HomeVideos

"We Don’t Mean What We Say" - Steven Pinker Part 1

Now Playing

"We Don’t Mean What We Say" - Steven Pinker Part 1

Transcript

482 segments

0:00

[Music]

0:05

[Applause]

0:08

Well, hello everyone.

0:11

This is very exciting. Um, Steven's book

0:14

is all about uh common knowledge. It's

0:17

called When Everyone Knows That Everyone

0:20

Knows. dot dot dot. all about all kinds

0:24

of things, but including things like how

0:25

people coordinate on making sure they're

0:28

all kind of dressed appropriately in a

0:29

way that matches, uh, which we we

0:32

immediately failed to do. Um,

0:35

I I should say I'm I looking forward to

0:38

this evening, but I'm also uh I I view

0:41

it with some trepidation. The uh the

0:43

first time that Stephen and I were in a

0:46

room together was at a conference about

0:48

um about political violence. I seem to

0:51

remember. And uh Steven was amazing. And

0:55

uh at the end of the conference, there

0:56

was a conference dinner and the host of

0:58

the conference dinner took time to

1:00

explain how unhelpful he had found my

1:02

contribution. Um

1:05

that was the first time we were in a

1:06

room together. The last time we we

1:08

interacted, it was for a BBC radio

1:10

program. Um, Stephen had uh me and uh

1:14

the much missed investor Charlie Munger

1:18

uh talk about cognitive errors that we

1:21

had made. Charlie Munger told a little

1:23

story about a mistake he had made. I I

1:25

told a little story about a mistake I'd

1:27

made at which point Charlie Munger, who

1:29

I think was about 98 at the time, hooted

1:31

with laughter and said, "That is the

1:33

dumbest thing I have ever heard in my

1:35

life. I thought mine was stupid, but

1:37

that you you really must be an idiot."

1:40

So, um, anyway, so what I'm saying is

1:43

third time lucky. Uh, the bar has been

1:45

set low.

1:47

I I expect Steven to be excellent, but

1:49

who knows

1:51

what idiocies I'm going to commit over

1:53

the next uh hour and a and a hour and a

1:57

quarter. Hour and a quarter. Yes. Hour

1:59

and six minutes. Now, uh if I don't keep

2:01

if I keep talking like this, the um the

2:04

format is that I'm going to uh ask

2:07

Stephen questions about his wonderful

2:08

book for about 4550 minutes and then uh

2:13

you are going to send in questions uh

2:15

via this magic iPad in front of me uh

2:18

and I will put your questions uh to

2:20

Stephen and u most importantly at the

2:23

end of the evening uh Stephen in the uh

2:26

foyer out there is going to be signing

2:29

books for I think probably for every

2:31

single one of you. Um, he never gets

2:33

tired. He's indeathicable. He loves

2:34

signing books. So, please, um, make your

2:38

requests as complicated as you like.

2:40

Personalizations, poems, everything. Um,

2:43

he he loves that. So,

2:47

I'm having too much fun already. I

2:50

should probably let Steven say

2:51

something. Stephen. Um the so this is a

2:54

book about many things but in particular

2:56

it's a book about uh what is quite a

2:59

formal concept uh common knowledge

3:03

and I I wanted to start because no one

3:06

can escape now for the next hour and a

3:08

half. Um I wanted to start by exploring

3:10

the formal concept before we uh go

3:12

through all the interesting

3:13

ramifications

3:14

and you have I think the most elegant

3:18

illustration of a famous logic puzzle

3:20

the spinach problem. So may can we start

3:22

with the spinach problem and and um I'm

3:24

sure everyone in the audience will be

3:25

able to solve it.

3:27

>> This has been called the world's hardest

3:29

logic problem and it has uh it goes back

3:32

to the early 1950s. It comes in various

3:35

versions, some of them not safe for work

3:38

by 21st century standards. Sometimes

3:41

called the muddy children problem, the

3:43

barbecue sauce problem. My version is uh

3:46

the spinach and teeth problem. So

3:48

there's an academic dinner, a bunch of

3:51

psychologists

3:53

at a banquet. They serve um halalibet

3:56

with sauteed spinach. And so there's a

3:59

hazard that you might get a a bit of

4:01

spinach in your teeth. It's mortifying

4:03

if it happens to you, but there are no

4:05

mirrors around. Uh everyone is too

4:07

polite to tell anyone else that they

4:09

have spinach in their teeth, and no one

4:11

wants to uh pick their teeth if they're

4:14

in fact clean.

4:16

So, the department chair can't stand it

4:18

any longer. Uh, there are three people

4:20

with spinach in their teeth. She gets up

4:22

and she says, "Uh, at least at least one

4:26

of you has spinach in their teeth. When

4:28

I clink the glass, that would be a good

4:30

time to remove it." So, she clinks the

4:33

glass and no one moves. She clings the

4:36

glass a second time and no one moves.

4:38

She clinks the glass a third time and

4:40

the three diners with spinach in their

4:42

teeth all remove the spinach. The

4:44

question is how do they know it was

4:46

them? That's that that's the problem.

4:50

>> Any anybody want to

4:53

>> I said it was too early for audience

4:55

participation.

4:57

>> There will be I will be asking questions

4:59

later by the way. Um so ex ex how how

5:02

does this puzzle resolve itself?

5:04

>> By the way I could have made it that

5:05

there were um seven diners with spinach

5:08

in their teeth and on the seventh clink

5:10

of the glass everyone cleans their teeth

5:12

by the same logic. speak up.

5:15

>> Oh,

5:18

>> I will do that. Is this better?

5:21

>> Okay. So, um, imagine this simple case

5:25

where there's just one diner with with

5:27

spinach in his teeth. Then the when the

5:31

chairman says, "At least one of you have

5:32

spinach in your teeth," uh, and and

5:35

clinks the glass, everyone looks around.

5:37

Um the person with spinaches teeth sees

5:40

that everyone else's teeth is

5:43

clean knows that he has to be the one.

5:46

So that's easy. Now imagine there are

5:48

two people. The uh again department

5:52

chair clinks the glass. People look

5:55

around. Now at this point um no one can

5:58

know that it's them. They do see uh

6:02

everyone sees um the person with spinach

6:05

in his teeth sees one other person with

6:07

spinach in her teeth. But uh since all

6:10

the department chair said was that at

6:11

least one of you, he can't know that she

6:14

whether she's the only one or uh he also

6:17

has spinach in her teeth. However, she

6:21

did not remove her spinach at the clink

6:23

of the glass. That means it couldn't

6:25

have been that that she must have seen

6:28

someone else with spinach in his teeth

6:31

and so it must be him. So on the second

6:33

clink of the glass both of them clean

6:35

their teeth because she has gone through

6:37

the same uh coitation.

6:40

Now you can see that this logic can be

6:42

extended to any number of people with

6:44

spinach in their teeth after the

6:46

corresponding number of clinks and it

6:49

but it crucially depends on the

6:50

announcement being public. common

6:53

knowledge in the technical sense that

6:55

not only does everyone know it, but

6:57

everyone knows that everyone knows it.

7:00

And it is only the fact that the

7:01

announcement was public that could

7:03

allows you to deduce what the other

7:05

person uh didn't do uh and therefore

7:08

what the other person must have seen. If

7:11

for example the department chair had

7:14

slipped a note to everyone or whispered

7:15

in everyone's ear that at least person

7:18

one person has spinach in their teeth,

7:19

uh it wouldn't work. you couldn't come

7:21

to the same deduction. So it's a a

7:24

puzzle that depends on the concept of

7:27

common knowledge which sounds rather uh

7:30

obstruuse everyone knowing that everyone

7:32

knows that everyone knows something but

7:34

it's one of many examples perhaps the

7:36

the the most fish in which it actually

7:39

plays a role in our uh social lives.

7:43

>> So I I love the explanation of the

7:45

puzzle. I mean, they all these various

7:47

ones with people with blue eyes and with

7:50

black hats. And I mean, I'm I'm addicted

7:52

to logic puzzles. So, I' I've seen

7:54

versions of this puzzle uh over and over

7:56

again over the years. And I and I I

7:58

think this the spinach and the clinking

8:00

of the glass is is perfect. Um but it's

8:04

it's weird, right? This is a weird

8:06

puzzle.

8:07

>> This is a weird No one actually behaves

8:09

like this. No one actually thinks, "Oh,

8:12

everyone knows that. everyone knows that

8:13

everyone that you know no one goes oh

8:15

Steven knows that I know that Steven

8:17

knows that I know that Steven knows that

8:18

I know it's what the dot dot dot on the

8:20

cover of the book is for

8:21

>> which I had to fight for

8:23

>> yeah which well rightly so because

8:25

otherwise otherwise it's just talking

8:27

about one level of mutual knowledge

8:29

that's that's not the same thing at all

8:31

um but this is this is kind of this is

8:34

an odd kind of niche and so economists

8:40

like me who found out that you were

8:42

writing a talk about common knowledge,

8:46

we all have the same reaction, I think,

8:47

which is basically common knowledge is a

8:49

weird thing about kind of knowledge

8:50

partitions and odd logic puzzles and

8:54

strange corners of game theory and it's

8:57

all very cute if you like that sort of

8:59

thing, but it has no wider application

9:01

whatsoever. Um, reading the book, which

9:03

is a wonderful book, which you will all

9:05

enjoy, I realized that there's all the

9:08

applications in the world. So, so what

9:10

was the point at which you you got hold

9:13

of this

9:15

this sort of strange kind of infinite

9:17

regress of mental reasoning and realized

9:19

that there was something much broader

9:21

there. I came to it through my interest

9:24

in language. Um it enters into an

9:28

understanding of language at the

9:29

simplest level at um what language is

9:32

and how it works and even how a word

9:35

works. A word is a convention. Uh, a

9:38

rose by any other name would smell just

9:40

as sweet. But it the sound rose works as

9:44

a way of evoking the concept rose

9:46

because every English speaker tacitly

9:48

knows that that sound goes with that

9:50

meaning. But more important, they know

9:53

that everyone else knows it and everyone

9:55

else knows that everyone else knows it.

9:56

And so you can use it indiscriminately

9:58

throughout the community of English

10:00

speakers and even kids are sensitive to

10:02

that when they learn uh words. if they

10:06

hear a word say from their mother, they

10:08

just assume that's that that's English.

10:10

I can use it with anyone else. I can use

10:12

it with third parties. They don't stop

10:14

to think, well, that's what what mom

10:16

calls it. Maybe dad calls it something

10:17

different. So, it's necessary to get

10:19

language going. Um, and and some of the

10:22

philosophical work in common knowledge

10:25

originated by trying to understand

10:28

conventions such as every word in a

10:31

language. But it enters into it in a

10:33

more subtle way in the phenomenon of

10:36

indirect speech. What linguist sometimes

10:39

call implicature and that it refers to

10:42

the fact that we often don't uh say what

10:45

we mean and don't mean what what we say.

10:48

Uh it's if you simply took the meaning

10:50

of every word in a sentence and the way

10:52

the grammatical rules combine them and

10:55

then you kind deduce who did what to

10:57

whom from what's the subject and what's

10:59

the verb. you'd still be clueless as to

11:01

very often as to what the person

11:02

actually is trying to convey. Such as if

11:05

you could pass the salt, that would be

11:06

awesome. Okay? Now, if you parse that,

11:09

you're not going to get what everyone

11:11

understands the the meaning to be in

11:13

context. And that's one of the reasons

11:14

that it took so long to get computers to

11:17

understand language if they simply crank

11:20

through an algorithm of deducing the

11:22

meaning of a sentence from the

11:23

arrangement of words. Then if you ask

11:25

the computer, can you um tell me the um

11:29

the the cheapest airfare from London to

11:32

New York? It would answer, yes, I can

11:34

tell you the cheapest airfare from

11:36

London to New York. But it's kind of

11:38

missing the point that this is actually

11:39

a request. It's not a uh a uh assessment

11:44

of the uh the the algorithm's

11:48

capabilities. I mean, let's let's unpack

11:50

for a second the meaning, the literal

11:53

meaning of if you could pass me the

11:54

salt, that would be awesome. Literally,

11:56

what does that mean?

11:57

>> Literally, it's a it's a hypothetical.

11:59

You're musing on on on possible worlds.

12:03

Um, it's also hyperbolic. It surely

12:06

wouldn't be worthy of awe if the if the

12:10

salt was passed.

12:11

>> Yes.

12:11

>> U, but we effortlessly in context

12:14

understand that it is a request. We also

12:16

understand that it's a polite request.

12:19

That is presumably the reason that you

12:21

go through these this this this

12:24

indirection is that you don't want to be

12:27

barking commands at someone who is not

12:30

your servant like treating them as if

12:31

they were a servant. So there's

12:33

something polite, respectful. You're

12:34

maintaining a social relationship.

12:37

also counting on the ability of your

12:40

hearer to catch your drift, read between

12:43

the lines to figure why would that

12:45

person be musing on this hypothetical.

12:48

Well, if if the person were would not be

12:52

pleased to get the salt, then they

12:54

wouldn't be asking to pass the salt. The

12:56

fact that they say that they would be

12:58

pleased. Uh the only way to make sense

13:00

of that in context is that that they

13:02

want the salt. So there that's a a

13:04

phenomenal linguist studied for decades

13:05

with not much attention to why we go

13:07

through the bother. Why don't we just

13:10

blurt it out? And there are many other

13:12

examples of indirect speech. And I

13:14

devoted a chapter in a previous book,

13:16

the the stuff of thought, language is a

13:18

window into human nature to puzzle

13:21

puzzle out why so much of of our uh

13:25

language is roundabout. It's uh we use

13:29

we shilly shalley. we use weasel words

13:32

uh instead of just you know kind of

13:33

cutting the crap and saying what we

13:34

mean.

13:35

>> I mean there is a can I kind of

13:37

interject with a specific example here.

13:39

So just before we came on about half an

13:41

hour before we came on stage Stephen um

13:43

one of the organizers wonderful people

13:45

from how to academy thank you very much

13:47

how to academy for uh arranging this

13:49

evening. One of the one of the

13:51

organizers from how to academy said uh

13:54

would you like me to get you a coffee?

13:56

Uh and and I and I said um no I don't

13:59

want a coffee. Thank you. Uh, it keeps

14:00

me awake. And then I immediately thought

14:03

of George Costanza from Seinfeld. This

14:05

famous scene in Seinfeld where George

14:07

Costanza's he's been out on a date and

14:10

it's gone quite well. And at the end of

14:12

the date, the lady he's been uh uh spent

14:15

the evening with says, "Would you like

14:17

to come up for coffee?" And George says,

14:20

"Um, no, thank you. Um, I can't drink

14:22

coffee late at night. It keeps me

14:23

awake."

14:25

She

14:27

seems slightly annoyed. It's all a bit

14:29

awkward. She leaves. He drives off. And

14:31

then immediat and then and the next

14:33

morning he's talking to he's talking to

14:35

his friends and he's like, "Coffee

14:37

doesn't mean coffee. Coffee means sex."

14:40

And but then his his friend's like, "No,

14:42

no, George. Sometimes coffee just does

14:44

mean coffee." He's like, "No, coffee

14:46

doesn't." And but so the question is, is

14:49

it true that coffee does sometimes mean

14:51

coffee? Why didn't she say, "Would you

14:53

like to come up to my place for sex?" If

14:55

if that that is in fact what she wanted,

14:57

why did so why didn't she say that?

15:00

>> Yes. Um the

15:03

>> Well, I I have I I have studied this

15:05

problem.

15:06

>> We're not one of we're not talking about

15:08

soul now. Come on.

15:12

>> The uh and it's one of many examples

15:14

where in socially fraught situations, we

15:17

don't blurt out what what we mean. So a

15:20

a sexual comment is is an example. the

15:23

polite request, a veiled bribe. If you

15:26

were going into a restaurant and did not

15:29

have a reservation and it occurred to

15:31

you that perhaps you could slip a 20

15:33

note to the major D in exchange for

15:35

jumping the queue and being seated

15:37

immediately, uh, you probably wouldn't

15:39

say, uh, if I give you 20, would you

15:42

seat me right away? And we know that

15:44

this is true because this is actually

15:45

tried as a it was an assignment that an

15:48

editor of a food magazine gave to a

15:50

writer who dared him to do it in the

15:53

fanciest restaurants in Manhattan on a

15:55

Saturday night and write up his

15:57

experience for the the magazine. Well,

16:00

his first reaction was that this was as

16:03

he took the taxi uh on on the way to the

16:05

the first uh assignment. For one thing,

16:08

he was filled with with apprehension and

16:11

dread. He kept imagining the uh the

16:15

angry retorts of some insensed matraee

16:18

like how dare you insult me or do you

16:21

think you can get in with that or what

16:23

kind of establishment do you think this

16:24

is? That was what was running through

16:26

his mind. That was the terror if he just

16:28

blurted it out. And so instead he used

16:31

innuendos, weasel words. He said um he

16:34

first of all he would hold the note out

16:37

of in peripheral vision and say to the

16:39

major de I was wondering if you might

16:41

have a cancellation or is there any way

16:44

to shorten my weight or this is a really

16:47

important night for me by the way it

16:50

worked every time. So you now know that

16:53

at least in New York 100% of major G's

16:55

are bribable. But what was he so afraid

16:58

of? Uh and there are other examples. A

17:01

veiled threat is also usually issued

17:03

indirectly. I I reproduce a scene from

17:06

the Sopranos in which a member the

17:08

member of the family approaches a high

17:11

school acquaintance uh pays for his

17:13

candy at the checkout counter and he

17:15

says, "Danny, great to see you. I hear

17:17

you're on the jury of the Soprano trial.

17:20

It's an important civic responsibility

17:23

that we should all take part in. You've

17:25

got a wife and kids. We know you'll do

17:27

the right thing, which everyone

17:29

recognizes as a veiled threat.

17:32

So, what's going on? Uh what is the

17:34

difference between a an obvious innuendo

17:37

and the humor in the Seinfeld episode is

17:39

that in that case it did go over his

17:42

head, but most grown people know when

17:44

they're being propositioned even

17:46

indirectly. So isn't so much that there

17:48

is plausible deniability stanza

17:51

notwithstanding, but there's plausible

17:53

deniability of common knowledge. I

17:55

suggested that the difference between an

17:58

innuendo, even an obvious innuendo, and

18:01

blurting it out is that blurting it out

18:03

generates common knowledge. Which is to

18:05

say, let's say um uh we'll flip it.

18:10

Let's say uh George says to his date,

18:13

"Uh, do you want to uh come up for

18:15

coffee?" and she says no. Uh, well, you

18:19

know, she knows that she's turned down a

18:20

sexual overture and and he knows it. But

18:23

does she know that he knows? She could

18:25

still think to herself, well, maybe he

18:27

thinks I'm naive. Maybe he thinks I just

18:29

uh turned down invitation to coffee. And

18:32

he could think, well, maybe she thinks

18:34

I'm dense. So, there's no common

18:36

knowledge, the higher order knowledge

18:38

about the knowledge. And crucially that

18:41

allows them to return to their platonic

18:44

relationship or their supervisor

18:46

supervise relationship some relationship

18:49

other than a sexual relationship which

18:51

is a to use some jargon a different

18:54

coordination game. Yeah.

18:55

>> So the idea is that social relationships

18:58

are coordination dilemmas. Two people

19:01

are friends or lovers or a um

19:06

um a dominant and a subordinate. If each

19:09

one knows that the other one knows that

19:11

they are uh common knowledge can um

19:15

upend that common understanding and so

19:18

we avoid the common knowledge that

19:20

speech ordinarily generates by using

19:23

innuendo and euphemism.

Interactive Summary

The speaker introduces Steven Pinker's book "When Everyone Knows That Everyone Knows," which explores the concept of common knowledge. The discussion begins with a famous logic puzzle, the "spinach problem," to illustrate how common knowledge, defined as everyone knowing that everyone knows something (and so on), is crucial for coordination and deduction. The speaker then delves into how this concept applies to language, explaining how words, as conventions, rely on common knowledge for their meaning and use. Furthermore, the discussion highlights indirect speech and euphemisms, arguing that we often use indirect language to avoid generating common knowledge in socially sensitive situations, thereby preserving social relationships and allowing for plausible deniability. Examples like asking for salt indirectly or making veiled threats demonstrate this phenomenon. The speaker suggests that social relationships themselves can be seen as coordination dilemmas where avoiding the full generation of common knowledge through indirect speech is essential.

Suggested questions

4 ready-made prompts