← LSE: Public lectures and events
LSE: Public lectures and events
Rules and Representations: desire from an evolutionary point of view
01:28:02
Loading summary
Transcript73 lines
- [00:00]
Moderator
Hey, everybody, I think we'll go ahead and get started. Welcome to Philosophy at lse, a series of public lectures put on jointly by the Forum for European Philosophy, the CPNSS, and the philosophy department at lse. Today we have Armin Schulz, who is in our department working on the interaction of evolutionary biology with philosophy, social science and psychology. These kinds of questions. And today he's going to be speaking to us about rules and representations, desire from an evolutionary point of view. So I'll turn it over to Armin.
- [00:42]
Armin Schulz
Thanks very much. Let me fix it. Here we go. Okay, so, yeah, thanks very much. I really appreciate it, particularly since there is a philosophy party going on at the same time that some of you. In fact, I can talk, all of you are invited to, which is, you know, it also happens once the talk's over. But I do know the competition is stiff, so I'll do my best to keep up with it. Good. Yeah. So the title has changed slightly, but, you know, it's just the label that's going to be a theme anyway, as we're going through this a little bit. So what are we going to do? The two things I'll try to try to convince you of. The first is trying to talk to you, or talk to you about, convince you of, argue in favour of a certain way of doing philosophy. You might come here with a certain preconception about what philosophy is like or what it should look like, depending on what that view of philosophy is. I might try to sort of violate your expectations a little and try to suggest there's certain ways of doing philosophy that might be interesting and new and different from what you might expect it to be like. And particularly the kind of thing I want to suggest we'll do is something that might look in many ways a bit like a combination of biology and psychology and sort of an odd mixture of all of these different things. And you might sort of say, what does that have to do with philosophy? To some extent. And to some extent that's a good question to ask. And then I'll try to suggest, at the end of the day, that might not matter so much whether it's called philosophy or whatever. It's just interesting. That's my hope. And then the second thing is, that's basically what I'm going to be doing mostly is I'm trying to engage with a certain specific question about how our minds work, maybe our minds, maybe those of other animals as well, namely about the way they're structured. So that's the sort of the question I'll use as a way to address this. Okay, so here's a way to think about the issue I want to talk about. There's this model that underlies a lot of the work in the social sciences, a lot of work in economics, arguably some of the work in social science, certainly a lot of work in psychology, particularly about mental psychology, and a lot of work in cognitive science as well, and a lot of work in philosophy, which is. You could call that model the model of the belief desire psychology. The idea is that there's sort of two kinds of mental states that organisms use when they make decisions, belief states and desire states. The belief states are sort of there to tell the organism in some sense what the world is like. And the desire state saw states the world should be like to be good for the organism. On a famous metaphor, beliefs are like maps and desires are like places on a map that tell you where you want to go. It is also interesting. This is just a sign. I will briefly mention this later on as we go through this. This sort of picture arguably underlies the sort of ordinary, everyday psychology we used to make sense of each other. So when I ask you, why, you know. So your friend asks you, why did you go to this lecture? You, you know, it was, you know, it was a terrible lecture. Why would you go to that kind of thing? And you say, well, the reason is I thought it had desire in the title. It sounded really interesting. Turns out that was just a mistake. That is easily explained because I explained what you did by a false belief. I can also say, you know, you know, why did you go to that lecture? Because I just wanted to be bored for two hours. That's the kind of place that also explains why you're here. Then there's that sort of, that's meant to be happening every day, all the time sort of thing. Whether that's true or not is not something I want to touch on. That might not be true. It might not be, in fact the case that when we make sense of each other in everyday life that we actually use a psychology like this. This has become controversial and. And particularly what that psychology looks like is quite controversial. What I'm interested in is sort of the scientific side of this. I think there's this model of the mind that underlies the way lots of scientific theorizing works. And this model is based on this belief desire dichotomy. To what extent that model matches what's actually going on in everyday life. That's not a question I want to address now as it Turns out I think there are lots of ways in which people, people have studied this model trying to understand how the model works and how beliefs and designs are combined to make a decision and how these components work individually. And people have been doing this in psychology, have been doing this in philosophy, they've been doing it in cognitive science, doing mathematical modeling on this. And sort of recently one of the more fruitful approaches to this has been an evolutionary one. People have started to ask, look, minds, human minds as well as the minds of other organisms. In fact they're traits like other traits, traits like size. And if you're interested in how a trait is structured then it's often a good idea to just ask what can we say about this trait from an evolutionary standpoint? And so needless to say, people started to do the same thing when it comes to this sort of model of the mind. What is interesting about it however, is that typically when people have done this, they tended to talk about beliefs. They've sort of left their eyes out of the picture for one reason or another. So this, there's some body of work talking about the evolution of beliefs. Why would organisms evolve states that are belief like in some sense and you know, why would they do this and how do these belief like states look like given this and so on. But it's not a lot of work on desire so much. Why would you evolve desires independently of beliefs? If you can maybe ask that question. And that's a little odd obviously because it's, it's the belief desire model of the mind. It's got two parts. So it's just focused on one part. So what I'll try to do here is to answer or give one answer to the question about why desires might have evolved. And I'll try to suggest why or give you, convince you that this is an interesting question and that the answer to this question is interesting for our understanding of how our minds work. Also should say, I'm going to say a little bit more about what I mean by desire. And. Okay, here is sort of, here's where I eat my word. One of the very few attempts in the literature to even address that explicitly went out to address the question of why the zire like states might have evolved is by a philosopher called Kim Stolni in his book from 2003. There's earlier accounts that he's of his of sterlny stuff that he's sort of working off of. And there are other people in the cognitive ethology literature and so on that have been doing work on this. So he's not the only one working on this. His is for us, sort of the one that's most usefully focused on it. It's the most directly speaking, I think, to the issues that I want to address. And it draws on a lot of different other accounts. So it's sort of a nice synergistic account. And as I want to suggest, his account is interesting. I think it's interesting because on the one hand it does not quite right, but it's wrong in the right way to make it interesting. And there's lot you can, you can use it for. Sir Ali's account comes in two parts. There is a sort of a negative part which, which says that the reason why beliefs have evolved must be different from the reason why desires have evolved. So the kind of stuff that drove the evolution of beliefs must be different from the kind of stuff that drove the evolution of desire. The rough argument here is just that, look, beliefs are meant to be these kind of states that tell you what the world is like. They are meant to be. You know, they need to deal with an environment that is deceptive and certain. Typically, organisms have to deal with predators that don't want them to know that they're there, or they have to eat prey that doesn't want them to know that it is there. You have to deal with plants that sort of pretend not to be edible even though they're edible and so on, or plants that look edible even though they're not. And so you need to figure out how to deal with this. And so Zorani wants to say beliefs have to sort of respond to this kind of environment. But desires are certain. They're sort of responsive to your internal states. They say what it would be good for the world to be like. So, for example, if you're thirsty, might be good to get some drink, but that kind of stuff is signaled to you truthfully. And so these are very different environments that these two states are responsive to. So we need to look for different reasons why beliefs evolve from why desires evolved. And then he gives a positive story about why desires have in fact evolved, which is the idea that in some environments at least, desires have features that make it easier and more reliable for an organism to make decisions. What I want to do here, I'm not going to really say anything about the negative part. I think there is stuff to be said about it. I'm just going to assume for the sake of the argument that it's true, or at least what I'm going to assume is that it's useful to ask for a separate account of the evolution of desires from the evolution of beliefs. I'm not going to, not going to presume anything that the organism we're talking about here also has folate. Maybe they don't, maybe they do. I'm going to leave that open. I'm just going to ask, here's some organism, is there a reason to think that or what kind of circumstances would give us reason to think it should evolve? Something like the XYZ stir only then goes on to say, when you ask this sort of question, you need to sort of in some sense go in two steps. Again, the first step is you need to ask yourself if an organism does not use desires to make decisions, but how else would it make decisions? And his answer is clear. And this answer seems to be right. As we'll see in a second. If you don't use desires to make decisions, you're probably going to be some sort of drive based organism. Use some set of drives. Great. Now we've just got two labels as eyes and drives. What's the difference here? Roughly, drives are said to be states of an organism that dispose it to act in certain ways without having representational contents. These are like switches that are just triggered by the environment. So if your switch is on the left, it's like a thirst switch that says, you know, switch is on the left. If you're in the presence of water, drink, there's no further sort of processing going on. You're not representing the fact that you ought to be drinking now, you're just drinking when you're in the presence of. That's all that's going on. By contrast, if you're relying on designs, you're relying on space that represents what the world looks to be like for it to delete the organism, what it would be good for the world to be like for the organism, what the organism ought to do in the situation it's in. Something like this. So this thing is not representational, does not rely on making sort of drawing of present representation of the world. But this is sort of more like a trigger reflex and this relies on these representations. Next question then is of course to say, huh, great. What again is it to represent something? And I'm not going to answer this question because it's quite controversial, quite tricky and I don't have to do it, so I'm not going to. There are quite a lot of different accounts about, in the literature actually going back to certainly to Locke and Kant, but arguably even Further back. So this is an old debate. Here's just what matters for us. If on the motivational side I want to say you are representational, you're using desires to the extent that you A consult B some sort of explicit tokening of what the organism of what you ought to be doing in the situation that you're in in order to make a decision. That is to say you need to. When you make a decision and there's something in your mind that you can, that you can consult and say okay, what is it that I ought to do? And then you consult that thing and that thing tells you what it is that you ought to do. You're going through this mental state type thing that is making decisions in this representation away. If you just sort of act without going through this extra step, you're acting based on drives. It's an important distinction because it's not enough on this picture that you act as if you went through this sort of reasoning. You actually have to do it. We'll come back to this sort of distinction. That's going to be a key part of the picture. This is the story that this is in the background. Cerulean isn't hyper clear on this, but this is basically in the background. What Sorelli is wanting to say. I believe and it is not just surroundings account, this is from the background of the literature here. I believe it's sort of a minimal account. And I want to just agree with this in the doctors and here's again you might say, is this the same desire that I have in ordinary life when I say I didn't want to go to this talk, but my friend Navy then you know, is this the same kind of desire? Is this different? I want to leave this open. Our ordinary sense of the might be a bit cheer, might involve lots of other things. I'm just going to use this very basic version of it. Then Stranda goes on to say so this was the first part. If you don't rely on desires, you rely on drives. And he says when would you switch from drives to desires? He says, well, in some environments using desires is better than using drives. And there's sort of four key advantages that he wants to say desires have over drives. Firstly, if you're desire based, you have an easier time making adaptive decisions, decisions that are fitness enhancement, that are good for you biologically. You know, you can have more kids that way if you face a lot of different options in circumstances, right?
- [14:48]
Philosophy Student
So.
- [14:50]
Armin Schulz
If you have a fight or flight response, you can stay and fight or you can run away. These are your two options. Drivers might be good. They're only ones to say but if there are 14 different fighting behaviors, 14 different fleeing behaviors, it gets a little bit more tricky to figure out how these drives are meant to accomplish this task. With desires that's meant to be easier. Secondly, strongly wants to say design based organisms are better because you don't need to rely on a vast number of motivational states. In some sense you can see 2 as an answer to 1. 1. One way in which this drive based organism could make sense of a lot of behavioral options is to have a lot of drive have a lot of drives. You can make sense of a lot of possible options. Just have a different drive for every sort of option, as it were. But then arguably you get a lot of different drives. And it seems having that many drives, it's a little biologically impossible. You're dealing with, you know, vast numbers of these bizarre reflexes, reflex like things. And that. That's not possible. So what's this? Yeah. Desire based organisms are also sent to not to have to rely on these strange unreliable mechanisms for making decisions. Particularly what Cerroni has in mind here is this winner take all mechanisms. It seems that if there's, if you have in a certain situation several drives are active at the same time. So you're hungry and thirsty at the same time. Drive based organisms go with the strongest drive. Then you just drink or you just eat. But it seems lots of circumstances, you kind of want to do something in between, right? You want to drink and eat. And so a desire based organism can sort of do sort of something in between the options where the drive based organism is forced to do disease or. And that might just be the wrong thing. And then he says a desire based organisms are faster in changing in adapting to the world than the drive based organisms. Drive based organisms are basically stuck in evolutionary time because if you so that they have certain needs and so they have certain drives that help them, help them satisfy these needs. But then their needs change or the environment changes, gives them new ways to satisfy these needs. They have to then slowly get to have these new drives. It takes a long time. But with designs you can learn, you can say ah, you know, you have that you want to satisfy some design, you just acquire new beliefs. If you have beliefs or you make sense of the world in a different way now and that allows you to just say okay then now I want to do this other thing you can learn much quicker goes to developmental time. And so, and that's good. So you don't have to wait generations to get stuff sorted out. That's Durani's story in a nutshell. Here's why I don't think the story is all that far. I do think this is perfectly right that I think the way Surround sets up the account as desires versus drives, that's I think the right kind of contrast. That's not just his loss. I mean this is sort of, this is the contrast in literature. This seems to be the key debate. And at any rate this is certainly a key debate to be had here. Why should you rely on representations? Why should you represent the world when you could just sort of rely on drive without these representations? That's a key question we want to answer and I think that's right that we want to answer this question. What's not quite right is I think the way he wants to ask this question that might be too strong in some ways. I want to agree with lots of what Steroni is saying. I think he's just not giving the story really that he should be giving. He stops a little short, I think. So in particular sort of two sided problem I think for his account. Firstly, I don't think it's not clear why desire based organisms would avoid the problem he alleges to exist for drives. And it's not clear why drives have these problems in the first place. And so you can sort of see this isn't such a going to be such a plausible story. And I'm going to just tell you this briefly why I think A and B, these two things are true for these four reasons that he had. So here's the first reason. Design based organisms are said to have an easier time making adaptive decisions when we render range variable options. But in many ways that just seems to me that's just begging the question. Maybe that's right. But why on earth would that be so? I mean I can. I told you, I just kind of said that. But I didn't really give a reason. It's not. Sturani doesn't give a reason and it's not at all clear to me. So it's not obvious what the reason should be here. This doesn't seem to be the kind of thing you can just say you want sort of evidence of this, you want something to rely on me as a matter of fact, I think that's probably true. But you have to do some work to get there. So one way to think about what I'm going to do next is to try to flesh this out. Another way to think about it is that I'm giving a different account. Doesn't mean that, right? So on the one hand, what's the problem for drives here and why are desires better? With point 2, this becomes more obvious, right? So he says if you're relying on, on drives, this is this idea you just have. Maybe if you just have a lot of drives, then you can solve this problem somehow. And he says, well, that's implausible because lots of these drives, having lots of drives are implausible. But that's odd because why is it so problematic to have a large number of drives? I mean, how many drives is too many? What's the problem here exactly? I mean, is 50 drives too much? Are we talking 500? What's the right order of magnitude? And for that matter, why is it so much better to have a lot of desires when compared to having a lot of drives? I mean, on the face of it, it just seems like I can have this or I can have this. Why is this one better than this one? Thirdly, he says desire based organisms don't have to rely on these unreliable mechanisms for adjudicating amongst different active motivational states. They don't have to rely on these winner take all mechanisms. But it's just not clear to me why a drive based organism must rely on this. I mean, it's true empirically, often these, the kind of accounts that people give here typically rely on these. But I don't think there's any necessity on this. These are simple accounts that people employ is sort of a good way to make sense of some behavior. But there's nothing intrinsic to drive based organisms that requires them to use a mechanism like this, as far as I can tell. And by the way, the same problem seems to arise for desire based organisms who would also have to figure out a way, if you have different desires that are active, you both want to eat and want to drink. Now what you going to do? It's not clear to me that Ceroni just gets to say, oh, it's easier to solve that problem rather than to have two active drives. And to make it worse, it seems like the few people who've sort of looked at this in the philosophical literature and the psychological literature, even in the neuroscientific literature, come up with accounts to solve this problem that are precisely the winner, take all account that Sterling finds so implausible, so it seems this whole story is. And then fourthly, design based organisms are said to be sort of quicker. They can adapt in the developmental, not evolutionary time, but that seems to me not right. I mean, why would you think that acquiring new drive has to happen in evolution time? There's plenty of data that you can acquire in drives over developmental time. You can do it. You know, there's a lot of work in conditioning. There's a lot of work from Darwin onwards about how habits can start to turn into instincts, how you can an organism that can acquire something, and how habits and instincts are very much alike. So you can acquire new habits, you can learn how to drive a car. There's in many ways something that's quasi instinctual, quasi reflex, like quasi driveway. But you can adopt this and adapt this and change this in developmental time. So it's not clear that this is true. And by the way, it's not clear why desires must be the kind of things you can change developmental time. Why isn't it the case that the fundamental desires you have are acquired innately or you have them innately and if you have to change these, you have to go through evolutionary time as well. So it's just not clear that we have an answer to this question. So I say maybe we should look for a different account or account that clarifies these things so that can at least help us solve these problems. Well, maybe. Here's one thing one can say, main idea behind this account is fairly simple. Lots of times when organisms interact with the world, there's a certain pattern behind their action. They do stuff, but there is a certain kind of pattern in what they do. This is important because in many ways you can think about this pattern as some principle that organizes, drives the actions. Then if you could just represent, consult this principle when making a decision. You don't have to memorize every instance of the principle. You can just memorize the principle and apply the principle to the particular situation you're in and then you can make a decision. This way you don't have to worry about memorizing every instance. You just memorize the principle effectively. And that can be, can help you save various kinds of resources, energetic cognitive resources of various kinds and help you make more adaptive decisions. And that can be a fitness enhancing. That's the story. To get this, to get a bit of a handle on what I'm trying to say. Here's an example, sort of a stylized example, right? Here's an organism, some organism, and assume it's drive based and it's living in or finds itself in some mountainous region and doesn't have water, but it needs to drink. I assume also that there are several sources of water. In the vicinity, some higher in the mountain, some lower. We also assume that the organism sort of has information about these water sources. And then we also know that the way the organism is constituted, walking, moving uphill is twice as costly in terms of energy than moving down. Then when, if this, if you look at this kind of organism, if this organism has to, wants to act appropriately adaptively, and because it's drive based, it needs to sort of connect a particular action to every possible position on the mountain it might find itself in or can distinguish. And so then if, if it needs to decide what to do, it needs to detect where on the mountain it is. I mean, detect in a sense that it might falsely detect it, or it needs to sort of acquire information somehow about where it is on the mountain and then use that to go through it. Sort of this, it sort of has this table and it finds the appropriate entry on the table that says, okay, for this position, do this, and then you act accordingly. So graphically, right? So here's the mountain and it's a nice mountain because it's, because it's strictly monotonically increasing. So we can represent any position on the mountain by a certain number, which is quite nice. All mountains are like this, naturally. And there's here, these dots are the water sources, spots. And then here's the table that sort of organizes the decision making mechanism of the organism. So it has sort of all the possible locations, right? It doesn't distinguish in between these lines, it can just display, distinguish these lines or it does some rounding there, whatever. And then it associates some particular water source in line with the. Walking uphill is twice as costly as walking downhill. And so it associates a certain water source with every entry on that table. So if you're an organism, you might be here. So that's sort of about 6ish, right? Then what you're going to do is you're going to go through your table and you find yourself right here. You say, okay, what is it I ought to do? I ought to go to water Source on Level 4. I just wanted to show I can do this with PowerPoint. Maybe that's the reason why I. Anyway, if, however, this organism is desire based, not drive based, the situation looks quite different because if it represents sort of the principle behind this action, it can calculate what to do. It doesn't have to rely on a table, it doesn't have to memorize it. So it's like this. If you have the ability to represent the principle that generates your actions, you have a behavioral rule that you can follow. So you have this rule for example that says walk to the water source, that is this from here. So X is the nearest downhill source, Y is the nearest uphill Source and it's 20, it's faster to walk uphill than downhill. And you want to find the shorter of the two distances. That's the rule. And then you say ah, you know, presumably you know, you've got some information about the rock environment. You presume yourself to be at location three, you know, height three on the mountain. Then you can just plug this into this equation and compute the fact that the minimum of 12 times 1 is 2. You go to water source 2. That should probably be 1, shouldn't it? Yeah. Okay, well that's 1. But you know, if you can do the math right, that will work for you. Now the key idea behind this system is that organisms that rely on laws to make decisions are relying on representation and motivation. What I want to try to suggest is if you make decisions in this rule based way, in some ways that is akin to making decisions using design. So the ability to make decisions in a rule based way requires you to have desire like estates. The ability to make decisions in a rule based way might be a reason for the evolution of desire like states which are these rules. So you can see this, it's not if you're rule driven, you don't just react to the environment you're in, but you consult a representation, you consult explicit mental tokens of how votes react to environment. Right? So unlike the drive based organism who just says I'm here, let me do this, the drive based organism says I'm here, what is it that I ought to do? It goes through the rule, consults the rule, computes the solution and goes where it's ought to go. And so that's why I want to say this is the story of why they can be relying on desires rather.
- [28:49]
Philosophy Faculty Member
Than.
- [28:52]
Armin Schulz
Drivers to make decisions. Two points are important. Clarify this. Depending on your inclination, you might have heart attacks by now already several or you might not. So here's something I'm not doing that you might think I'm doing, but I'm not doing this, which is there's nothing here I want to say that says representing something is following a rule of sorts. There is a tradition and philosophy that connects representation and rule following by intimate or meaning content and rule following is closely connected. So if you're sort of neo Wittgensteinian, you might be, you might be jumping on this. There's sort of the Wittgenstein through Kripke Bogosian sort of line that does this kind of thing. And that's not what I'm doing. All I'm doing here is just that if you consult a rule in deciding what to do, you consult a mental representation of what you ought to do next. You're relying on a design. That's really all I'm doing. There's no intrinsic connection between colour, content or meaning and rules or something else. There might be one that's just not relying on. The other thing is I'm only. What I want to claim is that only desire based organisms can follow a rule. It's not that only desire based organisms can act in accordance with a rule. So this is a famous classic distinction in the philosophy of social science. Also you get this from the sort of, you get a sense of vaguely Wittgenstinian stuff, but he a well known distinction. And to get the sense of distinction, I can compute the square of a natural number. I can do this by consulting a long extensive list of natural numbers in their squares. So if you ask me the natural what's the square of three? I just go, is it one? No. Two? No, three. Okay. And then I go to nine. That's the answer. So for every natural number you give me, I'm just going to go through the relevant spot here. That's one thing I can do. I can also consult the function F of x is equal to X squared. I can just say, okay, you want me to tell you what the square of three is? Well that's easy. Three times three is nine. I can compute it. Now the important one is obviously this thing here. If you're consulting a long list of these sort of, these relations, in some sense you are, in some sense you are consulting the function F is equal to X squared because this list is generated from this function. But you're not, in some sense you're not calculating, not going through the calculation, just consulting. You're acting in accordance with the calculation, you're not actually following the calculation. And it's the same distinction I want to adopt here. So this is to say, what I want to say is if you're dry phase, you might act in accordance with the same rule that desire based organism is acting on. You're just not consulting the rule. You're just acting as if you followed that rule. And that's very different. So why is this important? I think it's important because the explicit consultation of a behavioral rule like this can save you energy and it can save you cognitive resources of various Kinds and it can be adapted in a different way too. So firstly, there's reason to think that if you're relying on rules like this, you probably require smaller memory store, quite likely require more smaller memory store than dry based organisms. There's some evidence that suggests that organisms that have to make, you know, rely on more mental states. For example, if you have a network, if you have to distinguish 700 different categories in the world, for example, you have a vocabulary lexicon of 700 different words to distinguish, your neural network has to be much larger to make sense of this than if you have to only do half that. Much larger neural networks are important and something similar. That's true here. You can either rely on one rule and a bit of extra information about the water sources, or you can rely on this potentially gigantic list of location, state of the world location and water sources, source action connections, which can be very, very long. So you have many, many more states to account for here. This is important because if you, if you can get away with a smaller memory store, a smaller neural network, smaller brain, that's typically good. Brains are really high source of energy. So if you can, if you can cut down on that energy, you can do the same stuff that the drive based guy can do. You just do it with the smaller brain. Good philosophy isn't actually good. It's also good because even if you don't want to move towards smaller brains, you've got now free space, as it were, that you can use for other stuff. Or if you don't have the space yet, you can expand your memory with other things. You can track more relationships amongst your peers or something like this. And then you could do some combination of these. It's also the case that if you rely on rules you have, there are fewer cases where you just don't know what to do because you're now in a situation where you haven't yet assigned an action to a particular location that you're in. You only assigned water sources to the location up to 12. And then you find yourself at location 14. And then you kind of like, I don't know what to do and you kind of have to randomize all the stuff or you don't do anything or something. And that's problematic with the rule because the rules at least potentially can be sort of complete. They're always applicable. You can always compute the row and answer. You're not in the situations where you sort of lack a piece of the puzzle. And so you're just, you're more, as it were, more Consistently able to act adaptively. This is sort of a point that if you're fabulous. Daniel Dennett says stuff like this a lot. And this is important because efficiency gains like this are known to be favored by natural selection. Natural selection is one of these drivers of evolutionary change made famous by Darwin and Wallace back in the 19th century. And there's plenty of evidence, most theoretical, that suggests that something like this would be true. But there's also, there's lots of empirical evidence now so far, for example, there's evidence that suggests that the evolution of bipedalism in humans, that humans walk on two feet, was driven by efficiency considerations in terms of the energy used in local molding between different places, the venation on leaf structure. So the way veins are distributed in leaves of various plants seems to have evolved by efficiency considerations. These veins are distributed according to maximal efficiency in these leaves and sort of further data like this. So there's good reason, empirical and theoretical reason, to think that natural selection is going to favor deficiency. That is cognitive efficiency, where it's available, is adaptive, is fitter, helps you have more kids. Now before, if you're a fan of this literature or if you know a little bit about this and so on. And here's something that's important to be clear on. I've tried to argue that there are fitness differences between two traits, desire based minds and life based minds. This claim is quite different from the claim or we need to say, even if this claim is true, you can't conclude from this that either we must assume that desires always must now evolve, or to the extent that we have reason to think that some organism uses desires to make decisions, that it must have been because of what I've just said, because of these fitness differences. We can't say that because natural selection is not the only factor influencing the evolution of many traits. For virtually all traits, I think factors other than the natural selection are relevant to their evolution. That's because most populations that you're looking at, if organisms are sort of at some point or in the valve and period, have been fairly small, such that just randomness in form of drift have played a role. For some traits, population have always been so large that you can basically ignore other stuff and natural selection is pretty much the only thing that's going on. But for a lot of traits at least, it's important to realize that other things other than natural selection might have influenced the evolution of these traits. Also, there's certain things like inheritance biases. So traits might not be transmitted faithfully. They might be transmitted in certain odd ways. Drives might be transisted much more faithfully than desires. So even though desire based organisms have more kids than drive based ones, the kids that the desire based ones have end up having a lot of drive based machinery instead of desire based ones. And that's not true for desire based order, something like this. So that's something we need to take into account. It's also true that muscle selection is constrained in various ways by genetic, developmental and environmental factors, including physical stuff. So sometimes natural selection cannot change the value of a trait so easily because that trait value is connected. It's a pleiotropic effect of the same gene complex. The same gene gives rise to two different traits. And then changing one trait isn't necessarily so easy because there might be selection on the other trait as well. There might be sort of developmental stuff, so psychological stuff that's in the cell or even later. So the same, you know, two traits might tap into the same developmental program. So if you, if you wanted to change one, you have to change the other. Or if you don't want to change the developmental program, you can't really change the one without also changing the other. There might be environmental constraints. You can't make organisms bigger, for example, if, if you don't feed them the right kind of foods, there might be physical constraints. Making organisms bigger makes them also have more mass, heavier, and you might not want to have them be heavier and so on. So there are all these constraints that you need to take into account. This is important to mention, particularly in this sort of debate, because what I'm trying to talk about is sort of the evolution of psychological traits of one kind or another, cognitive architecture. And a. When you're doing this, taking this into account is quite tricky. It's tricky because if you're looking at, if your soft tissue, skin and so on doesn't fossilize particularly well, cognitive architecture fossilizes even less well. Right, because you're not just looking at how a brain is fossilized through trying to get evidence for how brains are structured. It's not just a neuronaloid that matters, but of what kind of cognitive architecture that neural network realizes. And that's tricky to see. So getting evidence for these things is quite tricky. But because it's tricky, you might even say it is actually quite important to not forget that there is this evidence that we should be looking for. Even though it's not so easy to find, I should know it's not impossible to find evidence like this. For example, Stephen Meissen, a Paleobiologist has done work on decision making on hunter gezo societies using very ingenious methods to sort of determine the sizes of different foraging sites, the kind of prey that was hunted, the location of the foraging sites, what kind of tools they used and so on to determine what the decision making procedures were used in different hunter Gallo tribes in the Pleistocene and earlier. So there is, so it's possible, it's just kind of tricky. And the other thing to mention is that there is a history here in people doing this sort of stuff, evolutionary psychology, e things not taking these points into account and going from something like trait A is fitter than trait B probably to trait A must have evolved because it's fitter than this. So male, human males back in the, you know, back in the Pleistocene it was adapted for human males to be promiscuous. Therefore male promiscuity now is not a patient. It has arisen by natural selection because of this. That's not, that's completely ignoring this, not sort of taking into account, sort of any of the fact that we should take into account, not even trying to find evidence for these sorts of things. We have to be careful that we don't do this kind of thing. So what I want to say is I think it's important, it's important to be clear about the kind of conclusion that I want to derive and the kind of conclusion I don't want to derive. What I want to say is this. All I wanted to say is that there are fitness differences between two traits. And I think that's not making the next step into saying exactly which traits ought to have evolved. But even though I don't want to make that step, I don't want to say this is completely boring or irrelevant what I have to say, and here's why. So firstly, we need to know natural selection is important. So often just, you know, it's not sort of you say, okay, one trait is fitter, who cares? Well, it is a bit more important than this and it's particularly so for complex traits like cognitive architecture, sort of even people who are very sort of strongly in many senses, strongly anti adaptationist. So they don't think that selection is that important. Overall, they're quite comfortable with complex traits like mind designs and eyes, at least to a large extent having been designed by an selection. The reason is it's just very hard to imagine this stuff not having been influenced in some way at all by natural selection. So figuring out the fitter, I design the fitter mind design is important for that reason. And secondly, all I want to do here, and I think the same is true for sterility. And so that people were careful in this debate, all they want to say is the fact that one trait is fitter than another is evidence for that trait evolving, not more than that. It's not forcing us to accept one view or another. It's just saying here's evidence for this view. The evidence is that this trade is fair. That gives me a reason to think that this trade would evolve. Other things are going on here too. And the more I know about it, the more I might sort of say overall I should probably have reason to conclude this trade didn't evolve. But the fact that the trade is fitter is evidence for its evolution. That's all that I want to conclude. So in other ways I want to say here's a prima facie reason for the the evolution of desires, that they're more efficient cognitively. Prima facie reasons, just in virtue of what they are stuff can interfere with them. That's okay. There's still reasons and they should not be ignored just because they're prima facie reason. So it's a moderate conclusion, but I think it's still an interesting conclusion as we'll see. I hope it has interesting implications, but it is a bit more, bit more tricky than the sort of the knee jerk evolutionary psychology thing which we don't want to do anyway. So it's all good. Now at this point you might have sort of objections. You might say this is all wrong. And here are three objections that you might have that think why everything I said is just wrong. And I think these objections are. I picked these objections because I think they get an important stuff and I think I can respond to these objects. I think that they do raise important issues and that there's a lot of interesting stuff to be said in. So the first objection that obviously comes to mind is this. I used one sort of measure of cognitive efficiency which is sort of effectively quantifying the number of mental states you need to make decisions, right? The drive based one need a lot of drives. The desired based one need fewer motivational states. And that's something that's floating through surroundings discussion as well. And I'm just sort of piggybacking on this and using this. And as I suggested, this is not crazy. There's other people use some measures like this, but you could use other measures and you could create other measures that would have the relationship come out differently. Maybe drives are more efficient on some other measure of cognitive Efficiency, maybe one that takes into account different types of memory. Maybe if you're relying on desires, you need sort of declarative memory. If you're a relying on drives, you only need sort of process type memory. Distinctions like this, maybe these are important. And I want to agree, Erica, it's actually quite tricky to come up with clearly the good measure, the best measure for cognitive efficiency, that's tricky. And it's also obviously true that different measures of cognitive efficiency is going to order these two things differently. That's true. But I do want to say the measure I'm using is not correct. As I say, other people use it. It's not used just in this context, but it's used, for example, when you're doing linguistics and computational linguistics. People use measures like this, various forms of cognitive neuroscience. People do this kind of stuff. So it's not crazy. And. So there's certainly, I think this is, at the very least, I want to say this is a useful starting point for further work on this. Look, if it turns out differently, I'm finding things turning out differently. But at least let's sort of figure out cognitive efficiency is the thing to focus on here. Let's figure out a good way to measure cognitive efficiency. Maybe we need to take other things into account, but this is a good starting point at the very least. And particularly other things that people have come up with, it's not so clear that they work. So this whole process memory versus declarative memory distinction, that sounds really plausible, but then you look into the memory literature and then people aren't so clear exactly what the distinction is, which one is more costly at the end of the day, neurally they, neurologically they turn out to be not so strongly dissociated in some context and so on. So it's not so obvious exactly, exactly what we meant to say about these things. For us, I think the key is, yes, I think figuring out a good measure of copper proficiency is hard. My measure isn't crazy, it's got something going for it. It's a good starting point. And that's really all I need because remember, I'm trying to give you a prima facie reason for why desires have evolved. My conclusion isn't, you know, it isn't the conclusion. Anyway, this is definitely what happened. Here's one thing that might, you know, that speaks in favor, that would force the evolution of designs. And you know, I'm quite happy with us fiddling with this and saying, well, yes, drive based organisms, you need more drives. So that's Sort of, that speaks in favor of desires. But then other things speak in favor of drives as well. So maybe overall we come out differently. That's fine. But the conclusion I think as a habit still stands there. That's the first objection. Second objection also comes to mind. You might say, look, how do I distinguish again desire and drive based organisms? Because I said, I mean you can do the same stuff when you're drive based and desire based. They seem to look quite similar. And so how can I distinguish them? And sort of, you might sort of the more particular an institution like this, you might be more preparing like this. And you sort of say, look, I mean I want sort of, I want to be able to test this stuff and falsify it in some way. And I need data to do this. But if I can't distinguish desires from drives, how can I possibly falsify this account? And now here's what I want to say in response to this. I think a, obviously this is a wrong. It's not actually, I don't want to say it's a wrong. I do want to say this requires some work. But this is true for all accounts of desire, evolutionary accounts and all accounts of cognitive architecture. Effectively it's often really tricky to distinguish amongst different architecture. That is true, but secondly, it's not, that's not something that's impossible. I think there are ways to do this. And so the obvious way to do it is to do more cognitive ethology to come up with better models, more models about how animals make decisions, and then come up with the best way of figuring out how different animals make decisions. And over time these models should give us a clue about whether these organisms are better captured by a desired model or by a drive based model. This is exactly the kind of stuff that's in the last five, ten years been happening in sort of cognitive neuroscience where we try to actually look at how neural architecture realize what kind of systems you, what kind of process diagrams and so on are best used to capture the way certain animals make decisions. And then you sort of map these different diagrams onto each other and onto belief design, drive based organism architecture. And you get a way to answer this question. And the other thing I want to, even if you don't really. So this is sort of, I think empirical, getting a way to empirically distinguish these accounts is really important. But I think there's an interest even before you fully can say which organisms are desirable, which drive base, there's still an interest in this account because for example, I can say that there's something wrong with Sterling's account. Even though I don't have a clear list of which organization and which drive based I can do this because I can rely on other data and other theoretical principles to do this. So it's not so clear that we totally need the answer right now. The third issue is this picking up on the second point here we do notice here's what seems to be clear what sort of cognitive pathology says is that some organisms at least seem to be making decisions used on drive this is a model that's been used quite a bit fruitfully a lot of work in cognitive ethology. Amelia for example some mollusks mollusks. Not all of them however seem to be captured quite well by a drive based model. And this is important because you might say on my account having desires is fitter than having drives. So how is it possible there's still these drive based organisms remain? And here's this is important again also because this sharpens the conclusion I want to give and the anti the adaptationist worries you might have. So it helps us focus in on this. So the first two things I want to say is just the things that I said earlier about what one has to be worried about when it comes to natural selection stuff is firstly, natural selection is not the only factor that influences the evolution of Importantly, even for homologous traits of similar traits in different organisms, a natural selection can be differentially important because for example population sizes can be different. So it could be the case that even though having desires is sort of universally adaptive, natural selection as a drive of evolutionary change wasn't as important for all organisms, you know, in the same way. So we wouldn't expect drives the desires to evolve in more cases because natural selection wasn't equally important in all contexts. Secondly, there might be different constraints on how natural selection works in different organisms. Again, some organisms evolve desires, some drugs that's perfectly fine on my account. I'm not making this conclusion of me to assume this. All I'm just saying is this evidential claim and lastly I think the picture is more complicated anyway in that it's quite likely and I'm here this is something that also stood only at maids and there's lots but possibly people I think just say this kind of stuff and it is plausible certainly that if you're moving from a drive based cognitive architecture to a desired based code it seems that there's a period in between where you actually so drive based. Here's one more thing. Drive based organisms are at a local maximum. They're doing quite well. And if you have to move away from this, you're going to make bad decisions for a while, worse decisions than if you were to rely on drives purely. So you go through this period where you suppose of not as good as you were before, but not, you know, you're not yet in the desire based trajectory either. So you're sort of. Actually there's a period when natural selection actually selects against moving from drives to desire. So here's this graph. So if you're somewhere here, then moving away from drives a little bit forces you back to drives. Only if you get a sufficiently large jump over here are you pushed into the direction of desire. So this is also important because for the same organism not all of its decisions need to be desire or drive based. And one reason is just that for some decisions sort of moving, sort of, if you say you're Zorunya, you only face a fight and flight response in some context there's only two different actions you've got to do. Then sort of relying on a rule might not give you that many efficiency savings might not be that good for you so that the selection pressure is quite weak. But you already have yourself organized quite well. So any small change away from this in your cognitive architecture is going to be detrimental to you. Only if you sort of end up in some context where these sort of basins are quite small will you actually move over to desires. Or if you get a sufficiently large ship, if a lot of stuff happens, a lot of changes happen to your cognitive architecture. You might then end up saying well now a better adopt the side or something like this. It is also here interesting to note that looking back at Cerroni in the previous point I think, I'm sorry this is so far below but here's an interesting difference I think between my account and Cerroni's account because my account says desires are basically universally adapted. Taking all of these complications into account. Zralni's account doesn't say this. Zoralny says in some contexts having desires adaptive. He's sort of more cagey about it. So I think on the face of it I would expect that on my account we should expect more organisms to have desires rather than strands. Just as a first sort of thought about this and that's kind of nice. It's a way to actually distinguish these two accounts. For what it's worth, we're almost there. So good news, here's an implication. So if you buy this, all of this, you might say this all very interesting. Why should I care and you might care because you think it's inherently interesting, because you're a nice person. All of that also care because there's been a debate in cognitive science and philosophy about this whole business about why we're relying on representations in the first place. Right? Representation, you represent the world to you. But why would you do that? The world is sitting there. You can just, you know, as they say, the world is its own best market. Just use the world that's sitting right there. You don't have to rebuild a picture of the world in your mind. You can just use the world as it is in front of you. And isn't that going to be sufficient on this picture of this whole business about representation that's been popular for a while. It's an avoidable luxury that we should get rid of. And what I want to suggest is this is interesting because on my picture, in some ways my picture supports this sort of recent anti representational business. And in some ways it disconfirms it. Here's how I think it supports it. So how it disconfirms it on my account, what makes representations interesting is that they make it easier for organizations to make decisions. They streamline decision making. The difference between representation and non representational organisms isn't that the representation representation organism has to rebuild the model of the world, go through extra stuff, but it's rather that it has an easier time making decisions. It's quicker, it streamlines its decision making compared to driving. So that's exactly. It's almost the reverse of what you get in a lot of this stuff on embodied cognition. But on the other hand, I also think that a, because it's true that there might be this cost between transition period between drives and designs, and also because it's true that I think you can do pretty much the same stuff you can do with drives that you can do with designs. It is also true that you shouldn't jump to the conclusion that we should see everything, designs everywhere and every decision needs to be based on representations. You can get around the world quite well using drives. And there's often good reason to think that organisms cap this drive based model because it does quite well for them. Might be ultimately, if you could get to desires, you could do probably equally well, maybe slightly better with desires, but where you are is perfectly good. And in that sense I think the anti representationists are certainly onto something. And that's absolutely right. There's this idea that we shouldn't over emphasize this representationist business. Yay. Conclusion. What I've tried To give you is a prima facie reason for why something like desire based organisms have a wolf and have something to do with the, the efficiency in the organism's cognitive decision making mechanism. Particularly. I think desires are better because they sort of. You replace a lot of storage with a bit of computation. The computation doesn't need to be particularly tricky, it can often be quite simple. But this sort of replacing of computation with storage can be energy saving. This is interesting because it stands in contrast to Cerlani's account, which is maybe kind of like this, maybe he tries to gesture this. Maybe it's completely different, not entirely obvious, but it isn't quite Sterlly's account. And I think it adds something at least to Sternly's account. So what I'm trying to say is I hope to sort of add something to the debate about the evolution of our evolutionary understanding of ourselves, other organisms, our mind, particularly of designers, but also added something to the belief design novel itself in terms of the value of representation, the value of designers itself. And here's one of these things where I want to say, look, here's an interesting issue and you know, is this philosophy, is this biology, is it cognitive science, is it psychology? To some extent, who cares, right? Let's throw everything we got at this problem. Let's try to figure out how we can solve it. Let's do the best we can in solving this problem. That's what we're here for. Sort of how we label it at the end of the day. Is this still Carl philosophy? Is this sufficiently philosophical to be worth the name? Or is this just really bad data lacking psychology or something? You know, if that's what it is, then make it better. Then let's not worry about whether it has to remain philosophical at the end of the day. So that's my conclusion. Thanks very much.
- [58:23]
Moderator
Okay, we've got about 20 half hour for questions here, so I think I'll let Armin feel his own questions. But if you could keep them concise and efficient, maybe I'll get to my question at the end.
- [58:40]
Armin Schulz
1, 2, 3.
- [58:42]
Cognitive Scientist
So in terms of this efficiency argument, it seems to be you're equating drives with table lookup behavior. Yeah, so for instance, I could make a biochemical circuit with about three components.
- [58:54]
Armin Schulz
Of pixel square root.
- [58:56]
Cognitive Scientist
So it seems to me that I can have very efficient rule following, sorry, acting in accordance with rule implementation. That would be in general more efficient than rule following rule behavior. So for instance, almost any SO way of constructing a neural architecture that follows a rule is A special case for something that obeys a rule and has extra requirements in order to obey the rule rather than follow the rule. So it seems to me that rule following behavior would always be simpler and more efficient than rule. Sorry, Rule affordance behavior would always be more efficient than rule quality behavior.
- [59:39]
Armin Schulz
Okay, yes. So I'm not sure I fully avoid that question, but. So. I think. So I think I want to make a distinction between. So there might be different kinds of rule following that I've told you to some extent. If you think of it in your network of sorts, then in some sense there's going to be acting in causes for some rule anyway just because to get the network off the ground. That's fine. Then there's still a further question about the kind of rules that I want to talk about. The behavioral rules, how are they? So maybe the lookup table business is, you know, I don't know if that's helpful or not. So I completely agree. So you can make finite distinctions amongst dry placed organisms. You could label them differently, that's fine. But the distinction between relying on a lookup table versus relying on a rule, that's the key distinction. I want to rely on four behavioral rules. Not right. I don't want to say in many ways. I don't want to say the driveways desire based organisms doesn't rely on any drive like thing. Obviously it still does in many ways. One way to think about it is maybe I'm trying to locate the desire based organism in between the sort of full blown representation of stuff that you get in some philosophy and the pure drive based stuff but more in the middle to get better bolts.
- [61:07]
Cognitive Scientist
But I don't see the connection between saying it's drive based and saying it has a table lookup. I don't see that connection at all. It seems that you could Internet dry base organisms with a lookup table. But that's one of a million different ways of encoding drive based behavior. Right. And so to say that rule based behavior is more efficient than one specific way of encoding drive based behavior doesn't do any.
- [61:36]
Armin Schulz
This is why it's important. So I'm not wanting to say this is why this explicit consulting and explicit token is important. So the point isn't that the desire based organism is just a different way of encoding the same rule. The point is actually the architecture is differently organized in the two cases. So in one case there is no encoding of the rule. In the other case you encode the rule and you make a decision through the explicit consultation of the rule. So in many ways you can think about it. I'm comparing two different ways of implementing the rule. You can implement it explicitly or you can implement it implicitly. And I say the explicit representation. Explicit implementation has the benefit. When you're looking at several options, it is more efficient, not the simple options. But.
- [62:32]
Philosophy Faculty Member
You didn't mention the word sentience consciousness at all. So is it all right to assume that these desires are entirely unconscious or.
- [62:41]
Armin Schulz
That they could be? Yes.
- [62:43]
Philosophy Faculty Member
So you're using the word like desire, which suggests some kind of awareness, but in fact it can be totally explained in as well objective terms of some kind of teleological system.
- [62:56]
Armin Schulz
Absolutely. So that's exactly right. So this is one of these places where you might say ordinary notion of desire might be richer or might have aspects to it that are not captured by this. And that's fine. Then I'm using some sort of technical version of this or desire star or something that's okay. Yes, exactly.
- [63:11]
Moderator
All right.
- [63:12]
Philosophy Faculty Member
Okay. So this has nothing to do with the evolution of consciousness.
- [63:15]
Armin Schulz
No, exactly. Yes. This stuff might be conscious of. It might be separate account of this might be the second. Okay, thank you.
- [63:22]
Cognitive Scientist
Hi.
- [63:22]
Armin Schulz
Hi. Sort of drive organisms versus design organisms and brain biology, for example, in humans shows that we got to our primal brain and then on top of it we've got the sort of. The higher levels of functioning that kind.
- [63:33]
Evolutionary Biologist
Of gets going through at different levels.
- [63:35]
Armin Schulz
Could not we have both in one organism? Absolutely. You've got the drive and you've got desire coexisting. Yes, sorry, I meant this one quite quick. I meant to suggest that this is true. In fact, this is. I mean, this is clearly plausible. It seems to be clearly true. And the reason that this is that I don't want to say that, so I don't see it as such one wash that sort of. If you consider all the decisions the organism has to make, it's not that all aspects of the table, as it were, need to have the same potential for efficiency extraction than all the other ones. Some of that might be quite simple and you do quite well with drives. With drives then so you. You keep that and then other ones not and you want to move away. Maybe this sort of distinction between the older parts of brain and newer parts of brain might be good, but it's sort of, you know, sort of maybe emotions are still there, you know, an amygdala and an insula and so on because, you know, sort of that gets the job done quite quickly and you don't need to worry about it. But then you Know, there's certain other complex social decision making where you have to keep track of all these people and they want to just even go, it gets harder. And there sort of this table, if you do this with drives, gets quite long and then at some point it's just easier to switch to the wall. That's exactly right. Okay, I saw a hand over there and then.
- [64:55]
Philosophy Faculty Member
Yeah, I'm totally sold. Thank you. I notice your examples are almost entirely organic. Like they're plants, animals, humans, and extending into various forms of human design. I wonder if it's possible to extend this to non organic situations, like for instance, in the institutions such as business and government and economic groups. And whether or not governments have legendary, driven by reflex drives and desires for going to war or going to do something else. I might be pushing a metaphor too far here, but I'm not sure whether you can take it from that to another. But there seems to be something resembling a parable there.
- [65:49]
Armin Schulz
Yeah, so good. So, I mean, so here's a different. Before I come to your case, so here's a different extension that one might have in mind. So robotics, for example, I think, I think this is sort of an obvious place and I think that's exactly right. So one other implication one could draw out here is that if you're building robots for scientific exploration or something that has to deal with various sort of contingencies and has to make its decisions, this is the sort of story you could tell to say, here's one implementation of the design of the robot that, that's important to keep in mind. This is interesting because some of the people in the body cognition literature will say, forget about representation, they're coming from robotics, they're coming from this angle. And so one way to think about it is to say, well, here's one thing I have representation to do for you that you can't do without representations. They do allow you to streamline decision making. So I completely agree. Yes, there's a potential to extend this in other areas. Whether I wanted to extend it into things like groups of people, you know, governments and economic organizations, it's a bit more tricky because then you have to sort of, you have to attribute these sort of sort of states to the ensemble of people somehow. And that's not, I don't think that's totally crazy, anything people have done stuff like this and so on. So, you know, in principle I don't have a problem. I want to sort of say, let's see what work it does for us. But in principle, I don't have a. Have a major objection for this. Isn't it consequence then that the sort.
- [67:26]
Cognitive Scientist
Of being able to deal with novel.
- [67:29]
Armin Schulz
Environments or larger territories would sort of require desires instead of. I mean, would that be an empirical fact we can investigate?
- [67:36]
Cognitive Scientist
So it's like all right, this is a drive phase.
- [67:39]
Armin Schulz
There's only a certain size of the combination of that. That's really a reference to table situation. Yes, good. Yes and no. So here's the no version. So basically one of the things that Stirlni wants to emphasize is this novelty business is really important for Sturlni. He wants to say look, if you have to deal with these, it's quicker to change. With desires you can learn. And I want to say, well, you can do a lot of that with drives because you can acquire drives quite quickly. If you just add on drives, you can deal with a complex environment as well. Where I do want to say that's exactly right. Is that adding on to in some sense. That's exactly right. Because adding on drives at some point you run into the point at which switching to this representational model becomes quicker. That's exactly right. Larger landscapes, just the longer the. The more options you have, the longer it takes to go through all of the possible options just to figure the entry to keep track of all of these different. The quicker it is to switch. That's exactly right. So social environments are the kind of stuff that surrounding onstay drives the evolution of desires. I think it's right. It's just not quite right for the right reasons or it doesn't give you the reason. There is a hammer check.
- [68:58]
Philosophy Faculty Member
Does your approach imply that one's got to have a desire based organism has got to have a brain, not just a nervous system because that could react to tribes touch fire and you take your hand away without any computation. But it seems to me, look, to have to utilize desires, you've got to have computational power.
- [69:27]
Armin Schulz
Yeah.
- [69:28]
Philosophy Faculty Member
And that requires something larger than a nervous system needs a brain.
- [69:33]
Armin Schulz
Well, I mean it depends. I think you can, I mean. Right, so you can desert depending what kind of behavioral rule you're looking at, what kind of computation you need to do, how high you need to go is an open question. And sort of lots of quite simple nervous systems you might say can do quite complex computations. Yeah. So I mean, so I mean. So some people argue that sort of have to make quite complex computations and they're quite well studied. Same goes for sort of for certain kind of jellyfish which don't have a brain because they don't have a central oxygen system. But you know, they can still do certain kind of computations. Arguably. Maybe bacteria can do these kind of things to some extent. I want to leave this open depending on what the science will say is the right thing. Probably with bacteria, I don't think that's going to be a useful model. With bacteria I think we're doing very well with the drive based model. That's what I would expect.
- [70:37]
Philosophy Faculty Member
I was wondering whether when you get on an airplane they tell you how to get out in the event crash. What they're actually trying to do is to suppress your desire because in, well, there was a tremendous airplane fire in Manchester and they found that the people that survived just got out and other people got their luggage first. 50 odd people ended up dead. So what they were sort of saying was that people had gotten a desire which is to protect their property or take it off, whatever. But the, so that the idea of them saying when this happens, you do this is sort of as it were going, it's reverting from a desire to a drive.
- [71:23]
Armin Schulz
That's good. I mean, so in some sense that's one of these places where you once said, you know, it's this kind of story where if the options are sort of, if there's a clear cut right answer in this situation, there's one thing you want to do and that's get out. That's it. Then don't worry so much about calculating the answer to that question because it's not going to be hyper efficient. If you do the computational rule, well, it's going to get you the answer and it might not take all that much longer depending on how complex the calculation is. It's just that if you start out having this nice reflex going through, there's really little reason that I would see changing. So you expect natural selection to maintain that kind of system, particularly if it's costly to change. But then make it the case that you know, you're, you know, you're, you're in some pathways building with 14 different elevator shafts that go on different places and you have to figure out how to make your way out. And you sort of could be here, could be this at some point, you know, you want to say, ah, drives might not be so great. It's an odd analogy. I take that analogy back the other way. Help. But. So instead of.
- [72:32]
Moderator
Talking about desires versus tribes, I was expecting you to talk about desires versus habits, where by habits I mean stimulus response associations. Especially because Stereli actually talks about that work by Dickinson in the 1980s where Dickinson did some work in rats which seemed to show that in addition to having actions controlled by similar response associations, rats also had could sometimes control their actions by representations of what outcome the action was going to lead to. Representations of value of that. So yeah, the question is what's the relationship between the distinction between desires and drives and the distinction between a desire based system and a habit based system?
- [73:30]
Armin Schulz
Yeah, good. So partly, maybe this is an. I didn't mean to rule out habit based systems. I've classified as all under driver based systems. So I'm hoping not that that there's certain stuff that you can only say with habits somehow that I can't capture my dry place stuff. Maybe the talking lookout table might be more misleading here, but it's meant to be the same picture. In fact, there's some of the same discussion that he does with the norm based habit stuff, architectures with fittings and so on, and that, you know, I think the discussion there is exactly right. The contrast is right and might be even the conclusions right. The way you get to this conclusion is not quite right. And for example, what Dickinson sort of. Right. Sort of. Here's one. One after this debate, Dickinson says, well, if you want to make sense of how these routes act, you got to assume they represent the environment because otherwise you need to posit 400,000 drives. And that's insanely complex. So it's around the echosis. It's just it's biologically advisable to have that drive. Now when I say is that right and why is it then what's the obvious thing that desires do for you here that drives don't? And I want to say here's something that might be going on there. Sunrise.
- [74:49]
Philosophy Faculty Member
Your friend ever touched on something.
- [74:51]
Armin Schulz
Which I was doing today and that's.
- [74:52]
Philosophy Faculty Member
The concept of stimulus. I'm wondering whether in your model the idea of stimulus is similar for the desire based and drive based.
- [75:01]
Armin Schulz
Well, depending what you mean by similar. So often that's what I'm asking, what do you mean by. So if you understand stimulus and response in this classic sense, then I want to associate this with the drive based model. But if you mean by stimulus. So if you understand stimulus as the half of the stimulus and response package, then that's to me the drive based pack. So if you understand by stimulus just. Just something like a means of acquiring information about the world and that is perfectly consistent with the desire based organism in that I don't. So I don't want to because I don't want to make assumptions about whether the organism has beliefs or not represent its environment. So if a stimulus, you know, if you allow stimulus to be representational then you can stimulate an organism. And if that doesn't have to trigger a certain response response that it's just certain environment that you know certain information that it requires then that's perfectly consistent with it. So yeah, depending what you mean by citizenship in this debate, that's nothing.
- [76:08]
Evolutionary Biologist
It seems that we are speaking like about an evolution that it has of an aim kind of thing. And I think evolution.
- [76:21]
Armin Schulz
We have different.
- [76:22]
Evolutionary Biologist
Types of evolutionary pressures and evolutionary system even like genetic cliff where like normally the most efficient doesn't necessarily is the one that's the most successful. And it seems that in the. In the model that you're forcing also you are not incorporating heterogeneity. Like when you get like all the. All these decision making are a big network and normal evolution when the things are more heterogeneous when you have more possibilities of reacting to different and taking the decision through different systems is better. So I was wondering if you like not necessarily your OID human base or silo based. I think it's a big network and the genes that are incorporating in decision making are extremely difficult to measure. So I don't know how are you measuring this heterogeneity of decision making process.
- [77:30]
Armin Schulz
So far. So one thing of what you said so I completely agree and if I wasn't super clean until I redo a process. I do want to suggest that you don't. You're not either dry faced or desired. It's likely that you're both. And that's because it's just not. If you assume an organism starts out because all organisms start out simply and we say a simple organism is the dry faced organism as the ancestral status. Even some that doesn't dissolve some right for behavior one gives into a model. Maybe we'll start out. Let's assume that the ancestral state was the drive phase state. Then why would you move away from drive to desire as well it might not be you. One might not switch over completely from one to the other in the sense that all your decisions are switched over. You switch over in some aspects to it. You rely on some behavioral rules in some contexts but not others. You make decisions about where to go and find both sets using representations but at what to do when you're faced with a line making a different line. That's right. I want to agree with this. I think that's right. I want to agree with this. Based on Some sort of basic general principle of evolution favoring different solution. I don't know, but I want to agree with the conclusion. This stuff about epigenetic drift and Zone nine, I completely agree. There are other things other than natural selection. And that's why I try to say. I don't want to say this is what must have happened with the xyz. I wanted to say here's one thing that speaks of the evolution of the xylee states in some organisms, which is that they're fitter. Absolutely. If the population is small, if the fitness difference is not very large because it's a small, you know, they don't have that many states to distinguish, well then you would expect, you might well expect desires not to evolve or you might be diagnostic. So I want to agree. That's exactly right.
- [79:24]
Philosophy Student
On the distinction between drives and desires, as you mentioned, a drive could be the encoding of a rule in some sense within the architecture which does not involve representation.
- [79:41]
Armin Schulz
It has to be right, because at some point representation has to give way, otherwise we're ending up with this. So bedrock of unexplained representation. Yeah. So this whole thing is meant to rest on just on a neural network.
- [79:56]
Philosophy Student
Which is on drives very broadly.
- [79:59]
Armin Schulz
It's just, you know, action potentials of various kinds of.
- [80:05]
Philosophy Student
But the distinction still between this higher level or the level of desire and drive, which. The distinction then is it sort of hypothetical whether it's conscious or not. But as you said, it doesn't have to be conscious. But is it in a sense, or at least analogically, a sort of playing out of the rule before you actually play it out through a representation in your.
- [80:38]
Armin Schulz
On my other alternative, you're not playing out the rule in any way at all completely. So if you're drive based, you're just. There's a certain representation of the world and that triggers a certain response. This response is mediated by this intermediate step of what's the wrong. Now, this intermediate stuff is then encoded obviously in the network, which in some broad sense is not based on representation. So it can't be. But it's still one way to think about it is you. This is such a bad metaphor. A computer hardware, this computer chip, you can use in different ways to build different systems on top of. And at the end of the day, it's 0 and ones.
- [81:26]
Philosophy Student
At the end of the day, that's.
- [81:27]
Armin Schulz
Just what it is. But you can use different combinations of 0 and ones and hook them up in different ways to build different systems. You can capture this with sort of these Funky wiring diagrams about what sort of rule. This thing called.
- [81:43]
Philosophy Student
What'S happening then still in the desire is that you've built up ahead of your zenoism ones as a sort of model, if you like, a model of what is out, what is out in reality. And you're sort of engaging with that model in a desire.
- [82:01]
Armin Schulz
You're building more of what the world is like as and more of sort of what you ought to do to the world. What the rule is, what the thing to do is, what the world is like. That's the belief stuff, which you might not do.
- [82:13]
Philosophy Student
I see, I see. You're building a model of what you ought to do.
- [82:18]
Armin Schulz
The thing to do right now is this. And I consult that. I might also consult some sort of model of what I believe the world to be. I represent the world. That might be true. That's the belief component. And maybe that comes with desired component, maybe does not. But I want to leave that open. I have a story about that too. Looks quite like it. Does that help? Yeah. Yes, it's interesting. Yeah.
- [82:48]
Evolutionary Biologist
So we have a follow up question on that.
- [82:50]
Armin Schulz
So in a way your desire could be just to follow your drives then becomes a rule. In a way your drives are simply a subset. Each desire is a rule and the rule is following. So here's what you could do. If I understand correctly, you could have. Your rule is. And then you build. You know, the thing to do is if you're in a situation as one do this if you're in situation. So then you build this as a rule, your drive based architecture as into the rule in textiles. It would never be observable. You would never empirically know. Yeah, and that would be surprising if that were an efficient way to make a decision. Because then it seems. Because now you just. Right, you're just mimicking exactly the right. It's not clear why you're not. In what sense. You're still sort of. The kind of computation you're doing through this, relying on this kind of rule is quite minimal. So ideally you'd find sort of. You want to. It's like. It's a bit like you want to find the explicit function that you're relying on. It's not just some sort of implicit function or formula of this. Like it gives you a linear nonlinear differential equation. I'd like you to write down the exact proper equation for the variable at any point in time. Not just sort of. Well, there's a unique solution and I can approximate it somehow in terms of observing the difference between whether an Organ is what was desired or drives. That's almost impossible in case that's my problem. Yeah, that kind of would be. Exactly. Yeah, absolutely. I mean, the way I would expect you made this distinction is something like, look, you can. Here's. If I want to understand how this organism behaves and I can posit three different drives, and then I basically capture this behavior really well. And you know, at that point you start to think, oh, I could embody this rule. And then the gains from. From relying on this rule, that sort of, you know, this rule looks. Looks like it's basically just a redescription of the three drives in something put differently. If cognitive theologists find it really useful to use three drives to describe this organism, that's fine. To the extent that cognitive physiologists start saying, maybe like Dickinson, this isn't the way to go, we ought to really much better capture this behavior if we say it relies on this rule. Certain classic mistakes you would make if you have. If you miscalculate sometimes, that would explain patterns of mistakes better. That would be a good reason to say desires versus diet.
- [85:26]
Moderator
I think we have time for one more question.
- [85:30]
Armin Schulz
Make it yours, maybe.
- [85:31]
Moderator
My mind's been asked in various forms.
- [85:33]
Armin Schulz
That's good. Okay. Well.
- [85:41]
Philosophy Faculty Member
Both desires and beliefs require tokenism, require representation. It seems odd why you and Ceremony want therefore. Therefore, it seems odd why you and Sorelli both want independent descriptions of the evolution of the two traits.
- [86:02]
Armin Schulz
Because in you just want to evolve representation. Yes, because.
- [86:10]
Philosophy Faculty Member
An organism won't. An organism that can have beliefs. Has to have some desires. And similarly, an organism that has desires.
- [86:25]
Armin Schulz
Has to have some beliefs. They don't have to.
- [86:28]
Philosophy Faculty Member
They don't have to direct correspond.
- [86:30]
Armin Schulz
Yeah, I think surrounding as a conceptual truth, he would deny it. And so I would agree with this. Right. So you could imagine an organism that has beliefs that sort of forms a picture of representation of the world. But then sort of every representation is just triggering one particular response. It doesn't represent a rule of what they ought to do. It just says it sort of comes up with, I believe it's raining our boo as wanting. I believe that it's knowing. It does the other thing. It doesn't go, okay, I believe it's raining. And so, you know, well, here's the rule. The thing to do is. And then you have some calculation and you plug in your belief. Well, that would be belief and desire driven. It goes also the other way around. You could have some rule that you're relying on, but the rule doesn't belong on beliefs, but it relies on certain sort of perceptual states. Certain stranger. If you. If you face this sort of pattern of colors and shapes, you know, or here's the rule, and the rule says manipulate patterns of retinal impressions in this way, or, you know, something like this, then that's desire based without being belief based. So it goes different ways. Now, do I need to. Durani really thinks they must be different. And I'm not sure that I want to go to the. I think there's probably good reason to think that if one evolves, the other evolves too. But conceptually, I think there's reason to think that they're different.
- [87:55]
Moderator
Thank you.
- [87:56]
Armin Schulz
Thank you.