3 ms·

What do we consider smarts? For example, with some examples there, like pretending to do a mating ritual for another group of spiders in order to lure the fema

What do we consider smarts? For example, with some examples there, like pretending to do a mating ritual for another group of spiders in order to lure the female out, this seems like it's something built in to their neural networks through thousands of years of evolution. It doesn't seem significantly smarter than the other group of species doing mating rituals themselves. They just have a neural network that reacts to certain input with this output that evolution created and honed through all this time of them being around such input and this output causing a better survival performance. I didn't see anything specific that could've indicated they could learn new things in new situations, so it seems like built in neural networks that come to them rather than them having special smarts to learn new things like humans have. > Although spiders can’t literally count one-two-three, the research suggests some jumping spiders have a sense of numbers roughly equivalent to that of 1-year-old humans. I wonder about conclusions like these. I don't know about 1-year-old humans, but it's understandable that they can measure strength levels depending on the amount of whatever they are observing and if situations change they should recalculate their new decision. I just don't think it's a sense of numbers, it would be more like some sort of amount of input (pheromones, image, whatever) that makes them pay attention that something has changed. And all in all it seems odd conclusion to compare sense of numbers based on that to 1-year-old humans. Trying to compare it to something, but the way they use those "skills" doesn't really make comparative sense.

pjmlp 5y ago
The issue is that we have millions of evolution for living beings, and even us in 2021 aren't the same that were running from Dinosaurs, so it is only natural that other species also enjoy some kind of intelligence, naturally they aren't writing books, however it is a scale from DNA programmed behaviours to some level of self conscience, ability to learn or even develop some kind of culture that is passed across generations. There is so much to learn from other species on our own planet.
- mewpmewp2 5y ago
  While neural networks themselves are fantastic, are those species anything smarter than that? There's so much to learn from them yes, but I don't think based on this article that spiders would be "smarter than you think". I don't think they are necessary self-conscious in an intelligent way. I think all of this behaviour can be explained from evolution honing them to have some sort of biases and weights towards this input causing that output. And it's absolutely fascinating to learn to understand how they became to be the way they are. How environment affected, how all of it was stored in DNA so it can't be reproduced continuously and continuously. I think level of spider's intelligence doesn't differ that much from what Tesla FSD is for example. Tesla FSD needs quite a bit of honing still of course.
- falsaberN1 5y ago
  We never ran from dinousaurs, but it'd have been kind of cool.
  - pjmlp 5y ago
    Right, was more to put it perspective actually.
  - MauranKilom 5y ago
    "We" as in "mammals", I guess.
gpderetta 5y ago
> What do we consider smarts? I read one of the papers were they experiment with Portia spiders (it might have been shared here on HN months ago). The paper actually went in great details to catalogue the kinds of intelligence being studied (it didn't invent anything, it was a good recap of the literature) and to design experiments to fit the portias in this hierarchy. It was quite good.
Kim_Bruning 5y ago
Human social interaction is also something built into their neural networks through thousands of years of evolution too. They have a neural network that reacts to certain input with this output that evolution created as well. Of course, biological neural networks are self learning too, so the output gets tuned and optimized in situ just a little. But there's really no difference in that sense. Sure: you shouldn't anthropomorphize lest you end up drawing completely wrong conclusions; this was drilled into me as well. However, a strictly mechanistic view of something that is clearly software++ (biological neuroplasticity) might be overcompensating a bit too far in the other direction again. I guess what I'm trying to say is try to treat each organism as their own thing. Don't overestimate or anthropomorphize; but don't take an over-linearized super-mechanistic view either, since that can lead you astray just as much.
- mewpmewp2 5y ago
  Yeah, agreed, but I think one of the highest differentiations for whether someone or something is intelligent is how it learns about new unseen concepts and can use intelligence to adapt to a completely new environment. E.g. if that spider species was put to a completely different environment and during its lifetime it could learn to adapt to some new species mating ritual and fake out their female as well. This would make me consider the title to be valid "spiders are much smarter than you think".
  - Cybiote 5y ago
    Portia spiders do demonstrate trial and error learning, quoting the abstract of [1]: > All species from the jumping spider genus Portia appear to be predators that specialize at preying on other spiders by invading webs and, through aggressive mimicry gaining dynamic fine control over the resident spider’s behavior. There is evidence that P. fimbriata, P. labiata and P. schultzi derive signals by trial and error. Here, we demonstrate that P. africana is another species that uses a trial and error, or generate and test, algorithm when deriving the aggressive-mimicry signals that will be appropriate in different predator–prey encounters. It turns out that the species with more variation in encountered prey types are more likely to rely on search, varying over possible patterns until a response is received. Other papers show their ability to learn generalizes beyond mimicry of vibrational patterns. They are also capable of deriving and maintaining situation specific attack routes and plans. I'll also argue that a fully instinctual repertoire, even without learning, should count as intelligence if flexibly deployed. Consider: despite an inability to learn or adjust, a Nash equilibrium approximating poker bot or an Alpha Zero neural network can be described as encompassing a deep instinct of the game that enables intelligent action selection. [1] https://link.springer.com/article/10.1007/s10164-010-0258-5 https://link.springer.com/article/10.1007/s10164-010-0258-5

pjmlp 5y ago

The issue is that we have millions of evolution for living beings, and even us in 2021 aren't the same that were running from Dinosaurs, so it is only natural that other species also enjoy some kind of intelligence, naturally they aren't writing books, however it is a scale from DNA programmed behaviours to some level of self conscience, ability to learn or even develop some kind of culture that is passed across generations. There is so much to learn from other species on our own planet.

mewpmewp2 5y ago
While neural networks themselves are fantastic, are those species anything smarter than that? There's so much to learn from them yes, but I don't think based on this article that spiders would be "smarter than you think". I don't think they are necessary self-conscious in an intelligent way. I think all of this behaviour can be explained from evolution honing them to have some sort of biases and weights towards this input causing that output. And it's absolutely fascinating to learn to understand how they became to be the way they are. How environment affected, how all of it was stored in DNA so it can't be reproduced continuously and continuously. I think level of spider's intelligence doesn't differ that much from what Tesla FSD is for example. Tesla FSD needs quite a bit of honing still of course.
falsaberN1 5y ago
We never ran from dinousaurs, but it'd have been kind of cool.
- pjmlp 5y ago
  Right, was more to put it perspective actually.
- MauranKilom 5y ago
  "We" as in "mammals", I guess.

gpderetta 5y ago

> What do we consider smarts? I read one of the papers were they experiment with Portia spiders (it might have been shared here on HN months ago). The paper actually went in great details to catalogue the kinds of intelligence being studied (it didn't invent anything, it was a good recap of the literature) and to design experiments to fit the portias in this hierarchy. It was quite good.

Kim_Bruning 5y ago

Human social interaction is also something built into their neural networks through thousands of years of evolution too. They have a neural network that reacts to certain input with this output that evolution created as well. Of course, biological neural networks are self learning too, so the output gets tuned and optimized in situ just a little. But there's really no difference in that sense. Sure: you shouldn't anthropomorphize lest you end up drawing completely wrong conclusions; this was drilled into me as well. However, a strictly mechanistic view of something that is clearly software++ (biological neuroplasticity) might be overcompensating a bit too far in the other direction again. I guess what I'm trying to say is try to treat each organism as their own thing. Don't overestimate or anthropomorphize; but don't take an over-linearized super-mechanistic view either, since that can lead you astray just as much.

mewpmewp2 5y ago
Yeah, agreed, but I think one of the highest differentiations for whether someone or something is intelligent is how it learns about new unseen concepts and can use intelligence to adapt to a completely new environment. E.g. if that spider species was put to a completely different environment and during its lifetime it could learn to adapt to some new species mating ritual and fake out their female as well. This would make me consider the title to be valid "spiders are much smarter than you think".
- Cybiote 5y ago
  Portia spiders do demonstrate trial and error learning, quoting the abstract of [1]: > All species from the jumping spider genus Portia appear to be predators that specialize at preying on other spiders by invading webs and, through aggressive mimicry gaining dynamic fine control over the resident spider’s behavior. There is evidence that P. fimbriata, P. labiata and P. schultzi derive signals by trial and error. Here, we demonstrate that P. africana is another species that uses a trial and error, or generate and test, algorithm when deriving the aggressive-mimicry signals that will be appropriate in different predator–prey encounters. It turns out that the species with more variation in encountered prey types are more likely to rely on search, varying over possible patterns until a response is received. Other papers show their ability to learn generalizes beyond mimicry of vibrational patterns. They are also capable of deriving and maintaining situation specific attack routes and plans. I'll also argue that a fully instinctual repertoire, even without learning, should count as intelligence if flexibly deployed. Consider: despite an inability to learn or adjust, a Nash equilibrium approximating poker bot or an Alpha Zero neural network can be described as encompassing a deep instinct of the game that enables intelligent action selection. [1] https://link.springer.com/article/10.1007/s10164-010-0258-5 https://link.springer.com/article/10.1007/s10164-010-0258-5