（评论）

（评论）
(comments)

原始链接: https://news.ycombinator.com/item?id=40438510

本文讨论基因和文化对行为和智力的影响。作者分享了她的家庭成员对某些职业和爱好有着无法解释的迷恋的个人经历，这表明了遗传的影响。她思考智力的定义和测量，质疑人类和机器之间的区别。作者回忆了小时候大小便时避开潜在危险地形的经历，引发了对假设和认知局限性的反思。最后，她引用了一篇有关西澳大利亚州州长的文字中看似不一致的声明，引发了对“尚未”含义的质疑。总体而言，本文探讨了遗传、感知和智力的主题。

What amazes me about spiders is their complex unlearned behavior. No trapdoor spider learns how to build a trap. No orb weaver learns how to spin a web. Trap/web building behavior is entirely programmed in their DNA.

I think about this a lot in the context of conversations about intelligence. If spiders can have complex behaviors hard-coded, humans certainly do too. (In other words, the Tabula Rasa theory is wrong.) The ability to learn language and emotion are certainly two examples. We are pretty good at certain things (learning language, picking up social cues) and relatively bad at others (calculating an 18% tip).

So if you’re going to measure “intelligence” the first thing you’ll need to do is choose what to measure. What questions do you ask? You might be inclined to pick things we humans think are important. But then that’s not an objective universal measurement, at best it’s yardstick for human cognitive abilities.

I wonder if the process of building a complex web can be reduced to a simple set of rules, almost like a cellular automata.

The simple rulesets which lead to complex outcomes in nature is amazing to me.

Many years ago when I was working IT for a public school district, I was unpacking a large number of identical HP PCs from cardboard boxes. I noticed that a large portion (it must have been half or more) of the boxes had the exact same type of spider in a nearly identical web in the exact same spot in the cardboard box. It was a small cardboard cubby taped onto the inside of the large exterior box, for storing a power cord and other small accessories. It must have been just the right shape and size, and accessible in just the right way, that this type of spider absolutely loved it.

simple nets for sure, but spiders can easily deal with an uncountable number of corner cases, I somehow think it's not reducible to all-too simple rules (in the 'game of life' sense) and more an interaction of several non-trivially complex systems like their various perception systems.

obviously their neural networks are limited in size and therefor complexity has to remain somewhat simple relatively to larger animals).

I kind of wonder whether certain cultural artifacts might be genetically inspired. My son, when he was young, without any prompting wanted to be an architect (like his great-grandfather whom he had never met, nor even heard anything about before this) and was obsessed with guitars (like me, although at that point, my own guitar had been in the basement for years and he had never seen me play it). This is not to say that these particular behaviors are explicitly encoded in genetics (both guitars and architecture as a discipline are much younger than evolutionary time), but rather that some intrinsic predisposition towards them exists (so, perhaps had we lived in fifteenth-century Persia, we would be oud-obsessed, to choose one posseible alternate reality).

> If spiders can have complex behaviors hard-coded, humans certainly do too.

Depends what you mean by complex... All creatures get born with an idea how to get food. Human babies have never seen breast or anyone sucking breast, yet they can go through the process pretty much immediately. That's a very clever movement / sensation following combination, but it's there from the beginning.

Babies are definitely hardwired to be attracted by the contrast of skin / areola on the female breast. Once they have targeted this contrast, they will do a very typical (also hardwired) "search" for the nipple: they come very close to the breast, open their mouth, do exploratory sucking via fast breathing, and quickly move their head left and right until they have "docked" on the breast. They start doing this minutes after birth (they are also hardwired to search for the breast when placed on the mother right after birth), and stop once their sensory systems are fully developed and they have learned how mom looks, how mom smells, and where mom's breast is (after 1-2 months). On my own child, I even noticed how it once mistook the snout of a teddy bear for a female breast.

The "will suck at anything" instinct is IMHO a fall-back mechanism if for some reason a female breast is not available.

Here's a source for the above claim that I found after looking this up:

>During pregnancy, the nipple has become more pigmented 68 and is easy for the newborn infant to discover (Fig. 1C). We have observed that soon after birth, the areola expands and takes a bulb-like shape (Fig. 1G). The Montgomery glands also become more pronounced (Fig. 1G). The scent of areolar secretions has been linked to behavioural responses, such as head turning 69 and directional crawling in newborn infants 70. This release of the breast odour by the Montgomery glands is known to help the newborn infant find the nipple 17, 27, 69. The newborn infant recognises the scent of the mother's breast from the amniotic fluid 71, touches the breast and transmits the taste of the breast to the mouth (hand-to-breast-mouth movements) 17. This stimulates rooting and crawling movements in the newborn infant to reach the nipple. The connection between the taste of amniotic fluid and the scent of the breast from the Montgomery glands highlights a biological survival mechanism – a pathway of flavour with lifelong consequences. (Fig. 3) (17, 71-80) When knowing of this sensitive odour-dependent mechanism, it might be wise not to interfere with unfamiliar hands.[0]

[0] https://onlinelibrary.wiley.com/doi/10.1111/apa.14754

> If spiders can have complex behaviors hard-coded, humans certainly do too.

That does not follow. It's a logical fallacy. ('Spiders can have eight legs, so humans certainly do too.')

It's only a logical fallacy if it is taken to be the entirety of the argument rather than an abbreviated representation of a longer chain of reasoning. I don't think anyone would claim that sentence is self-evidently true!

in the limit the distinction between something that is learned and something that is programmed genetically is somewhat a matter of definitions. if a species genes makes it so they create a complex social web that teaches all its individuals certain skills, is that genetically programmed or not?

free will wouldn't even change that, since it's presumably also given by our genes.

Carl Jung held this view too. And he knew his teacher Freud disagreed. For instance a Carl Jung thought a humans fear of spiders in innate, and not learnt.

The same as how a bird can be scared on snakes. Interestingly native birds in Australia aren’t scared of humans.

> For instance a Carl Jung thought a humans fear of spiders in innate, and not learnt.

I don't have fear of spiders (was surprised by how many people do fear spiders). When I was young, I "hosted" a bunch of them under my table. Thinking of it, I think most people have their fears transmitted through their mothers. My mom certainly didn't fear spiders but she hated their "dirty" webs.

For example I do have a strong reaction to cockroaches till this day; a harmless insect that my mother fear. I can't even look at the beast. My cousin do not and conversely do fear spiders.

We struggle to define intelligence, much less measure it.

When I was at school a million years ago "smart" kids were ones who learned to read fast, did numbers well, and had good memory recall.

Then along came computers, which had perfect memory, did numbers really well. So "intelligent machines?"

Obviously not, so we tweaked our measure of intelligence. Perhaps to include writing well, or learning quickly. Which, um, computers now do.

But we still view the computers as "not intelligent" but I'm not sure what measure we'll use now. Something to do with art or music? Hmm, no, wait, that's no good either....

(PS schools still mostly measure numbers and memory....)

Schools deem a child smart when that child excels at tasks that the school assigns to them. I've those tasks were simply to read and add. Then those tasks included writing and remembering. What tasks do schools demand of children today?

It seems that intelligence is a term that we apply to people who consistently excel at the mental tasks assigned to them, no matter what those tasks entail.

Who struggles to do what for which purpose?

Sometimes the word “intelligent” is used by someone who sells IQ tests.

Sometimes the word is used by someone who sells the idea of a computer that can think like a human.

I don’t know if people in general struggle or care about the definition.

"But we still view the computers as "not intelligent" but I'm not sure what measure we'll use now. Something to do with art or music? Hmm, no, wait, that's no good either...."

We could focus on whether the student has actually understood the topic and is not just able to produce nice sounding words and some quotes about it.

How do you differentiate between a student who actually understands the topic and one who is just able to produce a convincing essay about it?

More generally, how do you measure ‘actual understanding’, if not by judging the student's output?

"if not by judging the student's output"

Where did I say to not do that?

And someone who can produce a convincing essay usually has understood the topic (unless he copy pasted it). And when there are doubts, one can ask that student specific questions.

> Where did I say to not do that?

You asserted that the student should “not just [be] able to produce nice sounding words”. What do you suggest output is if not words (nice sounding or not)?

> And someone who can produce a convincing essay usually has understood the topic (unless he copy pasted it).

Then, what is the difference between a convincing essay (‘nice sounding words’) produced by a human, and similar words produced by an LLM?

After all, even today, students are passing ChatGPT’s work off as their own. Does this then demonstrate understanding by the student or by ChatGPT?

> And when there are doubts, one can ask that student specific questions.

If an LLM can answer specific questions about a subject, does this demonstrate true understanding? If not, why not?

"If an LLM can answer specific questions about a subject, does this demonstrate true understanding? If not, why not?"

If it could, I would say there is understanding. But in every area where I have expertise, I see LLMs failing my questions. Just like a bad student would, who memorized textbooks, but did not connect or process the knowledge in any way.

Curiously, though, we don't really know it is coded in DNA.

Michael Levin has been discovering lately that biology is more complicated than we thought. His YT videos are mind-boggling. Flatworm genetics are a dog's breakfast that the cells work around by means nobody understands.

What do you mean by this specifically? What claims does he make? It's hard to respond to this just based on what you wrote. I don't think biologists have claimed that biology isn't complicated, and I don't think anyone claims that DNA isn't complicated either. Unless you are referring to genetics? That is still caused by DNA, just not by gene expression.

I saw a project on GitHub some time ago where someone trained a neural net with only a single, shared weight to play a driving game.

They used a gentic algorithm to evolve the network, its connections and activation functions (per node).

It got quite good.

Perhaps not that surprising that one can encode a lot in the graph structure itself, and given how DNA is quite good at dictating complex organs, I don't see why something similar couldn't be at work for insects and mammals alike. After all, nobody taught the newborn kangaroo babies how to crawl up into the pouch of their mom either.

They don't just build one. You may mean they don't learn from observation of others or from their mother. They may still acquire a sense of a better trapdoor and make better trap doors by experience. I can't think of an experiment to test the theory, but it might be around qualities like "thickness" or "wind resistance" if you could make a condition of test which put the trapdoor they do un-biassed under specific stresses.

Still, even if that's true, it does not refute the point. The knowledge necessary to build the very first burrow or web, or trapdoor (which will be definitely more than good enough otherwise spiders would no longer exist) is hardcoded, not externally acquired.

I agree. An individual may develop improvement over time (43 years!) but there's no evidence of transmission across generations. Still, ability to refine might be heritable?

No, learning from parents is common for the young of most mammals and many birds, at least.

Primates are on another level compared to most other animals, and humans are definitely outliers, but learned behaviour is far from unique to humans.

I’ve thought about differing levels of innate behaviors in different species as being somewhat akin to different sorts of programming languages. So dogs and cats are more akin to, say, Pascal¹ which provides a high level of functionality at the language level, but are limited in capability² while humans are more like C where there’s less functionality out of the box, but the ability to accomplish unique tasks is less bounded.

⸻

1. This is a theory I came up with as a high school student in the 80s so my choices of programming languages reflect that.

2. Insert obligatory Turing-completeness disclaimer here.

Story time.

I'm from Western Australia and spent a lot of time in its deserts in my youth.

I have a very distinct memory of walking out among the spinifex as a child, needing to pee. Its hot, my bare feet are burning, the ground is unforgiving and I already have spinifex'ed ankles, so naturally I tread as carefully as I can, trying to avoid the sand and the spikes, only walking on the slightly less heated organic mats that surround some of the spinifex, a kind of carpet of twigs and leaves, gathered in pockets by the wind.

The natural urge is rising, and I stop paying attention and get a bit of pace, as relief must be had. I see a particularly appealing area, covered by this organic dust, and I leap over to it, get my body parts aligned, and begin the process of relief.

A few seconds into it, I feel something moving. The entire mat that I was standing on, is in fact, a field of trap doors. Maybe a hundred of them. They begin to open.

I have never levitated as hard and as fast in my life.

So my Tabula Rasa has given me this lesson: don't go into the spinifex to pee. Ever.

the children of time series explores this concept in spiders. What happens if you were able pass down all of your knowledge into your children through genetics and how would that change society.

I think the biggest issues involved around knowledge/skills/intelligence being potentially linked to genetics is that it often draws the eugenics proponents into the discussion and it's obvious how that conversation ends.

Sometimes I feel like my fear of spiders is baked into my DNA. I think I found some evidence to support this, can't remember where though...

It has been well argued that "The Blank Slate" argument is wrong, or at least more complex and subtle than that.

In Pinker's book The Blank Slate" in particular.

We are born blank of our mother tongue, but are wired to acquire one. Not wired with the skill itself, but with the propensity to acquire it. Both nature and nurture are deeply important to the outcome.

> What amazes me about spiders is their complex unlearned behavior. No trapdoor spider learns how to build a trap. No orb weaver learns how to spin a web. Trap/web building behavior is entirely programmed in their DNA.

Do we really have evidence for this rather than observation and communication from parents to offspring. I'm deeply skeptical.

Spiders that hatch without their parents around build the same webs as any other spider from the same species. This behaviour is really not something that leaves any space for being sceptical.

I have no idea about this, but, how hard would it be in theory to test? Just remove a spider from any other spiders and see if they develop that behavior.

I don't know if the parent comment represented the expert consensus correctly, but if they did, I wouldn't automatically be skeptical of it.

>For her 40th birthday, research assistant Leanda Mason wanted to give the spider a mealworm, but Main denied the request since it would interfere with the study.

I have to admire the level of professionalism it takes to not give a single mealworm to a spider you have been watching for forty years.

Reminds me of "Red XIII", a character in Final Fantasy 7, who was a specimen in a study when encountered by the player. The name is probably chosen to highlight the heartless practice of giving a number to a sentient being instead of using a name.

This is an outcome of Jane Goodall challenging the convention of the day to use code numbers for animal subjects. At the time it was considered apostasy.

The respect is of other researchers who have to consume the data at a later date. "Number 16" is far more valuable a label than "Brenda", since the former implies the specimen is one of at least 16 other data points - whereas the latter implies that the researcher involved is simply bored of their work and just wants to have fun.

Anyway, only humans use names. Is "Brenda" a human or a spider? Sure, you can answer this now - but can you answer it in another 10 years?

Humans use names for non human things as well and using a number doesn't imply spider anymore than anything else. Is #16 a spider or a sheep? Did anyone ever confuse Dolly for a human?

It also doesn't tell us that it's one of at least 16 data points. From the article we only know that it was the 16th spider they recorded as being present but we don't know that any of the other 15 made it into the study. The fact that you're reading something we don't know into it from the name #16 actually highlights a failing in this system for me. Surely if it matters to you then you'll be reading the study which will these details explicitly.

Meh. As another commenter pointed out, https://en.wikipedia.org/wiki/Alex_(parrot) . Also I find the idea of "academic seriousness" very weird. I would never ever want to wake up with the thought of doing "serious science" all day. I want science (in my personal case, math) to be interesting, and thus fun.

"Man is most nearly himself when he achieves the seriousness of a child at play." - Heraclitus

Thats fine, just keep your personal preferences out of my labs' white paper submissions.

"Number 16" is valuable, in that it communicates the fact that the specimen involved is one of a set of at least 16 individual data points. "Brenda" doesn't communicate anything much more than, perhaps, the researcher is bored with their work and wants to have fun instead of doing real science. Besides which, this is a human name, so is it referring to a human or a spider?

Scientific rigour is important. Fanciful notions of cultural joviality, less so.

what if the spider dies shortly after? You'll have no way of knowing if the worm was the cause of the death.

what if the spider doesn't die shortly after? You'll have no way of knowing if the spider would have died of hunger without your intervention

Yeah, it’s got a bit of that “sci-fi books that HN recommends “ syndrome in that it’s a cool idea for a book but not actually a good book. More interesting to have described to you by someone who has read it than to read.

But probably worth forcing your way through the poor writing, character building and storytelling to get those cool ideas out of it.

It’s a good book to have read, but not to read, I guess. Sadly, the second one squanders an even more interesting concept, and isn’t worth reading.

I read it before I became a daily reader of HN.

I always thought and still think it's a great book. It may miss some writer's instruments to make it "a good book" in the classical sense, but to me it wasn't missing at all.

If anyone here neglects this book due to being a "sci-fi booksthat HN recommends", please give it a try!

I was a huge Asimov fan as a kid. Re-reading some of my favorites, I noticed they lacked a lot of classical "good book" elements. But they are still great in their own way.

I’m so glad to hear you say that, i was carrying it around while slogging through it and everyone else who saw exclaimed how good they’d heard it was. I do think the sequel is better, I can suspend my disbelief more and found parts of it honestly thrilling

> Yeah, it’s got a bit of that “sci-fi books that HN recommends “ syndrome in that it’s a cool idea for a book but not actually a good book.

Since you seem to have a higher bar than many, what are some examples of sci-fi books that you think are actually good books?

> it’s a cool idea for a book but not actually a good book

This was my impression of K. J. Parker's work. (I read the Scavenger trilogy of his.)

Intricate idea, uninteresting writing.

A neighbor of mine went from the most brilliant and kindest person I've ever met, to not recognizing his own wife at home in a short time span of few months.

The fact that he would not recognize me after talking together every week for a decade felt slightly odd.

But what broke the camel's straw was after one morning he assaulted his own wife coming back with groceries thinking she was an intruder.

Living 50 years together and not remembering her at all was awful.

Broke her and his heart too when he had to be taken to a proper elderly care facility as he was too dangerous (70+ but still very fit and strong) to his own family. Obviously, luck wanted that he was lucid when they were taking him out the house. He was aware of what was happening and why for few minutes. But when it came to say goodbye it felt like Shutter Island finale, you couldn't say whether he was now faking a few moments later or if he was doing it because aware of his condition and danger.

it's neat that the spider outlasted the arachnologist's career.

I wonder how long it would have lived without the wasps' interference.

> Australian arachnologist Barbara York Main ... returned to the site annually, sometimes more frequently, for more than four decades.

Huge respect. Always amazed by scientists sticking with studies like that.

One has to consider that if the longest lived spider ever, is one out of possible millions, chosen for a particular study, that maybe 43 years isn't actually that long for a spider. It seems unlikely that at random they just happened to choose the spider that would live longer than them all.

> There's a tortoise living on the Seychelle's governor garden since 1882 and has outlived every single governor yet.

Is this just a fancy way to say that the tortoise is alive, and whoever was governor in 1882 is dead?

No, actually it is saying that every governor since 1882 (except, by implication of “yet”, the current one implied to exist) is also dead, not just that the 1882 governor is dead. Of course, in the limit case if there were no governors after 1882 (except the current one, if any) it would just say the one in 1882 was dead, but there were, in fact, 24 or 25, depending on exactly when in 1882 the tortoise arrived (not sure on that), subsequent governors before the position was abolished with home rule which quickly transitioned into independence.

Which makes the “yet” weird, but I suppose it admits the possibility that the office might be restored before the tortoise dies, and the tortoise might not outlive the potential future governor.

Sure, the "yet" seemed to imply that one governor has taken over. There's no further information regarding the number of governors, and I refrained from speculation. All I could deduce was that the 1882 governor is dead, the current one is alive, and there is a nonnegative whole number of intervening governors (zero or more).

Imagine the life of a spider. Imagine it for forty three years. Lurking in your burrow, waiting for a snack to wander by. Getting fucked by spider boys and maybe eating them. Do trapdoor spiders eat their mates? I’m not sure. Laying eggs and watching the spiderlings go running off. Again and again for forty three years.

That just feels like a long time to be a spider.

Black widow spiders last something like a thousand eggs, and the hatchlings murder and easy one another until only a couple survive... Who go in to live for as long as black widows live for. The average lifespan is likely quite low, however.

Very few people seem to be able to distinguish between "the largest number on record" and "the largest thing that ever happened", even if they know perfectly well that records of whatever the phenomenon is are never kept.

In my opinion, the logical assumption is that this spider species can't live for much more time than exactly that interval of time

The interesting point here to me is that it was not killed before. This suggests that it was able to fend off the wasps attacks for fourty years. I assume that hen died it was too weak (too old) to fight back. Another possibility is that it had moult recently and was too soft to defend itself.

Insects die soon because adult wings can't moult twice. Spiders don't have wings so they can moult indefinitely and became much much older than insects --but-- is easy to predict that (unlike jellyfishes), spiders can't live forever

We can be 100% sure of this empirically. We even could predict mathematically the life time span of each species, also if we would want it (again, in my opinion).

（评论） (comments)

（评论）
(comments)