How we know what isn’t so – cognitive factors in false beliefs
description
Transcript of How we know what isn’t so – cognitive factors in false beliefs
How we know what isn’t so – cognitive factors in false beliefs
Dr. Jeff Saunders
Dept of Psychology
Hong Kong University
False beliefs
Topic: cognitive factors in false beliefs How we acquire false beliefs?
What cognitive tendencies lead us acquire mistaken beliefs about the world?
Why do false beliefs persist? What tendencies inhibit us from learning from
experience and correcting false beliefs
Demo: try to outwit the professor in a card guessing game
Card game strategy?
‘Strategy’: choices were entirely random! But might have perceived that that there
was some causal strategy behind choices No way that opponent could have done
better or worse than chance But might have perceived that there was
some strategy that would have increased the likelihood of winning
Seeing structure in randomness
One factor in acquiring false beliefs: tendency to see structure in randomness
We are very good at seeing meaningful patterns in the world
Useful and impressive ability! But so pervasive and automatic... can
sometimes lead us astray
Clustering illusion
Distribution of bombs in London during WWII Was Regent’s Park targeted? No, just random
Clustering illusion
These are randomly generated points Our minds readily see structure in random
Home remedy: radishes!
Your friend’s remedy for common cold: eat large amounts of radishes
One day when you are really sick, the friend brings you a bunch of radishes to eat
You think: oh well, might as well try it Next day: you are feeling a lot better! Wow, do radishes really work? Or … could this just be coincidence?
(sounds unlikely to you)
Health varies over time, semi-randomly, and most illnesses get better on their own
Seek treatment when health at low points Likely to get better regardless of treatment Misinterpretation of spontaneous recovery
could lead to false beliefs about remedies
Spontaneous recovery
Summary - misperceived causation
We are very good at seeing patterns in the world and generating causal hypotheses
But this tendency can lead us to mistakenly see causal relations in randomness
Clustering illusion Misinterpretation of streaks or clusters that
occur naturally by chance Regression fallacy
Misinterpretation of regression to the mean
Why do erroneous beliefs persist?
Suppose we have an false belief Due to clustering, regression, coincidence ...
In principle, exposure to counterevidence might allow one to correct the error
However, some cognitive factors interfere with correction of false beliefs…
Neglect missing information…
Example: selection criteria
A manager at a company has unusual method for choosing who to hire
Top 8 candidates compete in chess tournament, winner gets the job
Manager: “Everyone I’ve hired with this method has been excellent!”
Example: selection criteria
Hiring method: chess tournament If previous employees performed well –
does that mean that hiring method is good? Or are we missing information?
Manager: “Everyone I’ve hired with this method has been excellent!”
Example: selection criteria
Problem: missing data about how rejected candidates would have performed if hired
Only get feedback about the performance of candidates that were hired
If strong candidate pool, even flawed system would select good performers
Some rejected candidates might have been even better, but no way of knowing
Home remedy: radishes! (again)
Your friend convinced you that radishes are a good treatment for colds, which is false
Will you learn from experience that radishes do not really help?
Problem: missing information If you belief in treatment, then every time
you are sick you will eat radishes Never get to see how quickly you would
recover without radishes
Confirmatory evidence more noticeable…
What evidence is noticed?
Example: belief that your roommate never remembers to wash their dishes
Lots of relevant evidence – every meal at home, either does or does not wash dishes
But which cases will be noticed?
“Roommate never washes dishes”
Monday Tues
Weds Friday
Sunday
Thurs
Again??!I need a new roommate!
What evidence is noticed?
Example: belief that your roommate never remembers to wash their dishes
Pos / neg evidence not equally noticeable Salient evidence would only reinforce belief
Unpleasant event, highly salient!
Going smoothly is non-event, not salient
Reinforce belief(even if rare)
Example: prophecy
Using my psychic powers, I predict: On the first day of Lunar New Year in 2013, … there will be a major earthquake in China
Example: prophecy
Prophesy: On the first day of Lunar New Year in 2013, major earthquake in China
Specific, falsifiable prediction But - what outcome would you notice?
Prophesy: On the first day of Lunar New Year in 2013, major earthquake in China
Example: prophecy
WOW! He really is psychic!
Non-event, unlikelyto remember prophesy
Earthquake!New Year
No earthquakeNew Year
Prophesy: On the first day of Lunar New Year in 2013, major earthquake in China
Example: prophecy
WOW! He really is psychic!
Earthquake!New Year
No earthquakeNew Year
Noticeable outcome would confirm psychic powers, not disconfirm
http://xkcd.com/628/
Bias in seeking information…
Exercise: rule discovery
Goal: to figure out an unknown rule for sequences of three numbers
Some sequences of numbers satisfy the rule, some sequences do not
Here is one sequence of numbers that satisfied the rule: 2-4-6
Now you can suggest numbers for testing I will tell you “yes” or “no”
Exercise: rule discovery
Demo: test cases to discover rule Initial example chosen so that you would
likely have some guess about the rule Tendency: test additional examples that
would also satisfy rule Problem: did not get opportunity to learn
that your guess was wrong Restricted test cases could only reinforce
mistaken belief about rule
Application: medical diagnosis
How might confirmation bias lead to misdiagnosis by doctors?
Example: test for extroversion
Goal: scale for measuring extroversion Items are self-reflective statements
“I often feel that …” “I generally do not …” Each item is rated on scale agree/disagree
1 – ‘strongly agree’ 2 – ‘agree’ 3 – ‘neither agree not disagree’ 4 – ‘disagree’ 5 – ‘strongly disagree’
Exercise: everyone write down a possible item for an extroversion scale
Example: test for extroversion
In principle, items could test for either presence or absence of extroversion
Presence: “I am often the life of the party” Extroverts would “agree”
Absence: “I often keep to myself at parties” Extroverts would “disagree”
In your sample items, would extroverts be expected to agree or disagree?
Expected result: mostly “agree” items
Seeking confirmation
Extroversion test example: tended to seek information that confirms not disconfirms Look for presence of a trait not absence Look for +extroversion not -introversion
If asked to make a test for introversion, would have chosen different statements
… even though these are assumed to be opposites along the same continuum
Wason selection task
Hypothesis: “Cards with an odd number on one side have a circle on the other side”
Which cards need to be flipped to evaluate this hypothesis?
Wason selection task
To evaluate hypothesis “if odd, then circle” Typical answer: (a) and (c)
Correct answer: (a) and (d)
Wason selection task
To evaluate hypothesis “if odd, then circle”
Does not matter
If odd, supports hypothesis
If even, does not refute!
If odd, refuteshypothesis
If not circle, would refute hypothesis
Wason selection task
To evaluate hypothesis “if odd, then circle”
Does not matter
If odd, refuteshypothesis
If not circle, would refute hypothesis
Could only confirm, never refute
Wason selection task
To evaluate hypothesis “if odd, then circle”
Obviously irrelevant
Incorrectly treated as strong evidence
Relevant but neglected
Obviously relevant
Wason selection task
Hypothesis: “If drinking beer, then over 18” Which people need to be checked? Logically, exactly same as previous, but
people tend to get this version correct “Permission” schema seems to help
Wason selection task
Hypothesis: “If odd, then circle” Logically, need to check: odd, square But typically drawn to check circle card,
which could confirm but never disconfirm Confirmation bias in seeking information
Summary - Confirmation bias
When evaluating a belief, we tend to seek and focus on confirmatory information
As a result, less likely to be exposed to evidence that might refute a false belief
Example demonstrations: Rule discovery exercise Look for stereotypical extroverted traits Wason selection task
Expectations influence interpretation of evidence…
Example: knowledge -> perception
By R. C. JamesOnce you know what it is, looks different!
Bias due to expectations
Bias due to expectations can allow us to interpret highly ambiguous information
Perceptual example: experience allows us to see Dalmatian from degraded image
Adaptive use of knowledge and experience Is there a downside to this? Potential problem: re-enforce false beliefs
Example: Referees’ judgments
Referees in sports often have to interpret ambiguous information to evaluate penalties
Susceptible to bias from expectations
High tackle or not?
Depends: are you an All-Blacks fan??
Example: Referees’ judgments
Frank & Gilovich (1988) –effect of black uniforms on referee judgments
Referees evaluated possible penalties from videos
Identical situations except varied uniform color
Finding: more penalties for players wearing black
Explanation: stereotype of black as “bad guy”
Lack of source memory…
Trivia quiz
Answer trivia quiz
Trivia quiz
Compute the number of “true” responses for even numbered statements
Actual: equal true/false Even numbered statements were repeated
from previous trivia quiz (if you took it) Prediction: more “true” for statements that
were repeated
Illusion of truth effect
False True0.3
0.4
0.5
0.6
0.7
test1
test2Pe
rce
nt j
udg
ed
“tru
e”
Data from my HKU class:
For identical statements, more “true” judgments on second test
Change due to just one prior exposure
Trivia quiz
Why might there be a bias toward “true”? Repeated statements were more familiar General bias: familiar -> perceived true “Validity effect” or “Illusion of truth effect”
Hasher, Goldstein, & Toppino (1977)
Example of a failure of source memory
Source memory
Take some fact that you know is true: “The largest ocean in the world is the
Pacific Ocean” How do you know this is true? Can you answer: when and where did you
learn this fact? You know that this fact is true, but you don’t
remember the source of knowledge
Hard!
Source memory
Human memory is very good at recognition … but memory is not as good for source We know many things… .. but not necessarily how we know
Source memory
Real life example where poor source memory might lead to persistent false beliefs?
Urban myths
Lack of source memory can explain persistence of “urban myths”
Suppose we have heard a myth many times Hard to remember whether or not we heard
from a reliable source (Also, we don’t tend to question how we know something – we just do)
So if myth is familiar, tend to believe it Then we might repeat the myth to others,
increasing familiarity for them, and so on ...
Home remedy: radishes!
I’ve discussed a false remedy: radishes Right now, you remember that this is just a
silly example for illustration But years from now… … may only remember hearing something
about radishes curing the cold Sorry for planting a future false belief!
Summary – persistence of false beliefs
Tend to seek confirmatory evidence Neglect missing information Confirmatory evidence more noticeable Biased interpretation of evidence Lack of source memory Effect of these factors: can be very hard to
change a false belief!
Avoiding bias?
Try to consider possibility of coincidence We tend to seek causal interpretations, and
neglect the possibility of coincidence Helpful to keep this in mind, and recognize
some specific situations eg. regression to mean
Avoiding bias?
Be aware of missing information Information required to evaluate a belief is
often unavailable or unnoticeable How well would rejected applicants have performed? If roommate does reliably washes dishes, would I
notice?
Would the condition have improved even without treatment?
Noticing these situations can cue you to be skeptical about beliefs
Avoiding bias?
Be aware of limited source memory We tend to perceive familiar as true Try to reinforce good information more than
bad information
Avoiding bias?
Strategy: study science! Thinking like a scientist can help overcome
some of these biases Idea of testing falsifiable predictions Appreciation of randomness and statistics
Thanks for your attention