
However, saying "unless you can prove it" shows ignorance. Not everything is known. Lack of proof does not indicate a hypothesis' invalidity. and I'm not aware of solid proof against it either.
Leaving babies to cry or separating them from their mothers for extended periods of time is primarily a health issue which I believe should concern society. [1][2][3] Unfortunately, it fits our economic model very well -- so there is that.
[1] http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.432...
[2] https://www.cambridge.org/core/journals/development-and-psyc...
[3] http://champagnelab.psych.columbia.edu/docs/champ11.pdf
If you want it to concern society perhaps first you need to make clear what it is you are trying to avoid.
I don't see any evidence which proves some larger societal problem.
For sleep training? It’s a somewhat rough 2-7 nights when they’re in the 8-12 week age range, or way, way worse later and a ton of bad nights until then. Probably some kids don’t take to it, but that’s true of anything you try. We’re 3-for-3 on it, with the worst taking three nights. A+++++, would not accustom my kids to someone coming in and rescuing them from learning how to chain REM cycles every time they wimper again. Only way I’ve seen it go wrong with ours or with friends’ kids is 1) inconsistency on the part of the parents, or 2) illness getting in the way.
"In children, we think slower epigenetic ageing could reflect less favourable developmental progress"
Between the cell's nucleus, where complex enzymes transcribe to RNA in a chemical process, resulting in gene expression, that concludes with the construction of proteins, your claim is that high stress at early developmental moments, not only has no long term effect, but no effect at all?
It sounds like you've conducted a study. Is your sample size N=1 where 1 equals you?
There are multiple reasons that could have happened: one is that the scientists didn't think of it (which to me would undermine their credibility as it is really obvious as a possibility), a second is that this study was much cheaper and they weren't sure there would be a correlation at all (so now there is enough evidence to warrant a better study), and a third is that they didn't care about "why" and were only interested in "that".
To the scientist's credit, the article quotes them as saying that they don't know the cause, which is great, and they say "too early" so they might intend to keep going. To their discredit, they gave enough of an explanation based on stuff they weren't sure of to the reporter to guide it to the point where the headline is just guesswork, not real science.
On the other hand, you certainly... and apparently the scientists :/... are simply assuming your explanation and then using a study that is incapable of discerning the the difference between these two hypotheses to dismiss the idea, which is clearly not OK.
[I am sorry if someone saw a draft of this comment or if it appeared under the wrong parent for a moment: I was having issues replying on my iPhone today :/.]
Not at all. What I am saying is a standard criticism of overhyped epigenetics studies which follows from the lazy design which does not control for very large known confounds (pretty much all effects are smaller in a within-family comparison) or actually demonstrate it's not reverse causation, which as in OP results in such wild overhyping of causal implications. This is as much 'zero effort' as noting, of the latest nutrition study claiming that fat (or carbs, or red meat, or protein, or milk, or eggs, or...) will definitely kill you, that the methods simply do not work for what they are claimed to prove. If you find the criticisms tiresomely cliche & predictable, that's an indictment of the researchers or funders, not the criticisms. (I know I find it tiresome to point out the same problems every time and respond to comments shocked that correlation!=causation, or that most human traits are heritable, or that p-values mean nothing interesting, estimates are inflated by publication bias, one shouldn't control for intermediate variables, gains fade out etc. I suppose I do it out of stubborn perversity. I keep resolving not to get into arguments about these things, and keep failing.)
It has been demonstrated many times that epigenetics are heritable, affected by gene-environment correlations, and are not themselves causal for many outcomes they have been claimed to be but passive indicators of environment; so OP is, at best, possible but not probable and the speculation that the epigenetics changes will have any causal effects later on to be taken with considerable skepticism... (as they are more likely to simply indicate the manifestation of the childrens' genes as inherited from their parents). Epigenetics is of tremendous biological interest as it is often how genetics exerts its effects, but unfortunately, that's not what people want epigenetics to be. You may remember the scientific reaction to Siddhartha Mukherjee's New Yorker article on epigenetics last year which similarly took likely-confounded & reverse causation epigenetics claims at face value.
As someone who has been involved in multiple longitudinal studies, although not in epigenetics, I can say that they are more expensive.
In terms of what I said about prior evidence, I was coming from a cognitive psych background where we do not typically deal with record-based data. We deal with participants directly. It is expensive and time consuming. Longitudinal studies usually follow cross-sectional evidence.
Alright, since you refuse to think, I'll give it to you for free: do an intent-to-treat design with incentives for additional cuddling for either the parents or hired helpers like nannies, or passive controls; as no infant receives less physical affection than they would otherwise, no one is hurt by the randomization and it is not unethical, but you still get a causal effect. (Not all of them will use it, and the ones who do, the difference will be smaller than if you could remove cuddling, but that's merely a matter of power.) Such designs are often used in economics, and are applicable to many topics where randomization is considered infeasible for 'ethical' reasons: if you have a limited supply of X, it can be allocated by lottery; if you want to measure the impact of health insurance, surely this is unethical? How could you ever study it? But no, you have limited Medicare funds, so you simply allocate by lottery. Charter schools, lead poisoning, vaccines, basic incomes, daycare, enrichment programs, cuddle subsidies etc. The winners get the additional treatment while the losers serve as baseline controls, finessing the ethical problems. As it is usually the case that there's not enough money to deliver the claimed-beneficial intervention to all possible beneficiaries, it's quite a general strategy.
> As someone who has been involved in multiple longitudinal studies, although not in epigenetics, I can say that they are more expensive.
I didn't say that they weren't, just that they were only somewhat more expensive, and they can be done cheaply in a number of ways. There are considerable economies of scale in these things - look at the UK Biobank.
Uh huh.
> Second, you are correct that I didn't consider that design. This may be because it is a little stupid, or at least impractical without the right target population.
It's neither stupid nor impractical, and I gave several examples, many of which could or have been done in the US. Even healthcare can be randomized in the US this way - consider the Oregon Medicaid or the RAND healthcare experiment. (Both of them, BTW, deliver the profoundly surprising result that correlation!=causation and additional healthcare does little for health.) As it is a priori unlikely that all US children are above some hypothetical threshold for cuddling - in which case OP would be even more dead on arrival that it looks - the design would be perfectly adequate to investigate the causal effects of cuddling. One can screen for low cuddling levels if need be to increase power.
> but only if my sample contained a number of malnourished kids. If I took my sample from middle class America, and I assigned them to food and food+plusmorefood conditions (via force feed I guess?), I'd be studying obesity more likely than malnutrition.
This is a bad example as middle class American kids are not 100% micronutrient replete. Iron deficiency/anemia is common, and even iodine deficiency is actually quite common according to the urinary surveys. (If you look, you'll find at least two iodine RCTs claiming noticeable cognitive benefits in New Zealand college students, a country typically considered better off than the US on public health.) Vitamin D is another question mark - the official RDAs are very low but they're heavily criticized and if you accept some of the higher recommendations, then middle class American kids are quite likely to be deficient (all that time indoors).
> The effect between some and more affection, is probably qualitatively different than none and some affection.
That would be very unfortunate for OP, then, as I doubt any of the kids in their study had literally zero affection, as they were not recruiting from Cold War Romanian orphanages.
https://news.ycombinator.com/newsguidelines.html
Since expression of these tags result from external influences, I don't see what the relevance is. Can you explain? How does inheritance of these epigenetic tags, unexpressed, contradict the article?
I have understood that epigenetics refers to heritable changes in gene function without changes in the genome. Change in gene expression even if it last the rest of the life of an individual not epigenetic change. But even university press release says otherwise.
What I'm missing?
And now, the whole story is becoming even more confusing because those base modifications also happend in RNA (transcribed DNA) which lead to the term epitranscriptomics, which to be honstest, has nothing to do anymore with inheritance, it is just about how a RNA molecule interacts with other cell processes.
If DNA methylation happens while cuddling, would this always be a good thing for the baby (e.g. one of the parents has a genetic disease)?
(I literally only read the first two sentences)