Judgment Bias 1 (Pre lecture 19) Flashcards
What are cognitive heuristics?
Reflexive mental shortcuts used to:
- simplify complex cognition.
- increase speed and efficiency of thinking- when outcome of judgement or decision not know.
- can lighten cognitive load.
- but can lead to bias (or error) because not all relevant info considered.
Heuristics: mental strategies (‘rules of thumb’) used under uncertain conditions.
Used due to cognitive limitations- working memory and attention.
Used due to external constraints- eg. limited time.
Used due to motivational constraints- lack of interest in or understanding of task.
Why ‘heuristics and biases?’
Using cognitive heuristics involves not taking (sufficient) account of relevant information, which leads to biased judgements or decisions.
- relying too much on info that’s easy to retrieve from memory (eg. stereotype-consistent) rather than taking time to find out about characteristics.
Original approach/theory
Kahneman & Tversky (1970s & 80s):
- challenged dominance or normative (rational) models of thinking. (eg. subjective expected utility theory- making decisions using probabilistic reasoning).
- cognitive heuristics: reflexive ‘mental operations’ used to make complex tasks manageable- but heuristic nature typically leads to bias.
Kahneman, Slovic & Tversky, 82):
- notion that thinking under conditions of uncertainty due not to extensive information processing but use of a few simplifying mental strategies (shortcuts) ‘came of age’.
- propose cognitive model of bias in judgement and decision making (JDM)- bias due to simplified cognitive information-processing.
Spawned research beyond psychology- eg. business, economics, law etc.
Identified 3 cognitive heuristics (K&T):
- availability.
- representativeness.
- anchoring and adjustment.
Three heuristics
Availability: judging something based on ease of recall.
Representativeness: judgements based on similarity between seeming-related things (eg. personality and star sign).
Anchoring and adjustment: judgements based on numerical information- bias due to insufficiently ‘adjusting from anchor’.
Availability
Memory retrieval or mental simulation of event-consistent info leads to bias.
Research (K&T, 81; T&K, 82): ps that read newspaper article about house fire, overestimated chance of involvement in incident.
- fire-service data used as criterion against which judgement bias evaluated.
- reading article served as ‘memory-prime’ for event-consistent information, which led to overestimation.
Most ps wrongly judged that more english words start with R than have r as third letter- 3 times as many words with r as third letter.
- easier to think of examples of words starting with R.
- greater cognitive availability of such words because people tend to remember words based on their first letters.
Representativeness
Violate normative principles (eg. Bayes’ theorem) when reasoning probabilistically.
‘Taxi cab’ study (K&T, 72):
- ps given scenario about hit and run involving blue or green taxi.
- ps told proportion of taxis run by blue (15%) and green (85%) companies in city.
- ps also told of correctness of eye-witness’ identification of colour of taxi involved.
Results: ps overestimated probability that blue or green taxi involved in incident.
- judging on basis of info in scenario about colour rather than proportion of taxis; ps discounted proportion of blue and green taxis in city when judging.
- colour of taxi in scenario was taxi most ps said was involved in incident.
Linked to conjunction fallacy: thinking error (non-logical) occurring with probabilities.
- 2 or more not perfectly correlated things judged as being more likely to occur together than separately.
Violated ‘extension rule’ of probability theory- mathematically probability of 2 or more events occurring together has to be less likely than these events occurring on their own.
Representativeness and conjunction fallacy
‘Linda problem’ study (T & K, 83):
- ps given information about characteristics, job and interests of ‘Linda’.
- ps judge likelihood of information about ‘linda’.
- most ps judged two pieces of info would separately.
Anchoring and adjustment
‘Wheel of fortune’ study (T&K, 74):
Judgements of percentage of African nations in UN affected ‘random’ number derived from spinning wheel of fortune beforehand (numbers 1: low anchor; 64:high anchor).
Judgements higher when wheel stopped at high anchor value; lower with low anchor value:
- overestimation in high anchor condition.
- underestimation in low anchor condition.
Judgements biased in direction of anchor because of insufficient adjustment from it.
Two major biases
1) Overconfidence (eg. overestimating ability to do well on test):
- typically evident in probabilistic reasoning (eg. confidence in occurrence of future events).
2) Over-optimism (eg. underestimating time needed to write essay):
- both explained by neglect of base-rate data.
- both linked with case-specific reasoning.
Planning fallacy bias (K&T, 79)
Underestimate future task duration despite knowing that previous tasks overran.
Over-optimism regarding future task duration judgements.
Due to:
- neglect of base-rate data (eg. previous task performance).
- focusing on aspects of current task (eg. what’s unique/different about current task).
Underestimation evident on laboratory (eg. anagrams) and real world tasks (eg. essays)- but influenced by previous task experience.
Reduced by prior task experience:
- eg. Thomas, Handley & Newstead (2007)- task duration predicted more accurately when similar task performed previously.
Due to use of anchoring and adjustment heuristic (Thomas & Handley, 08): previous task duration used as anchor for predictions- bias due to insufficient adjustment from anchor value.
- underestimation with ‘anchor’ of shorter duration.
- overestimation with ‘anchor’ of longer duration.
Gambler’s fallacy bias (Gilovich et al, 2002)
Operational when judging repeated stimuli: eg. because roulette ball landed on red last 20 times, doesn’t mean it’ll land on black next time.
- same probability of ball landing on black or red (50/50) for each spin.
Tendency to look for order or sequences.
Misconception of ‘randomness’ or ‘chance’.
- based on representativeness heuristic.
- don’t consider probability of stimulus occurring- violation of ‘law of averages’.
Summary
Kahneman & Tversky’s original heuristics and biases approach/theory:
- cognitive model of error in thinking.
- heuristics: reflexive mental strategies used to make efficient judgements and decisions under conditions of uncertainty.
- bias: heuristic nature of such thinking.
- 3 heuristics (eg. ‘representativeness’)
- 2 biases (over-confidence & over-optimism).
