I get what you're saying, but isn't it also true that we as society pay no attention to domain experts' null findings?
For example, if a ghost expert concluded ghosts did not cause 9/11, scott wouldn't mention it. But if a ghost expert did conclude ghosts caused 9/11, he might.
The gut microbiome, though, has been (so far) a gigantic disappointment in terms of explanatory and treatment power. A decade or so ago it seemed that we'd be titrating bacteria balances to clear up depression, anxiety, and all flavors of GI tract disorders, and yet the main upshot (again, so far) is that my PCP suggests I eat probiotic yogurt for a few days after prescribing me an antibiotic. Maybe this is just a specific example of medicine being slow, and we will in fact have the golden future of titrated gut microbiomes, but I can't help but get similar vibes as with antioxidants (cause of cancer/heart disease/aging!) and other big-hype/small-results grand theories of health.
Also maybe useful to consider the opposite case, i.e. things that really were a big deal: immunotherapy(CAR-T, monoclonal antibodies), GLP-1 (though I'm a little skeptical of the strong version of the "GLP-1 theory of everything"), antiretrovirals, and, in the good timeline we aren't living in, mRNA vaccines.
I get why the lack of actionable solutions makes it hard to believe, especially if you're a Bayesian or a believer in the "efficient market" hypothesis.
If it's so promising, where are all the solutions?
The answer is that it's hard, and nobody's really tried it yet.
Have a read through my archives, let me know if you're still unconvinced.
Yeah, I know. I'm working on it. But it's a long and expensive process, and I've got other priorities at the moment. So unless you want to help, you'll have to wait.
You really think there's something to this and would like to distinguish your work from the bad science that gets published by researchers who don't care whether their work holds up or not. Of course, there are also cranks who really believe in bogus theories in addition to the real McCoy. But lots of bogus studies have been identified as likely to fail replication, and it would help to see best guesses for that in advance.
I do agree that treatments tend to be over-hyped initially. I think there's some mix of 'medicine is not doing what it can' and 'silver bullets get promoted, then turn out to be regular bullets. Still useful, but more constrained in utility than expected.
Though doesn't it seem like improving gut bacteria just doesn't fit well into currently accepted treatment modalities? By way of analogy, administering bacteriophage with antibiotics reduces antibiotic resistance in topical and gut infections and there are studies which show this. We really need to be reducing antibiotic resistance. We still don't use phage therapy in the US on humans directly because pure antibiotics are more 'assembly line' and easier to get right without really knowing what's going on. Phage would require some kind of cocktail of phages to fit the infection, with constant updates to that cocktail. Constantly updating a treatment used on people sounds a highway to regulatory hell.
Antioxidants are over-hyped but ascorbate *is* probably still under-used in medicine. It helps with wound healing, at the least. Though too much ascorbate can temporarily dull the immune system through reduction of free radicals.
I think that Pauling's big failing was that he neglected to account for the antioxidant effects of Uric Acid, so he dramatically over-estimated human ascorbate needs based on non-higher-primate animal models. I would still really like to see what happens if we reactivate GLU in diabetics, allowing them to produce their own vitamin C endogenously again. Most diabetics have low grade scurvy and re-activation could address that. Oral supplementation is not ideal for diabetics since diabetes also features decreased absorption of ascorbate.
Re: mRNA vaccines, there's some hint of minor downsides with increased use. Higher levels of IgG4 seem to lead to tolerance to spike proteins rather than immune response. In practical terms, may be best to make sure that your third COVID jab is not Moderna or mRNA. Though the first two jabs seem fine.
>Skolnick could add an epicycle: maybe identical twins have more similar microbiomes because their identical genes caused identical gastrointestinal ecologies. But then schizophrenia is determined by genes again (albeit indirectly) and it should show up as genetic in GWASes (ie they should pick up the genes for having a certain type of gastrointestinal ecology suitable to schizophrenia-causing bacteria, and identify them as schizophrenia genes).
It's funny you should mention, because this is exactly what we do see. All the genetic loci for schizophrenia risk worth talking about are in the major histocompatibility complex, an immune complex. And it has indeed been shown that identical twins have more similar gut microbiomes than fraternal twins.
Specific loci such as TCF4, GRIN2A, DRD2 (dopamine) and CACNA1C arent in there, nor are risk factors like 22q11 deletion (one of the largest effects) or SETD1A in there?
I think "worth talking" plays an outsized role in your description?
No, it plays an appropriate-sized role in my description: those loci are not worth talking about.
Read their methods. They mask MHC in their analysis because trying to study other risk loci without doing so is like trying to do astronomy at high noon.
Hard disagree with your statement of "not worth talking about"; this leads me to regarding your other claims and writing to be likely over-the-top to the point of being misleading. Of note, I've personally collected stool samples as part of studies in the microbiome in schizophrenia. Microbiome and associated fields (leaky gut; nutrition & diet) are likely to play a factor in schizophrenia, but you present it as the silver bullet, which Scott pushes against. Similar to the immune hypothesis, which is now delegated to subgroups of SZ, I find it very unlikely for the reasons mentioned by Scott that your hypothesis is correct, and would urge you to be modest and when presented with counter examples (i.e. non-MHC gene loci) don't disregard them.
I've read their methods - I work in the field. The paper makes a clear-cut argument that there are loci worth talking about - and recent reviews indeed talk about them, contrary to your assessment. Reviews point out the MHC as having strong evidence, but immediately also mention other loci. MHC has strong evidence, but is burdened by unique problems, and does not necessarily replicate (i.e. didn't replicate in two studies with Asian population.
>22q11 is not schizophrenia.
True on a surface level; similar to a statement like "a mutation in the MHC is not schizophrenia." However, the genetic mutation on 22q11 has the highest odds ratio that I know of for developing schizophrenia - i.e. https://www.nature.com/articles/s41380-023-02293-8/figures/1 with about 25-30% developing schizophrenia. And it's not on the MHC. How is that not evidence for a genetic risk factor worth talking about that is not on the MHC?
This doesn't seem like an argument that does much in terms of salvaging the model.
If you have a model that suggests that asbestos should be irrelevant to your risk of lung cancer, it's meaningless to point out that most people with lung cancer aren't exposed to asbestos, if a disproportionate number of people exposed to asbestos develop lung cancer.
Obviously, certain rare genomic aberrations can produce a syndrome which meets the diagnostic criteria for schizophrenia.
In your analogy, I am suggesting that maybe smoking causes lung cancer. I don't care that asbestos also causes lung cancer, because 0.1% of people have significant asbestos exposure, while 50% of the population smokes. Capice?
Also MHC I plays a synaptic formation/pruning role during development and regulates synaptic transmission. Many "immune molecules" have entirely different functions in the brain/nervous system. They are dubbed "immune" because they were discovered in the periphery first. This also applies to the role of complement in the developing brain, cytokines and chemokines etc. which are more akin to patterning molecules during neural tube formation . Thus, a link to MHC I could be considered as part of the synaptopathies grouping as much as the immune grouping. See work by Carla Shatz and many others: https://pmc.ncbi.nlm.nih.gov/articles/PMC2773547/
I agree with the person below who listed various non-MHC schizophrenia risk loci and the summary article saying that the risk variants are associated with neurodevelopmental disorders.
The majority of schizophrenia heritability is not in the MHC but even the MHC association does not fit the gut/immune model. The apparent causal mechanism in the MHC involves structural variants at the C4 gene complex, which has been linked to synaptic pruning in one of the most detailed experimental validations of a GWAS locus to date (Sekar et al. 2016; https://pubmed.ncbi.nlm.nih.gov/26814963/). Importantly, the C4-related mechanism that increases schizophrenia risk also *decreases* the risk for autoimmune conditions like Lupus and Sjorgen's (Kamitaki et al. 2020; https://pubmed.ncbi.nlm.nih.gov/32499649/); and it does so in a sex-specific way. The mechanistic hypothesis is that C4's clear out dead cells which leads to both: (1) aggressive pruning of neurons during development, increasing the liability of schizophrenia; (b) aggressive pruning of sites of injury during the life-course, decreasing an autoimmune response ("Because complement facilitates the rapid clearance of debris from dead and injured cells, increased levels of C4 protein probably attenuate interactions between the adaptive immune system and ribonuclear self-antigens at sites of cell injury, pre-empting the development of autoimmunity."). Neither this mechanism nor the sex specificity is consistent with the gut bacteria hypothesis, which is that more Ruminococcus gnavus should lead to more autoimmune flares, more lupus, AND more schizophrenia: the opposite relationship to what is observed in the MHC.
Moreover MHC I plays a synaptic pruning and plasticity role as well in neurons, so that association could also converge on synapse formation, pruning, and plasticity. See Shatz, Boulanger, McAllister and many others working in this area .
This reminds me a lot of the "autism is caused by dysfunctional microbiome" studies which failed to account for the fact that autistic kids are often picky eaters and have very different diets from non-autistic controls.
Would you not expect someone to become a picky eater, if certain foods caused them immense gastrointestinal and neuropsychiatric distress? Seems like a great way to Skinner box somebody into only eating a few few "safe" things.
That is correct. It is responsible for the vast, vast majority of protein-coding genetic diversity among humans. We are all 99.5+% identical to one another genomically, but two people can have as few as 50% of their gut bacteria in common. Please see stephenskolnick.substack.com/p/cheating-at-evolution for more
We are also 99% identical to chimpanzees, so... I mean, sure, maybe gut biota counts for something, but the genomic differences aren't phenotypically trivial.
Correct! Not saying they're trivial, just that any argument which applies to the genomic differences also applies to microbiomic differences, about 10,000 times over, especially for complex traits which we've been unable to find in the genome despite our best efforts.
Multiplied by 10,000 for the genomic differences in microbiome, and then multiplied again by the tiny fraction by which that translates into meaningful variation in human function. The human genome has a fairly straightforward and causal role in influencing how the human body functions and maintains itself. The gut microbiome mostly acts indirectly through the digestive system, so its impact on most biological functions is orders of magnitude smaller.
Even if the gut microbiome in total is way more impactful than most people realize, it should be obvious that if you swap out one gene on one species of bacteria in your gut, the total impact on the human host is going to be smaller than if you swap out one gene on the human directly. So the math needs to take that into account.
I'm willing to believe the gut microbiome is vitally important. But your rhetorical approach does not inspire confidence that you are willing to give enough weight to alternate hypotheses and evidence that contradicts your priors, such that it can adequately counteract your hammer/nail bias.
As I mentioned in another comment: I was a physicist. Ten years ago, I saw a big ol' fucking nail, and I have spent the intervening decade getting my hands on a hammer and learning to swing it.
We can talk all we want, but there's only one real way to find out whether or not I have hit the nail on the head here.
This area of research is vitally important, especially in view of the dietary percentage of fast, junk and processed food eaten where it is available. May be a reach here but what would people's pets microbiome look like if they were fed 70% Twinkies and corn chips and hot dogs?
Speaking as someone with an autism diagnosis who was a very picky eater in childhood, this doesn't seem to me to properly describe the direction of causation I experienced.
None of my pickiness was ever associated with any source of gastric distress (I often faked gastric distress to skip school growing up, but experienced it very rarely.) There were many foods that I was comfortable eating as individual ingredients, but wouldn't want to eat if they were mixed together. One common form of presentation which I personally experienced is an aversion to different ingredients on a plate "touching." Since they all get chewed up and mixed together in the stomach, it seems implausible that this would be relevant to their impact on the gut biome.
Also, I stopped being a picky eater in my teens, and this did not involve an abatement of my autistic tendencies in general; to the extent that those have abated, it's been a learned process.
Chiming in as someone else with an autism diagnosis to say that this aligns with my own experience extremely well. What stands out most in my memory is that I loved sweet things and most meats, but would not eat meat with any noticeable sweetness. I'd serve myself jellied cranberry sauce as a side dish for my turkey and eat it separately like a pudding.
Please try them? Particularly if you're talking vegetables, they're good for you? Or, tumeric, or half a dozen other "these will really help your life" products?
It seems a shame to go through life without even trying something to see if you'd like it. I'd never tried beets before I was 21 -- they're delicious~
Autists routinely shit their pants (having "gut issues" is considerably more normal for them, and I'm not sure if it's simply because the societal shaming over the issue doesn't hit them as hard as normies). I'm not convinced that "dealing with immense gastrointestinal distress" is enough to cause an autist to stop doing what they otherwise enjoy -- force of will with these folks is quite strong.
Again, you're probably not discussing "sheds entire intestinal lining" or actual "immense gastrointestinal distress" -- what most people think of as "immense gastrointestinal distress" really isn't that bad. If you can show me a lot of autists who "routinely shed their entire intestinal lining" I'd love to know about it.
Is this just a case of somebody who's an expert at something and seeks confirmation that it is The One Thing That Explains Everything? I find that particular subset of people tiring.
I was a physicist. I became a gut microbiome scientist when I learned about the strong heritability of the microbiome, because it seemed to me that this has the potential to be as big as mendelian inheritance for our understanding of human biology.
Fair enough! I have no training in any of this and am not a regular reader of your blog. I guess your best shot at proving the claim you made in your post is to spur direct research in this? What's your next step from here?
Well I have my hands pretty full trying to cure heart disease at the moment, but when I have a minute the next logical step is in a trial of FMT in schizophrenia. Fully wipe out the microbiome and replace it with that of a happy healthy person, and—if my hypothesis is correct—we should see durable remission.
You might still be weird, as you would if you had taken a magic mushroom/amphetamine cocktail every other day for the majority of your adult life, but presumably the hallucinations would at least stop.
This will be a hell of an IRB process, because the true test will involve taking them off their antipsychotics. But it might be possible.
The notion of queuosine deficiency as a plausible explanation for schizophrenia relied primarily on the "kynurenine shunt", and NMDA receptor antagonism by kynurenic acid, as an explanation for the positive symptoms. In light of this, Q deficiency seems like a better fit for classic MDD—although I should point out that gnavus is one of the most abundant gut microbes which scavenges the full end product from its environment, as opposed to e.g. Lactobacilli, which take up a precursor produced by Bacteroides and similar and perform the last few steps of the biosynthesis themselves.
It seems like if this was going to work, we would have some anecdotal cases of people who were schizophrenic and then were apparently cured after they got a serious infection and went on a bunch of antibiotics that massively disrupted their GI bacteria.
I don't think we've yet moved past this syllogism:
1. If your hypothesis is correct, schizophrenia is treatable with antibiotics.
2. Schizophrenia is not treatable with antibiotics.
3. Therefore, your hypothesis is not correct.
None of the Above asked, pretty reasonably, I think, for your method of moving past this. Is it (1) or (2) that you deny? Maybe both?
Unless I've missed some of your other writing here on in your post (which I read), it seems like the entire program you've offered so far for moving past this pretty obvious objection is the case of Susannah Cahalan, a women who was successfully treated for anti-NMDA receptor encephalitis (not schizophrenia) with immunoglobulins (not antibiotics), although when pressed you seem to have implied that an earlier course of antibiotics was what actually cured her.
So, I'm interested in anything else you want to offer that might help me move past this objection, which seems fatal, completely independently of the several other objections already raised, also fatal.
I scoffed when she announced it, but the Youtuber Lauren Kennedy who previously had the handle @LivingWellWithSchizophrenia has rebranded to @LivingWellAfterSchizophrenia following going into remission and reducing her meds all the way to zero while on a keto diet. That is the sort of thing you might expect to affect your gut, though the theory behind it just appears to be a metabolic one not involving the microbiome.
I dunno... I know two brothers with schizophrenia, and the one supposedly tried keto (for other reasons: he jumps on a lot of bandwagons), but it had no noticeable effect on his—pardon me French—craziness; the *only* thing that's helped is medication.
(On the other hand, the dude has enough trouble just complying with "take this pill regularly and stop smoking weed", so I'm not very sure how well he *actually* "tried keto"—and IIRC he didn't try for longer than about 1/2–1 month anyway, so... not exactly firm evidence.)
Ok, great, thanks. I put your name into Google Scholar and I see some review articles, but none seem to include any experimental work. That's why I am asking what sort of experimental work you do.
Greg Cochran is a physicist who I recall back in the day suggesting schizophrenia was caused by a pathogen because of the winter-birth effect. Now he's moved on to thinking it's mostly genetic, manifesting when someone is at the tail of a distribution of sanity (like low IQ, though I've pointed out retardation has some rare alleles of large effects we've identified).
This makes me feel old and stupid, but I'd value an explanation of what "heritability" means and why there isn't an exact theoretical formula relating it to these binomial probabilities of Schiz/Not and Twin/Not. Why do we have to do simulations? (Except for the good reason that sometimes it's easier to do a simulation than to find the exact theoretical formula.)
From the footnotes on Scott's recent article on missing heritability:
Geneticists distinguish between three related concepts:
Polygenic score r^2 is the degree to which our current best genetic models can predict traits. You might use this to discuss the accuracy of a genetic test or an embryo selection procedure.
Narrow sense heritability is the degree to which all normal, additive genetic variation affects traits. You might use this to discuss the effectiveness of breeding programs, or how much you expect a parent’s traits to affect their children.
Broad sense heritability is the degree to which all genetic variation, including interactions and rare mutations, affect traits. A correctly-done twin study should (modulo certain small issues) return the broad sense heritability. This is useful in resolving deep questions like “How much do genetic vs. social causes affect traits”, and acts as the limit for what we might be able to explain through some future genetic science.
I'm not sure I understand your question about simulations. "Simulations" are a thing we could/would do even in situations where we have _extremely_ good physical, mechanistic understanding of the processes involved.
"2: But the twin concordance rates are pretty low - if your identical twin has schizophrenia, there’s only about a 30%-40% chance that you get it too. Is that really what we would expect from a mostly genetic disease?
This is exactly the concordance rate you should expect from a polygenic condition where 80% of the variance is explained by genes."
This is false. It's what you'd expect in a liability-threshold model where 80% of the variance in liability is set by genes. But there's still the question of what the triggering threshold is.
How are these different claims? Isn't liability-threshold just a dichotomization of saying that some of the variance is genes and some is something else? In my discount simulation, I tried to cash out "80% genes, 20% environment" as neutrally as I knew how.
Nah, L-T is a specific model of how discrete diseases work. It happens to be a broadly realistic model, but if you think carefully, you will see lots of ways it could conceivably go wrong (in worlds I expect to be different from ours).
This doesn't mean Scott's claim is false, though, does it? I.e., an L-T model may be a better fit with schizophrenia, but the concordance rate is still consonant with a less-specific model (e.g. "a polygenic condition where 80% of the variance is explained by genes") too... no?
It's like the distinction between saying "height is heritable" vs "being an NBA player is heritable". The latter would be found true on liability scale but is obviously not biological in practice.
I'm not sure what you mean by that. Being an NBA player isn't entirely due to biology. Some people just decide they really want to play basketball and practice a lot. Sure there's steps between being tall and joining the NBA, but there's also steps between having a certain set of genes and having growth hormones, and between having growth hormones and being tall.
>there is no adoptive-parent-to-adoptive-child schizophrenia correlation
Do you have a source for this claim, or are you simply taking absence of evidence as evidence of absence? If so, please consider: they don't really let people with a diagnosis of schizophrenia adopt children very often, do they now?
In this case, shouldn't we expect to see lower incidence of schizophrenia among adopted people? This seems like a pretty straightforward thing to check in principle.
What disruptions to transmission would you expect would result in a reduction in incidence?
Live birth normally results in a transmission of the mother's gut bacteria to the child, but C-section birth does not. In cases where mothers with schizophrenia give birth by C-section, and then the child is immediately taken away, do you think we should see baseline levels of schizophrenia incidence?
What about cases where the father has schizophrenia, but isn't present? Under a gut biome model, that doesn't seem like it should result in elevated incidence of schizophrenia.
“According to the consensus theory of schizophrenia, this is because schizophrenia is a neurodevelopmental disorder that interferes with adolescent synaptic pruning”—I’m unclear on what the consensus is (if any), but this sentence seems different from what’s in the link https://pmc.ncbi.nlm.nih.gov/articles/PMC9870034/
Well, here’s how I would put it: There’s some strength with which the various different areas of the cortex communicate with and constrain each other. If the communication strength amount falls below some threshold, you get schizophrenia. The communication strength starts very high but goes down thanks to adolescent synaptic pruning. And that’s why schizophrenia starts manifesting around age 18-30.
Re this and the other day: would assume all traits are 100% genetic definitionally, even if there is any sort of environmental component to their expression - a building is made of bricks; if they vary it is because variance is allowed by the genome.
The environment can’t create information; things can be A or B because A or B is in the genome.
The genome goes back multiple generations so the percentage of inheritance is difficult to demonstrate. You can’t assume the value of any trait not expressed by your immediate ancestors is not inherited.
But presumably some genes are much more important than others in how the genome is expressed - making it difficult to compare genomes.
When you don’t know how something works, e.g. how information for complex traits is encoded in the genome - you can’t be too confident in making assumptions about inheritance.
This comment supplied as a car game by my fellow road trip passenger.
The idea of "synaptic pruning" is a new one to me. I tried to follow the link but it was very dense and I had trouble understanding it. Is there an ELI5 for "what is synaptic pruning, and how confident are we that this is a real thing that exists?"
(I guess I could ask an LLM, but this seems like the sort of thing where it would hallucinate and I'd never know the difference.)
Also, if "insufficient synaptic pruning" is a cause of schizophrenia, does that mean you can test for schizophrenia by doing a brain scan and measuring how pruned people's synapses are?
The consensus understanding of human brain development is that, at first, neurons are highly interconnected (many synapses). Each neuron has a high neighbourhood size, i.e., many other neurons can be reached in just a few hops. Human brain development is a two-phase process, where first the brain grows rapidly with many inter-connections forming, and second the brain gets rid of a lot of the less active inter-connections (the pruning step). To speak in terms of purpose, the first phase allows young humans to synthesize lots of information in order to build a general portrait of the world they live in, while the second phase allows mature humans to build the habit of determining what information is relevant to a given situation and make use of that information in a consistent way.
If the second phase doesn’t happen, you’re basically stuck synthesizing lots of information, noticing patterns, but lacking the cognitive stability to act on those patterns or filter out what’s not useful. This is a rough portrait of a person in the schizotypal cluster.
As for how we know it’s real, I believe the main evidence comes from two sources: microscopy of brain tissue at different phases of development (not just from human beings since this pattern of development is apparent in other mammals such as rats and chimpanzees); and hormones. Hormones affect the growth of neurons so it’s possible to get a sketch of the development of the brain (or any other organ) based on which hormones are dominant at which time.
To answer your last question: synaptic pruning is difficult or impossible to assess directly without direct samples of brain tissue. It affects both long and short-range connections at every scale of the brain’s organization. Brain scans are only good at detecting macro-scale phenomena (brain waves, for example, can be detected because the EM flux of millions of synapses firing in concert adds up to the point that it becomes measurable). Since pruning tends to primarily affect synapses that were quiet anyway, a difference in the degree of pruning is probably not sufficient to cause a difference in activity that would show up on modern equipment.
I'm confused. You seem to be describing schizophrenia as resulting from a brain that *doesn't* prune synapses during adolescence, but the linked paper attributes it to *excessive* pruning. Wouldn't that imply a brain that's too "adultlike" rather than too "childlike"?
Also, I'm confused about the timing. An 18-year-old is clearly cognitively different from a 35-year-old, but they're clearly cognitively closer to that 35-year-old than to a 2-year-old (if it's strictly linked to adolescence, substitute in "14-year-old" and the point still holds: high schoolers are kinda dumb but they're not dumb the way toddlers are).
Is the idea that a prodromal schizophrenic initially has the normal neurological processes of adolescence occur, but then these processes continue even after adolescence? Then at some point the pruning crosses some threshold and brain function degrades to the point where schizophrenia results.
Well, I guess I wiffed it including that paragraph about schizophrenia at all because my intuition was that it went the opposite way.
It looks like Scott shares my error—I don’t know how else to interpret the sentence “schizophrenia is a neurodevelopmental disorder that interferes with adolescent synaptic pruning”. It’s hard to read that as saying the pruning is positively, rather than inversely, correlated with symptoms.
I’m just skimming, but my gloss of the paper is that there are several information-rich correlations in the literature but a lack of consensus as to how they all come together to paint a causal picture. Maybe prodromal schizophrenics have a latent defect that is exposed by pruning; maybe pruning is overly aggressive and that directly causes the disorder; maybe the real problem is a breakdown of communication between the prefrontal cortex and thalamus, and the synaptic pruning follows after (because areas that lack activation are the first to be pruned). Perhaps a combination.
The upshot I’m reading here is that pruning is involved *somehow*, and more pruning is correlated with more schizophrenia, but the mechanism is not yet clear and there is likely more than one distinct etiology for schizophrenia.
This is not what I had thought previously and also not, to my ear, how Scott summarized the paper he cited. So maybe the consensus theory does not actually consensus that hard.
With surrogate pregnancies it's possible to get the benefits of identical twins without the risks of resource depletion by having the pregnancies be carried by two different mothers. I've never heard of this being done but maybe it should start being encouraged just for the obvious benefits to science.
Surrogacy costs too much money and is too controversial itself to use it for experimental embryos or (if the splitting process were well-established) to populate twin studies.
This probably hasn't been tried with human embryos but the splitting process is probably literally you take it when it's two cells and pull it in half. Pulling off a single cell when it's just a few is already how genetic testing is done. The benefits of identical twins also extend to the twins, they like having each other.
And let's not forget, Leprosy and Tuberculosis also ran in families. Being able to predict a disease with genetic tests does not necessarily confirm a true genetic disease because you could also be looking at an unusually high genetic vulnerability to an unrecognized endemic infectious disease.
As I commented above, Cochran previously though schizophrenia was likely caused by pathogens. He now thinks it's mostly genetic. I can't quickly find where I asked him about what changed his mind, but in this post https://westhunt.wordpress.com/2018/07/22/more-theory/ he writes "Again: you find that having more mutational load, more deleterious mutations, increased your chance of schizophrenia, or autism, or low IQ"
Wow, that's interesting; I used to read Cochran pretty regularly,* but didn't know he'd reconsidered this particular hypothesis. Thanks for bringing this up! I wonder if he's reconsidered the "gay germ" hypothesis as well...
*(can't recall why I stopped—think just because he stopped updating as regularly & I forgot to keep checking? heh)
I've banned this person because I literally did not think of Cochran once while writing this post, and resent being accused of cowardice and straw-manning for something that I don't think any reasonable person would consider related to anything I wrote. If someone wants to argue that this wasn't out-of-bounds hostile and aggressive, I'll reconsider.
Not #1. He talks much more about homosexuality being caused by that. Also various cancers and heart disease I think were discussed in that Atlantic article with Paul Ewald.
Those interested in this subject might consider an adjacent field that tackles a parallel debate. In Yap et al., 2021 Cell, the authors find that behavioural traits (especially restricted eating patterns tied to autism) largely explain observed microbiome differences--especially the reproducible findings of less diversity. Microbiome variation looks more like a consequence of neurodevelopmental phenotype than a cause. I think this is a cautionary tale about directionality in these brain-gut associations everyone is so hyped on. https://www.cell.com/cell/fulltext/S0092-8674(21)01231-9?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0092867421012319%3Fshowall%3Dtrue
Couldn't the causation work both ways? Autistic restricted eating leading to reduced gut bacteria which in turn intensifies some autistic traits. I saw one paper that made the argument (https://pmc.ncbi.nlm.nih.gov/articles/PMC8784764/
2). This would match up with studies that show autistic kids having both improved gut and a little less severe traits after fecal transplants (like https://pmc.ncbi.nlm.nih.gov/articles/PMC8560686/).
The Yap study is the largest human study so far (247 Australian children with shotgun-sequenced stool) which found that restricted-interest eating explained up to 58 % of the variation in microbial genes, while the autism diagnosis itself explained ~7 %, a signal that vanishes once diet is in the model. In other words: behaviour → food choice → microbiome.
• Who was treated? 40 Chinese children with autism and co-occurring chronic GI complaints; no placebo arm.
• What changed? Parent-rated GI scores fell 35 % and stayed down; core-autism severity (CARS) dropped ~10 % but half that gain was gone eight weeks later.
• What didn’t change? Overall microbial α-diversity; transplanted microbes faded unless they matched the child’s diet.
Relieving constipation is great, but the data do not support a microbiome-mediated alleviation of repetitive, communicative, and social behaviours as claimed in the mouse model.
Why FMT can’t stick in severely selective eaters is illustrated by a classic study of Tanzanian Hadza hunter-gatherers https://www.science.org/doi/10.1126/science.aan4834: when the menu shifts with the seasons, the gut community shifts with it: diet, not the person, is the dominant architect. A child who eats only one brand of yoghurt and four additional beige foods cannot support new microbes delivered by capsule; these microbes will soon disappear.
A deeper issue is degeneracy: biology often achieves the same function with very different internal states so multiple, even non-overlapping, microbiomes can all count as healthy. Until we map that normal phase-space, many disease/disorder associations may simply flag rare-but-healthy corners. (Though obviously not the case for the kiddos who have ARFID; their microbiome cannot reflect a healthy state given the level of dietary restriction and they need immediate and extensive nutritional support).
Meanwhile, regulatory reality favours hype: probiotics already skirt FDA oversight, and “live biotherapeutic products” could follow with modest evidence (a placebo-friendly business model!)
So can causation run both ways? Biologically, yes. Empirically, the only arrow that survives careful adjustment points from diet to the gut. Treat constipation because it improves comfort; use behavioural feeding therapy to widen diets (the surest path to a stable microbiome); identify any other co-occurring GI issues and treat those (allergies, food sensitivities, autoimmune disorders etc. .) but we don’t have the evidence to support all the hype about gut therapeutics in my opinion. We are still very early days in this field; I am open to changing my view with new evidence:)
Isn’t this an easy test to setup? Give multiple antibiotics with wide range/coverage that eradicates all gut bacteria and then see if the schizophrenia recedes over time?
Koch's postulates! But no I'm pretty sure the claim that subject would have to be sterile at birth to account for the whole course of neurodevelopment in childhood.
Not so, Cochran's model of the condition postulates live microbes modulating neurotransmitters, not a developmental influence. It doesn't explain the presentation at early adulthood, except maybe through a two-strike theory.
That said it would be interesting to give a non-schizoid person a light treatment of schizophrenia drugs and see what happens to their gut flora.
His theory of homosexuality involved an insult to the brain rather than live microbes. Similarly, polio is a gut infection which reaches a dead-end in the nervous system, but causes damage when it wanders off course to there.
> (during residency, I met a schizophrenic patient who, whenever he wasn’t restrained, would eat as much glass as he could until he got internal bleeding and had to go to the emergency room. I don’t know what kind of microbiome disruption this causes, but I bet there aren’t many non-schizophrenics who have it.)
Sounds a lot like Pica to me. Plenty of people have Pica without being schizophrenic.
I assume the patient had other symptoms of schizophrenia and they didn't just diagnose it from a compulsion to eat glass - even as a resident I think Scott would have recognized if the latter were the case
Sure. I'm not disputing that the guy was schizophrenic; just that "I bet there aren't many non-schizophrenics who have [a compulsion to eat stuff that's not food]" sounds odd if you're aware of Pica.
It wasn't just "[a compulsion to eat stuff that's not food]" but a compulsion to eat broken glass despite the obvious and immediate terrible effects of doing so. I'm not sure if ordinary pica ever stretches into broken glass.
Schizophrenia is fascinating because it makes the brain go wrong in such odd ways: it's not just a meltdown of all capacities. Either we'll learn lots more about the brain before we understand schizophrenia, or understanding schizophrenia will clarify many other things about the brain.
Isn't schizophrenia more common in men than in women? That would seem to strongly contra-indicate a gut microbiome cause, given that babies crawl around scraping up whatever microbes are around on a purely egalitarian basis.
It's bizarre to read that Skolnick describes microbiological inheritance as "you picked it up when you were crawling around and putting everything in your mouth." It's reasonable to think that you might have acquired some of your mother's gut microbes in utero -- I'd think that would be closer to inheritance than Skolnick's explanation. Or did I miss something?
Okay. But then what would be the mechanism by which microbial inheritance occurs? I mean, if I'm a kid scrabbling around on the ground eating dirt with my similarly aged next door neighbor, they'd be "inheriting" the same microbes I was. That sounds more like nature than nurture.
Point of clarification worth mentioning here: Most of the bacteria that make up a healthy human gut microbiome are not really found anywhere else on Earth aside from the mammalian gut. There was some research a few years back suggesting that live bacteria are transmitted in milk, which would be a fantastically elegant explanation—but I'm somewhat skeptical of that notion now. Most of the bacterial DNA that we find in breast milk does appear to be contamination from skin surface, backflow from child's mouth, etc.
But what we do know, from deep-sequence longitudinal tracking of mothers and their infants over the first year of life, is that you definitely do get your parents' strains.
The how, as much as I hate to say it, is still sort of a mystery. It may come down to stray farts, or picking your nose and eating it. (Weirdly well-conserved behavior. Why? Hmm.) This is an incarnation of the notion that "everything is everywhere and the environment selects"; our bodies put a lot of effort into creating the right niche, so that when you are born basically sterile, all it takes is a few microbial cells getting into the right place in order to take root and persist for life, while the immune system sculpts away most of the potential pathogens.
Note that this does allow for substantial non-vertical transmission; I spent probably three days a week at my friend David's house through most of elementary school, and I strongly suspect that if you sequenced our guts, he and I would have a few strains in common. Maybe one in my gut came from me using the bathroom right after he took a shit, while the flush-cloud was still in the air.
Maybe I got a strain from grandma Zhang's red bean buns, because one time she didn't wash up before making the dough. As best I can tell, it's the billion little ways the world is dirty (but growing less so all the time).
I want to respectfully push back on your point with a parallel that might be getting overlooked:
ChatGPT points out that monozygotic (MZ) twins can have more similar gut microbiomes than dizygotic (DZ) twins even beyond their genetic similarity—possibly due to shared epigenetics, early developmental environment, or immune/mucosal traits.
I wonder if this same logic might help explain part of the missing heritability problem. Could subtle, non-genetic but deeply shared developmental factors be amplifying trait similarity in MZ twins in ways that our models still treat as purely genetic?
In that case (replying to your last paragraph) the explanation is going to be Jungian type (say, MBTI), presumably shared by MZ twins (given they look so similar), but not showing any particular similarity for DZ twins.
(however, if I understood last week's missing-heritability post correctly, the message of twin studies is confirmed by adoption studies)
Isn't Jungian type of the form: "they look alike, so they might be treated similarly or develop shared self-concepts based on appearance"? If that's what you mean, then no, my query is unrelated to that type. I'm asking about developmental factors like epigenetics, shared intrauterine environment, developmental synchronization, microchimerism, and maternal adaptations for having identical twins.
This is another example in a line of examples of some guy confusing (weak) correlation with causation. Or of invoking mono-causality in the absence of evidence.
It is widely claimed that autoimmune diseases are more common in women. Is there a theory of this?
But this depends on what you label autoimmune. Ongoing autoimmune diseases like lupus are a lot more common in women, but acute autoimmune diseases like narcolepsy are slightly more common in men. Schizophrenia, with its abrupt onset appears in the second case. Indeed, it is more common in men. Moreover, when it occurs in women, it occurs maybe 10 years later. Do theories of sex differences in ongoing autoimmune diseases shed light on acute ones?
If testosterone down-regulates the immune system, than males would have more infections and less autoimmune diseases. But this doesn't explain why they have more narcolepsy and schizophrenia. And if it's as simple as down-regulation, why isn't there some other mechanism to up-regulate it? Maybe not all the way back and maybe with a lot of noise, but wouldn't this predict more autoimmune diseases?
For a serious disorder to be polygenic with a wide variety of environmental triggers is frustrating, but our frustration is not an argument that matters are otherwise.
This is a very interesting take down of Skolnick. Thank you. The one nitpick I have is that 'cat exposure,' which you mention, is an example of microbiologically mediated schizophrenia via toxoplasma gondii CNS infection. So it supports the hypothesis that some instances of schizophrenia are microbiologically mediated. Though not by R. gnavus or gut bacteria specifically in those instances.
So Skolnick thinks gut bacteria are behind schizophrenia, huh? Do they serve snacks too? Anyway, if you're looking for something to spice up your discussion on microbial theories, check this out for some star symbols to shine a light on those wild claims <https://starsymbols.io/
I wonder how much being a gut microbiome expert predisposes one to claim that the gut microbiome is responsible for any given condition...
I get what you're saying, but isn't it also true that we as society pay no attention to domain experts' null findings?
For example, if a ghost expert concluded ghosts did not cause 9/11, scott wouldn't mention it. But if a ghost expert did conclude ghosts caused 9/11, he might.
The gut microbiome, though, has been (so far) a gigantic disappointment in terms of explanatory and treatment power. A decade or so ago it seemed that we'd be titrating bacteria balances to clear up depression, anxiety, and all flavors of GI tract disorders, and yet the main upshot (again, so far) is that my PCP suggests I eat probiotic yogurt for a few days after prescribing me an antibiotic. Maybe this is just a specific example of medicine being slow, and we will in fact have the golden future of titrated gut microbiomes, but I can't help but get similar vibes as with antioxidants (cause of cancer/heart disease/aging!) and other big-hype/small-results grand theories of health.
Also maybe useful to consider the opposite case, i.e. things that really were a big deal: immunotherapy(CAR-T, monoclonal antibodies), GLP-1 (though I'm a little skeptical of the strong version of the "GLP-1 theory of everything"), antiretrovirals, and, in the good timeline we aren't living in, mRNA vaccines.
FMT is still done especially in the context of C. diff infectious but yeah these are still rare cases where the microbiome effect size is huge.
I get why the lack of actionable solutions makes it hard to believe, especially if you're a Bayesian or a believer in the "efficient market" hypothesis.
If it's so promising, where are all the solutions?
The answer is that it's hard, and nobody's really tried it yet.
Have a read through my archives, let me know if you're still unconvinced.
Fundamentally no amount of substack essays will convince me; what will change my mind is an RCT that's pre-registered on clinicaltrials.gov.
Yeah, I know. I'm working on it. But it's a long and expensive process, and I've got other priorities at the moment. So unless you want to help, you'll have to wait.
This seems like a good opportunity for applying betting to science:
https://www.overcomingbias.com/p/let-gambling-save-sciencehtml
You really think there's something to this and would like to distinguish your work from the bad science that gets published by researchers who don't care whether their work holds up or not. Of course, there are also cranks who really believe in bogus theories in addition to the real McCoy. But lots of bogus studies have been identified as likely to fail replication, and it would help to see best guesses for that in advance.
Out of curiosity, is there some way to help that does not involve sending money?
I do agree that treatments tend to be over-hyped initially. I think there's some mix of 'medicine is not doing what it can' and 'silver bullets get promoted, then turn out to be regular bullets. Still useful, but more constrained in utility than expected.
Though doesn't it seem like improving gut bacteria just doesn't fit well into currently accepted treatment modalities? By way of analogy, administering bacteriophage with antibiotics reduces antibiotic resistance in topical and gut infections and there are studies which show this. We really need to be reducing antibiotic resistance. We still don't use phage therapy in the US on humans directly because pure antibiotics are more 'assembly line' and easier to get right without really knowing what's going on. Phage would require some kind of cocktail of phages to fit the infection, with constant updates to that cocktail. Constantly updating a treatment used on people sounds a highway to regulatory hell.
Antioxidants are over-hyped but ascorbate *is* probably still under-used in medicine. It helps with wound healing, at the least. Though too much ascorbate can temporarily dull the immune system through reduction of free radicals.
I think that Pauling's big failing was that he neglected to account for the antioxidant effects of Uric Acid, so he dramatically over-estimated human ascorbate needs based on non-higher-primate animal models. I would still really like to see what happens if we reactivate GLU in diabetics, allowing them to produce their own vitamin C endogenously again. Most diabetics have low grade scurvy and re-activation could address that. Oral supplementation is not ideal for diabetics since diabetes also features decreased absorption of ascorbate.
Re: mRNA vaccines, there's some hint of minor downsides with increased use. Higher levels of IgG4 seem to lead to tolerance to spike proteins rather than immune response. In practical terms, may be best to make sure that your third COVID jab is not Moderna or mRNA. Though the first two jabs seem fine.
There's no such thing as a ghost expert. There's just people who claim to be one.
I think this is great, each individual contributing their own POV and providing the signal that makes sense to them is how markets work as well.
>Skolnick could add an epicycle: maybe identical twins have more similar microbiomes because their identical genes caused identical gastrointestinal ecologies. But then schizophrenia is determined by genes again (albeit indirectly) and it should show up as genetic in GWASes (ie they should pick up the genes for having a certain type of gastrointestinal ecology suitable to schizophrenia-causing bacteria, and identify them as schizophrenia genes).
It's funny you should mention, because this is exactly what we do see. All the genetic loci for schizophrenia risk worth talking about are in the major histocompatibility complex, an immune complex. And it has indeed been shown that identical twins have more similar gut microbiomes than fraternal twins.
>All the genetic loci for schizophrenia risk worth talking about are in the major histocompatibility complex
Huh? https://www.nature.com/articles/s41586-022-04434-5 - most of the risk is spread.
Specific loci such as TCF4, GRIN2A, DRD2 (dopamine) and CACNA1C arent in there, nor are risk factors like 22q11 deletion (one of the largest effects) or SETD1A in there?
I think "worth talking" plays an outsized role in your description?
No, it plays an appropriate-sized role in my description: those loci are not worth talking about.
Read their methods. They mask MHC in their analysis because trying to study other risk loci without doing so is like trying to do astronomy at high noon.
22q11 is not schizophrenia.
Hard disagree with your statement of "not worth talking about"; this leads me to regarding your other claims and writing to be likely over-the-top to the point of being misleading. Of note, I've personally collected stool samples as part of studies in the microbiome in schizophrenia. Microbiome and associated fields (leaky gut; nutrition & diet) are likely to play a factor in schizophrenia, but you present it as the silver bullet, which Scott pushes against. Similar to the immune hypothesis, which is now delegated to subgroups of SZ, I find it very unlikely for the reasons mentioned by Scott that your hypothesis is correct, and would urge you to be modest and when presented with counter examples (i.e. non-MHC gene loci) don't disregard them.
I've read their methods - I work in the field. The paper makes a clear-cut argument that there are loci worth talking about - and recent reviews indeed talk about them, contrary to your assessment. Reviews point out the MHC as having strong evidence, but immediately also mention other loci. MHC has strong evidence, but is burdened by unique problems, and does not necessarily replicate (i.e. didn't replicate in two studies with Asian population.
>22q11 is not schizophrenia.
True on a surface level; similar to a statement like "a mutation in the MHC is not schizophrenia." However, the genetic mutation on 22q11 has the highest odds ratio that I know of for developing schizophrenia - i.e. https://www.nature.com/articles/s41380-023-02293-8/figures/1 with about 25-30% developing schizophrenia. And it's not on the MHC. How is that not evidence for a genetic risk factor worth talking about that is not on the MHC?
>How is that not evidence for a genetic risk factor worth talking about that is not on the MHC?
It is not relevant to 99%+ of schizophrenia cases.
This doesn't seem like an argument that does much in terms of salvaging the model.
If you have a model that suggests that asbestos should be irrelevant to your risk of lung cancer, it's meaningless to point out that most people with lung cancer aren't exposed to asbestos, if a disproportionate number of people exposed to asbestos develop lung cancer.
Your analogy doesn't check out.
Obviously, certain rare genomic aberrations can produce a syndrome which meets the diagnostic criteria for schizophrenia.
In your analogy, I am suggesting that maybe smoking causes lung cancer. I don't care that asbestos also causes lung cancer, because 0.1% of people have significant asbestos exposure, while 50% of the population smokes. Capice?
Also MHC I plays a synaptic formation/pruning role during development and regulates synaptic transmission. Many "immune molecules" have entirely different functions in the brain/nervous system. They are dubbed "immune" because they were discovered in the periphery first. This also applies to the role of complement in the developing brain, cytokines and chemokines etc. which are more akin to patterning molecules during neural tube formation . Thus, a link to MHC I could be considered as part of the synaptopathies grouping as much as the immune grouping. See work by Carla Shatz and many others: https://pmc.ncbi.nlm.nih.gov/articles/PMC2773547/
I agree with the person below who listed various non-MHC schizophrenia risk loci and the summary article saying that the risk variants are associated with neurodevelopmental disorders.
Also, I feel like https://www.astralcodexten.com/p/but-vs-yes-but is relevant here.
The majority of schizophrenia heritability is not in the MHC but even the MHC association does not fit the gut/immune model. The apparent causal mechanism in the MHC involves structural variants at the C4 gene complex, which has been linked to synaptic pruning in one of the most detailed experimental validations of a GWAS locus to date (Sekar et al. 2016; https://pubmed.ncbi.nlm.nih.gov/26814963/). Importantly, the C4-related mechanism that increases schizophrenia risk also *decreases* the risk for autoimmune conditions like Lupus and Sjorgen's (Kamitaki et al. 2020; https://pubmed.ncbi.nlm.nih.gov/32499649/); and it does so in a sex-specific way. The mechanistic hypothesis is that C4's clear out dead cells which leads to both: (1) aggressive pruning of neurons during development, increasing the liability of schizophrenia; (b) aggressive pruning of sites of injury during the life-course, decreasing an autoimmune response ("Because complement facilitates the rapid clearance of debris from dead and injured cells, increased levels of C4 protein probably attenuate interactions between the adaptive immune system and ribonuclear self-antigens at sites of cell injury, pre-empting the development of autoimmunity."). Neither this mechanism nor the sex specificity is consistent with the gut bacteria hypothesis, which is that more Ruminococcus gnavus should lead to more autoimmune flares, more lupus, AND more schizophrenia: the opposite relationship to what is observed in the MHC.
Moreover MHC I plays a synaptic pruning and plasticity role as well in neurons, so that association could also converge on synapse formation, pruning, and plasticity. See Shatz, Boulanger, McAllister and many others working in this area .
This reminds me a lot of the "autism is caused by dysfunctional microbiome" studies which failed to account for the fact that autistic kids are often picky eaters and have very different diets from non-autistic controls.
Would you not expect someone to become a picky eater, if certain foods caused them immense gastrointestinal and neuropsychiatric distress? Seems like a great way to Skinner box somebody into only eating a few few "safe" things.
Oh, so you don't just think that Schizophrenia is caused by gut biome, you think that gut biome is the explanation for a whole host of issues.
That is correct. It is responsible for the vast, vast majority of protein-coding genetic diversity among humans. We are all 99.5+% identical to one another genomically, but two people can have as few as 50% of their gut bacteria in common. Please see stephenskolnick.substack.com/p/cheating-at-evolution for more
We are also 99% identical to chimpanzees, so... I mean, sure, maybe gut biota counts for something, but the genomic differences aren't phenotypically trivial.
Correct! Not saying they're trivial, just that any argument which applies to the genomic differences also applies to microbiomic differences, about 10,000 times over, especially for complex traits which we've been unable to find in the genome despite our best efforts.
Multiplied by 10,000 for the genomic differences in microbiome, and then multiplied again by the tiny fraction by which that translates into meaningful variation in human function. The human genome has a fairly straightforward and causal role in influencing how the human body functions and maintains itself. The gut microbiome mostly acts indirectly through the digestive system, so its impact on most biological functions is orders of magnitude smaller.
Even if the gut microbiome in total is way more impactful than most people realize, it should be obvious that if you swap out one gene on one species of bacteria in your gut, the total impact on the human host is going to be smaller than if you swap out one gene on the human directly. So the math needs to take that into account.
I'm willing to believe the gut microbiome is vitally important. But your rhetorical approach does not inspire confidence that you are willing to give enough weight to alternate hypotheses and evidence that contradicts your priors, such that it can adequately counteract your hammer/nail bias.
As I mentioned in another comment: I was a physicist. Ten years ago, I saw a big ol' fucking nail, and I have spent the intervening decade getting my hands on a hammer and learning to swing it.
We can talk all we want, but there's only one real way to find out whether or not I have hit the nail on the head here.
Only one? What is it?
(I appreciate you taking the time to engage everyone's thoughts here.)
This is why people say physicists shouldn't be allowed near any other field. (There's an obligatory xkcd here but I'm sure everyone's seen it.)
This area of research is vitally important, especially in view of the dietary percentage of fast, junk and processed food eaten where it is available. May be a reach here but what would people's pets microbiome look like if they were fed 70% Twinkies and corn chips and hot dogs?
I don't think most humans have a diet like that.
Twinkies and other heavily processed foods are known to be addictive. They're also notably bad for the gut biome.
Not exactly, but ultra-processed foods make up about 60% of the average diet in the US & UK.
Speaking as someone with an autism diagnosis who was a very picky eater in childhood, this doesn't seem to me to properly describe the direction of causation I experienced.
None of my pickiness was ever associated with any source of gastric distress (I often faked gastric distress to skip school growing up, but experienced it very rarely.) There were many foods that I was comfortable eating as individual ingredients, but wouldn't want to eat if they were mixed together. One common form of presentation which I personally experienced is an aversion to different ingredients on a plate "touching." Since they all get chewed up and mixed together in the stomach, it seems implausible that this would be relevant to their impact on the gut biome.
Also, I stopped being a picky eater in my teens, and this did not involve an abatement of my autistic tendencies in general; to the extent that those have abated, it's been a learned process.
Chiming in as someone else with an autism diagnosis to say that this aligns with my own experience extremely well. What stands out most in my memory is that I loved sweet things and most meats, but would not eat meat with any noticeable sweetness. I'd serve myself jellied cranberry sauce as a side dish for my turkey and eat it separately like a pudding.
Oh, I love sweet meats (Lebanon Bologna, a good beef stew, pastitsio), but cranberry sauce? That's a "by itself" sort of thing for me too.
I'm autistic and that is absolutely not why I'm a picky eater. Most of the stuff I won't eat I have literally never tried before.
Is it a texture thing for you?
Please try them? Particularly if you're talking vegetables, they're good for you? Or, tumeric, or half a dozen other "these will really help your life" products?
It seems a shame to go through life without even trying something to see if you'd like it. I'd never tried beets before I was 21 -- they're delicious~
Autists routinely shit their pants (having "gut issues" is considerably more normal for them, and I'm not sure if it's simply because the societal shaming over the issue doesn't hit them as hard as normies). I'm not convinced that "dealing with immense gastrointestinal distress" is enough to cause an autist to stop doing what they otherwise enjoy -- force of will with these folks is quite strong.
Again, you're probably not discussing "sheds entire intestinal lining" or actual "immense gastrointestinal distress" -- what most people think of as "immense gastrointestinal distress" really isn't that bad. If you can show me a lot of autists who "routinely shed their entire intestinal lining" I'd love to know about it.
Is this just a case of somebody who's an expert at something and seeks confirmation that it is The One Thing That Explains Everything? I find that particular subset of people tiring.
I was a physicist. I became a gut microbiome scientist when I learned about the strong heritability of the microbiome, because it seemed to me that this has the potential to be as big as mendelian inheritance for our understanding of human biology.
Fair enough! I have no training in any of this and am not a regular reader of your blog. I guess your best shot at proving the claim you made in your post is to spur direct research in this? What's your next step from here?
Well I have my hands pretty full trying to cure heart disease at the moment, but when I have a minute the next logical step is in a trial of FMT in schizophrenia. Fully wipe out the microbiome and replace it with that of a happy healthy person, and—if my hypothesis is correct—we should see durable remission.
You might still be weird, as you would if you had taken a magic mushroom/amphetamine cocktail every other day for the majority of your adult life, but presumably the hallucinations would at least stop.
This will be a hell of an IRB process, because the true test will involve taking them off their antipsychotics. But it might be possible.
How does this tie in with your previous thoughts on queuosine?
The notion of queuosine deficiency as a plausible explanation for schizophrenia relied primarily on the "kynurenine shunt", and NMDA receptor antagonism by kynurenic acid, as an explanation for the positive symptoms. In light of this, Q deficiency seems like a better fit for classic MDD—although I should point out that gnavus is one of the most abundant gut microbes which scavenges the full end product from its environment, as opposed to e.g. Lactobacilli, which take up a precursor produced by Bacteroides and similar and perform the last few steps of the biosynthesis themselves.
Interesting. Your contention is that genetics is necessary but not sufficient, then?
Disclaimer: I'm very far from my expertise here.
It seems like if this was going to work, we would have some anecdotal cases of people who were schizophrenic and then were apparently cured after they got a serious infection and went on a bunch of antibiotics that massively disrupted their GI bacteria.
Read "Brain on Fire: My Month of Madness".
That wasn't schizophrenia though, that was encephalitis.
I don't think we've yet moved past this syllogism:
1. If your hypothesis is correct, schizophrenia is treatable with antibiotics.
2. Schizophrenia is not treatable with antibiotics.
3. Therefore, your hypothesis is not correct.
None of the Above asked, pretty reasonably, I think, for your method of moving past this. Is it (1) or (2) that you deny? Maybe both?
Unless I've missed some of your other writing here on in your post (which I read), it seems like the entire program you've offered so far for moving past this pretty obvious objection is the case of Susannah Cahalan, a women who was successfully treated for anti-NMDA receptor encephalitis (not schizophrenia) with immunoglobulins (not antibiotics), although when pressed you seem to have implied that an earlier course of antibiotics was what actually cured her.
So, I'm interested in anything else you want to offer that might help me move past this objection, which seems fatal, completely independently of the several other objections already raised, also fatal.
I scoffed when she announced it, but the Youtuber Lauren Kennedy who previously had the handle @LivingWellWithSchizophrenia has rebranded to @LivingWellAfterSchizophrenia following going into remission and reducing her meds all the way to zero while on a keto diet. That is the sort of thing you might expect to affect your gut, though the theory behind it just appears to be a metabolic one not involving the microbiome.
I dunno... I know two brothers with schizophrenia, and the one supposedly tried keto (for other reasons: he jumps on a lot of bandwagons), but it had no noticeable effect on his—pardon me French—craziness; the *only* thing that's helped is medication.
(On the other hand, the dude has enough trouble just complying with "take this pill regularly and stop smoking weed", so I'm not very sure how well he *actually* "tried keto"—and IIRC he didn't try for longer than about 1/2–1 month anyway, so... not exactly firm evidence.)
Yeah, she also claimed she tried it earlier but didn't do it sufficiently & for long enough to see results the first time.
Are you a practicing gut microbiome scientist with a lab?
https://drive.google.com/file/d/1nHK_yNCaYY9Ie8EzF55bzSgvqG8rTdHL/view?usp=drivesdk
Read 'em and weep bruv
Sorry, I am not going to click on a random link. I assume those are papers. Can you post a direct link to the papers? Thanks.
Lmao it's a photo of me, in my lab, holding up a piece of paper with "Hi, lightlyseared!" Written on it.
Ok, great, thanks. I put your name into Google Scholar and I see some review articles, but none seem to include any experimental work. That's why I am asking what sort of experimental work you do.
Greg Cochran is a physicist who I recall back in the day suggesting schizophrenia was caused by a pathogen because of the winter-birth effect. Now he's moved on to thinking it's mostly genetic, manifesting when someone is at the tail of a distribution of sanity (like low IQ, though I've pointed out retardation has some rare alleles of large effects we've identified).
You're about a decade late to have hit at peak microbiome hype
This makes me feel old and stupid, but I'd value an explanation of what "heritability" means and why there isn't an exact theoretical formula relating it to these binomial probabilities of Schiz/Not and Twin/Not. Why do we have to do simulations? (Except for the good reason that sometimes it's easier to do a simulation than to find the exact theoretical formula.)
From the footnotes on Scott's recent article on missing heritability:
Geneticists distinguish between three related concepts:
Polygenic score r^2 is the degree to which our current best genetic models can predict traits. You might use this to discuss the accuracy of a genetic test or an embryo selection procedure.
Narrow sense heritability is the degree to which all normal, additive genetic variation affects traits. You might use this to discuss the effectiveness of breeding programs, or how much you expect a parent’s traits to affect their children.
Broad sense heritability is the degree to which all genetic variation, including interactions and rare mutations, affect traits. A correctly-done twin study should (modulo certain small issues) return the broad sense heritability. This is useful in resolving deep questions like “How much do genetic vs. social causes affect traits”, and acts as the limit for what we might be able to explain through some future genetic science.
I'm not sure I understand your question about simulations. "Simulations" are a thing we could/would do even in situations where we have _extremely_ good physical, mechanistic understanding of the processes involved.
"2: But the twin concordance rates are pretty low - if your identical twin has schizophrenia, there’s only about a 30%-40% chance that you get it too. Is that really what we would expect from a mostly genetic disease?
This is exactly the concordance rate you should expect from a polygenic condition where 80% of the variance is explained by genes."
This is false. It's what you'd expect in a liability-threshold model where 80% of the variance in liability is set by genes. But there's still the question of what the triggering threshold is.
How are these different claims? Isn't liability-threshold just a dichotomization of saying that some of the variance is genes and some is something else? In my discount simulation, I tried to cash out "80% genes, 20% environment" as neutrally as I knew how.
Nah, L-T is a specific model of how discrete diseases work. It happens to be a broadly realistic model, but if you think carefully, you will see lots of ways it could conceivably go wrong (in worlds I expect to be different from ours).
This doesn't mean Scott's claim is false, though, does it? I.e., an L-T model may be a better fit with schizophrenia, but the concordance rate is still consonant with a less-specific model (e.g. "a polygenic condition where 80% of the variance is explained by genes") too... no?
It's like the distinction between saying "height is heritable" vs "being an NBA player is heritable". The latter would be found true on liability scale but is obviously not biological in practice.
I'm not sure what you mean by that. Being an NBA player isn't entirely due to biology. Some people just decide they really want to play basketball and practice a lot. Sure there's steps between being tall and joining the NBA, but there's also steps between having a certain set of genes and having growth hormones, and between having growth hormones and being tall.
"for example, they die about 15 - 20 years earlier, have of heart problems."
I think there's a typo here.
>there is no adoptive-parent-to-adoptive-child schizophrenia correlation
Do you have a source for this claim, or are you simply taking absence of evidence as evidence of absence? If so, please consider: they don't really let people with a diagnosis of schizophrenia adopt children very often, do they now?
In this case, shouldn't we expect to see lower incidence of schizophrenia among adopted people? This seems like a pretty straightforward thing to check in principle.
No. I’d expect that schizophrenia in birth parents is a strong predictor that their kids are up for adoption
What disruptions to transmission would you expect would result in a reduction in incidence?
Live birth normally results in a transmission of the mother's gut bacteria to the child, but C-section birth does not. In cases where mothers with schizophrenia give birth by C-section, and then the child is immediately taken away, do you think we should see baseline levels of schizophrenia incidence?
What about cases where the father has schizophrenia, but isn't present? Under a gut biome model, that doesn't seem like it should result in elevated incidence of schizophrenia.
This literature is complicated, but see https://www.sciencedirect.com/science/article/abs/pii/0022395687900896 for the study I think is most useful and representative.
“According to the consensus theory of schizophrenia, this is because schizophrenia is a neurodevelopmental disorder that interferes with adolescent synaptic pruning”—I’m unclear on what the consensus is (if any), but this sentence seems different from what’s in the link https://pmc.ncbi.nlm.nih.gov/articles/PMC9870034/
Well, here’s how I would put it: There’s some strength with which the various different areas of the cortex communicate with and constrain each other. If the communication strength amount falls below some threshold, you get schizophrenia. The communication strength starts very high but goes down thanks to adolescent synaptic pruning. And that’s why schizophrenia starts manifesting around age 18-30.
(That’s my take, which is close to but not identical with the link. For more of my own opinions see https://www.lesswrong.com/posts/tgaD4YnpGBhGGbAy5/model-of-psychosis-take-2 and links therein.)
Re this and the other day: would assume all traits are 100% genetic definitionally, even if there is any sort of environmental component to their expression - a building is made of bricks; if they vary it is because variance is allowed by the genome.
The environment can’t create information; things can be A or B because A or B is in the genome.
The genome goes back multiple generations so the percentage of inheritance is difficult to demonstrate. You can’t assume the value of any trait not expressed by your immediate ancestors is not inherited.
But presumably some genes are much more important than others in how the genome is expressed - making it difficult to compare genomes.
When you don’t know how something works, e.g. how information for complex traits is encoded in the genome - you can’t be too confident in making assumptions about inheritance.
This comment supplied as a car game by my fellow road trip passenger.
The idea of "synaptic pruning" is a new one to me. I tried to follow the link but it was very dense and I had trouble understanding it. Is there an ELI5 for "what is synaptic pruning, and how confident are we that this is a real thing that exists?"
(I guess I could ask an LLM, but this seems like the sort of thing where it would hallucinate and I'd never know the difference.)
Also, if "insufficient synaptic pruning" is a cause of schizophrenia, does that mean you can test for schizophrenia by doing a brain scan and measuring how pruned people's synapses are?
The consensus understanding of human brain development is that, at first, neurons are highly interconnected (many synapses). Each neuron has a high neighbourhood size, i.e., many other neurons can be reached in just a few hops. Human brain development is a two-phase process, where first the brain grows rapidly with many inter-connections forming, and second the brain gets rid of a lot of the less active inter-connections (the pruning step). To speak in terms of purpose, the first phase allows young humans to synthesize lots of information in order to build a general portrait of the world they live in, while the second phase allows mature humans to build the habit of determining what information is relevant to a given situation and make use of that information in a consistent way.
If the second phase doesn’t happen, you’re basically stuck synthesizing lots of information, noticing patterns, but lacking the cognitive stability to act on those patterns or filter out what’s not useful. This is a rough portrait of a person in the schizotypal cluster.
As for how we know it’s real, I believe the main evidence comes from two sources: microscopy of brain tissue at different phases of development (not just from human beings since this pattern of development is apparent in other mammals such as rats and chimpanzees); and hormones. Hormones affect the growth of neurons so it’s possible to get a sketch of the development of the brain (or any other organ) based on which hormones are dominant at which time.
To answer your last question: synaptic pruning is difficult or impossible to assess directly without direct samples of brain tissue. It affects both long and short-range connections at every scale of the brain’s organization. Brain scans are only good at detecting macro-scale phenomena (brain waves, for example, can be detected because the EM flux of millions of synapses firing in concert adds up to the point that it becomes measurable). Since pruning tends to primarily affect synapses that were quiet anyway, a difference in the degree of pruning is probably not sufficient to cause a difference in activity that would show up on modern equipment.
So interesting. I did not know any of this!
I'm confused. You seem to be describing schizophrenia as resulting from a brain that *doesn't* prune synapses during adolescence, but the linked paper attributes it to *excessive* pruning. Wouldn't that imply a brain that's too "adultlike" rather than too "childlike"?
Also, I'm confused about the timing. An 18-year-old is clearly cognitively different from a 35-year-old, but they're clearly cognitively closer to that 35-year-old than to a 2-year-old (if it's strictly linked to adolescence, substitute in "14-year-old" and the point still holds: high schoolers are kinda dumb but they're not dumb the way toddlers are).
Is the idea that a prodromal schizophrenic initially has the normal neurological processes of adolescence occur, but then these processes continue even after adolescence? Then at some point the pruning crosses some threshold and brain function degrades to the point where schizophrenia results.
Well, I guess I wiffed it including that paragraph about schizophrenia at all because my intuition was that it went the opposite way.
It looks like Scott shares my error—I don’t know how else to interpret the sentence “schizophrenia is a neurodevelopmental disorder that interferes with adolescent synaptic pruning”. It’s hard to read that as saying the pruning is positively, rather than inversely, correlated with symptoms.
I’m just skimming, but my gloss of the paper is that there are several information-rich correlations in the literature but a lack of consensus as to how they all come together to paint a causal picture. Maybe prodromal schizophrenics have a latent defect that is exposed by pruning; maybe pruning is overly aggressive and that directly causes the disorder; maybe the real problem is a breakdown of communication between the prefrontal cortex and thalamus, and the synaptic pruning follows after (because areas that lack activation are the first to be pruned). Perhaps a combination.
The upshot I’m reading here is that pruning is involved *somehow*, and more pruning is correlated with more schizophrenia, but the mechanism is not yet clear and there is likely more than one distinct etiology for schizophrenia.
This is not what I had thought previously and also not, to my ear, how Scott summarized the paper he cited. So maybe the consensus theory does not actually consensus that hard.
So what makes phenothiazines work?
Then again, chlorpromazine remains the treatment of choice for brain-eating amoeba, and nobody knows why or how it works in this instance.
Didn't know that chlorpromazine killed them wow.
With surrogate pregnancies it's possible to get the benefits of identical twins without the risks of resource depletion by having the pregnancies be carried by two different mothers. I've never heard of this being done but maybe it should start being encouraged just for the obvious benefits to science.
Did you mean fraternal twins?
No I mean splitting a blastocyst in two and implanting into two different surrogates
Surrogacy costs too much money and is too controversial itself to use it for experimental embryos or (if the splitting process were well-established) to populate twin studies.
This probably hasn't been tried with human embryos but the splitting process is probably literally you take it when it's two cells and pull it in half. Pulling off a single cell when it's just a few is already how genetic testing is done. The benefits of identical twins also extend to the twins, they like having each other.
Funny that Skolnick's theory bears a passing resemblance to Hal's condition in Infinite Jest....
Well, if you think the childhood mold dosing is what did it. I was always a subscriber of the wraith-jim-applying-DMZ-to-toothbrushes theory.
And let's not forget, Leprosy and Tuberculosis also ran in families. Being able to predict a disease with genetic tests does not necessarily confirm a true genetic disease because you could also be looking at an unusually high genetic vulnerability to an unrecognized endemic infectious disease.
As I commented above, Cochran previously though schizophrenia was likely caused by pathogens. He now thinks it's mostly genetic. I can't quickly find where I asked him about what changed his mind, but in this post https://westhunt.wordpress.com/2018/07/22/more-theory/ he writes "Again: you find that having more mutational load, more deleterious mutations, increased your chance of schizophrenia, or autism, or low IQ"
He writes "what we call a given type of mental illness, like schizophrenia, is really a grab-bag of many different syndromes" because "the majority of the genetic variance in IQ is the product of mutational load, and the same may be true for many psychological traits". https://westhunt.wordpress.com/2017/06/06/happy-families-are-all-alike-every-unhappy-family-is-unhappy-in-its-own-way/
Wow, that's interesting; I used to read Cochran pretty regularly,* but didn't know he'd reconsidered this particular hypothesis. Thanks for bringing this up! I wonder if he's reconsidered the "gay germ" hypothesis as well...
*(can't recall why I stopped—think just because he stopped updating as regularly & I forgot to keep checking? heh)
I've banned this person because I literally did not think of Cochran once while writing this post, and resent being accused of cowardice and straw-manning for something that I don't think any reasonable person would consider related to anything I wrote. If someone wants to argue that this wasn't out-of-bounds hostile and aggressive, I'll reconsider.
You should have thought of Cochran. Schizophrenia is his #1 disease most likely to be caused by an infection.
Not #1. He talks much more about homosexuality being caused by that. Also various cancers and heart disease I think were discussed in that Atlantic article with Paul Ewald.
The Cochran-Ewald paper has explicit rankings.
Those interested in this subject might consider an adjacent field that tackles a parallel debate. In Yap et al., 2021 Cell, the authors find that behavioural traits (especially restricted eating patterns tied to autism) largely explain observed microbiome differences--especially the reproducible findings of less diversity. Microbiome variation looks more like a consequence of neurodevelopmental phenotype than a cause. I think this is a cautionary tale about directionality in these brain-gut associations everyone is so hyped on. https://www.cell.com/cell/fulltext/S0092-8674(21)01231-9?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0092867421012319%3Fshowall%3Dtrue
Couldn't the causation work both ways? Autistic restricted eating leading to reduced gut bacteria which in turn intensifies some autistic traits. I saw one paper that made the argument (https://pmc.ncbi.nlm.nih.gov/articles/PMC8784764/
2). This would match up with studies that show autistic kids having both improved gut and a little less severe traits after fecal transplants (like https://pmc.ncbi.nlm.nih.gov/articles/PMC8560686/).
The Yap study is the largest human study so far (247 Australian children with shotgun-sequenced stool) which found that restricted-interest eating explained up to 58 % of the variation in microbial genes, while the autism diagnosis itself explained ~7 %, a signal that vanishes once diet is in the model. In other words: behaviour → food choice → microbiome.
Hsiao’s “why-not-both” reply grants those numbers but points to mouse experiments from her group where a single bacterial strain or cocktail “rescued” autism-like behaviours. Those dramatic effects have yet to be reproduced by independent labs (as far as I know), and the same investigators hold equity in start-ups developing microbiome therapeutics potentially cause for healthy scepticism. See the myriad caveats with the rodent models here: https://pubmed.ncbi.nlm.nih.gov/30188509/; https://pubmed.ncbi.nlm.nih.gov/32335198/; https://pubmed.ncbi.nlm.nih.gov/29166585/; https://pubpeer.com/publications/B521D325772244D8F656F1ED193ACA?utm_source=Chrome&utm_medium=BrowserExtension&utm_campaign=Chrome
What about the human FMT study?
• Who was treated? 40 Chinese children with autism and co-occurring chronic GI complaints; no placebo arm.
• What changed? Parent-rated GI scores fell 35 % and stayed down; core-autism severity (CARS) dropped ~10 % but half that gain was gone eight weeks later.
• What didn’t change? Overall microbial α-diversity; transplanted microbes faded unless they matched the child’s diet.
Relieving constipation is great, but the data do not support a microbiome-mediated alleviation of repetitive, communicative, and social behaviours as claimed in the mouse model.
Why FMT can’t stick in severely selective eaters is illustrated by a classic study of Tanzanian Hadza hunter-gatherers https://www.science.org/doi/10.1126/science.aan4834: when the menu shifts with the seasons, the gut community shifts with it: diet, not the person, is the dominant architect. A child who eats only one brand of yoghurt and four additional beige foods cannot support new microbes delivered by capsule; these microbes will soon disappear.
A deeper issue is degeneracy: biology often achieves the same function with very different internal states so multiple, even non-overlapping, microbiomes can all count as healthy. Until we map that normal phase-space, many disease/disorder associations may simply flag rare-but-healthy corners. (Though obviously not the case for the kiddos who have ARFID; their microbiome cannot reflect a healthy state given the level of dietary restriction and they need immediate and extensive nutritional support).
Meanwhile, regulatory reality favours hype: probiotics already skirt FDA oversight, and “live biotherapeutic products” could follow with modest evidence (a placebo-friendly business model!)
So can causation run both ways? Biologically, yes. Empirically, the only arrow that survives careful adjustment points from diet to the gut. Treat constipation because it improves comfort; use behavioural feeding therapy to widen diets (the surest path to a stable microbiome); identify any other co-occurring GI issues and treat those (allergies, food sensitivities, autoimmune disorders etc. .) but we don’t have the evidence to support all the hype about gut therapeutics in my opinion. We are still very early days in this field; I am open to changing my view with new evidence:)
I appreciate a good "no this isn't it" explainer.
Isn’t this an easy test to setup? Give multiple antibiotics with wide range/coverage that eradicates all gut bacteria and then see if the schizophrenia recedes over time?
Koch's postulates! But no I'm pretty sure the claim that subject would have to be sterile at birth to account for the whole course of neurodevelopment in childhood.
Not so, Cochran's model of the condition postulates live microbes modulating neurotransmitters, not a developmental influence. It doesn't explain the presentation at early adulthood, except maybe through a two-strike theory.
That said it would be interesting to give a non-schizoid person a light treatment of schizophrenia drugs and see what happens to their gut flora.
His theory of homosexuality involved an insult to the brain rather than live microbes. Similarly, polio is a gut infection which reaches a dead-end in the nervous system, but causes damage when it wanders off course to there.
Pretty sure this has been done. Pretty sure it's medically unethical and "not allowed for research purposes" but ... it's been done.
> (during residency, I met a schizophrenic patient who, whenever he wasn’t restrained, would eat as much glass as he could until he got internal bleeding and had to go to the emergency room. I don’t know what kind of microbiome disruption this causes, but I bet there aren’t many non-schizophrenics who have it.)
Sounds a lot like Pica to me. Plenty of people have Pica without being schizophrenic.
I wondered for years why I'd sometimes have such strong cravings to chew ice; turns out I'd been anemic due to undiagnosed ulcerative colitis.
I assume the patient had other symptoms of schizophrenia and they didn't just diagnose it from a compulsion to eat glass - even as a resident I think Scott would have recognized if the latter were the case
Sure. I'm not disputing that the guy was schizophrenic; just that "I bet there aren't many non-schizophrenics who have [a compulsion to eat stuff that's not food]" sounds odd if you're aware of Pica.
ah ok, sorry, I misunderstood
It wasn't just "[a compulsion to eat stuff that's not food]" but a compulsion to eat broken glass despite the obvious and immediate terrible effects of doing so. I'm not sure if ordinary pica ever stretches into broken glass.
perhap there be no schizo
I really want the cat one to be wrong.
Schizophrenia is fascinating because it makes the brain go wrong in such odd ways: it's not just a meltdown of all capacities. Either we'll learn lots more about the brain before we understand schizophrenia, or understanding schizophrenia will clarify many other things about the brain.
Isn't schizophrenia more common in men than in women? That would seem to strongly contra-indicate a gut microbiome cause, given that babies crawl around scraping up whatever microbes are around on a purely egalitarian basis.
It's bizarre to read that Skolnick describes microbiological inheritance as "you picked it up when you were crawling around and putting everything in your mouth." It's reasonable to think that you might have acquired some of your mother's gut microbes in utero -- I'd think that would be closer to inheritance than Skolnick's explanation. Or did I miss something?
It is not reasonable to think that you acquire your mother's gut microbes in utero. Best evidence to date supports the notion of womb as sterile.
Okay. But then what would be the mechanism by which microbial inheritance occurs? I mean, if I'm a kid scrabbling around on the ground eating dirt with my similarly aged next door neighbor, they'd be "inheriting" the same microbes I was. That sounds more like nature than nurture.
Point of clarification worth mentioning here: Most of the bacteria that make up a healthy human gut microbiome are not really found anywhere else on Earth aside from the mammalian gut. There was some research a few years back suggesting that live bacteria are transmitted in milk, which would be a fantastically elegant explanation—but I'm somewhat skeptical of that notion now. Most of the bacterial DNA that we find in breast milk does appear to be contamination from skin surface, backflow from child's mouth, etc.
But what we do know, from deep-sequence longitudinal tracking of mothers and their infants over the first year of life, is that you definitely do get your parents' strains.
(https://www.sciencedirect.com/science/article/pii/S1931312823000434)
(https://www.cell.com/cell-host-microbe/fulltext/S1931-3128(24)00176-8)
The how, as much as I hate to say it, is still sort of a mystery. It may come down to stray farts, or picking your nose and eating it. (Weirdly well-conserved behavior. Why? Hmm.) This is an incarnation of the notion that "everything is everywhere and the environment selects"; our bodies put a lot of effort into creating the right niche, so that when you are born basically sterile, all it takes is a few microbial cells getting into the right place in order to take root and persist for life, while the immune system sculpts away most of the potential pathogens.
Note that this does allow for substantial non-vertical transmission; I spent probably three days a week at my friend David's house through most of elementary school, and I strongly suspect that if you sequenced our guts, he and I would have a few strains in common. Maybe one in my gut came from me using the bathroom right after he took a shit, while the flush-cloud was still in the air.
Maybe I got a strain from grandma Zhang's red bean buns, because one time she didn't wash up before making the dough. As best I can tell, it's the billion little ways the world is dirty (but growing less so all the time).
Thank you, that's very informative.
sniffling and swallowing seems more common and persistent than picking and eating, for the same effect and smaller gross-out factor
I want to respectfully push back on your point with a parallel that might be getting overlooked:
ChatGPT points out that monozygotic (MZ) twins can have more similar gut microbiomes than dizygotic (DZ) twins even beyond their genetic similarity—possibly due to shared epigenetics, early developmental environment, or immune/mucosal traits.
I wonder if this same logic might help explain part of the missing heritability problem. Could subtle, non-genetic but deeply shared developmental factors be amplifying trait similarity in MZ twins in ways that our models still treat as purely genetic?
In that case (replying to your last paragraph) the explanation is going to be Jungian type (say, MBTI), presumably shared by MZ twins (given they look so similar), but not showing any particular similarity for DZ twins.
(however, if I understood last week's missing-heritability post correctly, the message of twin studies is confirmed by adoption studies)
Isn't Jungian type of the form: "they look alike, so they might be treated similarly or develop shared self-concepts based on appearance"? If that's what you mean, then no, my query is unrelated to that type. I'm asking about developmental factors like epigenetics, shared intrauterine environment, developmental synchronization, microchimerism, and maternal adaptations for having identical twins.
This is another example in a line of examples of some guy confusing (weak) correlation with causation. Or of invoking mono-causality in the absence of evidence.
It is widely claimed that autoimmune diseases are more common in women. Is there a theory of this?
But this depends on what you label autoimmune. Ongoing autoimmune diseases like lupus are a lot more common in women, but acute autoimmune diseases like narcolepsy are slightly more common in men. Schizophrenia, with its abrupt onset appears in the second case. Indeed, it is more common in men. Moreover, when it occurs in women, it occurs maybe 10 years later. Do theories of sex differences in ongoing autoimmune diseases shed light on acute ones?
Testosterone seems to impair the immune system.
There's a lot of different kinds of impairment.
If testosterone down-regulates the immune system, than males would have more infections and less autoimmune diseases. But this doesn't explain why they have more narcolepsy and schizophrenia. And if it's as simple as down-regulation, why isn't there some other mechanism to up-regulate it? Maybe not all the way back and maybe with a lot of noise, but wouldn't this predict more autoimmune diseases?
This paper argues that testosterone selectively down-regulates the more metabolically costly immune responses:
https://pmc.ncbi.nlm.nih.gov/articles/PMC5075254/
So it's not up-regulated because that would be imposing the cost that the down-regulation is intended to avoid.
For a serious disorder to be polygenic with a wide variety of environmental triggers is frustrating, but our frustration is not an argument that matters are otherwise.
This is a very interesting take down of Skolnick. Thank you. The one nitpick I have is that 'cat exposure,' which you mention, is an example of microbiologically mediated schizophrenia via toxoplasma gondii CNS infection. So it supports the hypothesis that some instances of schizophrenia are microbiologically mediated. Though not by R. gnavus or gut bacteria specifically in those instances.
Are we allowed to post memes in the comments? By point 5 I really felt a "stop! stop! he's already dead" would be appropriate.
In any case, a very convincing takedown.
So Skolnick thinks gut bacteria are behind schizophrenia, huh? Do they serve snacks too? Anyway, if you're looking for something to spice up your discussion on microbial theories, check this out for some star symbols to shine a light on those wild claims <https://starsymbols.io/