Week 10: Reasoning Flashcards
Reasoning and Decision Making
Judgement:
estimating magnitudes/probabilities of some characteristic
Reasoning: drawing conclusions
Decisions: making choices between alternatives
Forming an opinion on a particular thing
Finding magnitude of certain characteristics
Taking available evidence that we have and concluding
Looking at available alternatives and choosing among them and acting in that certain way
Conclusions help us make decisions
Deductive Reasoning
Starts broad principles to make logical predictions about specific cases
Determining whether a conclusion logically follows from premises
Syllogistic reasoning
Two statements called premises
Third statement called conclusion
Categorical syllogism
Describe relation between two categories using. all, no, or some
Do not confuse “validity” with “truth”
Syllogism is valid if conclusion follows logically from its two premises
Truth refers to a relationship that holds in reality
Syllogisms can have validity but not the truth, and can have truth but not validity
Deductive
Situations where we start with broad principles and try to make a logical prediction about specific case
General to specific
Remember, don’t confuse validity with truth
Trying to determine if a syllogism is valid
Truth is what we usually think about
Categorical Syllogism
All men are mortal. (major premise)
Socrates is a man. (minor premise)
(therefore)
Socrates is mortal. (conclusion)
Logic and Eular
Assess syllogistic logic
Have category A represented by a circle, show a relationship with A and B by circle overlap
All A are B
No A are B
Some A are not B, Some A are B
Goal isn’t to find a way that sylllogism is true, but false
False makes it invalid
Abstract
All A are B (major)
All C are B (minor) therefore, All A are C (conclusion)
All A are not C in the diagram
True or False (valid)?
Concrete
All Liberals are human. (major premise)
All Conservatives are human. (minor premise)
(therefore)
All Liberals are Conservatives. (conclusion)
True or False (valid)?
Misinterpretation: All A are B = All B are A
Biggest error people make is that they feel like they can rearrange the equation
You can’t just flip it around
Form vs. Content
All dogs are animals. (major premise)
Some animals are pets. (minor premise)
(therefore)
Some dogs are pets. (conclusion)
True or False (valid)?
Dog category is independent from pet circle so syllogen is false
Don’t focus on content, focus on form of diagram
Using sharks make it easier to conceptualize
Belief bias: if content pushes you towards one interpretation, makes it hard to evaluate it logically
Reasoning Error
Belief bias (Henle, 1962): The tendency to think that a syllogism is valid if its conclusions are believable
Therefore, beware of syllogism’s conclusions that are true or agree with your beliefs
All of the students are tired
Some tired people are irritable
Some of the students are irritable
Atmosphere effect (Woodsworth & Sells, 1935): global impression or feel of the premises (e.g., same form)
All + All = All ?
All birds are mammals
All mammals sleep
All mammals are birds
Confirmation bias: the tendency to selectively look for information that conforms to our hypothesis and overlook information that argues against it
Reasoning Error
Some of the students are irritable is incorrect based on diagram
Atmosphere effect: idea that if the premises give a general feeling of particular form so that you have all As or Bs or Cs, that sort of gives the impression that it must be true
All + all = all
Since all mammals are not birds
Tendency that we have to prove that something is true due to expectation
What we want to do in logic and generally approaching things and we try to find ways that something is untrue
Conditional Syllogisms
A logical determination of whether the evidence supports, refutes, or is irrelevant to the stated relationship
Conditional clause (premise 1) – antecedent (if clause)
consequent (then clause)
Evidence (premise 2)
Conclusion?
Psyc 221
Psyc 221
First premise is the (if then) part
Second premise is the evidence support, refutes or is irrelevant to the first premise
Then you have to determine if you can reach some type of conclusion
Modus Ponens (97%)
Modus Tollens (60%)
The only time that you can come up with a valid conclusions is if you are affirming antiselent or denying the consequent
Example
Raining (antecedent)
Domique gets wet (consequent)
Evidence: it is raining
Conclusion: therefore, dominque gets wet (modus ponens)
Therefore its not raining (modus tollens)
Tricky since someone could just dump a bucket of water on her
Example 2
If i live in vancity
Then i live in BC
Evidence: i live in vancity
Conclusion: therefore, i live in BC
Or i do not live in BC, therefore, i do not live in vancouver
Or i live in BC, therefore, i do not live in vancouver
I dont live in vancouver, but i do live in BC
Wason Four Card Problem (1966)
To determine which cards needed to be flipped over to test the rule
One side had number and other had letter
If vowel then it was an even number
7 would go against the rule
E and 7 would need to be flipped
People have the most difficulty with this
Becomes easier when stated in terms that you can actually think of
Concrete terms will help
Agrees with the schema, a permission schema which agrees with the rules of society
Drinking is allowed to certain individuals in society
16 yr old would invalidate the question, drinking beer would also
Falsification principle:to test a rule, you must look for situations that falsify the rule
* Most participants fail to do this
* When problem is stated in concrete everyday terms, correct responses greatly increase
Wason Four Card Problem (1966)
To determine which cards needed to be flipped over to test the rule
One side had number and other had letter
If vowel then it was an even number
7 would go against the rule
E and 7 would need to be flipped
People have the most difficulty with this
Becomes easier when stated in terms that you can actually think of
Concrete terms will help
Agrees with the schema, a permission schema which agrees with the rules of society
Drinking is allowed to certain individuals in society
16 yr old would invalidate the question, drinking beer would also
Falsification principle:to test a rule, you must look for situations that falsify the rule
* Most participants fail to do this
* When problem is stated in concrete everyday terms, correct responses greatly increase
Permission schema: if A is satisfied, B can be carried out
* Used in the concrete versions * People are familiar with rules
Inductive Reasoning
Reasoning that is based on observation
* Generalizing from observations of specific cases to more general conclusion
* Reaching conclusions from evidence
* Conclusions are suggested only, with varying
degrees of certainty.
* Strength of argument
* Representativeness of observations * Number of observations
* Quality of evidence
Used to make scientific discoveries
* Hypotheses and general conclusions
* Usedineverydaylife
* Make a prediction about what will happen based on observation about what has happened in the past
Inductive Reasoning
This is the opposite of deductive
Particular observation and determining the general conclusion
Based on evidence
We are trying to figure out what is being suggested
We still want as much certainty as possible and there are things that will strengthen the argument
You want a representatives sample
E.g., all crows are black in particular city
But if you only have the crows in one city, you cannot generalize all crows to the ones you have seen
Need higher amount of observations, the stronger the conclusion will be
Is there multiple sources that point to your conclusion
Such as articles with similar observations
Essentially what is used in science
Constantly looking at certain situations, taking these data points and taking hypothesis and determining general conclusions
Everyday, you have these expectations in the world that things will behave the way you expect, these are the inductions
Accounts of Reasoning/Decision Making
Normative: how things ought to go; what people should do
Descriptive: how things are; what people actually do (do)
Heuristic: a short cut; a strategy that risks error to gain efficiency (speed)
Algorithm: a guaranteed route to an outcome, which may be more tedious and effortful
Normative is how things should work in society
Descriptive is what people actually do
Heuristic is a short cut that allows us to reach a decision faster
Algorithm is sitting down and actually thinking about what we actually need but is very tedious
Kahneman and Tversky
pioneered the research on judgment under uncertainty
* emphasized the heuristics we use to process evidence and make judgments
* early work done at Hebrew University of Jerusalem
This work started in the 70s at University of Jearusalm
Proposed theory about decision making called prospect theory
Largely done in collab with Tversky
We typically rely on heuristics since we can’t get all the information and even if we could, it would take forever
Representative Heuristic
assume that each member of a category is representative of that category (like the prototype?), and has all its traits
* also assume the reverse—if something has a lot of the traits of a category, it probably belongs to that category
* be willing to draw conclusions from a quite small sample: “seen one, seen ‘em all”
Representative Heuristic
Argued that when we think about members of categories, we make the assumption that every member is like everyone else in the group
We also make the opposite as if something has alot of the traits of the category, it may be apart of it
If you have a situation where you are seeing something, you have a lot of info of that object since you are assuming that objects category therefore you know all the information about that item
Coin flip example: coin is random, assuming that this is random event and there are 2 possibilities, when things are random, we expect them to change
If you flip a coin 8 times, any of these 3 possibilities are equal due to chance
Each throw is independent
Gamblers Fallacy: feel like random events, the next thing should be a particular outcome
E.g., flipping a coin 4 times and 3 times it landed on heads, bet on last one, most will bet for a tails
People will think tails is more likely even though this isn’t true
Kahneman and Tversky (1973)
Talked to a bunch of students as samples in different groups
1 group: given a bunch of majors that they may be pursuing, what probability will students be in each category
Were given a prompt about a person
Asked questions about his similarity to other grads, likely a grad student, and a personality sketch of highschool and likelihood of being a student
More likely to be seen as comp sci or engineering student, likelihood highest as that as well
This proved that people were judging, people assumed his characteristics to people who were probably inclined with those majors
Even happened in the final group, even though things may have changed, people still said that he was more likely to be a comp sci/engineering student since he was similar to them
His test carried a lot of weight into the future
E.g., linda being judged as a feminist accountant doesn’t make sense since all accountants have the same category
This is the conjunction fallacy: 60% of being accountant 20% feminist, 12% of being feminist accountant
Probaility of conjoinment of 2 events is the multiplication of the two probabilities
Hospital birthrates: probabilities of which gender is born more
Large vs. small
10 births vs. 100 births
Any given day, fluctuations can happen so 10 is better and much more likely
Kahneman and Tversky (1973)
Likelihood judgments closely mirrored the similarity judgments rather than base rate
* How likely is it that Tom is a comp. sci. student versus humanities student?
* 95% pick comp. sci.
* base rate estimations by other subjects had humanities as 3 times more likely
* Even when based on projective tests that are unreliable, and things change from high school to grad. school
English
More r words or words with r as third letter?
Twice as many words with third r
Why do people think first one is right?
Fairly easy to generate words with r and harder to generate words with third r
Hence why we use availability heuristic
Availability Heuristic
scan quickly through memory seeking relevant instances
* if instances come quickly to mind, they are likely to be frequent in experience
* availability or accessibility in memory serves as a proxy for frequency in experience
* but the organization of memory can create biases
Scan through memory to see how often this particular event has happened
Use as proxy to say it must be more like;y
The more likely something is rehearsed, better chance of being in LTM
Salience, emotionality of an event
Organize in particular way
Slovic, Fischhoff & Lichtenstein (1976)
Subjects asked to estimate frequency of various causes of death.
Homicide Drowning Asthma Asthma Appendicitis Appendicitis Auto-train crash Botulism Tornado Pregnancy
Students in the US
Gave people a bunch of potential causes of death and asked to estimate of how many people die
Per 200 million US
First 2: hugely overestimated from tornado
Last 2: underestimated probability
Due to not hearing it as often
Tornado being on news, boy dying from asthma is not on news
These come to mind more, so we overestimate them
Hindsight Bias
hindsight is 20/20”
* the “knew it all along” effect
* looking at a situation retrospectively (after the fact), we saw all the signs leading up to this particular outcome
– going on 3rd (or 4th) and 1 in football
– knowing the outcome of an experiment before it is conducted
– knowing a relationship would fall apart
20/20
Idea that you know the outcome of the situation, it is easy to see the outcome occurring
E.g., football fan, team going 3rd and 1 in football
Before the play, it would be 50/50
If they mess it up, they were stupid
When people tell you the results of the experiment, not uncommon that people will say “obviously”
Bystander effect: diffusion of responsibility
Knowing that a relationship will fall apart, as saying “i could see that coming”
Anchoring
people are influenced toward a possible anchor value, even if they should know better
* Prelec & Ariely (2006) – MIT students bid on auction items (told price was last 2 digits of Soc. Sec. #)
– 80-99 paid $26, 00-19 paid $9 for trackball
* Annoying sounds – paid either 90 or 10 cents
– Would take 73 or 33 cents to hear it again
People get influenced into believing something
Prelec and Areily:
Bunch of items for sale and said that cost was the last 2 digits of SS number
Could be $1 or $90
Asked what the max price would be for item
Willing to pay 3x more for same item
Only difference was where they started
Paid users either 90 or 10 cents
Asked if they were willing to do it again
10 cent group was willing to do it 33 cents
Class Example
If given 65, say 45
If given 10, say 25
Actual answer was 54/56 so 96%
1 - 8 multiplication -> 512
8 - 1 multiplication -> 2250
This is the anchoring effect: actual answer is over 40k
Decision Making
Reasoning/choosing among alternatives * What clothes to wear
* Who to date/marry
* What school to attend
* Even relatively simple decisions are cognitively demanding
* Donders – ~100ms to decide left of right * Posner & Boies(1971)
We do it all the time
When you have these decisions, a lot of them feel really difficult
You have to think about them to make them
Posner & Boies (1971)
- Used a dual task technique
- secondary task measures resource demands orsparecapacity(likeathermometer)ofthe primary task at various points
- primary task = letter matching (Aa, AA, AB)
- secondary task = turn off tone on some trials; short RT = low resources in primary task
Duel task technique
Two things at the same time
Idea that the secondary task will basically let you know how resource demanding the first source was
How quickly they could do secondary task
If quickly, means they had a lot of resources
If resources were being allocated to resource task, it would take longer and cause errors
Measured how long it took to turn off tone
Capacity Decision
When first letter came up and tone came up same time, people were able to turn it off faster
If it was inbetweem, then rt went up
At second letter, got slower
At key, rt was high but not as high
Process Account
Task
1. wait
2. warning
3. first letter
4. delay
5. second letter
Process
1. none?
2. alerting
3. LTM retrieval
4. WM rehearsal
5. a) LTM retrieval
b) trace matching
c) decision (response choice)
d) motor output
Prior to trial, we don’t know whats going on
Triggers system to be ready to respond
LTM retrieval is automatic
The delay is due to WM rehearsal
Even if you find a match, you still keep on going because it would take longer to make a decision to every letter rather than making one decision at the end
Decision Making
each decision has costs (taking us farther from our goals) and benefits (moving us toward our goals; providing our values)
* we must weight the costs against the benefits
* goal of making decision to maximize utility
Every particular situation, making one decision will have a cost associated with it
This decision will also have benefits though
People might do a cost-benefit analysis
Expected Utility
expected utility = (Probability of an outcome) X (Utility of that outcome)
* Assumption: People are rational
* If they have all relevant information, they will make a decision that results in the maximum expected utility
Evaluation of risks
What you will get out of a situation is going to be the product of how likely a particular outcome is X how much that outcome means to you
Requires us to have good estimates of these concepts so you have the relevant information about probability of outcome and how valuable that outcome is
Advertising (likelood) for lottery tickets
Office pools for lottery tickets, talkin to others
Serotonin release when thinking about if you won
(these are the utility), you think about how good the situation is
Expected utility goes up, likelihood goes down, makes seem like a better deal than it is
Lottery Example
lottery ticket = $2
* likelihood of winning = 1 out of 14,000,000
* prize = $4,000,000
* expected utility = (1/14M) X ($4M) = $0.29
* therefore, you are giving away 7 times what
you can expect to receive each time!
Decision Making
Decisions depend on how choices are presented
* Opt-in procedure
* Active step to be organ donor
* Opt-out procedure
* Organ donor unless request not to be
* Status quo bias
* The tendency to do nothing when faced
with making a decision * Framing of alternatives
Depend on how choices are presented
Organ donation - various ways that this can be put towards people
Filling out form to opt in for donation
Some countries are fixed organ donors unless requested not to
Status quo bias: we have a bias to do nothing
Companies keep giving deals as you won’t leave since you are already apart of the company and you have to do work to switch
Other companies will give other deals to entice you to join them
Framing
Assume yourself richer by $300 than you are today. You must choose between
– a sure gain of $100
– a 50% chance to gain $200 or gain nothing
* Assume yourself richer by $500 than you are today. You must choose between
– a sure loss of $100
– a 50% chance to lose nothing or lose $200
People choose different things according to how they are phrased
If you are going to be loosing something, you will throw the dice and hope that you will not loose much
As you are gaining something, people don’t wanna loose what they are gaining
More Framing
Imagine that Canada is preparing for the outbreak of a new disease, which is expected to kill 600 people
* Two alternatives have been suggested in labwork
A: 200 people saved
B: 1/3 probability that all 600 people will be saved, but a 2/3 probability that none of the 600 will be saved
Contrast this with these alternatives
A: 400 will die
B: 1/3 probability that no one will die but a 2/3 that all of the 600 will die
Framing
phrasing of the decision affects our choice * framing in terms of losses tends to make
us risk-seeking
* framing in terms of gains makes us risk- averse
Decision Making
Emotions affect decisions
* Expected emotions
* Emotions that people predict that they will feel concerning an outcome
* Explain risk aversion
* People inaccurately predict their emotions
* Overestimate negative emotions
Emotions affect decisions
People are not good at predicting on how they will feel about a situation
Kermer and colleagues
Gave people $5 and flipped coin, 50% you will gain 5 and 50% you will loose 3
First tell how you will feel if you win or lose
Lose, down by 4 points, win, go up 2 points (happiness)
Prediction is twice is bad of negative feeling if i lose
After filler test, people who lost were down by a point and people who won went up a point
So winning does make you happy but the difference seems to be about the same
Not what people estimated
Emotions
Incidental emotions: Emotions that are not specifically related to decision-making
* May be related to one’s general disposition or personality, some recent experience, or one’s general environment or surroundings
* Can affect one’s overall decision making processes
Emotions that are from other factors can effect DM
Study done and simple task was given
A highlighter, how much are you willing to sell it for
Sad, disgust and neutral
How much are you willing to buy it for
Neutral: buy low, sell high
Disgust: buy low, sell low (just wanna stay constant)
Sad: buy high, sell low (just want change)
Unconscious Bias
Prejudice
* Implicit Association Test (IAT)
Response to pictures
White or black person
Or bad or good
Face and words
Idea is that there is an unconscious assocaitation of the words and face
Green, Carney, Pallin, Ngo, Raymond, Iezzoni & Banaji (2007)
Measured explicit attitudes
* Measure IAT
* Read a vignette about Mr. Thompson a 50- year-old male presenting to the emergency department with chest pain and an EKG suggestive of anterior myocardial infarction. It is stated that primary angioplasty is not an option and no absolute contraindications to thrombolysis are evident.
Study in hospitals in Boston, went to medical residents
Questionaire on their explicit attitudes on black and white people
Given a prompt
Results
No explicit bias reported
* Diagnosis of Coronary Artery Disease:
Very likely - 29.8% white vs. 40.1% black
* Were not anymore likely to treat blacks with Thrombolysis
No difference in explicit bias
Low on IAT, they were more likely prescribe thrombolysis to black people
High on IAT, more likely to give to white people
Predominant patients were black but still showed bias
Context
Number of response options
* Redelmeirer & Shafir (1995) – found that physicians were less likely (53%) to prescribe arthritis meds when there were two medication options than when there was only one (72%)
Simonson & Tversky (1992) – more likely to buy $240 camera when more expensive option added ($170, $240, $470 instead of just $170 and $240)
Past History: Shen et al. (2010) – physician cesarean section decision affected by seriousness of preceding cases
Context
Number of options
Had to decide which medication to give so did not give any
50% of people bought more expensive camera at 2 options
People chose $240 with 3 options
More likely to buy expensive option when more expensive options are present
Probability that physician ordered a c-section if beforehand cases were easy, opposite caused c section went down
1600 more people died after 9/11 from driving more often
Rationalization (Tversky & Shafir, 1992)
You can buy a highly desired vacation package after passing an exam. Do you do it?
*Most of us (~60%) will because we’re celebrating
*But most of us (~60%) will also buy it after failing
—consolation
*However,if we don’t know how we did on the exam, only about 33% of us will buy it
*Pass/fail information isn’t relevant,but having some information so that we can justify it is obviously important
Passing or failing exam would result in 60% of taking vacation for different reasons though (pass/fail)
Only 33% if exam results were unknown
People were able to justify themselves when more info was given
Sanfey et al. (2003)
UltimatumGame
* Proposer decides how to split money and responder decides whether to accept
*From a utility stand point should always accept
*Responder often rejects low offers because they became angry that offers were unfair
*Does not happen when dealing with a computer
Sandy
Game where situation is essentially one person is given 10$ and decide how they want to split it up between others
Participants were the receiver so they were asked to chose money or reject
As long as the person is offering something, you should take it
Found when they thought they were playing with another person, wouldn’t accept the deal unless it was 3 dollars or more
35% or less accepted 1 dollar offer when they were dealing with another person
If it was computer, then it was nearly 70%
They thought it was unfair
Willing to sacrifice own benefit to punish others, but computer cannot be punished so there is no effect
fMRI was used
Prefrontal cortex was also involved with decision making process
Activation in right anterior insula correlated with likelihood of rejecting offer
Need to justify is just as important as utility
If we are put in situation, we would consider the best solution
People wouldn’t accept a bribe as it goes against their morals/nature
When others ask for your decisions, you should be able to justify them
How Do We Decide?
utility is not the only goal, or even necessarily the most important one
Social and emotional factors are also important considerations
need for reasonable choices, with integrity
justification is critical—need to make sensible and defensible decisions
no accepted normative theory
Two System (Processes)?
System 1:
– Intuitive
– Fast
– Unconscious
– Automatic
System 2:
– Reflective
– Slow
– Conscious
– Controlled
