what is bounded rationality =
a human decision making process in which people seek a decision that will be good enough rather than optimal
what is a rational decision maker
- have consistent transitive preferences
- maximise their utility
- use all information to make decisions
- obey Bayes rules and the laws of probability
- description invariance - answers remain unchanged by a transformation
- emotions dont affect logical thinking
why are probabilities important in decision making
when making decisions - required judgement of probabilities is needed
- people make decisions based on their perceived probabilities
what do rational people do with probabilities - when forming beliefs
- they obey the law of probabilities
- have all information
- use Bayes rule to asses probabilities of events
- bayes rule uses conditional probability
Casscells (1978)
- do humans apply Bayes rule
- asked 60 harvard medical students - a question to solve the conditional probability
results from getting students to use Bayes
- half got the obviously wrong answer
- only 18% got the right answer
- people ignore Bayer rule and ignore base rates
- peoples subjective beliefs are very different from the real probabilities - overestimating
How are people bounded rationally?
they use heuristics rather than making fully rational choices (rational = use Bayes theorem when forming probabilistic judgements)
- people use mental shortcuts that dont overload cognitive thinking to simplify problems
what are the 2 models of thinking
system 1 = quick and intuitive thinking, effortless, difficult to control, silly mistakes = intuition
system 2 = consious reasoning, slow, calculating, effortful, complex = reasoning
why does system 1 override 2
system 1 is fast - easier to reach a conclusion = plausible answer - comes quickly to mind
what are the 3 reasons system 1 intutions come to mind fast
- accessibility
- information is readiliy accessible - come to mind fast
- if it comes faster than working out hard calculations
- depends on context and reference of things - perceptions
- system 1 is stronger than system 2 - think answers that come naturally are right - priming
- subconscious clues that hint us to give a certain answer = primes only work on system 1
instead of using Bayes rule people use heuristics to …
make judgements and estimates of probabilities because
- simple cognitive complexity
- speed up decision making
- but LEAD TO SYSTEMATIC BIASES
what are the three important types of heuristics
- representativeness
- the probability of x belonging to y is judged on the basis of how similar x is to y - availability
- probability is estimated by how easy it comes to mind - anchoring
- a quantity is estimated by starting from a convenient anchoring and adjusting appropriately from there
Kahneman (1973)
- peoples rankings of probability and similarity turn out to be the same - they use similarity as a proxy for probabilistic thinking
- 3 groups
1. rank 9 degrees by frequencies of students (elicit base rates) - education 20%
2. gave a summary of Tom - asked students how similar Tom is to the typical student of each degree - computer science highest
3. how likely is Tom to be a student of each subject rank - computer science highest
what are the results of getting to rank students into degrees
- they ignore the base rates that they think 20% of students choose education
- instead peoples probability judgement are the same as there similarity scale
- system 1 mistake - easy to use stereotypes instead of working out answer
what are 3 other biases related to representativeness = kahnemann
- insesitivity to sample size
- the likelihood of a sample result by the similarity of the results to the properties of the population - misconception of chance
- law of small numbers - think that population expected values will show in small samples - regression to the mean
- outliers move back towards the mean in next trials - they draw incorrect inferences
example of insensitivity to sample size
- large hospital = 45 babies
- small hospital = 15 babies
- which hospital would you expect more boys to be born
- right answer = small hospital - wrong answer = about the same
- as sample size grows mean gets closer to theoretical mean
example of misconception of chance
- people expect sequences of events that will represent the essential characteristic of large sequence
- HHHTTT < HHTHTT
- lottery - given numbers drop off
- store is lucky - more people get tickets
examples of regression to the mean
- if extreme variable is observed in first measurment it will tend to be closer to average in second measurement
- people think previous trials are representative of further trials
- if my child does bad on test and i punish then they do good
- if my child does good i reward and then they do bad
availability heuristic example
- frequency substituted with how easy they come to mind
- ease of recall used to estimate probabilities
anchoring example
- will wrongly adjust probabilities according to anchor
- acts as a prime
- anchoring as priming = swade you to answer according to the prime - system 1 problem
