Experiments and Authors - PSYC3211: Cognitive Science UNSW

(Kahneman & Tversky, 1979)

prospect theory.

concave for gains and convex for losses, steeper for losses than for gains. discard events of low probability and treat high probability as certain

Rakow, Demes, & Newell (2008)

Decisions from Description - Outcomes and their likelihoods are clearly and unequivocally specified to the decision

maker

Decisions from Experience - Outcomes and their likelihoods are initially unknown to the decision maker, and must be

learned by observation or experience

Results: Description-Experience Gap

○ Underweighting the 20% of the rare event in the experience condition

○ Overweighting the 20% of the rare event in the description condition

Hertwig & Erev (2009)

● Full Feedback , seeing both results. Because you are

seeing the possibility, you tend to go for the safe option but

will be attracted to the risky option from time to time

● Sampling & Partial , the probabilities of choosing risky

options drops. This is presumably because you never see

the result of the risky option.

● Description

Jessup, Bishara, & Busemeyer (2008)

● Give people a high or low probability gamble

● IV = whether they got feedback trial by trial.

Results:

● High probability condition show that No Feedback is essentially indifferent to both choice, gamble or sure option

● But when given feedback, people have a preference for risky option even though this is a high probability game in the domain of gains, which should suggest a risk aversive behaviour, but not in this case

○ THIS IS SIMILAR TO THE REPORT PROBABILITY MATCHING OR MAXIMISING

● No feedback group starts indifferent but slowly leans towards taking higher risk. This group is consistent with OVERWEIGHTING a rare event, because you overweight the percentage of the rare event

Hotaling, Jarvstad, Donkin, Newell (2019)

Information Format & Cognitive Algorithms

● Goal is to pick out balls from a box, difference colour balls have different payout prices on them (green vs yellow)

Conditions:

Standard (Exploratory) - After drawing a ball you know what the balls are worth. Sampling vs Choice stage.

Value Ignorance (Description) - Do not know what the balls are worth in sample, only know the colour. Value is shown during choice phase.

Results:

● The results to the Prospect Theory

● Just by changing which stage of the experiment people receive information, you can see the weight people put on those events

● How information is presented changes the cognitive algorithms people use to make choices

Camilleri & Newell, 2011

A = Description B = Sampling C= Partial Feedback D = Full feedback

Erev et al., 2020

Reckless = not bothering to physically distance or isolate Responsible = following distancing guidelines, isolating

Both reactions could be due to a tendency to rely on small samples of similar experiences when choosing between behaviours

Renato et al.

Propensity: "Are you generally a risk-taking person?" Frequency: "How many cigarettes do you smoke per day?" Behavioural: (DFD)"Do you prefer $3000 for sure or $4000 @0.8?" (DFE) "A or B"

Results:

● Behavioural measures tend to sit outside of these propensity measures and frequency measures

● Even though behavioural measures sit outside of the mind map, it also shows how different information formats affect people's preferences

Gigerenzer, (1996)

"the heuristics... are Rorschach ink blots: a researcher can read into them what he or she wishes"

Circularity of availability, representativeness, etc

Arkes et al.

NO EVIDENCE of person becoming money pump

Houston et al., 2007

How to maximize chances of overwinter survival?

Is intransitivity (always) costly?

"optimal strategy always involves taking fewer risks in terms of predation as reserves increase"

Arkes et al., 2016

"the ecological-rationality research strategy does not equate rationality with following a coherence rule. Rather, it measures rationality in terms of success in the world , such as making competitive decisions and accurate predictions. This success is not measured by adherence to a coherence norm, but by the "match" between a cognitive strategy and the structure of the environment ."

Hertwig et al., 2008 Exp 3

Causal reading of 'and'

Most chose X&Y. Suggests that causal reading of 'and' might explain conjunction effects here (e.g., prob (effect Y|cause X) > prob (effect Y))

Johnson et al. 2012

Beyond Nudges: Tools of a Choice Architecture

Johnson and Goldstein (2003)

Study: Defaults and Organ Donors

Defaults work, even across cultures

Bryan et al.

Heterogeneity revolution

Krijnen et al., 2017

"the month after the bill passed the Dutch House of Representatives...the number of residents who registered as nondonors spiked to roughly 40 times the number observed in previous months".

Krijnen, Tannenbaum & Fox (2017)

PreDICT Checklist

Preference Uncertainty

Distrust

Importance

Change

Transparency

Osman et al.

Learning from Behavioural Changes that Fail

Taxonomy of Failure of Behavioural Change

Ridder, Kroese & van Gestel (2021)

The relation between individual preferences and the effectiveness of nudges seems best represented by an inverted U curve.

People with less developed preferences (because they are ambivalent or in doubt about their choice) can be nudged toward a specific option.

Those on the extreme left end, who have a clear preference for the alternative, will not be affected by the nudge. At the extreme right end of the inverted U shape, we would find people who have strong preferences in line with the nudge; for this group, nudges tend to be redundant because they would make the desired choice regardless of the presence of a nudge.

Hertwig and Grune-Yanoff (2017)

Nudging and Boosting: Steering or Empowering Good Decisions

The objective of boosts is to foster people's competence to make their own choice-that is, to exercise their own agency.

Thaler & Sunstein

Libertarian Paternalism - People should be able to make their choices freely but its ok for choice architects to try and influence people for their own good

Nudge - any aspect of choice architecture that alters behavior in a predictable way without forcibly removing options

Humans err in predictable ways that can be addressed.

Edwards (1968)

CONSERVATISM: not extracting as much information from the sample as required by Bayes's Theorem

Von Neumann & Morgernstern (1947)

Expected Utility Theory

Mathematical focus: not intended to describe how people actually but how people would behave, behave if they follow requirements of rational choice

• Von Neumann & Morgenstern interested in what people do not but axioms provide a 'benchmark' against which to evaluate the 'rationality' of people's choices