Heuristics

Transcript

Heuristics
Ce.R.D. - Center for Research on Risk and Decision
DPSS - University of Padova
http://decision.psy.unipd.it/
Heuristics
Behavioral Economics Class
Faculty of Economics - University of Padova
Academic year 2010/2011
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Instrumental rationality
Nowadays, most researchers working on rationality define it as «instrumental
rationality»:
Decision processes are rational if they help a decision maker to
achieve his/her objectives.
Under such framework, rationality does not concern the correctness
of a decision maker cognitive processes or inferences, but only the
efficacy of his/her actions.
Does the decision maker achieve the planned goal? Yes/no.
How he/she achieves it is not relevant.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality
As we saw in the first lecture, an important contribution to the analysis of
psychological processes in the economics field came from Herbert Simon.
Simon was among the first to suggest that the analysis of rationality
should investigate:
Non only the results of a decision (substantive rationality).
But also the procedure people use when making a decision
(procedural rationality).
Ce.R.D.
Bounded rationality
Centro di Ricerca sul
Rischio e la Decisione
Simon suggests that cognitive limitations, together with the context in which the
decision is made, determine the procedures actually applied by decision
makers:
Cognitive limitations depend mainly on our attention and memory, as
well as emotional feedbacks.
Memory span.
Selective attention.
Mood.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality
To overcome their cognitive limitations people can use different reasoning
strategies depending on the context of the decision:
Often, «environmental» factors can interact with people’s cognitive
limitations further reducing their ability to make optimal decisions;
among the most relevant «environmental» factors there are:
The number of available alternatives.
The amount of information available for each alternative.
Temporal pressure.
Others’ judgment about the decision maker’s decisions.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality and heuristics
Simon proposed two useful concepts in order to describe more accurately the
way people reason and decide in the real world:
The notion of «satisfaction»:
Seldom, people in their everyday life apply the procedures required
to choose the alternative maximizing their expected utility.
Those procedures are cognitively demanding (energy consuming)
and, therefore, individuals simply aim to choose something that is
at least satisfying.
The notion of «heuristic»:
Heuristics are simplified procedures that do not guarantee to make
the best choice, but allow to make satisfactory decisions saving
time and effort.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality and heuristics
Heuristics are for the most part automatic cognitive processes, that is strategies
that people apply unconsciously (without being aware of following them).
They are the result of the evolutionary processes involving the human
brain.
Using heuristics decision makers accept an implicit trade-off between
effort and accuracy:
They accept a lower degree of accuracy in their judgments and
decisions.
In turn, decisions are made more easily and without sapping the
individual cognitive resources.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality and heuristics
In this way a decision maker can:
Minimize the exploration of all possible solutions.
Face decision contexts in which there is a high load of data/
information to analyze.
Many of Simon’s studies have analyzed decisions in the real world, like:
Decisions in organizational environments.
Analyses of chess players’ strategies.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality and heuristics
Therefore, Simon’s approach adds:
To the limits identified by the neoclassical economics theories (for
instance, limits in the monetary resources).
The limits that characterize the cognitive system.
Simon does not reject the economic notion of rationality and accepts its
centrality for the analyses of economical behaviors.
However, he identifies a disagreement between rational behavior and
the behavior of real decision makers.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
Bounded rationality and heuristics
Research identifies many different heuristics, three of them are the most
relevant of all (Tversky & Kahneman, 1974):
«Representativeness» heuristic.
«Availability» heuristic.
«Anchoring» heuristic.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic
This heuristic is used to answer questions like:
What is the likelihood that object A belongs to class B?
What is the likelihood that event A is caused by process B?
On many occasions, these probabilities are judged on the basis of how much A
is representative of B.
If A is extremely representative of (similar to) B, people will judge the
likelihood that A is caused by B to be high.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic
People make mistakes when judging whether an event belongs to a particular
process or not.
When doing this type of judgments people...
do not consider information like:
Base rate (sample size).
Rules of statistics (i.e., the law of large numbers).
Ce.R.D.
«Representativeness» heuristic
Centro di Ricerca sul
Rischio e la Decisione
Example (Tversky & Kahneman, 1974):
Imagine to throw a fair coin six times. Which of the following two
sequences of outcomes is more likely?
TTTHHH
THHTHT
H = Head
T = Tail
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic
Usually, the majority of people judges the THHTHT sequence more likely than
the other one.
This is the sequence that is more similar to our perception of a
random outcome originating from the throw of a fair coin.
This error is called «gambler’s fallacy»:
The randomness of an event can only be found for long series of
event (i.e., 100 throws rather than only 6).
This is «the law of large numbers».
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic
Many gamblers think that the law of large numbers applies to short
series of events, too.
This is what Tversky and Kahneman labeled the «law of small
numbers».
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic
In a similar way lotto players think that if a number has not been extracted for a
long time its likelihood to be extracted the next time increases.
Actually, in games like lotto every extraction is like a new event,
independent from previous ones since all numbers are reinstated in
the urn after each draw.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic:
Sample size
Representativeness heuristic induces people to neglect the «base rate» of an
event.
«Base rate» is the impact of a phenomenon on a population or even
on a specific sample.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic:
Sample size
EXAMPLE (Tversky & Kahneman, 1974):
Imagine that in a town there are two hospital. In the larger hospital
about 45 babies are born each day, and in the smaller hospital about
15 babies are born each day.
As you know, about 50 percent of all babies are boys. However, the
exact percentage varies from day to day. Sometime it may be higher
than 50 percent, sometimes lower.
For a period of 1 year, each hospital recorded the days in which more
than 60 percent of the babies born were boys.
Which hospital do you think recorded more such days?
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Representativeness» heuristic:
Sample size
It is more likely that the hospital in which the percentage of boys is more often
above 60% is the smaller one.
Smaller samples are less representative of a population than larger ones,
therefore it is easier to find instances that deviate from the general distribution
of the population in smaller rather than larger samples.
Only 20% to 30% of respondents answer correctly to this question
(and other similar ones).
Most people think there should not be any difference between the
two samples.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Availability» heuristic
This heuristic is used to estimate the likelihood that an event will happen:
While making such judgments people use their knowledge about
similar events and try to recollect as many examples as possible from
their memory.
More examples they find and more likely they will judge the event.
However, memory is not infallible and do not follow the rules of statistics:
It makes associations by similarity.
It makes perceptive errors (some events struck people’s imagination
more easily and are remembered better than others).
It is selective and memories are modified as time goes by.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Availability» heuristic
EXAMPLE (Tversky & Kahneman, 1974):
Participants in this experiment were asked which type of word is more
frequent in English (they were given 20s to answer):
Words that have the letter “R” as the first one.
Words that have the letter “R” in third place.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Availability» heuristic
Most people answer that there are more English words starting with
“R” than words having the “R” as the third letter.
Actually, in English the opposite is true.
This depends on the fact that it is easier to recollect from memory
words that start with the letter “R” than words having the same letter in
third position.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Availability» heuristic
EXAMPLE (Tversky & Kahneman, 1974).
Often people are influenced by how relevant or familiar stimuli are.
Relevant and familiar stimuli are usually judged as more frequent.
The authors of this study prepared two lists of famous names. The two
lists were made with names of famous people from the show
business.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Availability» heuristic
LIST 1: 19 names of famous women and 20 of well-known men from the show
business.
LIST 2: 20 names of well-known women and 20 names of famous men from the
show business.
Example of a well-known person (in Italy): Carlo Conti.
Example of a famous person (in Italy): Pippo Baudo.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Availability» heuristic
In the first case (LIST 1), people judged the names of women as more frequent
than names of men, despite they were 19 (but more familiar and easier to
memorize) versus 20.
The opposite happened in the second case when the names of men
were less, but more familiar, than the names of women.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Anchoring» heuristic
On several occasions, when people are asked to provide a numerical estimate
(i.e., estimate the future price of a stock) they tend to anchor to some numerical
value to then adjust their answer in the direction of what they believe is the
correct answer.
When people do not know the correct answer, whatever numerical information
is available becomes the starting point to begin to reason about a plausible
answer.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Anchoring» heuristic
Tversky and Kahneman (as well as many other researchers) showed that
people use anchors even when:
They are told that the numbers are extracted randomly.
The anchors are implausible numbers (extremely high or extremely
low).
There are no anchors available (self-generated anchors).
Ce.R.D.
«Anchoring» heuristic
Centro di Ricerca sul
Rischio e la Decisione
The anchoring heuristic is strongly influenced by a phenomenon called
“confirmation bias”, that is people’s tendency to find a confirmation for their
initial opinions (or hypotheses).
EXAMPLE:
Are the genes of the fruit mosquito, called Drosophila, more or less
than 2?
Ce.R.D.
«Anchoring» heuristic
Centro di Ricerca sul
Rischio e la Decisione
The anchoring heuristic is strongly influenced by a phenomenon called
“confirmation bias”, that is people’s tendency to find a confirmation for their
initial opinions (or hypotheses).
EXAMPLE:
Are the genes of the fruit mosquito, called Drosophila, more or less
than 2?
How many are the genes of the fruit mosquito called Drosophila?
Ce.R.D.
«Anchoring» heuristic
Centro di Ricerca sul
Rischio e la Decisione
The anchoring heuristic is strongly influenced by a phenomenon called
“confirmation bias”, that is people’s tendency to find a confirmation for their
initial opinions (or hypotheses).
EXAMPLE:
Are the genes of the fruit mosquito, called Drosophila, more or less
than 2?
How many are the genes of the fruit mosquito called Drosophila?
Correct answer: 13600 genes
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Anchoring» heuristic
When people are asked to answer the second question reported in the
example...
They start from the hypothesis that the number reported in the first
question corresponds to the correct answer.
Subsequently, they follow a reasoning which tries to confirm such
hypothesis rather than falsify it, therefore making the number highly
influential in their final answer.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Anchoring» heuristic
The more people think and the more they tend to search for information that
confirms their initial hypothesis therefore increasing the anchoring to the
number they used as a starting point.
This is why we talk about anchoring. The final estimate is usually
insufficiently adjusted (or distanced) from the number used as anchor.
This is a huge problem with serious consequences in fields like finance
which rely heavily on numerical estimates about the future performance of
the market.
Unfortunately, experts are prone to anchor their estimates to recent
market performance and this can have huge impact on their
judgments.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Affect» heuristic
In the last two decades many researchers have shown that emotions are used
as an automatic source of information that guides people’s decisions (Damasio,
1994; Loewenstein, Weber, Hsee & Welch, 2001; Slovic, Finucane, Peters &
MacGregor, 2002).
Slovic and collaborators used the label «affect» heuristic to define the
role played by emotions in the decision process.
Emotions influence our behavior because they are associated with:
The objects inducing them.
The mental images of those objects.
Ce.R.D.
«Affect» heuristic
Centro di Ricerca sul
Rischio e la Decisione
In our case the term emotion is used to indicate a series of affective reactions
(and feelings) that are quite weak and automatically associated to stimuli (i.e.,
choice alternatives)...
These reactions are very general and communicate that an alternative
is:
Good versus bad.
Attractive versus awful.
These affective reactions are experienced at the unconscious level and
influence people’s behaviors without them being aware of it:
They provide a feedback about approach/avoidance strategies to
follow in a specific context.
Ce.R.D.
«Affect» heuristic
Centro di Ricerca sul
Rischio e la Decisione
Emotions activate semantic networks that induce individuals’ mind to think
about other concepts and mental images whose emotional content is directly
related to the initial stimulus.
CHOCOLATE
Birthday
Vacation
Cake
Friends
Sacher
Diet
Eastern
Ce.R.D.
«Affect» heuristic
Centro di Ricerca sul
Rischio e la Decisione
Emotions activate semantic networks that induce the individuals’ mind to think
about other concepts and mental images whose emotional content is directly
related to the initial stimulus.
CHOCOLATE
Birthday
Vacation
Cake
Friends
Sacher
Diet
Eastern
Ce.R.D.
«Affect» heuristic
Centro di Ricerca sul
Rischio e la Decisione
Emotions activate semantic networks that induce the individuals’ mind to think
about other concepts and mental images whose emotional content is directly
related to the initial stimulus.
CHOCOLATE
Birthday
Vacation
Cake
Friends
Sacher
Diet
Eastern
Ce.R.D.
«Affect» heuristic
Centro di Ricerca sul
Rischio e la Decisione
Emotions activate semantic networks that induce the individuals’ mind to think
about other concepts and mental images whose emotional content is directly
related to the initial stimulus.
CHOCOLATE
Birthday
Vacation
Cake
Friends
Sacher
Diet
Eastern
Ce.R.D.
Euristica dell’«affettività»
Centro di Ricerca sul
Rischio e la Decisione
Affective reactions have a huge impact on our evaluation of risks and benefits
A
Investment in stocks
B
Investment in stocks
Affective
reaction
+
Inference:
Risk is low
Information:
Benfit is high
C
Affective
reaction
+
Investment in stocks
D
Investment in stocks
Affective
reaction
_
Affective
reaction
_
Information:
Benefit is low
Inference:
Benefit is high
Information:
Risk is low
Inference:
Risk is high
Information:
Risk is high
Inference:
Benefit is low
Ce.R.D.
«Affect» heuristic
Centro di Ricerca sul
Rischio e la Decisione
It is possible to distinguish between two different sources of emotional influence
on behavior (Slovic et al., 2002):
Affect or feelings:
This is a specific quantity of good or bad experienced as an
emotional state (i.e., happiness, anger and so on).
It is induced by external stimuli.
It indicates the positive versus negative quality of a stimulus.
Mood:
It is the psychological state of a person.
It is independent from the quality of a specific stimulus.
It is induced by internal reactions.
It can be generalized to many different stimuli.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Affect» heuristic
Previous studies on the role of emotions have led to the development of socalled dual-process theories that state the existence of two different typology of
thinking systems:
Experiential (intuitive) system
Analytical (conscious) system
1. Holistic
1. Analytic
2. Affective: happiness - pain
2. Logic: Reasoning
3. Associative connections
3. Logical connections
4. Behavior is regulated by reactions
to past experiences
4. Behavior is regulated by
conscious valuations of events
5. Stores reality in images and
concrete metaphors
5. Stores reality in abstract symbols,
words and numbers
6. Faster processing: oriented to
immediate reactions
6. Slower processing: oriented to
delayed action
7. Self-validating: “Try to believe”
7. Requires validation through
reasoning and logical trials
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions
Thanks to repeated experience people learn to automatically associate their
affective reactions to stimuli and can develop automatic (implicit) decision
strategies.
We talk of automatic strategies because they are applied on the basis
of an affective feedback and without the conscious analysis of which is
actually the more convenient action to make.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions
The use of automatic strategies has been demonstrated by a series of studies
run by Bechara, Damasio and their colleagues (1997) in which participants
were presented with four decks of cards («Iowa Gabling Task»).
Unaware to participants:
2 decks were «disadvantageous» (they offered big gains, but also big
losses).
2 decks were «advantageous» (they offered lower gains, but also
lower losses).
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions
Over the long term, while playing, «disadvantageous» decks lead to an overall
loss, whereas «advantageous» decks lead to an overall gain:
Participants start the game with an endowment of €100 and they can
win more money or lose them depending on the cards they draw from
each deck.
Participants must draw one card at the time (for a total of 100
drawings).
At each drawing, participants are acknowledged of their result (i.e.,
«you just won/lost €50).
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions
Iowa Gambling Task
«Disadvantageous»
decks
A
B
Every 10 cards drawn the
balance is negative by - €250
«Advantageous»
decks
C
D
Every 10 cards drawn the
balance is positive by + €250
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions
Participants must draw 100 cards, one at the time.
Usually, after drawing 50 cards most participants start selecting cards
only from the two «advantageous» decks.
However, they are unaware that this is actually the best strategy,
they simply feel it!
After drawing about 80 cards participants are also able to recognize
and explain that there are two decks that are more favorable and two
decks that are less favorable in terms of overall performance.
The researchers recorded «skin conductance» and «heart beat» noting that as
the game advanced people showed increasing stress reactions (higher skin
conductance and faster heart beat) before drawing a card from the
disadvantageous decks.
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions
Ce.R.D.
Centro di Ricerca sul
Rischio e la Decisione
«Implicit learning»:
Experience and affective reactions

Documenti analoghi

Lecture 2

Lecture 2 Because of loss aversion, as well as the tendency to code outcomes in terms of gains and losses, people are more able to make comparative evaluations rather than absolute ones: When choosing betwee...

Dettagli

Libertarian Paternalism

Libertarian Paternalism Often this small incentives have a great effect, despite individuals had already incurred huge losses because of their

Dettagli