Decision making: Theory and
practice
Lecture 1 – Introduction: the study of judgment and
decision making – 17/04/2018
Practcal inforaaton
There are fie regular lectures and three guest lectures (legal decision making, hydrological
forecastng, and medical decision making). The exam is on oonday June 18 th and the resit on July 6th.
It’s an electie course in which book chapters are discussed in the regular (non-compulsory) lectures
and papers are addressed in the guest lectures. The (guest) lectures do not fully discuss the book
chapters and the corresponding papers so make sure you read them thoroughly and in adiance so
you can ask questons during the lectures. During the last meetng there will be a Q&A opportunity.
The exam are detailed multple choice questons about the book chapters (Chapter 1-6, 8, 9, 11, 15-
17) (70%) and open essay questons about the papers related to the guest lectures (30%).
The lectures are aiailable as ‘web lectures’ untl the next one (one week only!!) – this has now
changed.
Today’s lecture: an introducton to the feld of decision aaainn
Today we will be introduced to the theory of decision making. We will explore the philosophical side
of it, the historical side of this feld, how it became a feld of study (scientfc feld). The goal of this
course is to introduce us to the basics, the essental parts of this feld, which can be useful in all kinds
of professions and all kinds of problems we might haie to resolie. It’s really about the basic side of
decision making. We will ask questons like: What is a decision? What is a judgment? What is the
difference between these two? And what is the difference between how we should make decisions
and how we do make decisions? What implicatons do these differences between how we should and
how we do make decisions haie?
The foundatons of the feld of decision aaainn and judnaents
The foundatons of the feld are important to well appreciate the rest of the lectures, the rest of the
materials and to properly understand the book. We will frst look at the psychology of deciding and
judging, and the differences between these two processes. Next, we will go through a small bit of
history of the feld and then we will look at the psychological stages the whole process of judgment
and decision can be diiided in oier tme.
1. The psychology of deciding and judging
In this part of the lecture we will look at different things:
Judgments is Decisions
The focus on errors – Normatie theories
Framing (one of the most important factors that affects the way we make decisions)
Our intuitons about ‘good’ decisions (which can be iery different from what scientsts or
theories would predict to be good decisions)
1.1 Judgments vs Decisions
,We can ask ourselies questons like: Will it rain tomorrow? Will my camera be stolen during my
holiday? Or, how nice is this man? These questons are answered with judgments, an estmaton, a
kind of trial about the answer. One partcular feature common to these questons is that they haie a
kind of uncertainty in them: we don’t know exactly, eien if we answer these questons in whateier
way, we can neier be sure. It is about the future, about informaton out there which we don’t haie.
These are judgments that we make – but we are sure that we’re unsure about the answer. This is a
iery important ingredient of the whole feld of decision making: uncertainty about giiing judgments.
We can also answer questons like: Should I take my umbrella? Should I buy a traiel insurance?
Should I marry this man? The answers to these questons sound like decisions. What is the difference
between these types of answers? The second kind of answers are more about “What will /do I do?”
while the frst kind of answers are more about “What do I think?”.
So, summarizing aboiementoned:
Judnaents = Estmatng the probability of outcomes (rrobability Theory, chapters 5 and 6)
Decisions = Choosing a course of acton (Expected Utlity Theory, chapters 8 and 9)
Decidinn is choosing a course of acton on the basis of how likely I estmate its result to be
(judgment) and how much that result means to me.
These are two ingredients which are important aspects of a decision. So, there’s something about the
result meaning something to you. For example studying psychology iniolies a kind of predispositon
that you like to be a psychologist, that it has some ialue to you to become a psychologist. It also
iniolies some judgment (some estmaton under uncertainty) that you will be able to follow it – that
the result will be good in the end. Thus, meaning and likelihood are iniolied in decisions.
As said, decisions are almost always made under uncertainty. At least, these situatons in which we
are uncertain about the outcomes are the most interestng ones to study. If you are sure about an
outcome, there is no point in doubtng about your decision or haiing an adiice about what decision
to make – you know what to do because you are sure about what the result will be and how much
you like that result.
A situaton in which we are uncertain about the outcomes for example is the weather forecast.
Weather forecasts are judgments – weather forecasters tell
you what their estmaton / predicton about the weather is.
They represent their uncertainty in terms of ranges.
Something later in the future is more uncertain than
something more nearby, as can be seen in the graph.
1.2 The feld of psychology of uudgment and decision making shows and ex plains errors
The feld of psychology is a specifc part of the feld of judgment and decision making. This feld is
iery much focused on how we make decisions and especially on how we make errors in making
decisions. Let’s look at an error (according to mathematcal theory of expected ialue). According to
the predictons of this theory taking a traiel insurance is an error.
We can calculate the expected ialue according to the theory for not buying decisions and buying
decisions. We can then calculate which course of acton yields the highest expected ialue. So, for
traiel insurance for your camera you haie two optons: don’t buy it or buy it.
Opton 1: don’t buy. The expected ialue of this opton is calculated as follows: let’s assume
the probability of your camera getng stolen is 1% and the ialue of your camera is 400. You
can then multply these ialues [camera stolen = 0.01 * -400]. There’s also the opton that
, your camera is not stolen, which chance is 99% and the ialue of the status quo is 0: [camera
not stolen = 0.99 * 0]. You can then add the expected ialues of both outcomes [0.01*-400] +
[0.99*0] = -4.
Opton 2: buy – here, we add the price of the insurance. The ialue of your camera stolen will
then be reimbursed by your insurance. So then you get the formula for expected ialue:
[camera stolen = 0.01 * -400+400] + [camera not stolen = 0.99 * 0] – [price of insurance = 1 *
30] = -30.
The expected ialue of not buying an insurance thus leads to a higher expected ialue (i.e. a beter
opton). That is how mathematcal theory of decision making would tell you to make your
calculatons when you consider your different outcomes. Of course this is a relatiely easy example,
with only a camera and you will know how much a camera costs, but in many situatons this becomes
iery difcult. Howeier, this is what you should do according to mathematcal theory. This calculaton
leads us to conclude that the expected ialue of not buying a traiel insurance is beter than buying
one.
Howeier, we do buy traiel insurances on a regular basis. Can we explain why we do afer all buy
traiel insurances? And what exactly do we buy? What do we pay for when we buy these things? One
explanaton could be that insurance companies do not sell a naked, business like insurance, but they
sell you a good feeling: if something happens, it’s insured, eierything is fne, so you will feel beter.
This is something insurance companies really focus on (e.g. in their adiertsements: “Het
Zwitserleien Geioel” with a man on the beach; “TraielSafe, traiel with peace of mind”). They thus
want to put in your mind that this feeling is worth some money and is thus worth buying the
insurance.
If we would add this to the equaton (and it’s not in the traditonal mathematcal approach of
decision making) as psychologists, we would end up with the following expected ialue of the opton
buy: [camera stolen and reimbursed = 0.01 * - 400 + 400] + [insurance price = 1 * 30] + [ialue of
“feeling safe” = 1 * -30] = 0. This is an example of framing and of putng emotons in the equaton of
decision making and a iery quantfed and clear way which explains why people afer all make
decisions that are irratonal according to expected ialue theory.
1.3 Framing
So, insurers frame the choice, to infuence the expected ialue of consumers (so that they buy the
insurance). Expected ialue theory is thus not (always) compatble with how we actually make
decisions in our daily liies. This feld (the psychology of decision making) is iery much oriented on
showing errors. Why is that? And, how is it possible that we make so many errors? Besides adding
emotons and things like framing to the equaton, another possible explanaton for the focus on
errors could be science itself. It could be that we, as scientsts, create these cognitie illusions
(decision “errors” are ofen called cognitie illusions). This is an argument (or critcism) that has been
used iery much against for example Kahneman. The critcism says that when you bring people to the
lab, they behaie in a certain way. And the problems they are presented with, are problems they
rarely come across in real life. So, if we compare the decision situatons to real life what sometmes
can be shown is that you elicit unnatural behaiior which can explain these errors.
An example of such an error in iisual illusions is the table example. When
asked which table is longer, and giie them the choice between these two
tables, people say the lef table. They are actually the same size – it is an
illusion. Then, why do people say the lef table is longer?
What people do, is that they interpret intuitiely, spontaneously and immediately 2D pictures as
representng known 3D objects. You thus do not see a 2D surface here, you see in your head a true
table according to the laws of perspectie, and thus answer the queston as if you would see them in
3D. That’s iery smart, and the thing you should do if you want to suriiie! Our iisual system is
, oriented on suriiiing and is tuned to giie us informaton that is releiant to us (thus, 3D, for example
if a lion approaches you). Interpretng 2D paterns (drawings) as representatons of 3D things, is the
most “natural” interpretaton in daily life. The normatie answer here, howeier, is the 2D shapes. So,
if you bring people to the lab you conclude they suffer from this iisual illusion error. And this is also
what happened in some of the literature of decision making – it was iery dependent on situatons,
stmulus materials and the questons that were asked, specifcally in the experimental lab situaton.
Answers there are counted as “wrong” but might be iery good answers in real world situatons. This
is for example also true with the “Linda” problem – you are giien a descripton of a person who looks
like a feminist, and are then presented with the queston whether Linda is a feminist and a bank
teller or a bank teller only. What people tend to do, is say that it is more probable that Linda is a
feminist and a bank teller (because of the descripton, but probabilistcally it is more likely that she’s
only a bank teller).
Illusions and errors can arise when the iisual or cognitie system operates under unusual or
unnatural conditons (e.g. when it is in the lab) or infuenced by social and emotonal biases.
Fraainn can infuence (the accuracy) of decisions. reople haie argued that if you present people in
the lab with problems in which you menton probabilites (e.g. the probability that Linda is a feminist
is .6), these are iery abstract. rrobabilites are part of our mathematcal system, and do not come
naturally to us. Frequencies, on the other hand, are natural – we can count! And in nature we can see
differences in frequency, but we cannot see differences in probabilites naturally. This is why the
frequency frame has been called the “natural frame”. He claims that when you say to people “One
out of 1000 men has the same shoe profle as the suspect. Is the suspect guilty?” (natural frequency
frame) or you ask them “The probability that any man has the same shoe profle as the suspect is p =
0.001. Is the suspect guilty?” (unnatural probability frame) – what happens is that people on aierage
assign higher probabilites to the unnatural probability frame conditon as compared to the
frequency frame conditon. The frequency frame leads to fewer ‘guilty ierdicts’ than the probability
frame. In our minds, “one-in-1000” is smaller than p = 0.001.
Then we can ask the queston how ofen we make errors under normal conditons. In most situatons,
we are iery ratonal, actually. roletek points out that another explanaton of this focus on errors is a
kind of publicaton bias – a scientfc bias. We as scientsts come up with conclusions about how
people think and reason, and we like to show the errors in this reasoning. There is not much to say
about people who reason exactly according to the normatie theories. rublishing something about
errors is much more fashy than showing that people do exactly what theories predict.
There are two iideos on Blackboard. One by Tim Harford and one by Dan Ariely who represent the
two streams about thinking about errors. Dan Ariely is a psychologist, Tim Harford is an economist.
Dan Ariely argues that people are basically irratonal whilst Tim Harford argues that we’re basically
ratonal.
Video Tim Harford
Video Dan Ariely
2. History: from maths to economics to psychology
How did this feld become science? In the 17 th century there were philosophers like Blaise rascal and
Christaan Huygens. They had this intuiton about how to calculate hope. They wanted to know how
to quantfy the idea that we hope something, how big our hope is. They found intuitiely that two
ingredients were important: How likely is the eient? (the judgment part of the decision) and What
does the event mean to me? That last one was iery difcult to quantfy back then. They came up with
the basic formula:
practice
Lecture 1 – Introduction: the study of judgment and
decision making – 17/04/2018
Practcal inforaaton
There are fie regular lectures and three guest lectures (legal decision making, hydrological
forecastng, and medical decision making). The exam is on oonday June 18 th and the resit on July 6th.
It’s an electie course in which book chapters are discussed in the regular (non-compulsory) lectures
and papers are addressed in the guest lectures. The (guest) lectures do not fully discuss the book
chapters and the corresponding papers so make sure you read them thoroughly and in adiance so
you can ask questons during the lectures. During the last meetng there will be a Q&A opportunity.
The exam are detailed multple choice questons about the book chapters (Chapter 1-6, 8, 9, 11, 15-
17) (70%) and open essay questons about the papers related to the guest lectures (30%).
The lectures are aiailable as ‘web lectures’ untl the next one (one week only!!) – this has now
changed.
Today’s lecture: an introducton to the feld of decision aaainn
Today we will be introduced to the theory of decision making. We will explore the philosophical side
of it, the historical side of this feld, how it became a feld of study (scientfc feld). The goal of this
course is to introduce us to the basics, the essental parts of this feld, which can be useful in all kinds
of professions and all kinds of problems we might haie to resolie. It’s really about the basic side of
decision making. We will ask questons like: What is a decision? What is a judgment? What is the
difference between these two? And what is the difference between how we should make decisions
and how we do make decisions? What implicatons do these differences between how we should and
how we do make decisions haie?
The foundatons of the feld of decision aaainn and judnaents
The foundatons of the feld are important to well appreciate the rest of the lectures, the rest of the
materials and to properly understand the book. We will frst look at the psychology of deciding and
judging, and the differences between these two processes. Next, we will go through a small bit of
history of the feld and then we will look at the psychological stages the whole process of judgment
and decision can be diiided in oier tme.
1. The psychology of deciding and judging
In this part of the lecture we will look at different things:
Judgments is Decisions
The focus on errors – Normatie theories
Framing (one of the most important factors that affects the way we make decisions)
Our intuitons about ‘good’ decisions (which can be iery different from what scientsts or
theories would predict to be good decisions)
1.1 Judgments vs Decisions
,We can ask ourselies questons like: Will it rain tomorrow? Will my camera be stolen during my
holiday? Or, how nice is this man? These questons are answered with judgments, an estmaton, a
kind of trial about the answer. One partcular feature common to these questons is that they haie a
kind of uncertainty in them: we don’t know exactly, eien if we answer these questons in whateier
way, we can neier be sure. It is about the future, about informaton out there which we don’t haie.
These are judgments that we make – but we are sure that we’re unsure about the answer. This is a
iery important ingredient of the whole feld of decision making: uncertainty about giiing judgments.
We can also answer questons like: Should I take my umbrella? Should I buy a traiel insurance?
Should I marry this man? The answers to these questons sound like decisions. What is the difference
between these types of answers? The second kind of answers are more about “What will /do I do?”
while the frst kind of answers are more about “What do I think?”.
So, summarizing aboiementoned:
Judnaents = Estmatng the probability of outcomes (rrobability Theory, chapters 5 and 6)
Decisions = Choosing a course of acton (Expected Utlity Theory, chapters 8 and 9)
Decidinn is choosing a course of acton on the basis of how likely I estmate its result to be
(judgment) and how much that result means to me.
These are two ingredients which are important aspects of a decision. So, there’s something about the
result meaning something to you. For example studying psychology iniolies a kind of predispositon
that you like to be a psychologist, that it has some ialue to you to become a psychologist. It also
iniolies some judgment (some estmaton under uncertainty) that you will be able to follow it – that
the result will be good in the end. Thus, meaning and likelihood are iniolied in decisions.
As said, decisions are almost always made under uncertainty. At least, these situatons in which we
are uncertain about the outcomes are the most interestng ones to study. If you are sure about an
outcome, there is no point in doubtng about your decision or haiing an adiice about what decision
to make – you know what to do because you are sure about what the result will be and how much
you like that result.
A situaton in which we are uncertain about the outcomes for example is the weather forecast.
Weather forecasts are judgments – weather forecasters tell
you what their estmaton / predicton about the weather is.
They represent their uncertainty in terms of ranges.
Something later in the future is more uncertain than
something more nearby, as can be seen in the graph.
1.2 The feld of psychology of uudgment and decision making shows and ex plains errors
The feld of psychology is a specifc part of the feld of judgment and decision making. This feld is
iery much focused on how we make decisions and especially on how we make errors in making
decisions. Let’s look at an error (according to mathematcal theory of expected ialue). According to
the predictons of this theory taking a traiel insurance is an error.
We can calculate the expected ialue according to the theory for not buying decisions and buying
decisions. We can then calculate which course of acton yields the highest expected ialue. So, for
traiel insurance for your camera you haie two optons: don’t buy it or buy it.
Opton 1: don’t buy. The expected ialue of this opton is calculated as follows: let’s assume
the probability of your camera getng stolen is 1% and the ialue of your camera is 400. You
can then multply these ialues [camera stolen = 0.01 * -400]. There’s also the opton that
, your camera is not stolen, which chance is 99% and the ialue of the status quo is 0: [camera
not stolen = 0.99 * 0]. You can then add the expected ialues of both outcomes [0.01*-400] +
[0.99*0] = -4.
Opton 2: buy – here, we add the price of the insurance. The ialue of your camera stolen will
then be reimbursed by your insurance. So then you get the formula for expected ialue:
[camera stolen = 0.01 * -400+400] + [camera not stolen = 0.99 * 0] – [price of insurance = 1 *
30] = -30.
The expected ialue of not buying an insurance thus leads to a higher expected ialue (i.e. a beter
opton). That is how mathematcal theory of decision making would tell you to make your
calculatons when you consider your different outcomes. Of course this is a relatiely easy example,
with only a camera and you will know how much a camera costs, but in many situatons this becomes
iery difcult. Howeier, this is what you should do according to mathematcal theory. This calculaton
leads us to conclude that the expected ialue of not buying a traiel insurance is beter than buying
one.
Howeier, we do buy traiel insurances on a regular basis. Can we explain why we do afer all buy
traiel insurances? And what exactly do we buy? What do we pay for when we buy these things? One
explanaton could be that insurance companies do not sell a naked, business like insurance, but they
sell you a good feeling: if something happens, it’s insured, eierything is fne, so you will feel beter.
This is something insurance companies really focus on (e.g. in their adiertsements: “Het
Zwitserleien Geioel” with a man on the beach; “TraielSafe, traiel with peace of mind”). They thus
want to put in your mind that this feeling is worth some money and is thus worth buying the
insurance.
If we would add this to the equaton (and it’s not in the traditonal mathematcal approach of
decision making) as psychologists, we would end up with the following expected ialue of the opton
buy: [camera stolen and reimbursed = 0.01 * - 400 + 400] + [insurance price = 1 * 30] + [ialue of
“feeling safe” = 1 * -30] = 0. This is an example of framing and of putng emotons in the equaton of
decision making and a iery quantfed and clear way which explains why people afer all make
decisions that are irratonal according to expected ialue theory.
1.3 Framing
So, insurers frame the choice, to infuence the expected ialue of consumers (so that they buy the
insurance). Expected ialue theory is thus not (always) compatble with how we actually make
decisions in our daily liies. This feld (the psychology of decision making) is iery much oriented on
showing errors. Why is that? And, how is it possible that we make so many errors? Besides adding
emotons and things like framing to the equaton, another possible explanaton for the focus on
errors could be science itself. It could be that we, as scientsts, create these cognitie illusions
(decision “errors” are ofen called cognitie illusions). This is an argument (or critcism) that has been
used iery much against for example Kahneman. The critcism says that when you bring people to the
lab, they behaie in a certain way. And the problems they are presented with, are problems they
rarely come across in real life. So, if we compare the decision situatons to real life what sometmes
can be shown is that you elicit unnatural behaiior which can explain these errors.
An example of such an error in iisual illusions is the table example. When
asked which table is longer, and giie them the choice between these two
tables, people say the lef table. They are actually the same size – it is an
illusion. Then, why do people say the lef table is longer?
What people do, is that they interpret intuitiely, spontaneously and immediately 2D pictures as
representng known 3D objects. You thus do not see a 2D surface here, you see in your head a true
table according to the laws of perspectie, and thus answer the queston as if you would see them in
3D. That’s iery smart, and the thing you should do if you want to suriiie! Our iisual system is
, oriented on suriiiing and is tuned to giie us informaton that is releiant to us (thus, 3D, for example
if a lion approaches you). Interpretng 2D paterns (drawings) as representatons of 3D things, is the
most “natural” interpretaton in daily life. The normatie answer here, howeier, is the 2D shapes. So,
if you bring people to the lab you conclude they suffer from this iisual illusion error. And this is also
what happened in some of the literature of decision making – it was iery dependent on situatons,
stmulus materials and the questons that were asked, specifcally in the experimental lab situaton.
Answers there are counted as “wrong” but might be iery good answers in real world situatons. This
is for example also true with the “Linda” problem – you are giien a descripton of a person who looks
like a feminist, and are then presented with the queston whether Linda is a feminist and a bank
teller or a bank teller only. What people tend to do, is say that it is more probable that Linda is a
feminist and a bank teller (because of the descripton, but probabilistcally it is more likely that she’s
only a bank teller).
Illusions and errors can arise when the iisual or cognitie system operates under unusual or
unnatural conditons (e.g. when it is in the lab) or infuenced by social and emotonal biases.
Fraainn can infuence (the accuracy) of decisions. reople haie argued that if you present people in
the lab with problems in which you menton probabilites (e.g. the probability that Linda is a feminist
is .6), these are iery abstract. rrobabilites are part of our mathematcal system, and do not come
naturally to us. Frequencies, on the other hand, are natural – we can count! And in nature we can see
differences in frequency, but we cannot see differences in probabilites naturally. This is why the
frequency frame has been called the “natural frame”. He claims that when you say to people “One
out of 1000 men has the same shoe profle as the suspect. Is the suspect guilty?” (natural frequency
frame) or you ask them “The probability that any man has the same shoe profle as the suspect is p =
0.001. Is the suspect guilty?” (unnatural probability frame) – what happens is that people on aierage
assign higher probabilites to the unnatural probability frame conditon as compared to the
frequency frame conditon. The frequency frame leads to fewer ‘guilty ierdicts’ than the probability
frame. In our minds, “one-in-1000” is smaller than p = 0.001.
Then we can ask the queston how ofen we make errors under normal conditons. In most situatons,
we are iery ratonal, actually. roletek points out that another explanaton of this focus on errors is a
kind of publicaton bias – a scientfc bias. We as scientsts come up with conclusions about how
people think and reason, and we like to show the errors in this reasoning. There is not much to say
about people who reason exactly according to the normatie theories. rublishing something about
errors is much more fashy than showing that people do exactly what theories predict.
There are two iideos on Blackboard. One by Tim Harford and one by Dan Ariely who represent the
two streams about thinking about errors. Dan Ariely is a psychologist, Tim Harford is an economist.
Dan Ariely argues that people are basically irratonal whilst Tim Harford argues that we’re basically
ratonal.
Video Tim Harford
Video Dan Ariely
2. History: from maths to economics to psychology
How did this feld become science? In the 17 th century there were philosophers like Blaise rascal and
Christaan Huygens. They had this intuiton about how to calculate hope. They wanted to know how
to quantfy the idea that we hope something, how big our hope is. They found intuitiely that two
ingredients were important: How likely is the eient? (the judgment part of the decision) and What
does the event mean to me? That last one was iery difcult to quantfy back then. They came up with
the basic formula:
