Week 11+12+13 (M4) Flashcards
Is judgement solely dependent on formal logic?
No! That is idealized reasoning
Normative vs descriptive
Normative = ideal
Descriptive = reality
What people actually do is FAR from ideal reasoning
Inductive reasoning
+ top-down or bottom-up
Observations to general conclusion/theory
Bottom-up
specific to general
Deductive reasoning
General premise to conclusion about a specific case
Top down
General to specific
Availability heuristic
Judge frequencies of events (probabilities) using availability
- relies on memory
ex. more n or ing words
What impacts availability heuristic
Imageable things
- increase likelihood estimates
- we fear what we perceive as uncontrollable
The Von Restorff effect + what heuristic
aka isolation effect or distinctiveness principle
- memory is better for things that stand out; things that are distinctive, isolated, humorous, bizarre, etc.
availability bias
Anchoring bias + issue
Estimates of frequency - start from a point (ANCHOR) and move up or down
People even use a completely unreliable number as an anchor and go from there
Problem - hard to put aside original estimates - whether right or wrong, and even if subjects
know it is not reliable, will serve as an anchor
Heuristic
quick and easy way to judge something (not always accurate)
Representativeness heuristic + relates to…
- assume homogeneity (that all members of a category are the same) and that each member of a category is representative of that category
stereotypes
Conjunction fallacy + what heuristic
bank teller
aka support theory
People use similarity to a prototypical example, rather than probability, as a basis for judgment
Representativeness
ex. bank teller fem mvmt lady
Sunk cost fallacy
“Throwing good money after bad”
The gambler’s fallacy
+ what heuristic
the belief that prior outcomes can
influence the outcomes of probabilistic
events
“due for a crash”
representativeness heuristic
Law of large numbers
things do tend to even out in the end, proportionally, over a LOT of trials. BUT, this does not extend to small samples
Reasoning from a single case to an entire population
AKA “The man who” argument
My grandma smoked every day but died in her sleep…
- representativeness heuristic issue
- assumption of homogeneity
Hindsight bias
People think after the fact that they would have known something before the fact when they really wouldn’t have. Turning vague statements into solid predictions after the fact.
- thought you were certain in the past
I knew it all along; I told you so; hindsight is 20/20
Hindsight bias - doctor example
The doctors in Group 2, who already knew the diagnosis, were told to IGNORE it, and to just do the task on the basis of the evidence at hand.
Results - Doctors who already knew the diagnosis assigned probabilities to that diagnosis THREE TIMES HIGHER than did the doctors who didn’t know the diagnosis.
Even though the actual diagnosis was the LEAST likely disease, given the symptoms.
The Better-than-average Effect
AKA: Illusory Superiority
People, on average, think that they are better
than average.
– Of course, it is impossible for everyone to be
better than average
– Exception is the clinically depressed
Covariation
Illusions of covariation
People project their own prior theories onto the data - see only the patterns you expect to see.
e.g. interpreting Rorschach tests
Data vs. theory driven covariation interpretations
Data-driven
- when participants
have NO prior
expectations or beliefs
about the things being
judged
- systematic
- conservative
(low estimates
unless correlation is
strong)
Theory-driven
- when participants DO
have prior expectations
or beliefs about the
things being judged
- inaccurate,
variable
- over-estimate
Confirmation bias - when detecting covariation
– People don’t look at all the evidence, just a
subset
– More attentive to evidence that CONFIRMS our
beliefs than to evidence that might falsify them
When detecting covariation issues with consulting your memory schemata
– helps to remember examples that fit your prior belief, don’t remember counter examples – leads
to availability effect for further reasoning
Base rate + diagnositc info
Edgar is a quiet man, who likes to read poetry. Is he more likely to be an English professor or a truck driver?
how DIAGNOSTIC is this? (what proportion
of English professors fit this description, and
what proportion of truck drivers fit this
description)
AND
what is the BASE RATE for English professors
and truck drivers
Base rate neglect
+ what heuristic do we use instead
we tend to use representativeness and ignore the base rate
You need to ask: Out of how many?
ex. the bank teller feminist thing
Bayes theorem base rates
…
Base rates + diagnostic info = which do we pick
If given JUST the base rates, people use them effectively.
If given JUST the diagnostic information, people use it (following categories, stereotypes, heuristics…)
BUT give BOTH base rates and diagnostic information, people tend to ignore the base rates
Dual system hypothesis
System 1 = heuristics
System 2 = less biased but more effortful
Data format
- Framing effects
- Triggering statistical knowledge
Training in statistical reasoning works
What is reasoning
Testing and adjusting our beliefs
Logical vs. realistic reasoning
Logic (ideal)
– Logical reasoning
– Rational thinking
– Utility theory
What people do
– Biases
– Heuristics
– Errors
– Irrational thinking
– Pragmatic reasoning
Issues with adjusting our beliefs
- overconfidence
- belief perseverance
- confirmation bias
Overconfidence
Both experts and novices are more confident in their judgments than is justifiable
Belief perseverance
We don’t want to let go, even in the face of strong evidence
Confirmation bias
people test their hypotheses by choosing tests that confirm them, rather than tests that would disconfirm them
Categorical syllogism
A form of deductive
reasoning consisting of a major premise,
a minor premise, and a conclusion
either valid or invalid
ex. human mortality thing
Conditional syllogisms
- These begin with an “IF – THEN” structure
- If P, then Q
ex. Modus Ponens and Modus Tollens are two
examples of conditional syllogisms
Modus Ponens - valid
affirming the antecedent
Major premise: If antecedent, then consequent
Minor premise: antecedent
Conclusion: consequent
Modus Tollens - valid
denying the consequent
Major premise: If antecedent, then consequent
Minor premise: NOT consequent
Conclusion: NOT antecedent
Are people better at modus ponens or tollens
People are usually pretty good at modus ponens
Much worse at modus tollens – people find it harder to see why the conclusion follows
Inverse error / denying the antecedent
invalid
If P, then Q
NOT P
= NOT Q
“If P” does not say anything about “If NOT P”
Converse error / affirming the consequent
invalid:
If P, then Q
Q
= P
P may not be the only condition for Q
valid:
If AND ONLY IF P, then Q
Q
= P
bidirectional: P implies Q, and Q implies P
Wason’s four card task
Each card has a letter on one side and a number on the
other side. Which cards must be turned over
to check the following rule?…
“If a card has a vowel on one side,
it must have an even number on the other side.”
flip A and 7
Pragmatic reasoning schemata
General schemata rather than “if-then” rules
- Permission
- Cause and effect
Are people good at syllogisms
No they’re terrible
Belief bias
- If people believe the conclusion, they tend to think that the argument is valid.
- If people don’t believe the conclusion, they tend to think that the argument is invalid.
Atmosphere errors
- Some conclusions “seem” more appropriate, given the context.
- If the premises use the word “all”, people are more likely to accept a conclusion that also uses the word “all.”
Conversion errors
People often treat:
All A’s are B as equivalent to…
All B’s are A
People often treat: Some A’s are not B as equivalent to…
Some B’s are not A
Improving logical thinking
- Formal training in logic does not help much
- Training in the use of reasoning schemata does help – triggers a pragmatic reasoning schema
Utility Theory
- What do you value?
- How does each option
move you towards what you want? - Every choice has costs and benefits
Maybe just pick the option that minimizes costs and maximizes benefits
Utility theory - issues
Sometimes costs and benefits are really difficult to compare
ex. good porch vs. closer to bus line
Uncertainty
ex. possible that housemates might add another roommate next year
Subjective utility theory
Figure out the SUBJECTIVE UTILITY of each factor
- What value does it have for YOU
- Subjective utility can also be negative (“disutility”)
To make a decision, sum the subjective utilities of each choice. Higher utility score wins.
Utility theory calculation
Expected utility = (probability of outcome) x (utility of outcome)
probability times value
The certainty effect
UT
issue with “ideal” utility theory
“A bird in the hand is worth two in the bush”
People prefer sure gains
aren’t “utility maximizers”
Non-transitive relationships
non-transitive relationships can be worked on even though they are irrational according to UT
Mental accounting
Would you drive 30 minutes across town to save $20 on a $10,000 car?
Would you drive 30 minutes across town to save $20 on a $100 vacuum cleaner?
Psychological utility
isn’t the same as dollar value
e.g. the psychological difference between $0 and $10,000 is much larger than between $1,000,000 and $1,010,000
Framing effects
If dealing with possible losses, decision makers are RISK SEEKING - maybe you could avoid the loss
If dealing with possible gains, decision makers are RISK AVERSE - hold on to what you have (c.f. certainty effect)
Prospect theory
It’s more about value assigned to gains and losses
We are less willing to gamble with profits than losses
Choice varies by framing the problem as a gain or loss
Loss aversion
Given a choice between avoiding loss and acquiring gains: Strongly prefer to avoid losses
Sunk cost reasoning + UT
Don’t want to give up something you already own
According to Utility theory, it makes no sense to pay attention to sunk costs
Rational vs. justifiable
Rational = max. utility
Justifiable = reason-based choice
Satisficing rather than optimizing
make a choice that is good enough rather than the optimal decision
optimal decision may require more time or more effort than the decision is actually worth
Regret theory
avoiding regret is a big motivation (might contribute to the certainty effect)
Improvement over time
We like things to get better over time
ex. hand in cold water thing
Prisoner’s dilemma + game theory vs. reality
Game theory (Nash Equilibrium): defecting is dominant
strategy because it has a slightly higher payoff.
Overall, the best outcome is mutual cooperation.
People show a systematic bias toward cooperation
Tragedy of the commons
There is a community of shepherds whose sheep graze on a common ground. If each herder restrains the number of sheep feeding on the commons, then the commons will not become exhausted from overgrazing.
Rational choice: compete for the resource
Better than rational choice: cooperation to conserve the resource
Problem solving
- You are not always presented with choice between options
- You frequently have to CREATE those options yourself
Problem solving as a search
- general problem solving method
- search through a “problem space”
ex. like a maze
Tower of Hanoi
number of steps 2^n - 1
n = number of disks
Operations vs. problem space
operations: the relevant things you can do
problem space: the SET of all the relevant things you can do
4 aspects of general problem solving methods
- Initial state:
– knowledge and resources you already have
– The “givens” - Goal state:
– Where do you want to end up? - Operators:
– tools and actions that can change your current state
– Means of transforming conditions - Path constraints:
– limitations on what moves you can make
– obstacles
Brute force solutions
Lay out the entire problem space
- go through each possible solution
takes way too much time
Problem solving strategies (6)
- hill-climbing
- means-end analysis
- working backwards
- building mental models
- representation
- analogy
Hill climbing heuristic + problem
always go uphill
- take one step closer to your goal
problem: sometimes solving problems requires that you take steps away from your goal
Means-end analysis
What is the difference between my current state and my goal?
how to reduce that distance
breaks the problem into more easily solvable sub-problems
involves both forward and backward search
Working backwards
start at a goal, work backwards from there
ex. water lillies covering lake
Generally, working backwards is helpful when the number of directions backward from the goal state is small, and the number of possible directions from the initial state forward is very large.
It can also provide a “new” perspective when you are stuck.
Mental models and imagery
Sometimes, solving a problem requires finding a different REPRESENTATION of the problem
- can use mental models to help reason out categorical syllogisms
(picture it)
Representation - the problem space
- Solving a problem requires problem representation and problem execution
The wrong representation can interfere with analysis:
- Incomplete information
- Computational complexity
- Somethings easy to solve in one format are difficult to solve in another
Analogy + issues
- Ability to draw analogies is a central intellectual tool
Challenges:
- Spontaneous, uninstructed use of analogies seems to be quite rare
- A suitable analogy isn’t always obvious
- Sometimes you need to find one yourself = difficult
- The analogy has to connect with the deep structure of the problem, not just surface properties
ex. tumor
Expertise
- Memory advantage leads to problem solving advantage
- Being able to see the larger structure also helps problem solving
- Experts tend to think about the deep structure of problems more than novices do
- Experts are better at knowing what information is irrelevant - not following so many blind alleys
- Experts are also good at CHUNKING into meaningful bits
Experts have…
MORE information
DIFFERENT information
- Knowledge of patterns (ex. Symptoms all connected to one disease), etc.
CROSS-REFERENCED information
- More connections linking information in memory
AUTOMATIZED procedures
- Autopilot frees up other mental resources
Limits of expertise
- Experts sometimes don’t remember details of problems very well
- Understood for gist, forgot the rest
- More likely to make INTRUSION errors (assumptions based on prior knowledge)
- Experts are no better than novices outside of their domain
Functional fixedness
Stuck or rigid in one way of thinking about an object, and what its function is
Ex. tack, match, candle task
Problem solving set
- Most problem spaces are huge - larger than we even realise
- Don’t realise how huge they are because of the heuristics and problem solving sets we use
- There are costs and benefits to problem solving sets
Einstellung
“attitude” = general term for rigidity in problem solving
based on starting assumptions about a problem (helpful or harmful)
Four elements of creative performance
- Preparation
- Incubation
- Illumination/insight
- Verification
AHA moment
- Might still be wrong
- Even though you think you’re getting closer and closer
What makes an idea or a person “creative”?
Divergent thinking
- Think of ideas that are not normally associated
Finding new connections between ideas
- Similar to divergent thinking
- Remote associations task …
Underlying idea - creativity is based on coming up
with new, unanticipated approaches to a problem