So what's a sapient species anyway?

[Ares Land]

We were talking sapience with WeepingElf over at the Bug thread, and I thouught the question deserved a separate thread...

I don't really have a good working definition of sapience... How would you define it?


Sapience -- at least in SF -- is something of a mess between technological capability, self-awareness and intelligence. But neither of these is really satisfying. Ants domesticate animals, use tools and practice agricultural, yet they're pretty clearly not sapient. We could imagine aliens very intelligent but not self-aware. Conversely a very dull animal that happens to be self-aware is conceivable...

Even tool use isn't that great a criteria. Would we even recognize an alien tool? (Even with human societies, that's a problem. The West blissfully ignored evidence of agriculture in the Amazon basin and Inca technology).

I guess my question is best expressed as a thought experiment:
To take a thought experiment, if you encounter an alien intelligence, how would you judge whether they're sapient or not? What would be the criteria?

(To give some context, I'd like my Bugs to be sapient, but not obviously so... In a word, alien enough that it would take quite some time to realize they're not just aggressive animals acting on instinct.)

[Salmoneus]

Well, you can define the word how you like - it's not a word with a particularly robust, consensus definition.

I think the core of the concept is extensive self-awareness. Things like:
- recognising images of the self as self
- recognising images of the self in the past as self
- distinguishing between evidence of self's past actions and evidence of the past actions of others (i.e. being able to remember or deduce whether they knocked over an object or something else did)
- showing extensive knowledge of self's abilities and limitations (this will vary between zero, through to awareness of obvious limitations like how high it can jump, through to more abstract things like predicting whether it will be able to remember something or not)
- a recognition that other beings have subjective experiences, and the ability to deduce what another being may be experiencing (eg to know what another being can and can't see, and act accordingly)
- recognition that the subjective experiences of others are of a kind with their own (eg demonstrating empathy and altruism in the face of the suffering of others)
- the recognition that just as they perceive others, others also perceive them, and that just as they form theories about the minds and characters of others, so too others form theories about their minds and characters (eg demonstrating subtle dissimulation)
- the recognition that others are also continuing beings (eg the ability to form long-term alliances, and not only when continually reinforced by rewards*)
- the recognition that others have particular wants at particular times
- the recognition that others have their own obligations and allegiances, beyond simply whether they are self's ally or enemy




In terms a broader concept of "intelligence", you could combine things like the above (social intelligence) with other forms. I would suggest that most important are the use of allegorical thinking (extrapolating from a usual situation to an unusual but analogous one), and problem-solving (particularly multi-phase problem-solving).



I don't think intelligence is one thing - it's bound by family resemblances. It's possible to imagine an intelligent creature that lacks one of the above, but not, I think, one that lacks all of the above.


[FWIW, ants may be more sapient than you think. They're one of the few animals, and the most reliable animals, to pass the mirror test (recognising themselves in a mirror). ]




*to explain my point here:
- if you take, say, two lions, one lion may realise that if they don't attack the other, the other won't attack it. Not being attacked is an instant reward, and continual rewarding can lead to continuing cooperation. This is very common in nature. And likewise, if the other lion hunts, and shares its kill (because there's more than enough so why start a fight), the first lion may do likewise, and may eventually learn that if it shares its kill more amicably, the other lion does likewise. Extensive social behaviours can therefore develop - but these don't require much understanding that the other lion is actually intelligent, just that it's reasonably predictable. What's much rarer is when one animal does something that it wouldn't do in a non-social situation, and that is to its own detriment, in order that the other animal does something in return at an entirely different time. This can be distinguished both from cases where the reward is immediate, and from cases where the reward is remote but the cost is minimal. For instance, mutual grooming isn't that impressive, because the animal would have to groom anyway - it just learns that mutual grooming is a more effective way of grooming. By contrast, if a chimpanzee recognises that another chimpanzee needs a tool, and goes and finds a tool and brings it to the other chimpanzee, and in exchange the other chimpanzee helps fight off the chimpanzee that's bullying the first chimpanzee, that's a much more convincing demonstration of the chimpanzee's awareness of each other's identities...

[Ares Land]

Thanks. That's very clear.

[FWIW, ants may be more sapient than you think. They're one of the few animals, and the most reliable animals, to pass the mirror test (recognising themselves in a mirror). ]

Which would mean that the mirrot tests correlates with complexity of social organization. Which makes a lot of sense!
But maybe the test actually measures sociability rather than sapience?

(As an aside, I've always been a little bothered with the mirror test. Now, I understand the point, there aren't a lot of ways to measure self-awareness, but it seems to me that it's testing for ability to figure out how a mirror works rather than self-awareness... A sapient species with very poor vision might well fail the test. )

Well, you can define the word how you like - it's not a word with a particularly robust, consensus definition.

I just thought of a possible definition... On what grounds would you extend human rights to an alien species?

[alynnidalar]

- distinguishing between evidence of self's past actions and evidence of the past actions of others (i.e. being able to remember or deduce whether they knocked over an object or something else did)

Interestingly, my first reaction on reading this was to think of a hive mind in which individual members may not actually differentiate between their own actions and the actions of another member. Yet I suppose the definition still holds--we may say that individual elements of the hive mind are not sapient, but the hive mind as a whole is. (and such an intelligence would be very difficult for us to recognize indeed, and likely would have a very hard time recognizing us as intelligent in return!)

[zompist]

The problem is that we have a sample size of 1, and absolutely no idea of the variance. There's no way to do science on this. But we can do science fiction, and over the last 200 years there's been all sorts of ideas on what aliens or just other minds might look like.

Theories which posit an absolute unbridgeable difference between humans and other animals don't look as good as they used to. I'd say a better question than "how do we know X is a sapient species?" is "how advanced does X have to be before we have to leave the planet alone?"

[Pabappa]

Would anyone here consider other animals on Earth to be sapient? Dogs are very humanlike in emotion if not in intelligence. Meanwhile some dolphin species might actually outrank us in intelligence. Elephants are known for emotions as well, and elephants seem to understand that humans are unlike all other animals, and perhaps see us as a lot like them. Lastly there are examples of birds such as parrots showing emotional attachment to humans along with high intelligence.

[Ares Land]

Theories which posit an absolute unbridgeable difference between humans and other animals don't look as good as they used to. I'd say a better question than "how do we know X is a sapient species?" is "how advanced does X have to be before we have to leave the planet alone?"

Yes, exactly. And that question is really difficult.
To take a science-fiction example... the titular Aliens in the Aliens series are clearly sapient and much, much brighter than we are. Yet any suggestion that humans try to communicate with them is met with 'shut up, evil corporate android'
Amusingly, Michael Crichton's raptors are an incarnation of absolute evil, yet they're so absurdly smart (*) they would certainly qualify as sapient.
(*) Except when plot requires them to be dull as bricks. I re-read the book sometimes ago, and man, they would totally figure out the poisoned egg trick. My cat would figure the poisoned egg trick out.

Would anyone here consider other animals on Earth to be sapient? Dogs are very humanlike in emotion if not in intelligence. Meanwhile some dolphin species might actually outrank us in intelligence. Elephants are known for emotions as well, and elephants seem to understand that humans are unlike all other animals, and perhaps see us as a lot like them. Lastly there are examples of birds such as parrots showing emotional attachment to humans along with high intelligence.

Corvids are really smart. Raven show emotional attachment to humans, use tools and make plans against each other.
Even crows are pretty smart. They've displaced all competition from the neighborhood, and you can tell they're brighter than other birds. (They observe us in a very human-like manner and are very curious about what humans are doing).

[xxx]

the only true difference between humans (the only species recognized as sapient (by itself ...)) and other animals is their viral infection by language ...

[Ahzoh]

I think the ability to think symbolically is an important feature of sapience.

[Pabappa]

Teppala is supposed to have five sapient species, but I havent written about the other four in years and I have rationalized it by saying that the different species rise and fall, and that the focus times I write about are the times where humans are dominant. Other animals peek in and out even within the humans' reign, ... for example a war lasting from 2662-2668 took an unexpected turn when an army of crabs that had been allied to one of the human armies defected and started invading all of the humans' areas indiscriminately. The humans that had until then been losing surrendered immediately to the other humans and then the war became one of all humans against all crabs. The humans eventually won and no animals played a major role in any human conflicts until another very brief appearance of firebirds in a war in the year 4138.

But Im interested .... Has anyone played with the idea of a conworld in which humans arent the most intelligent species?

My conworld doesnt really qualify because I dont really think about animal intelligence much ... the animals have their own languages, so theyre at least somewhat on par with humans, but not above us. This study started the debate about whether some dolphins might actually be *smarter* than humans, and just dont show it because nature cruelly trapped them in the sea and gave them a body without flexible limbs.

I could imagine a world in which dolphins are the unquestioned intellectual champions of the planet, with humans firmly in second place, and therefore those few humans who learn to speak the dolphins' languages, and win their trust, become the unquestioned intellectual champions of the human race. Thus, dolphins remain trapped in the sea, but wield power through the humans they befriend, and all sorts of complicated things arise from this as humans do material favors for dolphins in return for the dolphins' intellectual favors, and thus some dolphins are against other dolphins, and the humans are too inept to understand what's going on, and get caught in dolphin vs dolphin fights, and .... well, you can follow this down a million different paths, really.

[zompist]

But Im interested .... Has anyone played with the idea of a conworld in which humans arent the most intelligent species?

I think this is pretty common. In Tolkien, for instance, the elves are more advanced that humans. In Charles Stross's Laundry books, we share Earth at a distant third rank behind ocean-bottom and molten-lava creatures. In Hitchhiker's Guide, mice and dolphins are smarter than us. And the idea goes back at least to Swift's Houyhnhnms.

[bradrn]

But Im interested .... Has anyone played with the idea of a conworld in which humans arent the most intelligent species?

I feel certain that this happens in many works of science fiction. I’m not too familiar with that genre, but as an example, in Terry Pratchett’s book The Dark Side of the Sun, humans are not the most intelligent species; I’m certain that other science fiction would qualify as well. Still, it would be interesting to explore this idea in a non-science-fictional conworld such as the one we make on this board…

[Salmoneus]

Would anyone here consider other animals on Earth to be sapient? Dogs are very humanlike in emotion if not in intelligence. Meanwhile some dolphin species might actually outrank us in intelligence. Elephants are known for emotions as well, and elephants seem to understand that humans are unlike all other animals, and perhaps see us as a lot like them. Lastly there are examples of birds such as parrots showing emotional attachment to humans along with high intelligence.

No, dolphins are not smarter than humans.

No animals have demonstrated their sapience, and probably none should be considered sapient. However, I think there are some we can't yet rule out, who do show some signs of sapience. Other than chimpanzees and bonobos (and to a lesser extent orangutans), bottlenose dolphins are way out in front of that line. Dolphins appear to have language, though its extent has not been explored fully yet - they at least seem to be the only species other than humans who have fixed personal names, and it appears that some dolphins can 'translate' relatively complicated human instructions into dolphinese to teach other dolphins that didn't hear the original instruction. That's disputed, but it's certain that bottlenoses (and orcas and some other dolphins iirc) demonstrate culturally-learnt behaviour, which is exceedingly rare. They're extremely smart at understanding human-set puzzles, as well.

Corvids are probably next in line. Ravens particularly, but also crows, jackdaws and magpies can be disconcertingly smart. Two traits of ravens? They appear to do things just for fun: ravens have been observed teaching other ravens to repeatedly slide down snowbanks, for no apparent reason. And they appear to have some sort of symbolic language: they're the only vertebrate species, and the only non-eusocial species, that have demonstrated the ability to give abstract directions to one another (that is, one raven can make noises that allows another raven to navigate to an object that neither raven can see at the time of the communication). It's easier to do tests on crows - they're more numerous and more willing to jump through hoops for humans - and they've been shown to have better problem-solving than a lot of human children (iirc we think some crows are similar to approximately 9-year-olds in this regard). Two famous examples: when you give an animal a treat dangling from a long string, lots of animals are smart enough, eventually, to realise they can pull up the string. But only crows have realised that if the string is too long, they can pull part of it up, make a loop, hold it under one foot, and pull more of it up. And crows have demonstrated understanding of Archimedes' principle: when something they want is floating on water they can't quite reach, they've worked out they can drop in pebbles to make the water level rise further.

There was another finding just this week, actually: jackdaws have been shown to label individual humans as good or bad, learn these attributions from one another, and remember them. So, lots of animals make noise when humans approach, to warn others. Really smart animals learn that some humans are OK and others aren't, and make different noises for "OK guy coming" and "badguy coming!!". But REALLY smart animals like jackdaws can hear those noises, see the person they're describing, and remember for themselves whether that person is a goodguy or a badguy. That is, with smart animals, individuals can develop expertise (in this case, learning which people are dangerous), and other individuals can defer to the wisdom of the expert ("human-interpreter-bird says a badguy is coming! I don't recognise her myself, but human-interpreter is never wrong, so I'll trust her!"). But with really smart animals, the non-experts can learn from the experts, gaining expertise without having to actually replicate their experiences themselves ("hey, a badguy is coming! I know she's a badguy because human-interpreter told me so last month!").

Anyway, parrots are obviously on that list too. And then probably elephants.

Dogs and the like I think we can rule out as sapient - they're in some respects very intelligent animals (they have great learning skills, in particular), but they don't seem to be on the level of corvids or dolphins.

[Ares Land]

But Im interested .... Has anyone played with the idea of a conworld in which humans arent the most intelligent species?

I think this is pretty common. In Tolkien, for instance, the elves are more advanced that humans. In Charles Stross's Laundry books, we share Earth at a distant third rank behind ocean-bottom and molten-lava creatures. In Hitchhiker's Guide, mice and dolphins are smarter than us. And the idea goes back at least to Swift's Houyhnhnms.

I'd add the ilii to the list
And also, Well's Martians and the various Lovecraftian eldritch horrors are said to be smarter than humans.

I think the real problem is depicting greater-than-human intelligence convincingly. What would it be like? Peter Watt's starfish things are able to compute primes faster. Ho hum. Well's Martians, of course, are incredibly advanced and unaware of germ theory.
In Accelerando, Stross's superhuman AI-cat supposedly has mental pictures of people as complex as the original. Meaning, if you try to best it, it won''t guess at what your next move will be, it will run complete simulations of yourself and check what their reactions are. OK, that would be impressive... but isn't that overkill, and how do they gather the necessary data anyway?

So authors usually do the sensible thing: they tell us that the aliens are really smarter than we are, but we only get rare glimpses of the fair creatures/eldritch monsters.

[zompist]

There was another finding just this week, actually: jackdaws have been shown to label individual humans as good or bad, learn these attributions from one another, and remember them.

Just this week? Konrad Lorenz described this behavior in jackdaws 67 years ago, in King Solomon's Ring (which, go read, it's great).

[Pabappa]

I think the real problem is depicting greater-than-human intelligence convincingly.

I think I ran into this problem during my attempt to write a novel .... i worked on it from roughly age 10 to 12, and looking back i see that all of the adult characters were unreasonably stupid, perhaps because I couldnt conceive of anyone possibly being smarter than the lead character, who was based on myself. This applied to both the friendly and hostile characters.

Regarding dolphins .... I think they might be smarter than we can measure. The study's authors gave their argument for why dolphins might not be smarter than humans, but even they admitted it was just speculation, and that if we assume that dolphin brains follow the model we have found to be true in every other mammal, they would indeed outrank us in intelligence. As I said above, it may just be that nature is so cruel that the world's most intelligent animal happens to be one born without flexible limbs and confined to an inconvenient habitat. So why arent dolphins winning the intelligence tests we come up with? I dont know, but the species we have in captivity are the most abundant ones, which are not necessarily the most intelligent. It's possible that Globicephala melas outranks Tursiops truncatus but we just dont know it because we have only the latter species in captivity.

edit: That said, though, I would not be surprised if even T. truncatus is, on an objective scale, smarter than humans as well. It may just be that our measurements incorporate human biases. In an ideal world, dolphins would also be able to run intelligence tests on us, and who knows what they might come up with?

If it's true, for example, that dolphins are able to understand human language and then repeat things we've said to other dolphins, but yet we humans haven't the vaguest idea how dolphin languages work, they're ahead of us already.

[Moose-tache]

The problem of measurement remains daunting when studying animal intelligence. Cats, for example, have to be motivated to show signs of their understanding, but not being social creatures this is not always easy. Take, for example, the question of whether or not my cat recognizes himself in the mirror.

My cat can recognize shapes and faces. He knows what a face is, anyway. You can draw two circles on a paper bag and it will become 100 times more likely to make him jump and run away. So it seems unlikely that he could see a reflection of a cat and not realize it is a cat. But at the same time, he is terribly anxious around other cats. If a strange cat came into our house, our cat would freak out. I've seen him have similar reactions to cats seen in the distance, much too far away for smell to be part of the equation. So our cat knows what a cat looks like, and doesn't react well when he sees one. Why then, when I carry him in front of a mirror, does he have no reaction? There are only three possibilities:

1) He is unable to recognize the shape of a cat when it is flipped on its vertical axis.
2) He is unconcerned with foreign cats as long as they are kind enough to only appear in mysterious holes in the wall and shadow his movements.
3) He recognizes that the cat in the mirror is not another cat, but merely a reflection.

Now, option 3 seems like the obvious choice to me, but there are two sub-options.

3a) He realizes that the cat in the mirror is an illusion caused by the mysterious wall boxes, to be dismissed alongside other unexplainable natural phenomena like vacuum cleaners and sink faucets. This would be similar to when you see a shape in the clouds and realize it's not really a bunny rabbit.
3b) He realizes that the cat in the mirror is an illusory copy of himself, replicated in the mysterious wall box.

The difference between 3a and 3b is probably the line between full "sentience" in human terms and mere animal intelligence. But I can't begin to imagine how we would go about measuring it. In terms of visible cat behavior, this distinction is incredibly subtle.

[zompist]

A couple more hypotheses:

* The image doesn't smell, so it's not a real cat. (He does react to cats in the distance, but "far movements seen by eye" may be processed differently from "near movement that ought to smell".)

* The mirror registers as a caged-in area. Lorenz again: animals are much less agitated by another, close animal if they feel there's a safe barrier in between, like a fence.

[Ares Land]

As I said above, it may just be that nature is so cruel that the world's most intelligent animal happens to be one born without flexible limbs and confined to an inconvenient habitat.

Oh, I don't know about that. Maybe they're actually happier than we are

A couple more hypotheses:

* The image doesn't smell, so it's not a real cat. (He does react to cats in the distance, but "far movements seen by eye" may be processed differently from "near movement that ought to smell".)

* The mirror registers as a caged-in area. Lorenz again: animals are much less agitated by another, close animal if they feel there's a safe barrier in between, like a fence.

This makes me want to try experiments on the cat. (Mine makes that frustrated growl sound at animals on the other side of a window, so presumably we could exclude the caged-in hypothesis?)
I think it's probably something to do with the smell hypothesis, and with motion as well. Cats react more strongly to motion, perceived both visually and by the whiskers...
Isn't it curious, by the way, that birds (who consistantly rely on sight) are better at the mirror test than mammals (that are more smell-oriented). Would we even notice, if we came across something that mimicked our smell?

(For that matter, dogs bark at their reflections)

[alynnidalar]

Whether or not cats recognize themselves in mirrors is an interesting question to me. My cat exhibits zero reaction to either mirrors or phone/computer screens, both images and sound. (e.g. he will react to the dog across the road barking, but I've never seen him react to the sound of a dog barking from my phone) It seems like he must have some recognition that the thing he can see/hear in this flat object must not be real/is not actually present.

I'd almost wonder if there was something that means he can't process images on screens at all--if not for all the videos I've seen of other cats reacting to things on TVs/computer screens (e.g. checking behind it if something runs off screen, or appearing to recognize a video of their owner).

I like the idea that smell is part of it, but I wonder if 3D vision also is a factor? That is, he can tell the thing on the screen/in the mirror is flat, and thus can't be a real cat (or dog, or whatever).

As for sound, I don't know about your cats, but with mine, I wonder if the reason he doesn't react much to recorded sound is because of some difference he can hear between real and recorded sounds? He very occasionally reacts to recorded sounds, but it's never when I expect it (e.g. he never reacts to animal noises, he doesn't seem to care about loud music or high volume, it'll be in the middle of a Youtube video or something random and he'll freak out for no discernable reason). As such I suspect he's reacting to something outside my range of hearing.