Argumentation and human reason Hugo Mercier Institut Jean Nicod - - PowerPoint PPT Presentation

SLIDE 1

Argumentation and human reason

Hugo Mercier Institut Jean Nicod CNRS

Argument mining workshop, 2019

SLIDE 2

SLIDE 3

What is reason?

SLIDE 4

Peter is 8, John is 12 John is older than Peter It’s going to rain It needs food and water

Intuition

SLIDE 5

• S=1+2+…+99+100 is equivalent to S=(100*101)/2

because...

• My coat is neither on the coat hanger nor in my

dressing, so I must have forgotten it at work

• You shouldn’t have any coffee, it will keep you

awake

Reason

SLIDE 6

Why do we reason?

SLIDE 7

The individualistic view of reasoning

SLIDE 8

Peter is looking at Linda Linda is looking at Henry Peter is married Henry is not married Is someone who is married looking at someone who is not married? Yes No We can’t tell

The Levesque task

SLIDE 9

Peter is looking at Linda Linda is looking at Henry Peter is married Henry is not married Is someone who is married looking at someone who is not married? Yes No We can’t tell

An intuitive mistake

SLIDE 10

Peter is looking at Linda Linda is looking at Henry Peter is married Henry is not married Is someone who is married looking at someone who is not married? Yes No We can’t tell

Reasoning saves the day

SLIDE 11

Reasoning can help the lone reasoner correct mistaken intuitions and arrive at better beliefs

SLIDE 12

Reasoning can help the lone reasoner correct mistaken intuitions and arrive at better beliefs

SLIDE 13

Peter is looking at Linda Linda is looking at Henry Peter is married Henry is not married Is someone who is married looking at someone who is not married? Yes No We can’t tell

The Levesque task

Trouche, Sander, & Mercier 2014

SLIDE 14

Reasoning can help the lone reasoner correct mistaken intuitions and arrive at better beliefs But it often doesn’t

SLIDE 15

Peter is looking at Linda Linda is looking at Henry Peter is married Henry is not married Is someone who is married looking at someone who is not married? Yes No We can’t tell

What reasoning actually does

We don’t have enough information We don’t know if Linda is married The answer depends on Linda’s status

SLIDE 16

Reasoning can help the lone reasoner correct mistaken intuitions and arrive at better beliefs But it often doesn’t Because it does precisely the opposite

SLIDE 17

The individualist view of reasoning

WHAT THEN?

SLIDE 18

The interactionist view of reason

Mercier & Sperber, 2011, 2017

SLIDE 19

Myside bias

Prediction 1

SLIDE 20

Selective laziness

Prediction 2

SLIDE 21

Sylvia: “We should go to Isami, it’s a good restaurant” Helen: “I don’t know, I’ve had Japanese last week already” Sylvia: “But this one is very original”

Making the best of feedback

SLIDE 22

Sylvia: “We should go to Isami, it’s a good restaurant” Helen: “I don’t know, I don’t have much money at the moment, and Japanese restaurants can be pricy” Sylvia: “But this one is quite cheap”

Making the best of feedback

SLIDE 23

Sylvia: “We should go to Isami, it’s original, the prices are good, the fish is fresh, the crowd is lively… ”

Other option: exhaustive argument

SLIDE 24

Sylvia, thinking: I wonder if Helen has been to a Japanese restaurant lately. And would she be bothered by high prices? Does she eat raw fish? Does she enjoy the kind of crowd you get in typical Japanese restaurants? Would she believe it’s in an inconvenient location?...

Other option: anticipation

SLIDE 25

People should typically start with a reasonable argument, even if it is a relatively weak, generic argument

Prediction in production

SLIDE 26

People should carefully examine other people’s arguments

Prediction in evaluation

SLIDE 27

First phase: NO REASONING

Trouche, Hall, Johansson, & Mercier 2017

SLIDE 28

In a fruit and vegetable shop which carries, among

• ther products, apples:

None of the apples are organic. What can you conclude for sure about whether fruits are organic in this shop ?

All the fruits are organic None of the fruits are organic Some fruits are organic Some fruits are not organic We cannot tell anything for sure about whether fruits are organic in this shop

“Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.”

SLIDE 29

In a fruit and vegetable shop which carries, among

• ther products, apples:

None of the apples are organic. What can you conclude for sure about whether fruits are organic in this shop ?

All the fruits are organic None of the fruits are organic Some fruits are organic Some fruits are not organic We cannot tell anything for sure about whether fruits are organic in this shop

“Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.”

SLIDE 30

In a fruit and vegetable shop which carries, among

• ther products, apples:

None of the apples are organic. What can you conclude for sure about whether fruits are organic in this shop ?

All the fruits are organic None of the fruits are organic Some fruits are organic Some fruits are not organic We cannot tell anything for sure about whether fruits are organic in this shop

“Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.”

SLIDE 31

Second phase: ARGUMENT PRODUCTION

SLIDE 32

In a fruit and vegetable shop which carries, among

• ther products, apples:

None of the apples are organic. What can you conclude for sure about whether fruits are organic in this shop ?

All the fruits are organic None of the fruits are organic Some fruits are organic Some fruits are not organic We cannot tell anything for sure about whether fruits are organic in this shop

“Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.”

SLIDE 33

In a fruit and vegetable shop which carries, among

• ther products, apples:

None of the apples are organic. What can you conclude for sure about whether fruits are organic in this shop ?

All the fruits are organic None of the fruits are organic Some fruits are organic Some fruits are not organic We cannot tell anything for sure about whether fruits are organic in this shop

“Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.”

SLIDE 34

In a fruit and vegetable shop which carries, among

• ther products, apples:

None of the apples are organic. What can you conclude for sure about whether fruits are organic in this shop ?

All the fruits are organic None of the fruits are organic Some fruits are organic Some fruits are not organic We cannot tell anything for sure about whether fruits are organic in this shop

“Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.”

SLIDE 35

86% stick to their intuitive answer

SLIDE 36

Not more likely to stick to their intuitive answer if it is valid than if it is invalid

SLIDE 37

Third phase: ARGUMENT EVALUATION

SLIDE 38

You answered Some fruits are not organic Someone else answered We cannot tell anything for sure about whether fruits are organic in this shop And the argument was: “There is not enough information to conclude about all the fruits in this shop.” If you want you can change your mind

SLIDE 39

You answered Some fruits are not organic Someone else answered We cannot tell anything for sure about whether fruits are organic in this shop And the argument was: “There is not enough information to conclude about all the fruits in this shop.” If you want you can change your mind

SLIDE 40

You answered Some fruits are not organic Someone else answered We cannot tell anything for sure about whether fruits are organic in this shop And the argument was: “There is not enough information to conclude about all the fruits in this shop.” If you want you can change your mind

SLIDE 41

You answered Some fruits are not organic Someone else answered We cannot tell anything for sure about whether fruits are organic in this shop And the argument was: “There is not enough information to conclude about all the fruits in this shop.” If you want you can change your mind

SLIDE 42

You answered Some fruits are not organic Someone else answered We cannot tell anything for sure about whether fruits are organic in this shop And the argument was: “There is not enough information to conclude about all the fruits in this shop.” If you want you can change your mind

SLIDE 43

You answered We cannot tell anything for sure about whether fruits are organic in this shop Someone else answered Some fruits are not organic And the argument was: “Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.” If you want you can change your mind

SLIDE 44

You answered We cannot tell anything for sure about whether fruits are organic in this shop Someone else answered Some fruits are not organic And the argument was: “Because none of the apples are organic, and an apple is one type of fruit, we can say that some of the fruits in the store are not organic.” If you want you can change your mind

NOT THEIR ANSWER THEIR ANSWER THEIR ARGUMENT

SLIDE 45

54% do not detect the inversion

SLIDE 46

58% reject their own argument

SLIDE 47

43% reject their own ‘good’ argument 69% reject their own ‘bad’ argument

SLIDE 48

Good argument evaluation skills

Prediction 3

SLIDE 49

People discriminate between fallacious and non fallacious arguments (at least when they disagree with the conclusion)

Good argument evaluation skills

Hahn & Oaksford 2007 (and many others)

SLIDE 50

People can accept strong enough arguments, even if the arguments challenge their beliefs Prior confidence and trust in source do not affect the evaluation of conclusive arguments

Good argument evaluation skills

Trouche, Shao, & Mercier in press

SLIDE 51

Leve sque, etc. Leve sque, etc.

Yes becau se…

Try to convince someone

Yes becau se… We don’t know becau se… We don’t know becau se… Trouche, Shao, & Mercier in press

SLIDE 52

To elicit low confidence: We ask that YOU ANSWER QUICKLY. To elicit high confidence: We ask that YOU THINK VERY CAREFULLY ABOUT YOUR ANSWER, and that you JUSTIFY IT WELL.

Confidence manipulation

Trouche, Shao, & Mercier in press

SLIDE 53

Confidence in wrong answer

1 2 3 4 5 6 7

low confidence high confidence

SLIDE 54

% accepting the correct answer

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

low confidence high confidence

SLIDE 55

SLIDE 56

Groups outperform individuals on reasoning tasks

Prediction 4

SLIDE 57

18%

0% 10% 20% 30% 40% 50% 60% 70%

20% 60%

Trouche, Sander, & Mercier 2014

SLIDE 58

18%

0% 10% 20% 30% 40% 50% 60% 70%

20% 60%

Trouche, Sander, & Mercier 2014

SLIDE 59

1 1

• 1
• 1
• 1
• 1

SLIDE 60

1 1

• 1
• 1
• 1
• 1

1 1 1 1 1 1

• 1
• 1
• 1
• 1

1 1 1 1 1 1 1 1

• 1

1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1

• 1

1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1

• 1

1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1

• 1

1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1
• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

• 1

1 1 1

• 1
• 1

1

• 1

1

SLIDE 61

18%

Trouche, Sander, & Mercier 2014; Mercier & al 2015; Castelain, Girotto, Jamet, & Mercier 2016

1 2 3 French Japanese Maya

SLIDE 62

18%

1 1.1 1.2 1.3

Lie detection Forecasting Biology problems Medical diagnosis Judicial decisions Klein & Epley 2015; Mellers & al 2014; Smith et al 2009; Hautz et al 2015; van Dijk et al 1994

SLIDE 63

18%

1 2 3 College students 10 year olds 5 year olds

Trouche, Sander, & Mercier 2014; Park & Lee 2015

SLIDE 64

Argumentation’s efficacy is not intuitive

1 2 3 4 5 typical sample psychologists of reasoning reality

Ratio of group to individual performance

Mercier et al. 2015

SLIDE 65

Individual reason is overrated Argumentation is underrated Reason is for interacting: exchanging arguments and justifications

Summary

SLIDE 66

Thank you!