The Cognitive Heuristics Flashcards
FAMILIARITY OF HEURISTICS
- present throughout whole module
- include: categorisation/attribution/stereotyping/attitudes
- a lot suggests we use rules for inferences
- idea of inferential beh conceptualised as choices from alternatives each w/designated value/occurrence probability
ARTIFICIAL INTELLIGENCE ARCHIVES
- social psych borrowed heuristic term from AI
- computers/algorithms/futile search for optimal solution (we don’t always find them)
MOSKOWITZ (2005)
- human rules = fairly rational BUT only useful if uncertainty exists/if too much effort required to arrive at complete/accurate judgement
- heuristic reliance when option of ^ accurate analysis exists/when uncertainty reduced by useful data presence -> heuristics = bias source
HEURISTIC EXAMPLE
Q: SHOULD I HAVE A RISKY FLING?
1. I could stay w/current partner.
2. I could abandon them for fling w/someone I met at work.
1 = safe/comfortable/easy BUT dull/predictable/tedious
2 = exciting/different/new BUT reckless/uncertain/potentially disastrous
WHAT WE SHOULD DO WHEN MAKING A DECISION
- assess available alternatives for likelihood/worth of promised outcomes (probability/value)
- calculate each outcome utility (value product/outcome probability)
- choose option maximising utility
- AKA. we make decisions most likely to deliver desired benefits
OPTIMAL DECISION PREVENTION
- we aren’t computers
- may be too much info to rationally sift through
- we rarely have time
- can’t be sure of outcome (may still be unhappy)
- life decisions do not come via crystal ball/guarentee
MARCH & SIMON (1958)
- mostly we are satisfiers > optimisers
- satisfiers = making adequate inferences/decisions
- optimisers = drawing best possible inferences -> reaching best possible decisions
KAHNEMAN & TVERSKY
- look at ways we satisfice relying on heuristics
using economic theory - looked at slow/fast thinking via 2 systems
- we can be blind to the obvious/our own blindness
THE 2 SYSTEMS
SYSTEM 1
- allows to orient to sudden sound source
- complete phrase “bread and…”
- answer to automatic 2+2
SYSTEM 2
- allows to brace for starter gun in race/look for woman w/white hair/fill out tax form
- requires attention/effort
2 SYSTEMS COGNITIVE HEURISTIC APPLICATION
- samples w/smaller numbers -> extreme outcomes ^ likelihood (ie. VERY high/low kidney cancer rates in rural USA areas)
- yes it’s really that simple
- larger samples = ^ reliable > small samples BUT…
- sparse populations stand out more/yield ^ extreme results = grab more attention
EXAMPLES - London Blitz bombings believed as targeted BUT statistical analysis -> random process
AVUGOS ET AL (2013)
- meta-analysis of GLIOVICH ET AL (1985) aka. multiple basketball shots in row = random
- little documented evidence (ie. lit reviews) BUT meta-analysis = scientific/robust tool
- no “hot-hand” evidence found
SAMPLE SIZES TODAY
- sample size matters BUT we often fail to take account of it
- statistics produce many observations that seem to beg for causal explanations BUT are simple chance
- System 1 = thinking mode leaping on causal connections aka. runs ahead of facts/jumps to conclusions
SYSTEM 1 HEURISTICS
REPRESENTATIVENESS
AVAILABILITY
ADJUSTMENT/ANCHORING
REPRESENTATIVENESS: EXAMPLE
TVERSKY & KAHNEMAN (1974)
- eg. Tom = graduate at main state uni. Rank graduate specialisations in likelihood order (computer science/engineering/business administration/physical sciences/library science/law/medicine/humanities/social science)
- thinking about/deciding based on how popular course are = base rate info usage
- when given personal sketch of Tom, the results change based; stereotypes = System 1; disciplined/systematic consideration = System 2
- but what if info = based on uncertain validity? what about base rate info? does remembering this prevent System 1?
REPRESENTATIVENESS
- representativeness heuristic = mental shortcut whereby instances = assigned to categories based on how similar they are to category in general
REPRESENTATIVENESS: EMPIRICAL EXAMPLE
TVERSKY & KAHNEMAN (1974)
- estimate prob man = engineer/lawyer
- man sampled randomly from group of BOTH engineers/lawyers
- 2 (70% engineer/30% lawyer OR 30% engineer/70% lawyer) x 2 (personality profile/no profile) conditions
- no profile = estimate reflected base rates (ie. ^ engineers = engineer answer)
- profile = less rationale/ignored base rate/basically guessed (50/50 judgements)
WHY DO WE OVERLOOK THE BASE RATE INFO?
- sometimes there is truth to stereotypes BUT they can mislead us (ie. does a woman falling asleep on someone in a train have a college degree?)
- we use:
1. base rate info
2. predictive value (source credibility)
3. sample sizing (small = less reliable)
CAN WE OVERCOME TENDENCY TO OVERLOOK BASE RATE INFO?
SCHWARZ ET AL (1991)
- difficult BUT possible
- instructing people to think “like statistician” enhanced base-rate info use BUT think “like clinician” = opposite effect
- Q: doing task while puffing cheeks out VS frowning–what happens?
- frowning -> ^ vigilance = enhanced System 2 activation -> ^ base info rate use
AVAILABILITY: EXAMPLE
- Q: how many celebrities succumbed to plastic surgery?
- would you be systematic (conservative figure)?
- OR would you recall well-known celebrity instances that you know of?
- latter = availability heuristic
AVAILABILTY
- availability heuristic = cognitive shortcut allowing to draw upon info about how quickly info comes into mind about particular event to deduce frequency/likelihood in the future
- associative bonds = strengthened via repetition
- AH exploits inverse law form aka. uses association strength as frequency judgement basis
- BUT… not always this simple
BIASED ESTIMATES REASONS
- Not always about frequency ie. familiarity/salience
- Personal experiences w/things occurring frequently may be idiosyncratic
AVAILABILITY: EMPIRICAL EXAMPLE
TVERSKY & KAHNEMAN (1973)
- pps memories famous names list
- conditions = either men/women are more famous
- some asked to judge how many men/women in each list (equal numbers); others asked to recall names
RESULTS: gender w/^ famous names = ^ frequent (pps recalled 50% ^ of it)
- fame made names salient -> easier recall -> frequency overestimation of the group
WHY DO WE SUMBIT TO AVAILABILITY HEURISTIC?
- assumption of exemplar volume (content) correlates w/retrieval ease
- OR we feel that if info = easy in mind -> it must say how frequent it is
SCHWARZ ET AL (1991) - pps recall 6/12 assertive behs
- judge own assertiveness
- pps recalling 6 assertive examples = ^ assertive rating > 12 examples
- AKA. feeling of difficulty/ease of retrieval (SYSTEM 1) matters = numbers/content (SYSTEM 2)
ANCHROING/ADJUSTEMENT
- when making judgements under uncertainty you can reduce ambiguity by starting w/anchor
- EG: how many handouts should I print out?
- last time = 75% attendance BUT next tutorial = exam tips… 100% this time?
- we do this w/people too (correspondence bias)
WHEEL OF FORTUNE
- pps stood in front of wheel of fortune marked 1-100; asked to spin
- wheel rigged to stop at 10/65
- pps asked: what is best guess at percentage of African Nations in UN?
- 10 mean estimates = 25%
- 65 mean estimates = 45%
- aka. anchoring heuristic = most robust/reliable results in experimental psychology
ANCHORING/ADJUSTEMENT HEURISTIC
- anchoring/adjustment heuristic = cognitive heuristic making us place weight upon initial standards/schemas (anchors) -> we may not always adjust sufficiently far from anchors to reach accurate judgements
ADJUSTEMENTT HEURISTIC: EMPIRIAL EXAMPLE
ENGLICH ET AL (2006)
- participating legal experts shown realistic case materials involving alleged rape case; asked to provide sentencing decision
- pps received 1/3 anchors:
1. irrelevant source (journalist)
2. randomly chosen anchor
3. pps randomly decided on anchor themselves via dice toss
- anchor high (3y)/low (1y) for each case
- all 3 conditions = anchor constrained sentencing decisions
ADJUSTEMENT HEURISTIC: WHY?
- TVERSKY & KAHNEMAN didn’t agree here:
TVERSKY - traditional view
- deliberate BUT insufficient attempt to adjust from irrelevant value providing anchor
- System 2
KAHNEMAN - anchoring occurs via priming
- System 1
- OVERALL = both probably right in some ways
ADJUSTEMENT = EFFORTFUL
- AKA. System 2
EPLEY & GILOVICH (2006) - link w/correspondence bias
- cognitive load -> people adjust less
- insufficient adjustment = failure of weak/insufficient System 2
ANCHORING = PRIMING EFFECT
- AKA. System 1
MUSSWEILER & STRACK (2000) - anchor primes associated concept in memory
- “is annual mean temp in Germany ^/lower than 20C/5C?”
- 20C = easier recognition of sun/beach
- 5C = easier recognition of frost/ski
! SUMMARY !
- OVERALL: when faced w/difficult qs we oft substitute easier qs to answer
- we fall for System 1 thinking
LAW OF NUMBERS - fail to take into account sample size
- seek causality for random event
REPRESENTATIVENESS - probability judgement based on appearance
AVAILABILITY - judgement based on how easily it comes to mind
ADJUSTMENT/ANCHORING - estimate judgement via amending initial base valiue
