2) Decision Making

what is a decision

anything with more than one response option

risky decision

when outcomes are uncertain and where consequences have emotional aspects

what are the parameters that decisions can vary along - more relevant to gambling (2)

• size of the outcomes, gain and loss

• probabilities of the outcomes occurring, known as judgement

what are the parameters that decisions can vary along - less relevant to gambling (2)

• delay to receiving outcome

• effort to obtain outcome

how is risk defined (3)

• hard to define

• risk as variance

• risk as hazard

risk as variance (2)

• used in economics

• variability between things that might happen

risk as hazard (2)

• used in medicine and psychology

• potential for negative consequences

risk vs uncertainty (ambiguity) in decision making (2)

• decisions under risk is where there is known probabilities and outcomes

• decisions under ambiguity is where there are unknown probabilities and/ or outcomes and must be estimated or learned through experience

• e.g. risk: roulette, ambiguity: sports betting

expected value (2)

• EV = probability x value

• always choose the option that maximises EV

example of EV

expected utility (2)

• a common currency for choices

• total anticipated satisfaction (or "utility") a person gets from a choice that has uncertain outcomes

bernoulli effect (3)

• curve is plateauing

• those with low income will have a huge impact for the win but a multi- millionaire with the same win, will be worth less

• see diminishing marginal utility with increasing value

example of violations of expected utility

the tropical flu problem

the tropical flu problem (6)

• choose between two programs for tackling a dangerous disease that is expected to claim 600 lives

• randomly assigned to gains version or losses version

• gains version to choose between program A to save 200 ppl for certain or program B to have 1/3 chance of saving 600 people and 2/3 of saving none

• losses version to choose between program C for 400 fatalities for certain or program D for 1/3 chance of no fatalities and 2/3 chance of 600 fatalities

• found that in the gains version, most subjects are risk averse, more likely to choose A over B

• and that in the losses version, most subjects are risk preferent, more likely to choose D over C

what was used to explain the results of the tropical flu problem (2)

• framing effect which is a cognitive bias where people make different decisions about identical options depending on how the information is presented

• as all four programs were identical in their EV so in theory subjects should have no preference

what explains the framing effect

the value function

value function (5)

• relates objective value to utility (subjective)

• objective value of option A, 200 lives has a greater impact than option B, 600 lives due to curve plateauing

• the curve shows utilities, not expected utilities so when you multiply utility of B by its low probability, A will be better as it maximises expected utility

• objective value of option C, 400 fatalities has negative utility and is guaranteed but option D, 600 fatalities has worse utility but it is flattening and expected utility is higher than C as there is a chance that it might not occur

• depth or steepness of the loss curve is not important for the framing effect

framing effect in gambling(2)

• experienced gamblers rate their chances of winning as lower in “house edge” condition and higher in the “return to player” condition

• more optimistic with this frame

prospect theory 1 in Kahneman and Tversky (3)

• value function

• gains and losses are processed by different parts of the value function

• the value function has a reference point at “0” that separates gains from losses

prospect theory 2 - Kahneman and Tversky (3)

• weighting function

• true probability converted to a perceived (weighted) probability, via an inverted S shape

• overweight rare events and underweight common events

in the weighting function, which parts of the probability range are when humans are reasonable accurate

humans are accurate at 0 and 1 and at a point around p= 0.4

probability distortions (2)

• Lichtenstein looked at likelihood of different causes of mortality

• found that we overestimate rare causes of death and we underestimate common causes

bias vs error (2)

• bias is when they are systematically wrong but accurate

• errors have low accuracy

what factors might come into play when judging own person risk likelihoods (5)

• location

• family history

• personal experience

• attitudes

• optimism bias - people underestimate what might happen to them

economic approach to decision making (2)

• argues that people wants to maximise expected value or expected utility

• it is a through and systematic approach but it is effortful and requires access to alot of information

but what do people use to avoid cost- benefit calculations

heuristics

heuristics (5)

• they are shortcuts that lead us to a answer

• fast, don’t need to crunch numbers

• well suited to the everyday world

• often given an answer that is good enough, satisficing to meet the minimum requirements instead of searching for the absolute best choice

• but sometimes it can generate silly responses

two heuristics discussed in class (2)

• availability heuristics

• representativeness

the availability heuristics

used whenever we estimate frequency or probability by the ease with which examples could be brought to mind

representativeness heuristics (4)

• when the probability of an event is estimated from a similar situation

• two effects one mechanism

• base rate neglect

• conjunction fallacy

base rate neglect (4)

• when people make judgement based on similarity to a reference category, without considering the basic frequency of that category

• done through engineers and lawyers problem

• participants given base rates of engineers and lawyers and then a description of a person

• but participants neglected the base rate and said that they won’t know if the person is an engineer or lawyer

conjunction fallacy (4)

• people judge that the combination of two events is more common than one of the events by itself

• done through the linda problem

• given a statement and asked to ranked what would be true of linda

• concluded that linda is a bank clerk and active feminist more than just a bank clerk

heuristic in gambling

gambler’s fallacy

gambler’s fallacy (3)

• failure to appreciate the independence of turns and predicts that the other outcome must be due

• mistaken belief that past random events can change the odds of future independent events

• e.g. roulette, red, red, red, red, *will say the next one is black

what is gambling fallacy an example of

representativeness heuristics

explanation of gambling fallacy as the representativeness heuristics (2)

• people expect short sequences of outcomes to be representative of the overall population

• in reality, chance is lumpy and long runs of one event can and do happen by chance

heuristics in blackjack (4)

• Wagennar’s field study of experienced blackjack players

• looked at how real gamblers respond to runs of wins and runs of losses

• most gamblers increased their bet after runs of losses and decreased their bet after runs of wins - they rely on representativeness as a long run of losses or wins doesn't seem representative of what we expect randomness to look like

• a smaller number consistently did the opposite where they increased bets after a win and decreased bets after losses and that these players must be replying on availability