[Music]

Yeah. So I mean this is an important

idea in the book that you've got this

potentially infinite regress of people

thinking well Steven knows that I know

that Steven knows that I know that

Steven knows that I know forever and

that's it seems implausible that that

that is how people are actually thinking

even though that's the effect. Um but

you you suggest that there are certain

acts for example speech acts that just

you know it's it the bomb has been

detonated everybody can see it. So so so

tell us a bit about how different ways

in which you suggest that common

knowledge can be generated that don't

require

the departmental head tapping glasses

and almost reasoning. Well, yes, because

the the idea that the idea is that

common knowledge consists of an infinite

uh set of embedded propositions about

mental states, that coordination

requires common knowledge, and that

social relationships are coordination

games. At first, that might sound

completely implausible because common

knowledge, according to the definition,

would seem to be require superhuman

mental stamina. Your head starts to spin

after two layers of I knows that she

knows, let alone an infinite number

which couldn't even fit into a finite

head anyway. But common knowledge can be

granted at a stroke by a an an event

that is public that's conspicuous that's

that's out there that's self-evident

that you witness at the same time that

you witness other people witnessing it.

that can uh then the common knowledge is

implicit in that situation and we often

refer to it. It seems like a recendite

concept. I suggest that we we're all

aware of it. We just don't use the

language of common knowledge. We use the

metaphor of a conspicuous sound or

object. And you use the you just use the

metaphor of a bomb going off. Um that's

a great example of something that you

can't help but notice when it happens.

Therefore, if a bomb goes off and there

are several people witnessing it, it's

not just that people know the bomb went

off. They know that everyone else knows

that the bomb go went off. And to site

another Seinfeld episode just to show

how we are creative in coming up with

metaphors for the concept of common

knowledge, uh even if we don't reel it

out with all the embedded thoughts is a

case where again George Castanza is the

butt of the humor. He's been dating a

woman for a while and he's uh musing

with Jerry and Elaine whether it's time

to uh go to the next level and uh say to

her I love you. And Jerry says, "Oh,

that's a a big move, Georgie boy. Uh how

confident are you that you'll get the I

love you return?" Says, "50."

Jerry says, "Because if you say I love

you and you don't get the I love you

return, that's a big matzah ball hanging

out there." Uh, a big matabal hanging

out there is another metaphor for a

conspicuous object that can't help but

generate common knowledge. In this case,

a romantic relationship

uh like a friendship, like a sexual

relationship exists as common knowledge.

You're uh if two people are in love,

each one knows that the first one knows

that the other one knows. uh ratified by

in in most cases by the declaration I

love you followed by the refrain I love

you.

>> Yeah. Um you've done excuse me you've

done and you also describe other people

doing uh experiments performing

experiments

on

how common knowledge is generated.

Uh there's the the the burger bun and

the hot dog the hot dog bun and the hot

dog experiment. Um you also to describe

other researchers

these various um scenarios in which

there's an ice cream truck and the ice

cream truck starts moving around and I I

found it fascinating the different ages

at which people are able to think about

what other people know or what other

people don't know. So talk about talk

about that. Yes, there's a famous body

of research probably known to every

psychology student uh on u mentalizing

or theory of mind where in this case the

theory is not the scientist's theory,

it's the intuitive theory that we all

have of what's going on in other

people's heads. Um and to be a competent

social being, you have to

try to guess what someone else is

thinking or feeling. We don't treat each

other like like robots. um but we get

inside their others others heads. So

that's sometimes called mentalizing. And

in the book I talk about recursive

mentalizing which is getting inside the

head of someone getting inside your head

or getting inside a third party's uh

head uh reading the mind of a mind

readader. The famous experiment done by

uh um Joseph Perner and Hans Vimmer uh

involves false belief that being the

ultimate test of the ability to

differentiate what you know and what

someone else's know someone else knows.

Um the you act out with a three-year-old

child. You put a marble in a box and

then the uh Sally uh sorry, Sally puts

the marble in the box. Um leaves the

room. Then the experimentter puts it in

a basket. Uh Sally comes back. Where

does Sally think the the uh the marble

is? A three-year-old will say in the

box. Sorry. A three-year-old will say in

the basket. The incorrect answer because

Sally, not having witnessed the

transfer, has no way of knowing it. The

child knows it. uh and spreads his

knowledge on everyone else. Doesn't have

a separate record for what she knows and

what I know. Then later they did a um

kind of a second order version of that

experiment.

>> I'm sorry. And at what age do

>> Oh, I'm sorry. At four years old, they

get it right. I I forgot to mention

that.

>> They have a theory of someone else's

thoughts.

>> That's the That's the simplest

explanation. Okay.

>> Yeah. Um, and you can do it to the next

level where um Sally um wants to get uh

ice cream. The is told the ice cream

truck is at the church. The um however,

when she goes home to get the money, the

ice cream truck drives away. Where will

Sally go to buy ice cream? And just like

the the marble in the box, then you take

it to the next level. Um, while she went

home, while the ice cream truck went

away, someone told her the ice cream

truck went away.

Um, but uh, her friend John just saw her

leave. Um, where does John think that

Sally will go buy ice cream? If they

can't figure out that John J John's

knowledge of Sally's knowledge is

incomplete, they'll guess that he that

she will go to the church mistakenly

whereas in sorry that she will fail to

go to the church. sorry, she'll go to

the church mistakenly. Um, if the child

projects the knowledge that we attribute

to Sally to John, they'll think that

John um correctly knows that she will go

to the say to the park wherever the ice

cream truck went. So, it's a thought

experiment about um false uh belief

about someone else's false belief taking

it to the second order. Kids don't get

that until they're seven or eight. Um,

>> so, so at three they they don't they

don't get it at all.

>> They don't have a theory of of what

someone else might be thinking. At four

>> they do start to form thoughts about

what someone else might be thinking,

>> but they it starts to break down when

they have to think about what someone

else is thinking about what that person

is thinking. Yeah.

>> So they think

what do they know about what Johnny

knows about what Sally knows. So, it's

not just what they think about what

Sally knows, but you insert yet another

mind in in that um representation. And

that flumxes the four-year-olds who are

okay at figuring out what Sally knows

that they don't know, but what Johnny

knows that Johnny doesn't know that

Sally knows that they know. There it

starts to break down.

>> Yeah.

>> And but eight-year-olds can do can do

that.

>> Eight-year-olds can do it. Now all of

course um I had to struggle to

reconstruct it and I suspect some of you

are are

>> struggling. Everyone's following it

fine. Don't worry. Everyone's we've all

got it.

>> But this is it is mental work and

sometimes as with any kind of uh effort

it can be pleasurable. That's when we

have um murder mysteries, sometimes

comedies of manners, mistaken identity

plots where an author will kind of

stretch our ability to keep track of who

knows what about who knows what.

>> Yes. I ju I I just saw a uh play uh the

comedy about spies. And one of the one

of the setups is they're in a hotel and

there are various different spies

bugging other spies. Well, at least they

think they're bugging other spies. using

a radio to bug the other spies, but

somebody has his radio doesn't work. So,

he's gone to another room and taken

their radio and their radio is now in

his room. So, they're all listening to

and he's also he's an actor and he's

he's um auditioning for the part of

James Bond. So, so he so the comedy is

is trying to track all of these people

formulating all of these false beliefs

about all of these other people. And as

a member of the audience, you know more

than than the protagonists do, but you

you still don't know everything because

there are there are various twists, but

there's a lot of as you say, there's a

lot of pleasure to be derived from from

sort of tracking these mental states. So

So we we were talking about at what age

children form a theory of mind and then

a meta theory of mind. You've got two

levels or three levels. So what can

adults do? How many levels can adults

>> I don't know about comfortably. How many

adults can how many levels can adults

cope with?

>> Generally three or four. But it much

depends on how familiar they are with

the situation. Because uh another common

place of cog cognitive psychology is

that when we encounter a combination of

ideas repeatedly, we can uh put it into

a mental assembly. The technical term is

a chunk. Um I think there are various

equivalents in computer science like sub

routines, memoization, caching uh where

a complex data structure can be referred

to by a sing single symbol. Um so there

can be human scenarios like a scandal or

an affair where they're actually it's

defined by multiple levels of knowledge

about knowledge but they're familiar

enough that they become a cognitive

chunk and we can handle uh multiple

levels and there there even words like

say to humor someone to humor someone

means to you want them to think that you

are taking their thoughts seriously. ly

even though you know you aren't aren't

believing them. And so you can stretch

that limit um that ordinarily seems

comically difficult. If you're throwing

new actors and new relationships at

someone all at once, it seems

impossible. But if they're familiar

combinations, that's often how we deal

with them. Still, there are comedies

that poke fun at our inability or at

least our difficulty at keeping track at

multiple levels.

>> I I have the New Yorker cartoon here. Of

course, I care about how you imagined. I

thought you perceived I wanted you to

feel.

>> Yeah, that's that's funny if it's only

three, right? Because if it's only

three, you can you can follow it.

>> That Well, that exceed Yes, that's

right. If it's only three, then then

then generally you can follow it.

There's also the the scene that I I wish

I had included in the book I was

familiar with, but it did I didn't make

the connection. In the Princess Bride,

there's the battle of wits scene where

Wesley has uh poisoned one of two uh

goblets and Vitini has to figure out

which one he can drink from. And the way

he figures it out is because he fancies

himself as a genius. He said, "Well, you

want me to think that I'm putting it in

the far goblet, but I know since I know

that, I'm going to pick the near

goblet." But you know that I know that

you want me to think it's in the far

goblet. So, you're going to put it in

the far goblet after all. This goes on

for several iterations.

>> Yeah. And and and part of the joke is

that both of them are planning a double

cross that is makes the whole kind of

reasoning about other people's mental

states completely irrelevant. Fini

basically goes, "Oh, look over there."

Then swaps the goblets. And and

actually, yes, I won't spoil it. For

those of you who haven't seen Princess

Bride, you have to see it. Um,

so I wanted to just um can can I can I

get a volunteer from the front row, not

to come up, but just to to answer a

question. Would you would you mind

answering a question for me?

Thank

Okay, you could answer a question for

me. So here is here is your assignment.

Stephen, this is an assignment for you

also. Will don't say anything yet. So um

it becomes apparent that you need to

meet Steven Pinker in London tomorrow.

However, he doesn't have your phone

number. You don't have his phone number.

There's there is no way of of of

coordinating. So where do you go to meet

him? And when

>> um to the tower of the big Ben the

>> whatever is most conspicuous.

>> Yeah. Big Ben. When?

>> Uh midday. I don't know.

>> Steven.

>> Yes. I think I'd probably I'd return

here at the starting time of this event.

Uh

>> there you go. It's tricky because I

would have gone for that. But that also

makes perfect sense. Part of the problem

with these this is called a Thank you

very much by the way. Um this is a

shelling Thomas Shelling the economist

Thomas Shelling um first kind of started

talking about these games. It's called a

coordination game and uh part of the

problem is you're trying to think about

what is most salient and what context

matters. So you were thinking Big Ben is

the most famous landmark in London. It

has a clock and the most obvious time to

meet anybody is midday. It's not crazy.

You were thinking we were both at this

meeting. This is the only place that you

know we have met and the focal time

would be 7:30 the starting time of the

meeting. Also not crazy. Um it's turns

out to be quite a difficult game right

because there's more than one focal

point. But but what do we learn by by

reasoning about this sort of thing?

>> So this is a case where in the absence

of common knowledge we have to use this

next best thing which is common

salience. that is something that uh it

doesn't just pop out but that you can

guess would pop out to other people the

person you're trying to coordinate with.

In Shelling's example, it was the clock

in Grand Central Station at noon in in

New York um at at a time when a lot of

people use Grand Central Station. So a

common salience focal point sometimes

called a shelling point in his honor is

a solution to a coordination game. Now

the thing about a coordination game is

that it has two equilibria. That is

there are two situations such that if

you're in it, if the other if you make a

choice, the other person makes a choice,

there's no reason that either of you

should change your mind. But there's

another option. And if you made that

choice and the other person made uh that

choice, neither of you would change your

mind either. Should you drive on the

right? Should you drive on the left?

Either one is good. It's very important

not to avoid the cell where you drive on

the right, he drives on the left or vice

versa. And um so there's nothing in game

theory and mathematics in rational actor

theory that determines which of those

two cells you should try to get yourself

into. Instead, what shelling consist

says is this is where game theory leaves

off and psychology begins. that is your

theory of mind that is your guess about

what's going on in the other person's

head and vice versa can commend a

solution uh to you um there's no rhyme

or reason to it other than the fact that

it commends itself as a solution there

are many examples in in everyday life

and in actually in highly consequential

circumstances in which people are in a

coordination game and seeking a focal

point so they can get in a one of them

mutually beneficial equilibria. He gives

the example of um negotiation

bargaining. So let's say there is a car

salesman. Let's say there is a buyer.

Let's say there's a range of prices

where they can come to an agreement.

Namely the car salesman makes a profit.

The uh buyer is willing to pay that

amount of money for the enjoyment of

having the car. The question is where in

the range should they settle? any point

within the range is a an equilibrium but

nothing in the statement of the problem

can dictates where they should end up.

It is important that they end up

somewhere because if they walk away it's

uh both of them are worse off. So

shelling suggests that's why often in

negotiations people will um offer to say

split the difference uh when they when

they get near the end. Now splitting the

difference isn't inherently fair. Maybe

one of them started off with a a high

ball offer. Um maybe even anticipating

that they would split the difference and

gets and pulls it in his favor or vice

versa. Or he said they might settle on a

round number. He said the salesman who

announces that his rock bottom price for

the car is £30,0756

is fairly pleading to be relieved of

£7.56.

Namely, obviously the customer say, "Oh,

come on. Let's call it 30,000 even. Now

30,000 is not the fair price, the

logical price, the only price for the

car, but simply as a round number, it

pops up and so it's a possible focal

point for them to uh settle on