Judgment and Decision Making

simple experiment

process that lead to one possible outcome from a set of outcomes in a universal set

event

any set of sample points

elementary event

any event with one sample point

subjective probabilities

assignment of numbers to objects by a set of rules

dawes 4 rules in words

1. all probability are numbers between 0 and 1

2. probability of all events is 1, the probability of no event is 0

3. if two events are mutually exclusive, then the probability either event will occur is equal to the sum of their probabilities

4. conditional probabilities are equal to the probability either event will occur is equal to the sum of their probabilities

probabilities

numbers that rationally represent facts, individual beliefs, and relationships between evidence and belief

overestimation of conjunction

anchor on .9 to predict . 3

underestimation of disjunction

anchor on .2 to predict .7

thinking is thrown off by

setting up thinking incorrectly

use of probabilities does not

make thinking rational

how to make decisions based on probabilities?

statistical decision theory has been adopted by the scientific community

H0

no action required

H1

action required

null hypothesis

data suggest no action is needed

use of normal distribution

assign probability to observing samples with varying meaning

alternative hypothesis

we should take action

probability of observing a mean near H0 is

non-zero for both hypothesis

since observations are random

sometimes our decisions are wrong

founded by Gustav Fechner

philosopher and physicist

Gustav Fechner considers human as a measurement device involving

external stimuli, brain activity, conscious sensation

three relations of importance

external physical environment, brain activity- physiology/chemistry, conscious perception-inner psychophysics

Weber’s Law

the change in w, Delta(w), needed to obtain a unit step in subjective magnitude (R(w)) is a linear function og the initial intensity of w

just noticeable difference

magnitude of the difference in stimuli that the subject can detect, can be concatenated

hunt uses JND to say

the probability of the comparison being judged as greater than the standard

psychometric function

a model of detection probability, p(d), as a function of stimulus intensity

function forms for psychometric function

normal cumulative function of stimulus intensity

Weber fraction

(deltaw/w) measure of sensitivity

smaller deltaw/w

more sensitive the perception

larger deltaw/w

less sensitive the perception

theory of threshold

switch from performing at random chance level to demonstrate perception, deals with probability of detection

versions of the classical threshold theories

fixed and random

only purpose of covering the “fixed” version

help understand “random” version

we can expect that

stimuli below threshold will not be detected stimuli above threshold will be detected

mathematical model of sensation

small sensation increments are proportional to stimulus increments

sensation increments

assumed to be constant

variations on Fechner’s integration

1. constant change in stimulus and sensation

2. linear change in stimulus and constant change in sensation

3. constant change in stimulus and linear change in sensation

4. linear change in stimulus and sensation

weber/fechner measured

sensation scale

steven measured

response scale

weber/ fechner approach

subject who is acting as a measuring device senses difference in stimuli, measurement JND

stevens approach

subject is still acting as a measuring device the stimuli generates a sensation. participant generates a response with equal sensation, measurement-numerical stimuli intensity

local psychophysics

Fechner’s approach deals with many measurements near a specific stimulus

global psychophysics

stevens approach allows for participants to provide one judgment for a stimulus, allowing measurement of many stimuli

if we change Fechner’s assumption

we can derive different types of stimulus-response relationships

thurstone’s theory

produced a graphical representation for what he called “the discriminal process”

more discriminal distance

more sensitivity

less discriminal distance

less sensitivity

unit of measurement

distance between these stimuli relative to the amount of dispersion

decision utility

anticipated satisfaction of what will make us happy after we decide

experienced utility

subjective feeling of pleasure and pain

sum of momentary evaluations of pain

do not equal the summary remembered by experience

peak-end

only the memory of the most intense pain and the final level of pain predict the summary evaluation

duration neglect

people tend to be insensitive to the length of experience

emotion

reactions to motivationally significant stimuli and situation

mood

longer-duration background states

feelings

conscious perception of mood

evaluation

pleasure/pain good/bad judgment of consequences

emotions vs cognition

theorize that emotion is a cue to cognition, theorize that emotion is a response to cognition

emotions are theorized

to play an important adaptive role in decision making

logarithmic law implies

constant changes in unity of money requires a geometric increase in money, utility of money depends on the amount of money one already has

diminishing sensitivity

observed in many data sets and is the basis of much psychological theory

diminishing sensitivity can be viewed in several ways

geometric increase are required to generate constant increase in sensation, good thing satiate, and tendency to be risk adverse

psychological view of values

perceptual systems are attuned to notice changes, not absolute magnitude

economic view of value

final outcomes of choices are all that matter

diminishing sensitivity of economic utility implies

a person will be risk adverse

diminishing sensitivity to gains and losses implies a person will be

risk adverse for gains, risk seeking for losses

where might utilities come from?

suggest we predict decision utility based on past experiences and theory of decision by sampling

stewart et al provides results that mimic theory of subjective value by

using natural occurrence of gains and losses

economist theorize that preferences are

known

psychologists theorize that preferences are

constructed

attribution of choice options

tend to be negatively correlated and causes conflict

conflict

all choices are attempted to resolve conflict and is resolved by finding a balance in the costs and benefits

Herb Simon introduced bounded rationality

DM’s seek to be as rational as possible, given cognitive contraints

reducing cognitive demands of decision making

aspiration levels and heuristics

aspiration levels

model for simplifying the evaluation of outcomes, outcomes can be acceptable or unacceptable

heuristics

simple rules of thumb that allow decision makers to make quick decisions and judgments by overlooking certain information

heuristics cause biases

when decision contexts fall outside of their range of optimal performance

heuristics can be classified according to these principles

examining fewer cues, reducing the difficulty associated with retrieving and storing cue values, simplifying the weighting principles for cues, integrating less information and examining fewer alternatives

examining fewer cues

focus on most important/ validly cues and consider one cue at a time

retrieving and storing cue values

store result of a simple comparison and access information that is easier to retrieve

simplifying the weighting principles

ignore cue-validity information, or the predictive quality of each cue

integrating less information

stop once some reasonable amount of information has been integrated

examining fewer alternatives

DM’s can use heuristics that limit the number of alternatives that are compared simultaneously

strategies for choice

compensatory models- linear and addicitve difference non-compensatory-lexicographic and elimination by aspects.

kleinmuntz and Schade 1993

different models for displaying choice attributes-form, organization and sequence, DM’s apply different cognitive strategies to evaluate information

cognitive incentive system

display influence effort and accuracy of each available cognitive strategy

choices require consideration of

cognitive limits, context, choice architecture

decision theory

the study of how decisions are or ought to be made

normative theory

used to prescribe a course of action, it is concerned with optimal rather and actual choices, purely deductive

gamble

random mechanism that gives an outcome with a pre-defined probability

gamble has 2 attribute dimensions

outcomes and probabilities

simple model of gambles

construct a scale that each gamble can be mapped onto

Bernoulli’s utility

transforms dollars to “sensation of dollars” or “value of dollars”

why use utility?

using function that reflects diminishing sensitivity can help explain those results

diminishing sensitivity of money does not violate rationality but

predicts that people should be risk adverse

modern utility

construct a gamble between the two extremes

if the axioms are satisfied

then an appropriate function must exist

expected value

the long run worth of a gamble

expected utility

the long run worth of a gamble that reflects risk performance

a priori

means the probability is pre-defined ahead of time based on knowledge or theory

subjective expected utility

is an axiomatic theory, and if the axioms hold, then conclusions from the theory are normatively sound