mem and cog exam 3 Flashcards
How did scholars’ views of human rationality change during the 20th century?
we started to see human thinking as less rational than we had previously thought and as being influenced by biases and heuristics.
Give an example of the influence of availability on decision making
After all of the plane crashes on the news recently, someone may choose to drive to florida instead since the memories of plane crashes are more available to them than car crashes and thus seem more likely to happen.
How is “Harvard beats Yale, 29–29” an example of the anchoring heuristic?
Harvard was expected to lose and only in the last few seconds of the game tied with yale, which made them feel as if they had won when contrasted with their expectations.
How do factors involved in gathering evidence contribute to irrationality
our memories could be biased, and we’re more likely to rely on anecdotal information than statistical information
How are the representativeness heuristic and confirmation or myside bias examples
of irrationality?
We might go off of characteristics that we think are representative of a certain quality, but this is usually misguided (clinicians are more likely to make diagnoses based on how similar the patient is to a prototypical one rather than using DSM criteria)
We might seek out evidence that supports our predetermined ideas. And we are likely to weigh this information as more valuable than information that goes against our beliefs
How might one overcome confirmation bias?
Taking the perspective of how the other side is thinking
What is the problem with the reasoning from the “law of small numbers”
you may only have a few examples of something and then generalize to everyone. (i’ve taken two psych classes and both of the professors sucked therefore the department must suck)
Give positive and negative examples of people’s use of base rates
a doctor might show more concern for an elderly person vs a young person with heart issues
a doctor may be more likely to show concern for a man rather than a woman in having heart disease because he believes it’s more common in men, even though the prevalence is equal between genders.
Does the conjunction fallacy affect only statistically na ̈ıve students?
(believing the probability of event A and event B representative of the person is more likely than just event A happening) no even if they understood probability they still experienced it
How does framing affect people’s tendency to avoid risk?
people tend to be risk-averse in positive frames and risk-seeking in negative frames
deduction
one assumes first that certain pieces of information (premises) are true and then seeks to determine what conclusions follow from those premises
types of deductive reasoning
syllogistic and conditional
syllogism
a set of sentences that serves as the basis for reasoning. first two are the premises and the third is the conclusion.
conditional reasoning
If A (antecedent), then B (consequent)
modus ponens (MP)
method of affirming the consequent
modus tollens (MT)
method of negating or denying the consequent
are people more apt to accept MP or MT
MP
What did the cigarette syllogism study show?
the valid premises, believable consequent & invalid premises, believable consequent both had twice as much acceptance (92%) as the valid premises, unbelievable consequent. demonstrates belief-bias effect.
belief bias effect
when an invalid conclusion in a syllogism is supported by our beliefs, we are more likely to believe it is valid. belief bias influences judgements of invalid arguments more than valid ones
content effects
in syllogistic reasoning, belief bias makes people more likely to accept a conclusion that supports their pre-existing beliefs even when it does not follow from the premises
in conditional reasoning, determining the validity or invalidity of a conditional statement can be dramatically altered by the content of in which the problem is presented
formal rules theory
assumes that each of us possess a mental logic (internal set of abstract rules and a set of processes for using them)
- If the formal rules theory is correct then people should be
equally good on tasks with equivalent logical structure. - Content effect: People are not equally good on such tasks.
- Therefore, the formal rules theory is not correct.
what does deductive validity depend on?
only the template of the argument (syntactics), NOT the specific content (semantics)
memory cuing theory
Hypothesis: Each person has a large collection
of specific scenarios stored in memory.
* This theory has the opposite problem to the
formal rules theory. The scenarios are too
specific and cannot easily account for the
human ability to reason about situations that
are not familiar to them.
pragmatic reasoning schemas
We use pragmatic reasoning schemas instead of
episodic memory or formal rules of logic, for some
classes of learned situations:
* Schemas for permission, obligation, causality,
if you want to watch tv you have to clean first
Evidence for Pragmatic
Reasoning Schemas
Abstract Wason four-card selection task is
harder than “concrete” selection task
because abstract form does not fit any
pragmatic reasoning schema
* Training examples from familiar, concrete
schemas improved abstract selection task
performance more than training in abstract,
formal logic
When is an argument good? inductive vs deductive
- An argument is deductively valid if and only if
it is impossible that its conclusion is false while
its premises are true. - An argument is inductively strong if and only if
it is improbable that its conclusion is false
while its premises are true. The degree of
inductive strength depends on how
improbable it is that the conclusion is false
while the premises are true
relative frequency
To represent a probability
of A/B, draw the reference class
as a collection of B squares.
* Then single out A out of them.
conditional probability
If a certain statement, p, is known to be true,
this may affect the probability of another
statement, q.
* We write P(q|p) and read
“probability of q given p”.
* Example: P(star | red) denotes
the probability of a star given
the underlying square is red
P(q|p) =
P(q&p) / P(p)
base rate neglect phenomenon
a cognitive bias where individuals tend to ignore or underestimate statistical information (base rates) in favor of specific, often vivid, case-specific information when making judgments or decisions
Base rate neglect tends to exacerbate the tendency
to confuse P(q|p) and P(p|q).
Stanovich vs Gigerenzer views on human irrationality
Stanovich: Why are we irrational?
* Random errors
* Lack of cognitive capacity
* Lack of domain-specific
knowledge
* Heuristics and biases are
cognitive illusions
* People can be trained to
be more rational
* The glass is half empty
Gigerenzer:
Human reasoning doesn’t have to
be optimal, it just have to be
“good enough”
* Heuristics and biases are adaptive
* Fast and frugal heuristics: simple
strategies sometimes can make
decisions as well as complex
statistical models
* The glass is half full
Dysrationalia
Inability to think rationally
despite possessing adequate intelligence
* Rational thinking not entirely measured by I.Q.
tests (Stanovich, 1993, 2009, 2000: See Table 11.7!)
* Failures in cognitive processing (cognitive
miser) and content (knowledge of probability,
logic, and scientific inference)
bounded rationality
Decision making is constrained
by limited information, cognitive
ability, and time
* Satisficing (satisfy + suffice):
pick a strategy that meets our
standards for an adequate (not
optimal) solution
definition of a problem
- Present (or initial) state – not satisfactory
- Goal state – more satisfactory
- Operators (or moves) – change one state into a
different state - Constraints – restrictions, rules, budgets…
- Solution – a sequence of valid moves from the initial
state to the goal
well vs ill informed problems
- Well-defined problem: the initial state, goal
state, possible operators, and constraints
are all known: - Example: Chess board
- Ill-defined problem: one or more
components are not specified at the onset.
Part of the problem is to figure out the
missing elements: - Example: Mission to Mars
weak–strong continuum of problem solving
- Weak (domain-general) methods:
- Apply to broad classes of problems. E.g., search
- Because of their generality, weak methods do not benefit
from regularities specific to a given domain. - Strong (domain-specific) methods:
- Incorporate expertise relevant to a circumscribed class of
problems within a given domain, e.g. chess. - Pattern recognition, intuition, role of practice…
- Continuum – the more domain-specific knowledge,
the stronger the method
algorithms and heuristics
The world “algorithm” is used in two senses.
* Algorithm 1 = any information-processing procedure.
Includes algorithms 2 and heuristics.
* Algorithm 2 = an algorithm 1 that is guaranteed to
produce the correct solution to a problem of a given
class. Example: last lecture’s “recipe” for calculating
P(q|p) using frequency trees.
* Heuristic = a rule of thumb. Still an algorithm 1, but
does not guarantee a solution in all cases. It merely
facilitates the solution is some cases
