Animal Sentience
Are animals sentient? How do we measure sentience? What is sentience anyway?
For more information on Aleph Insights visit our website https://alephinsights.com or to get in touch about our podcast email podcast@alephinsights.com
Transcript
Hello and welcome to the Cognitive Engineering Podcast produced by Tell Me Studios for Aleph Insights. In this series of podcasts we take a look at interesting topics and discuss what we think they tell us about analysis and decision making. I'm Fraser McGruer and I'm here with Nick Hare and Chris Wragg of Aleph Insights and this week we're talking about animal sentience.
Speaker B: got the Animal Welfare Act of: Speaker A:to it. Essentially what it says. It does sound rather sloppy and sort of rather broad. To be
Speaker B:honest, this is sort of the worst excesses of sort of European rationalism. It's a big theory. Animals
Speaker C:are sentient. What are you sure? And I think that's the problem with the debate is it's become swept up in the Brexit debate and therefore it's been conflated with lots of other issues and there's political advantage to completely misconstruing what the law's about. But in essence, it does cloud what is quite an interesting debate, which is, you know, what is sentience? How do we measure it and how should we treat things which we regard as being sentient? Yeah, because it's worth saying
Speaker B:that this sentience clause has absolutely no legal ramifications. It's not as if we accept that they're sentient. That changes anything we're allowed to do. It is merely a word. So, you know, it doesn't
Speaker C:have doesn't actually have a legal implication. And in this sort of the what's called the EU animal sentience protocol, which is this particular sort of part of the legislation, what it describes is effectively an animal's ability to suffer or feel pain or experience things. And it, you know, other EU legislation talks about sort of freedom from five particular things, you know, the freedom from needing to suffer pain or harm, freedom to live a sort of fairly natural life and so on and so forth. But in essence, it doesn't really go
Speaker A:that much into what, what sentience is. So can you answer that question? And in fact, go back and answer a couple of questions you raised, which is, you said this clouds the debate about, well, what is sentience? And I think you had a second thing you said, and the third thing is, and how should we treat sentient beings along those lines? So can you answer those questions or start to
Speaker C:answer them? Yeah, so I think in terms of whatever our definition of sentience is, I think we can all agree that with current technologies, our measurement of it is decidedly imperfect. So, you know, whatever we decide sentience is, it's going to be very difficult for us to measure it and prove that something is is sentient. So the first thing is, we need to err on the side of caution. And if we decide that it is right to treat sentient beings in a particular way, we need to be broad in our definition of that, of sentience, because we know we can't definitely say whether something is sentient or not. Well, sorry, but that's sort of talking about
Speaker A:measurement. But can we start off by trying to answer then? Can we just do the easy bit first?
Speaker B: te in philosophy for the last: Speaker A:Well, there's no point having it as a topic if we can't have a go.
Speaker B:Well, let's start with a sort of thought experiment, really, which is the philosophical zombie, as it's known. Let's just say that we sort of think humans are sentient, right? No one's having a debate about that. But what do we actually mean by that? Now, is it possible that you could have a person who behaved exactly like they did? Like, take you, like you, exactly like you are now, and behaving exactly like you are now, but without any sentience, without consciousness? You're literally like a machine behaving like Fraser. Now, the two sort of strands of argument is one is, yes, that's perfectly possible, but we just think it isn't true. Or the other thing is, no, that isn't possible because of some physical reason. So in other words, we just think that in order to have something as complex as you, you have to have consciousness at the same time. And then, you know, there's also, in theory, could be a third strand where you're saying it's not logically possible to have someone who behaved like you and wasn't conscious, but I think that's probably not as difficult to argue. But so, yeah, so that's the question is philosophical zombies, can they exist? So could animals be philosophical zombies? Could they be behaving like they are, but without any consciousness or sentience?
Speaker C:And how would we know? And not only that, you know, could animals do it, but also could inorganic things do that? And I think that's probably the much, well, not the much more pertinent question, but it's certainly going to be an increasingly pertinent question. We're talking about AI here. We're talking about AI, yes. And, you know, what they are almost certainly going to be extremely good at is the zombie sentience that, you know, Nick was talking about. And the interesting philosophical issue is whether or not they'll achieve some sense of true, true sentience. You know, they'll definitely be able to ape our behavior extremely well and convince us that they are sentient, but will they be sentient?
Speaker B:Well, not necessarily ape our behavior. Let's say they'll be able to behave intelligently. Let's say more intelligently than us. And the question is whether or not, now it seems to me that you can have intelligence without consciousness. This is a sort of, so the reason that I'd sort of contend that is that you can make artificial intelligence, you can make artificial intelligence out of really unsentient things. So if we take something like AlphaGo, which we refer to a lot, it's a good example of a recent, you know, highly successful artificial intelligence. And, you know, it's very good in a certain domain, which is playing Go. Although it looks a bit like there's an analogy with a human because it's a machine, it's a computer and it has what we can think of as a brain. It's got the chip, which is processing information. It's got a memory, it's got sensors, you know. In fact, you could represent AlphaGo in theory with a very big lookup table. You could publish a really, really big table showing all the different board positions and what you do when you're in that board position. It would be bigger than the universe. So, you know, we can't in fact, but we could in theory and that would behave exactly like AlphaGo does. Now, would that book be sentient? I think I'd
Speaker A:want to say no. And just sort of approaching this from a different way, from a different angle. One of the things we were starting to talk about there was, well, what is human sentience? And what is animal sentience as well? Does it make any sense to ask those kind of questions? Or does it make any sense to differentiate humans from animals? Or is that what the defining difference
Speaker C:is meant to be? Well, I mean, you know, you could take the sort of Bentham position, which is that actually, you know, a fully grown adult horse or dog is more sentient than an infant child, for example. And so therefore, actually we do overlap on some sort of spectrum. But I think, you know, the key point kind of, you know, picking up on Nick's thoughts there about this distinction between sapience, you know, the ability to think and reason and sentience, the ability to experience, to think, what is it like to do something is quite important. And, you know, what we have to make sure we don't do is conflate the two and ascribe sentience to
Speaker B:something that is in fact merely sapient. Yeah, unless you try to construct an argument that sapience at a sufficiently high level implies sentience for some physical reason. So, you know, that actually, if you instantiate sapience, so if you have, you know, this would be akin to arguing, for example, that humans are, human consciousness merely just is the ability to take on board lots of information, process it, to have certain objectives that we want to achieve, and to have some sense of them being more or less achieved. And being frustrated is simply just not being able to achieve your objective. And it might be possible to make an argument along those lines to say, no, once you have all those things, you have consciousness by definition. But, you know, I think that would be an empirical claim. And if you look at the ways that people have proposed measuring consciousness, they all rest on a certain kind of empirical claims. So, in other words, you know, looking at physiological things, have you got pain receptors? You know, have we, then evolutionary arguments, well, if we've evolved in an environment to avoid predators and to avoid physical pain, then you'd expect us to, you know, evolve the ability to experience pain. But all of those rest on an assumption that we have identified that pain just is, or suffering just is some kind of neurological activity. And, you know, but could pain exist in other forms? Could it be instantiated electronically or artificially? Could we create an artificial pain box, which just consists of a consciousness experiencing pain? And if we could, would it be wrong to do so? Or would it be fun? Could you make thousands of them
Speaker C:and make the world an immeasurably worse place? But, you know, these kinds of, you know, thought experiments are what science fiction has been exploring for decades. And, you know, you look at novels like, Do Androids Dream of Electric Sheep? You know, fundamental to that, to what that novel explores is that very question. If you build up a certain level of sapience, does that imply sentience? And, you know, the whole ethical dilemma of, you know, Rick Deckard, who is having to go out and effectively test for sentience. Now, he does that indirectly through measuring androids sort of empathy levels. And, you know, it's supposed that they don't, they're unable to empathize because they perhaps don't have sentience and can't imagine another creature having sentience. But, you know, he then, you know, he then must execute these things and is having to view them, therefore, as something that isn't sentient. And so that novel plays around very cleverly with this idea that sentience is a human construct. We are biased in our view of what it is. We ascribe it to ourselves, but we suffer from speciesism. You know, we, hence the fact, you know, our treatment of animals over the past has, you know, possibly been different to our treatment of humans.
Speaker B:Yeah. And I think if you look at the, you know, what the, probably the first real paper in Artificial Intelligence, which was Alan Turing's Can Machines Think? Certainly one of the most influential papers ever written. That's where he proposed the Turing Test, which basically just smashes through all this discussion and says, okay, look, let's just agree that we will agree that machines think if a machine could fool someone into thinking it was a human. And I, and I think, you know, to sort of bring it back to the question of quite apart from philosophically what we might think consciousness is. Legally, if we have to have something that's measurable, which you do, you know, it's a bit like, we don't, you know, actually what we want is for people to be able to vote when they're old enough and mature enough to be able to make sensible decisions. But, but we can't write that into law. So we just say we've got to be 18. Likewise, I think we should expect to see laws about animals, but also in future about machines, having, having something like that, something like a Turing Test. So, I mean, you know, there's a test for animal self-awareness, for example. And I don't think we should think self-awareness is necessarily the same as consciousness, but it's quite a clever one, which is where you take an animal, you put it in front of the mirror and then you put a smudge of like dirt on its forehead. And if it tries to remove the dirt from its own forehead, you know, that it's worked out that, that it's the thing in the mirror. So, I mean, you know, tests like that are very useful because they, they either pass or fail, you know, they don't require a vast amount of judgment.
Speaker A:Look, we're going to have to finish there, but I enjoyed that. And we've just started to get somewhere, I feel. I just want to round things up. And that's why we're going to stop. Yes. Well, yeah. But I just want to round things off by talking about a cartoon. I don't know how relevant to our discussion this is, but you know, the cartoonist Gary Larson?
Speaker B:The farsighted guy.
Speaker A:The farsighted guy. And who hasn't, I don't know, I hope he's still alive, but he's not been doing anything for about 20 years or so. Anyway, one of my favourite cartoons of his is the scene is a city in chaos from, and you're looking at traffic and actually all the traffic streaming out of the city, as you might imagine, there's a mushroom cloud over the city in the background. And what you've got one, what the focus of the cartoon is, there's one dog looking out of a car window, looking at another dog, and they're both wagging their tails. And there's something about that I just, I just rather like, the sort of recognition of a dog knows, a dog definitely knows what another dog is. Even if he doesn't know what he, that he or she himself is a dog, but he definitely recognises another dog. And that's all that matters, you know.
Speaker B:And you're a dog guy, aren't you? I am a dog guy.
Speaker C:Yeah. Yeah. Yeah. I mean, on another level, you know, perhaps there's something there about actually, you know, the wisdom of animals as opposed to humans having just obliterated themselves.
Speaker A:Yeah. That's quite a nice fluffy, unusually fluffy remark from yourself, Chris. Okay, we're going to stop there. So I'm Fraser McGruer, been here with Chris Wragg and Nick Hare of Aleph Insights. Thank you as always to listening to the Cognitive Engineering Podcast. Until next time, goodbye. I'm off to have a dog sandwich.
