@NicoleCRust @summerfieldlab

The take-home message for me, personally, from results like this has always been to stop, think, and be deliberate.

The circumstances of the experiment are designed to provoke this kind of automatic processing and justification so it can be studied, which is awesome.

But it doesn't mean that we're necessarily totally trapped in reflexive rationalization in all cases. It does mean that if you haven't cultivated the conscious habit to avoid some of these traps, you will probably fall into them. (Some you will probably fall into anyway.)

@erinnacland @david_colquhoun @academicchatter

I do so wish that all science Ph.D. programs, at least, required a rigorous but highly applied statistics course (including data sets with a realistically complex structure) so that we would stop seeing literature filled with bad inferences from improperly-used statistics.

"Condition A has p < 0.05 compared to control, but condition B doesn't, so we can conclude that A and B are different" is a really common and wrong inference.

(Amusingly, or perhaps depressingly, ChatGPT does better than a lot of authors of Science and Nature articles.)

@joanagsa @ERC_Research

One of the really tricky things about this type of research--which isn't really covered in the highlight--is how to distinguish different models for why "disinformation" is shared.

In particular, the key question to me seems to be whether truth content matters, or whether it's purely a difference in style: disinformation tends to be styled a certain way, but if accurate information appears that bears superficial resemblance to that style, does that actually matter?

For instance, if you have a surprising important-if-true bit of information, does it matter to sharing patterns *whether* it is true? Of course disinformation has an advantage in that it can hit the surprising-and-important buttons harder (being unconstrained by reality and all). But if people who avoid sharing disinformation are actually just avoiding sharing *anything* that isn't common knowledge and widely accepted (or is only personal), that raises the question of how anyone would learn anything surprising to begin with, at least via social media.

Given that we exist in a world with both conspiracies and conspiracy theories, and with claims that fall into all four categories of widely-believed-and-true, widely-believed-and-false, counter-mainstream-and-true, and counter-mainstream-and-false, understanding what (if anything) selectively enriches for the true/false signal independent of the others is of the greatest benefit to society.

Rex Kerr boosted

Elegant bees–and minute mathematicians. Remember that these little critters understand the concept of zero [1] and can add and subtract [2].

Amegilla quadrifasciata, the white-banded digger bee inaturalist.org/observations/1

[1] Howard et al. 2018 "Numerical ordering of zero in honey bees" science.org/doi/10.1126/scienc

[2] Howard et al.2019 "Numerical cognition in honeybees enables addition and subtraction" science.org/doi/10.1126/sciadv

#iNaturalist #Catalonia #nativebees #Hymenoptera #entomology

@freemo Any phrase used as a tribal identity marker tends to attract stupidity because the point is allegiance not analysis; any topic of considerable depth discussed in 280 or less character chunks tends towards stupid because there isn't enough space to advance sufficiently nuanced ideas; and in this case the starting point is counterfactual, which poses serious challenges for non-idiotic support.

So, yeah, if it was anything other than a maelstrom of stupidity, we ought to be mightily surprised.

(I do note with displeasure that a lot of people more on the sane side of things tend to overestimate the degree to which one can conclude things like "vaccines are totally safe" from observations like "the anti-vax side is comprised of idiots". That's not how it works! Is there a talkorigins.org equivalent for Covid vaccines yet?)

@ceoln @rudyschwartz @freemo

I also think we can do better.

To be clear, the left's scientism here--as opposed to actual science--is in taking superficial and/or outdated perspectives, basing ritual behavior on them, and claiming it's science.

This has all sorts of pernicious knock-on effects that don't rise anywhere near the level of the huge death toll caused by the right's anti-vax anti-science perspective, but at this point I would argue are a comparable problem, (in part because it makes it almost impossible to combat the right's anti-science attitude).

Scientism is an anti-science perspective: it says that if you kinda go through the motions and say the right words, it's as good as continually refining your understanding through theory and experiment. Because it nominally is science-supportive, it exerts its negative effects by example rather than by exhortation. Because it's not actually based on anything solid, it has to be maintained by appeal to authority and/or social pressure rather than explanation, which also isn't good for a scientific outlook. In particular, if you oppose an anti-scientific political position with a scientism-based position, that it's not scientific will often be readily apparent, emphasizing that "the other side" doesn't actually care about the science either, reinforcing the original perception that something was off about science anyway.

Here are three examples of how this has played out in a negative way, two of which I've given, but I wanted to try to present the problems more clearly than I have previously.

(1) Ineffective vaccine requirements that are nonetheless requirements. As a hypothetical example, barring an unvaccinated 20 year old who has had Covid recently from university, while admitting an 80 year old who was (fully) vaccinated 20 months ago and hasn't had a booster, is what UO's policy calls for. If we think a lecture in a large lecture hall is a potential superspreader event (it is), and that attendees are in danger (they are), and that it's our responsibility to enforce behaviors that help mitigate the danger (debatable because enforcement always has downsides), then this is not a sensible way to do it. The students are probably okay anyway, and if we want to protect the 80 year old (professor?), that mandate isn't helping them at all. If everyone's very clear that this is some simple-to-enforce university policy that's rather divorced from how things actually are, sure, no worries. If criticizing the policy gets you shouted down for being "anti-science", that is a problem. (I don't know about UO specifically. I have witnessed substantial hostility directed against people who are critical of this sort of mandate.)

(2) Mildly effective masking requirements where all the focus is on masked-vs-not and not on effectiveness. Again, if it's clear that this is done for political expediency because "it probably helps some so we'll do it, but it's too much effort to do it well, so we'll do it poorly, and by the way we don't care if someone really hates masks", sure, no worries. If, however, the message is that the policy is "because of the science", no, it's loosely motivated by science and shouting people down who object pulls the wrong way. Unlike the vaccine one, where the problem is that having people vaccinated only protects them not everyone else (over the timescales that are relevant now), the problem here is that asking for actually effective masks *does* protect everyone else but we *don't* ask for it!

(3) Countering brainless anti-vax rhetoric with equally brainless anti-anti-vax rhetoric. Anti-vax sentiment is a real problem on two levels: one, it has led to a lot of unnecessary deaths, and two, it has led to a lot of unjustified skepticism of science. Unfortunately, the anti-anti-vax stance, while it might have been somewhat helpful in reducing deaths, isn't helpful in being pro-science. For instance, there's a very negative attitude towards reports of vaccine side-effects without regard for who is reporting it, how well documented it is, what the nature of it is, and so on. That's also anti-science! It also leads us, overall, to fail to ask important safety-related questions like: the CDC claims that it uses Rapid Cycle Analysis to detect vaccine-associated risks on a timescale of a couple weeks from VSD data; and people used VSD data to confirm myocarditis and GBS as rare side effects so the signal was in the data; but RCA was NOT how we learned about these side effects! So...is this system actually up and running?

You see another example with climate catastrophism developing on the left, to the point where I've had people tell me I'm anti-science because I'm quoting primary research and/or the relevant sections of the IPCC AR6 WG1 report to show why they're drastically overstating the case. It's bizarre. It's still pretty niche compared to climate denialism on the right, fortunately. But it's indicative of something really starting to go wrong with the left's relationship with science.

Note to self: this message is practically incoherent. When sleepy and rushed *just be quiet* rather than failing to express thoughts clearly.

@ceoln @rudyschwartz @freemo

Do you have solid evidence that at *this* point Covid anti-vax is actually a substantial medical problem? The immunological history of people is so complicated now w.r.t. Covid-19--keeping in mind all the asymptomatic cases that nobody ever detected!--that it's really hard to do a careful study. Nonetheless, exposure rates are getting close to 100% in populations where it's measured: cnbc.com/2022/03/29/cdc-majori

This suggests that the big gains in protectiveness--where you can use vaccination to avoid having immunologically naive people catch Covid, with concomitant risk of death and severe disease--have already been made, and what is left to do is perhaps try to reduce somewhat the impact of repeat infection by repeat vaccination.

Of course, the anti-vax-in-general people are still a problem, but at least where I am this cuts across political outlook, and if anything is more associated with a kind of kooky leftist natural crystal energy type perspective, or the view that the government has in the past taken harmful actions against some community so therefore this must also be some sort of trap or hostile action, than a no-gummint-gonna-tell-me-what-to-do-with-my-freedom irascibility.

@rudyschwartz @freemo
The degree to which it's amok may have been overstated, I confess.

The direction is wrong, though.

Having a mask mandate for small groups, but not an N95 mandate for shoppers, is a public health hazard, and indicative of scientism because it's something that looks good if you "believe science" but otherwise is poorly targeted to be effective.

I don't think getting tired of the fuss and trying to get life back to normal while taking largely ineffectual "science"-backed measures as an absence of scientism, nor do I think it has negligible health impacts--the velocity of infection has a sizable impact on the fraction of the population who gets infected, and even if people aren't dying right away, there's elevated all-cause mortality and long Covid to contend with.

@ceoln @rudyschwartz @freemo

Okay, that's fair. In part I was extrapolating from older behavior before I'd checked to see what it was currently (always a dangerous thing), and a few of the more egregious things that I'd remembered have since been fixed.

Also, the anti-vax movement was *really* bad. It's hard to compete with that. If we take it as water under the bridge and ask about the ongoing impact *now*.

But it isn't fair to pick a particular slice in time in the past, and compare it to the impact of right-wing anti-science right this second, and I think I was partially guilty of doing that. So, mea culpa. I should have gotten the timeline straight in my head first before giving these examples.

(They're still examples of pulling the wrong way, just well short of what was asked for if placed in proper historical context.)

@ceoln @rudyschwartz @freemo

I was talking about two different things.

(1) The UO requirements for vaccination are close to useless for safety and somewhat burdensome but not particularly harmful

(2) Personal anecdotes about medical exemptions NOT being available (e.g. you can't come on campus, and it's not possible to do this work remotely because, say, it's a lab class, so...).

@rudyschwartz @freemo
I was talking about masking, not post-vaccination behavior. Maybe you haven't witnessed large crowds of densely-packed cloth-or-medical-mask-wearing leftists during holiday shopping in the middle of a huge spike of Omicron. I have!

One of the biggest and sharpest spikes of Omicron anywhere in the United States was in the San Francisco Bay Area--and SF and surrounding communities saw fit to institute a mask mandate until the case counts went down (mercurynews.com/2022/01/27/san), but not mandate any particular type of mask.

You could argue that in a highly vaccinated population like in that part of California, given a less dangerous variant like Omicron, stronger measures weren't necessary...but...it's very weird to decide that a high-effort modest measure is necessary but a low-additional-effort very effective measure is not necessary.

@rudyschwartz @freemo
In terms of general pointlessness, the University of Oregon policy is a good example: hr.uoregon.edu/uo-covid-19-vac

You have to have been vaccinated twice, but boosters aren't required (any longer?!), and it doesn't matter how many times or when you've had Covid. This really doesn't have any basis in actual science. (But it doesn't have nearly the negative impact of an anti-vax stance.)

In terms of actually getting in the way of employment and delivery of important services, the cases I know about were earlier in the pandemic; I checked the places and they now, at least, list a reasonable policy. I don't have access to the people's personal correspondence, so I can't tell whether they were getting told different things privately than the institutions were saying publicly...but I do know that they weren't against vaccines in general, and also were impeded or unable to do their job. So, anecdote-level; take it for what it's worth.

Note, however, that what is being criticized is not leftism, but a leftist embrace of scientism--you can't assume the latter just because you identify the former. Since I don't live in B.C., I'm ill equipped to get a sense of things there.

@rudyschwartz @freemo
It's harmful because it gives the illusion of protection, lowering people's defensive behavior, causing them to catch Covid...and causing others to catch Covid because we only made them wear a cloth mask.
I don't think it's overall more harmful than the anti-vax stuff, but it's dangerous, and in the same high-Covid direction.

@rudyschwartz @freemo
No, they're not saying to disregard the advice of doctors.

They're saying you can't, for instance, teach unless you're vaccinated and boosted. If you shouldn't get boosted yet because your immunity to Covid is too high because you caught it, well, too bad for you. No teaching because "follow the science".

@rudyschwartz @freemo
Oh, oh, I've got another!

Gotta mask (sometimes)! Okay, sure, the mathematics and viral particle transmission studies and the epidemiology on that are convincing enough. The Bangladesh study is good!

The Bangladesh study, which showed that cloth masks don't do much, but surgical masks help some. Coupled that with the viral blocking studies that show that N95 masks are waaay better than surgical masks--going from medical to N95 is a much bigger improvement than going from nothing to medical.

So, in high-density environments, we tell people to wear N95 masks. Right? That's the requirement, isn't it? Leftists? Please? Maybe for visitors to hospitals with vulnerable populations? Please??

No! "Wear a mask!" That's it. You can't have any more refinement than that. That's the science for the general public: wear a mask. (Obviously medical staff wear appropriate PPE, including well-fitted N95 masks when appropriate.)

Just once, I saw one attempt at doing something better: you have to take off your mask--even if it's an N95 mask!--and wear our hospital-supplied medical mask. I don't even.....

If you care enough to make someone put a piece of cloth over their face that barely does any good, supposedly for my safety, why won't you make the same person do something that *actually* helps a lot with my safety?!

@rudyschwartz @freemo

I've got one (on Covid): the long-running refusal to alter policies about vaccination that were justified by the initial very high effectiveness of the mRNA vaccines in actually preventing acquisition of Covid in a population mostly naive to the Covid antigens.

It's not about saving people from giving it to each other, because we're not checking that everyone's had a booster within 3-4 months, and after that the protective effect size is roughly zero.

It's not about immunity, because for about 3 months after you've had Covid you're supposed to wait to get vaccinated so you get a better memory response (sooner than that and the residual antibody levels from the infection are already too high)...but...if you have a vaccination requirement, that you're too Covid resistant doesn't count as being vaccinated even though you're now *actually* less likely to give it to anyone else.

There are some people (including one I know) who really can't/couldn't take the vaccine (on doctor's orders!--significant but not life-threatening side effects on the first dose, for instance) who were providing health care or education who basically had to sit on their hands, not helping people because of this.

In terms of actual death toll, it's nowhere near the huge and terrible number caused by right-wing anti-vax sentiments, but on the other hand, we do generally provide pretty large allowances for people to suffer the consequences of their own stupidity.

In terms of completely pointless disruption checking something that isn't relevant and occasionally causing bad outcomes, with absolutely zero redeeming features, this one ranks pretty high. Medicine says: this person shouldn't take a vaccine right now. Research says: they aren't a danger to others and depending on details not themselves either. Left wing scientitism-driven policy says: "science says you gotta vax, so you gotta vax".

As Freemo says, the nuance is important. People die when you get important nuances wrong.

(Aside: I'm worried that the use of ivermectin to treat terrible nematode diseases in humans will be hampered by all the rhetoric about ivermectin being just a "horse dewormer". It's a really important medicine for people, too! Just not for Covid!)


I think the point is that the space is really really interesting. It has in-bounds points that nobody has sampled, yet which are interesting to people.

The trick is that most of the points in, say, the space of all reasonably grammatical sentences are completely nonsensical ("The car paints a loud idea."). However, there is am interesting manifold within that space along which human discourse lies, and being able to pick points from (close to) that manifold is often really interesting.

However, the manifold is so convoluted that prior to ChatGPT-type approaches, no machine learning method got anywhere close. Now, shockingly, we actually get points on or close to that manifold, and you can ask for them by giving text prompts that form a human-language context.

So in one sense they can't do more than what they've been trained with. But in another sense they can, because they've learned the shape of a super-interesting manifold and you can ask for it to pick parts of the manifold that nobody has ever produced before.

I don't think either extrapolation or interpolation is a good way to think about how this works. These intuitions are generally formed in extremely low-dimensional spaces (e.g. R x R), and those intuitions just don't translate to how ultra-high-dimensional spaces (with very high dimensional yet vastly vastly reduced dimensionality manifolds).

@NicoleCRust @albertcardona @matthewcobb

For example, if we're interested in cortex, we have to grapple with the fact that cortical connections are broadly specified during development, extensively modulated by activity, and may contain some intermediate representation and state of some computation that isn't easily connected either to outside stimuli or to goal-directed behavior. So we have a mixed problem of development, synaptic plasticity over multiple timescales, highly parallel input and output...and have a hard enough time even measuring the broad outlines of the basic rules like synaptic pruning, let alone how the system was built.

So, what is the operation of cortex, or a cortical column? Well, maybe the cortical architecture supports, as is, with known mechanisms of synaptic plasticity etc., a variety of general-purpose computations and the only thing we're really missing is the correct parameter ranges and initial network connectivity to support this function (but we'll never know without better data than we can get right now).

Alternatively, maybe the largely reductionistic approach taken by neuroscience has led us to missing systemic effects (LFPs, neuromodulators, etc.) that play a critical role in how the circuit functions, and if we had better measurements of such things (both what they are and the magnitude of their effects), we'd figure out the computation.

Or maybe we're not even conceptualizing the "computation" the right way at all. Maybe the analogy we draw to input-output devices like transistors or functions are poor ways to model how the brain works, and there's an emergent property of the system that will give us better explanatory power. But if we came up with a much better model, unless we magically manage to nail all the free parameters perfectly, would we actually agree with the experimental data we can collect any better than in the first case where we guess that each neuron is integrating dendritic inputs and producing spikes and that's basically all we need to know?

@NicoleCRust @albertcardona @matthewcobb

Do we *really*, though!

I'm very partial to Sydney Brenner's quote about progress in science: "Progress in science is made through new technologies, new discoveries, and new ideas, probably in that order."

To me, the biggest deficit--not that theory is unimportant--is in being able to experimentally address the problem at the scale that we know is important. It's kind of like we're trying to understand weather dynamics by giving everyone in London a thermometer (and not bothering to tell them to keep it outside), putting a wind-speed meter in Glasgow and in Miami, and observing the clouds from a lunar observatory.

We know for certain that there is lots of feedback, that connectivity is elaborate and important, and that gain control can be large--in short, we have every reason to believe that details ought to matter, and mostly we can't see them, which makes it very hard to discern how well we actually understand "how the brain works". If current (detailed) theories were fully adequate, I'm not sure we'd know!

So to me, we need ideas not about how the brain works but how to couple what we can actually observe with somewhat reducing the space of possible models. (Plus I second Albert's recommendation to invest effort in simpler systems.)

Show more
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.