CPSY 0220 - Making Decisions

Homo Economicus

completely rational, unbiased, choose what’s right for their interest

Choice Architecture

Changing the decision envrionment

Nudge

Alter decision environment to improve individual behavior without coercion

Liberal Paternalism

Encourage chioces that improve lives without restricting choices

Explicit consent

Opting in to donating organs, less donors

Presumed consent

Opting out to donating organs, more donors

Power of defaults

Default is a recommendation, requires less effort, and represents the status quo (loss aversion)

Human element

normal and faulty decisions/non-cooperation in daily life

Administrative burdens

more paperwork, more effort to get something done

Sludge

frcition that discourage individuals from making any/good decsions

The Long Tail Problem

only see popular things not the niche things that you might really like (ie. Netflix, Spotify)

Good friction

Waiting to buy a home, dangerous medical procedures

Fuel efficiency

best understood through GPM not MPG

Properties of Good Nudges

1. Don't focus on one outcome at expense of primary goal

2. Measure/Nudge the right outcome

3. Make sure the nudge isn't actually sludge

Optical acceleration calculation

move relative to the flying ball to catch it

Experts vs. Novices

Experts can do other things and succeed, novices have to focus really hard

Swoop Test

"Red" but green, "Green" but yellow, "Blue" but red, supressing automatic responses

Thin Slices

Only have to look for a second to judge competent and trustworthy candidates

Dual Processing Model System 1

fast, automatic, habitual, less effort, not memory

Dual Processing Model System 2

slow, deliberate, model-based, conscious

Chariot and the Rider

Plato's emotional horse and controlled horse and the rider that struggles with them together (dual processing)

Cognitive Reflection Test (CRT)

Deliberate vs. Intutitive reasoning, a baseball bat and a ball, lilypads doubling, “how long for x machines to make..”

Lower CRT

More intutitive less deliberate thinking (instant gratification, more device useage, less emotional regulation)

A not B task

Have to unlearn learned response, reversal, easy at 12 months

dlPFC

Needed for planning, prioritization, executive function, a not b

Cartesian Perspective

the rational gatekeeper - hear information, hold it as you analyze, then accept/reject

Spinozan

first accept as true then reject/continue to believe it

Cognitive Load and Unbelieving experiment

more things to do showed non symmetrical results, so more evidence for Spinozan

Hebbian Learning

neurons that fire together wire together (association/classical conditioning)

Blocking

agents learn little from redundant stimuli (add in an extra thing, only cares about first cue so extra cue blocked)

Backwards blocking

start with two cues, continue one with only A cue and B cue is blocked

Rescorla-Wagner Model

Neurons that fire unexpectedly wire together (prediction error)

Prediction Error

PE = V_obv - V_pred

V_new = V_old + PE

The Learning Rate

V_new = V_old + ɑ(PE) where ɑ tells how much you update

Reinforcement learning

Affects fast and slow learning over time, is shaped by rewards, helps actors choose the right things given current ideas

Goal-Directed Learning

“I want, so I’ll do the thing to get there,” associate state (color room) with acction (left/right) for reward

Temporal Difference Learning

When there aren’t immediate rewards, assign credit to cues that lead to future rewards

If the cue always leads to a reward

Dopamine at cue

If the cue sometimes leads to a reward

Dopamine at reward

Temporal Difference Learning Equation

V_new = V_old + ɑ(Reward + V_{best option} - V_old)

Model Free Learning

Habits

Fungible

my piece of paper is exactly the same as your piece of paper so I don’t care whose pice of paper I have

Non-Fungible

My mom’s wedding dress and your dress you bought at Nordstrom because I care which dress is which

Mental Accounting

how you organize, evaluate, and track where your money goes

Windfall spending

If you unexpectedly get more money, you’re likely to just spend it on fun stuff

Two-Step Task

Do one part and then the next (part a and b on exam)

Minimal Account

Both options save you $5 so you are indifferent

Comprehensive Account

Factor in overall level of wealth and gains and losses

Topical Account

compare by referencing the relevant mental account

Hedonic Accounts

Segregate gains and integrate losses

Weighted additive rule WADD

put each attribute on a common scale, weight each one depending on importance and trade offs, multiple value by weight, utility is the sum, assumes independence of attributes

Equal weight heuristic EQW

put each attribute on a common scale, the utility is the sum, assumes independence of attributes

Satisficing SAT

Look at options none at a time, choose the first one to satisfy cutoffs, if no wins lower cutoff and repeat, order matters, dating apps

Lexicongraphic order LEX

Choose the most important attribute, choose the option with best of attribute

Elimination-by-aspects EBA

Randomly choose an attribute to focus on based on weight, choose the one above your cutoff for the attribute

Frequency of good/bad features FRQ

Use cutoffs to determine good/bad values of each attribute, count all the good/bad attributes, choose the one with the most good

Habit Heuristic

choose what you chose last time, low effort, low uncertainty, may miss out, options can change

Compensatory strategies

WADD, EQW, FRQ

Non-Compensatory strategies

SAt, LEX, EBA, Habit heuristic

Ignoring information strategies

EQW, SAt, LEX, EBA, FRQ, Habit heuristic

Quantitative strategies

WADD, EQW, FRQ

Qualitative strategies

SAT, LEX. EBA

Selective information processing

Purposefully ignoring information about options, attributes, preferences

general optimism bias

undestimate chance of negative events and overestimate chance of positive events

overplacement

believing that you are better than others when you are not

overprecision

excessive faith that you know the truth

overestimation

thinking you are better than you are

people who do worse

overestimate performance

overclaiming false knowledge

claim knowing more, the less you know, the more you overclaim

help condition vs. delay condition

people mistake outside assistance as their own comptence

Predicting learning trajectories

people accurately predicted the exponential shape of learning curves but are overly optimistic

Game Theory

with fixed number of actors and best move depends on situation, what should you do

The Chicken Game

Cars go straight or swerve? What you do depends on the other

Balancing Selection

Eventually get to a point when ½ straight and ½ swerve

Cold War

USSR/US build nuclear supply, threats must be credible

Prisoner Dilemma

cooperate or defect? rationally: defect

Nash Equilibrium

if you know what everyone else is doing, what should you do? With stability (no regrets), self fulfiling (all will NE if they believe all is at NE), not necessarily optimal

Punishment

have people cooperate by another party punishing the defecters

Second order punishment dilemma

punishment is costly for the punisher so someone else punishes the punisher…

Reciprocity

be nice until the other person isn’t, encourages cooperation

Tit-For-Tat

repeated games, copy what the other person did last time

Surprise attacks

incentivizes attacking before they attack you

Second strike capability

if they attack, you can get them back, defensive build up

Cooperation game

must reach threshold value for desired outcome, public goods

More cooperation

Repeated games, your choice affects your Reputation, non-anonymity

Paradox of choice

rationally, more choices are better but in reality, more choices has less satisfaction and utility

Fancy jam study

more jam = less buys, less jam = more buys

Decision overload challenges

oppotunity cost (FOMO), regret/anticipated regret, expectations (more choice, more expectations)

Choice Set

The options you decide b/t

Set complexity

are all options good? is one dominant? are the options alignable (same kind of thing)?

Preference Uncertainty

I don’t know what I want, can’t rank choices, don’t have cutoff

Decision goals

it doesn’t have to be the perfect choice (satisficing)

Perception

an active process of learning

Motivated reasoning

believe evidence that confirms preexisting beliefs and discount information that counters beliefs, align with identity/beliefs/feelings/goals/information

Self serving theorires

View attributes that match myself more favorable

Evidence weighting

Evaluate evidence in line with motivations/interest more positively

Why motivated reasoning?

evidence weighting, memory recall, evidence innterpretation, evidence accumulation, evidence integration

Accuracy goals

effort, considering all evidence and alternatives

Directional goals

search for supporting evidence, biased interpretations