Re-release: Rational Irrationality

Speaker A: 00:00:00

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 Peter Coghill of Aleph Insights and this week we're discussing the philosopher Derek Parfit and rational irrationality. So Nick can you lead us in please?

Speaker B: 00:00:32

Yeah so the reason we're talking about this now is that Derek Parfit died a few weeks ago in January and he wrote a lot of very interesting things, did a lot of interesting work on some of the things which we're interested in, in Aleph Insights particularly about decision making and what constitutes kind of good decision making. And he wrote in particular one of his most influential books was called Reasons and Persons and it sort of goes through, it's a long series of short chapters each of which takes a specific sort of claim or runs with a particular thought experiment and some of them are really kind of bonkers, these thought experiments, but all of them are designed very rigorously to test various theories about you know how we ought to behave, what kinds of rules we ought to follow, what it means to say that we're the same person as we were yesterday which turns out to be very closely related to the issue of sort of good decision making. A lot of his thought experiments are about personal identity, you know it's a very famous thought experiment involving a transporter where you get destroyed on earth and exactly reconstituted on mars and you know the question of whether or not you're the same person which turns out again to be very important to thinking about you know when we make decisions about our future selves, how much should we care about our future selves, how much are they the same as us. Anyway, very influential philosopher, even he looked the part, he had a sort of big shock of crazy looking hair but he was extremely rigorous, you know, and very influential. But we want to talk about one specific thing he looked at which was this issue of, so some of the things he looked at were to do with sort of what are called dynamic decision problems, so where you have to make decisions which affect you over time, so usually decisions where you're looking at a situation right now and you know the effect will be immediate. They're usually quite straightforward but they get much more complicated when you start thinking about potential future consequences. So in particular...

Speaker A: 00:02:42

Give us an example.

Speaker B: 00:02:43

Yeah, so he talked about an example which had also been raised by Schelling who was a game theorist, where being rational, so rationality in philosophy and economics terms simply means, well it means that you try and make your beliefs true, right, so you want to believe true things about the world and you want to take actions which get you what you want to achieve, so you want to take actions that will deliver the things you want. Now that seems like a very uncontroversial claim, right, so the right thing to do is to do things that will deliver what you want. So that's a claim that says being rational is good, right, because being rational, following the rule that you should always try and do things that get you what you want is better. But he talked about a potential counterexample where a robber comes into your house and threatens you, threatens that he's going to murder your family unless you give him the... Keys to the safe? Well I was going to say the combination, so threatens to murder your family unless you give him the combination to the safe and obviously the right thing to do here is to give him the combination to the safe, that's in the short term what you want, you'd rather, and he knows that you're going to respond to that threat because you'd rather lose the money in the safe but keep your family safe. But Parfit said well wouldn't it be better to have a kind of irrationality potion where you could take the potion and you no longer cared, you know, you could just sort of say well you can threaten me all you like, I don't really, I'm not going to give you the combination of the safe, you can go and kill my family, I don't care, and if the robber knows that then he knows he can't threaten you, he can't rely on your rationality to get what he wants and you end up better off.

Speaker A: 00:04:29

Yeah I mean doesn't something like this happen in, actually happen in the natural world where What's His Face talks about it in...

Speaker B: 00:04:41

Ah What's His Face, the famous botanist, Sir Nigel What's His Face, yeah.

Speaker A: 00:04:45

I think it's Richard Dawkins who wrote The Selfish Gene, right, and in that I think one of he talks about is it's actually quite good to be a balance in an animal of something that may or may not attack. You don't want to be the kind of animal that will always attack and the one that will never attack, you want to be one where no one's ever quite sure what you're going to do. I'm not sure if I've...

Speaker B: 00:05:04

That is exactly bang on, yeah so it turns out that a lot of the tools economists develop to look at human decisions and human interactions can be applied really powerfully to evolution and game theory has got lots of explanations for apparently weird behaviour in the animal world and particularly self-sacrifice. So when animals fight to the death, the point is that it's better to be the kind of animal who will fight to the death even though at the point at which you're going to be threatened it would be better to run away. Yeah. So and the reason it'd be better is because other animals won't... Will stay away. Yeah well exactly so and that's and game theory kind of proves that there's an optimal equilibrium where you sort of sometimes fight to the death and sometimes don't.

Speaker A: 00:05:53

But also you could maybe there's two different types of rationality here because one where you're purposefully being irrational because you can because it makes sense to run away but actually it's a nice strategy to purposefully be irrational but sometimes you could just be irrational without reason as it were.

Speaker B: 00:06:12

Yeah well it's hard you can't actually be deliberately irrational that's one of the sort of weird things you can't choose to be irrational unless you are somehow completely preventing yourself you have to impose a constraint on yourself in order to stop being rational that's the weird thing you can't choose to be irrational if you could actually at some point choose to stop being irrational because then you would just take the short-term benefit you would give the robot the combination.

Speaker A: 00:06:44

So I mean I presume of everything we've been talking about five minutes now of everything I mean that's a platform because I doubt we've said anything there which is groundbreaking it's more of a summary right so let's sort of try and develop this a little bit and Peter what have you got to say at this point?

Speaker C: 00:07:00

Well I mean I'm not a I'm an engineer not much of a philosopher so I've really got questions for Nick to explore this a bit more. So presumably it changes if you have this the game with the robber and the safe is repeated frequently presumably there's a more of a repeated game changes the nature of the balance.

Speaker B: 00:07:25

Well yeah I mean so one of the biggest challenges I think to the kind of old conventional idea that if we all behaved in our self-interest we'd be better off you know the kind of traditional view that follow everyone following their self-interest will all gradually push us towards a world where we're all better off right that's kind of classical economics assumption completely undermined by the prisoner's dilemma I think we've discussed that before the prisoner's dilemma yeah but it but it's a situation where two people following their self-interest make they end up all worse off and there's loads of examples of that you know the kind of examples from real world stuff like over fishing where you know it makes each person has an incentive to catch as much fish as they can but if everyone does that then there's no fish left and you know and all the fishermen go out of business that's a tragedy of the commons yeah exactly tragedy of the commons is a kind of multiplayer version of the prisoner's dilemma so now when you put when you have a repeated prisoner's dilemma against the same player you open up other kinds of a solution where you can actually end up cooperating so again it's one of these things where when you have a dynamic element when you have repeated games when you add time to the thing you end up with lots of other potential solutions one of which is cooperating you know and then basing what you did on what the person did last time I'm not sure exactly how that applies to

Speaker C: 00:08:52

the robber situation how does it sort of happen then another question I have is how does it how does it apply if at all to negotiation practice and theory so assuming that there's some opportunity for dialogue between the robber and and the the the robbie is there a is there a is there a because they they're majoring on what they're they're they're talking about what they really want so that you're the the the bit hasn't been robbed is saying oh I'm I don't want my family killed and the robber saying I want the money in the safe but if there were if there was other scope for negotiation does that does this does that link to this yeah I mean that's negotiations

Speaker B: 00:09:33

where like it's probably the place where real life game theory happens the most so people

Speaker C: 00:09:38

actually do use game theory because I'm thinking of where in negotiations you would concentrate on the minutiae of things and not talk about the big issues to to to conflate and confuse your opponent

Speaker B: 00:09:51

yeah actually that's I mean and that's one of the findings from from sort of the game theory analysis of negotiation is what you should do is pretend to care about the things you don't care about and pretend not to care about the things that you really do care about because then you then you can make a what in your from your perspective is a very small sacrifice look like a big sacrifice and extract concessions from it but yeah I mean so so negotiations is another classic example where it would be better to seem irrational there's a there's other analysis of kind of games like where you have a trade union for example so trade union wants to be able to make a sacrifice it wants it's saying look we're going to actually impose a cost on ourself to try and gain some kind of advantage in negotiations they want to seem irrational they want to look like they would be willing to go on strike you know for ages and ages to cost the company lots of money now if they're being really rational they just wouldn't you know in some circumstances they wouldn't and you can see that a lot international negotiations people want to look like they're being irrational they want to look like they're going to bullheadedly ask for something and if they don't get it you know they're going to just pull the trigger and yeah I mean

Speaker A: 00:11:07

that's that's something that so can we turn this to I mean something that might be interesting is just apply this to the current political international political environment can we hang this off something any current negotiations or any current international issues going on at the

Speaker B: 00:11:23

moment yeah I think we can probably think of one we can because uh yeah I mean uh well there's obviously is Brexit where we we you know there's quite a lot of debate wasn't there about um well why don't we tell everyone what our negotiating stance is you know parliament moaning a bit about wanting to see what the plan was I mean you definitely don't do that and um um things like uh the the you know the EU nationals being in the UK um a lot of people you know well-meaning people saying we should take that off the table you know we should just say yes it's fine if they stay here well that's terrible negotiating strategy because it's something that we do have control over which the EU will care a lot about so we definitely you know but actually um it's also interesting think about uh the personality types of um of leaders like someone like Putin or Trump um I think people have drawn an analogy between them by the fact that they are very they're not utilitarians they don't they're not going to sit there and go hmm well the best thing to do here would be to back down they're very uh sort of victory driven they and they what they you want to look like you care about winning at all costs uh if you're in an international negotiation monopolizing the

Speaker A: 00:12:34

irrational card yeah that's the interesting thing well this reminds me it makes me think a lot of the two games of poker and chess um because something like poker is you definitely want people just to go what the hell is this guy doing and you know um and that's why it can be difficult sometimes playing against a um a fairly novice uh poker player because they tend to sort of be overconfident and they tend to go all in too much and you never know is it just because this guy's really got a great hand or are they a total idiot um and similarly in chess I'm I've when I used to play chess quite a lot I actually never became a particularly good player but I used to find it and it's quite well documented you know playing a quite a new chess player can be quite difficult because they do some very odd things responding yeah there's a fundamental difference between

Speaker C: 00:13:22

those games because chess you have you have all the information whereas in poker there's

Speaker B: 00:13:27

there's obscured information sort of although I mean there is the the problem that actually in practice you you don't really interest because it would be computationally unfeasible to to to compute the consequences of any given move so there is there is uncertainty there um different

Speaker A: 00:13:41

kind of uncertainty yeah yeah but um sorry I'm probably sort of slightly um taking a soft track from my question about the international environment at the moment so we talked a little bit about brexit we talked about trump and putin uh peter anything you want to bring

Speaker C: 00:13:55

in at this point uh well I mean maybe we maybe we could explore towards uh what's the optimal strategy so is it potentially um an awesome strategy is to be able to appear to be irrational when it best suits you so being able to sort of message in the right way that you're irrational whereas your your actual decision making process is still entirely rational yeah I mean we you want

Speaker B: 00:14:19

the so I just take it in as sorry in a slightly um engineering e-direction not necessarily for Peter's direction uh for Peter's benefit but because it's interesting is I don't know if you know about the Russian perimeter system do you know but I'd like to well it's a system uh called perimeter and um uh sometimes called the a dead hand uh firing system um which is designed to look for uh whether or not there's been a nuclear war um the principle being that if the all the decision makers in Moscow were knocked out uh they would still want to be able to retaliate now this is so this was this is a similar uh this is actually is it like the doctor strange love it's exactly it yeah so that and this is actually true so the Russians did have this system and I'm not sure it's known whether or not it's still uh active but um where you know they would there would be kind of automatic sensors looking for radiation um you know seismic activity looking for signs of a nuclear war and if they detected it with no humans in the firing chain at all they would just launch a massive retaliation um now that's brilliant from a from a sort of strategic point of view because there's you you've completely removed a rational response because you know most people think the rational response to being attacked with nuclear weapons is just to suck it up and go well we've lost you know there's no point in making the world worse by retaliating but what you really want is the opposite of that you want you want to be able to retaliate with 100 percent uh um you know with a 100 probability because then the other side

Speaker C: 00:15:51

won't start a nuclear war at all yeah yeah so does it undermine or strengthen the sort of mutually

Speaker B: 00:15:56

assured destruction balance that you get well the downside is that what if it goes wrong you have to put quite a lot of faith in the system not not going wrong and as we all know from lots of examples throughout sort of nuclear diplomacy there have been quite a few near misses we don't know how near because you know we saw them when they got stopped um but uh so that that's you know as usual with these real life things you know you the all of the kind of game theory does slightly go out the window when you're dealing with real life systems because of the uncertainty about how they'll actually perform where do we go from here that well i do want this is i'm trying to i'm trying to get a piece to talk about uh systems but particularly like systems where you take human decision making out so at the moment for example we have we have drones that um always have to have a human deciding whether to fire right but um it would be a lot scarier wouldn't it if they were uh designed to fire automatically uh when they saw certain you know examples of um of uh sort of terrorist type activity or whatever and i mean the question of the question of how much you design for humans and you know autonomous vehicles it's the same thing if you if you design autonomous vehicles so that they always swerve when they see a person in the road then people will just blithely wander about in the road weren't they in perfect certainty they won't get

Speaker C: 00:17:14

run over you get yeah you get a sort of safety system uh effect where uh seat belts make people drive faster um and more recklessly yeah no well i think i think the issue is it's fair to say that it's going to be the human in the loop where all the mistakes will come from however um they all they they they they are also the place that has as in the ability to accommodate new information that's not in this predefined pre-programmed way of working uh that will cause a nuclear fire chain to to to to to activate but so it i suppose it's there's some sort of balance required as to how much risk of um of getting it wrong when you want it to want something to happen versus it going wrong because the computer is just it's computer says fire computer says fire computer says yes uh when when you probably didn't want you there's an additional bit of information that can't be accommodated in that decision um that a human would say the human on the firing chain perhaps in the submarine would just not fire because he knows something that the computer

Speaker B: 00:18:19

can't take account of yeah i mean the uk's policy if you can call it that is to not have is to not tell anyone what our policy is when it comes to nuclear retaliation um so we we deliberately maintain this strategic ambiguity uh right up to the very um to the very top really and that um the i think one of the things that peter hennessey talks about is the first job of any prime minister is to write down orders for the submarine commanders for the nuclear submarines um you know the submarines carrying nuclear weapons uh about what to do if if you know the uk gets taken out by a nuclear strike um and so no one knows no one ever looks in those envelopes and they're destroyed um at the end of that prime minister's tenure so we never i hope that doesn't end up

Speaker A: 00:19:03

like the the the moment comes and the submarine commander opens it up and goes la la land well

Speaker B: 00:19:10

yes but i think i think well i think right maybe making this up i think harold wilson said that his his uh orders were to just sort of go to australia and find a nice beach somewhere which is a completely sensible i mean you want what you want is for the russians to think that you've written down you know massive retaliation against against moscow uh but actually um actually when the push comes to the shove you you don't want that to be the case because that

Speaker A: 00:19:32

would just make things worse you know um peter i mean we've still got a few minutes left but

Speaker C: 00:19:38

anything to say before i wrap up no no well one thing one thing i just just sort of just to try to explore perhaps more comedically this example so is there the the you've you've got this option of taking this potion to make you genuinely irrational is there not a danger then that the robber might be able to convince you to open the safe anyway yeah i mean actually what you what

Speaker B: 00:20:02

is really about is stopping yourself being able to take actions i mean being unconscious would be perfectly good taking out taking a taking a potion that just knocked you out for two hours would be just as good it's really about cutting off your ability to take options uh and i mean that's that's the that's the paradox really is that sometimes it's better to stop yourself being able to have a choice right which is which is um you know counterintuitive because you think well actually that surely being taking the optimal um the optimal option every point in time is going to be better uh but at the point about this example and one of the things that parfait talked about a lot was that sometimes it isn't uh that really uh throws a spanner in the works actually when we when we think about how we ought to behave it makes it makes the world much more confusing and more interesting and more difficult to manage um but i mean it does but does it does

Speaker C: 00:20:51

that mind does that um thought experiment also assume that uh the the robber has some put some sort of value on human life that he won't kill you just out of revenge or annoyance because you've you've you've you've swung the game in your favor um yeah i think the assumption is that the robber

Speaker B: 00:21:10

would sort of very slightly prefer not to have to be a murderer as well yeah uh but the point is that if you're being rational he doesn't have to be he doesn't have to have to be a murderer because you're going to open the safe right yeah okay understood yeah but actually so actually

Speaker A: 00:21:26

that's interesting because that requires the robber to be rational right and i'm just wondering what

Speaker B: 00:21:31

happens if you get to get an irrational robber and an irrational robby yeah nightmare totally utter breakdown of everything we hold dear maybe the robber ends up walking away with the kids

Speaker C: 00:21:44

and uh yeah and lives happily ever after the other guy locks himself in the safe or something

Speaker B: 00:21:48

yeah yeah that's probably what that's probably what the outcome would be yeah very briefly slightly

Speaker A: 00:21:54

off topic um what do you think are the largest irrationalities that you have in your day-to-day lives because i know i mean one of mine uh is probably i mean there i mean i started trying to think of one and i realized there are many but um one of them was probably uh smoking of you know of and it goes back to what we're saying with i always want to say rick parfait but um with derrick path rick parfait for anyone who doesn't know who was was or is one of the lead guitarists of of of the amazing didn't he die recently yeah yeah well that's why i get confused between the two but of the legendary rock band heavy metal rock band status quo but anyway um yeah and it ties in one of his things about i think what we were saying about identity and future selves is that i used to used to annoy me when i smoked because i knew that was such a bad thing for me and it made me question right how much do i care about my future yeah well i mean that's that's

Speaker B: 00:22:46

a good example where which is the real fraser right is it smoking fraser or fraser who wants to give up smoking and the whole point about trying you know doing using tricks like um throwing away your cigarettes or um you know or or taking uh even just just taking nicotine patches or even telling people you know i'm going to give up smoking um are all ways of constraining your future self because future fraser gasping for a fag uh it's a different person he's got different goals and you your goals right now might be to give up smoking and you've got you're playing a game against future fraser whose goals are to carry on smoking and you basically you're playing this game against yourself and you want to win so you're going to make make life hard for future fraser of course future fraser is going to is going to be really annoyed with past fraser because he wants a fag and he may well try and rig things so that when regretful fraser turns up again uh you know he won't be able to um so yeah i mean they're um it is something that crops up all the time it's a very very common uh situation and there are some some places there are kind of

Speaker C: 00:23:47

there are external systems to help you impose those constraints so a pension scheme is a way of locking money away that you would otherwise you know you might otherwise spend at the end of the month um that you because you know when you sign up to a pension scheme it's a good idea so past you has made this decision for future you to stop you from spending all your money yeah but

Speaker B: 00:24:07

i think my i i'm in terms of the things that we've been talking about i'm i am a real grudge merchant i do bear i bear grudges for years there are particular there are there are certain teachers that i will i'm still out to get somehow and when the chance comes i will um you know so i i mean that's that's completely irrational some sometimes uh you know people who just perhaps slight you a couple of weeks ago bloke barged past me on the bus instead of saying excuse me i i'd happily you know kill that guy uh so kind of um you know annoying uh and that's that's the kind of uh you know reptilian brain uh rational irrationality which which means that you know people don't mess with me unfortunately i don't know that when they're on the bus sure that's

Speaker A: 00:24:53

now he won't make that mistake you need to paint yourself like yellow and black or something so

Speaker C: 00:24:57

yeah yeah exactly peter well i suppose i i mean i like wearing nice leather shoes and off and off will go out wearing a shirt and i know that's really mad i know it really doesn't matter what i look like i could go out in trainers and jogging pants and it wouldn't matter what i looked like by and large but i kind of like to feel like maybe some people think i look smart

Speaker B: 00:25:23

even though that doesn't matter it sounds like that's a rational response to other people's irrational uh desire to make other people wear nice shoes i think i think you know ties we've

Speaker A: 00:25:32

talked about ties a lot but uh i think you should start with a haircut then if you're gonna don't worry about the shoes and i would mean i had a haircut two weeks ago this is this is short it's funny actually i've been thinking about doing something similar to what you've said which is i spend most of my time working from home and i just slob around in any old rubbish and i just i'd like i like the thought of my children as they're growing up and actually once they've grown up thinking back to this uh to their father oh yes he was always so smartly dressed in a nice jacket and a shirt always wore a waistcoat at home right yeah and i think i need to start doing this because um and also for my own sense of self-worth and uh and belief in what i am that i need to be starting to doing that i don't know if that's rational or irrational but um but all i would say is i don't want to end up uh being regretful fraser which i quite like that characterization um we're gonna have to stop there um so gentlemen i enjoyed that um thank you as always to nick hare and peter coghill of aleph insights you've been listening to the cognitive engineering podcast i'm fraser mcgruer thank you and until next time bye bye and thank you

Speaker C: 00:26:35

to Derek Parfitt for his excellent works and thank you to Derek Parfitt