Chapter 2 Decision-Making: Normative Models and Anomalies Flashcards
Decisions under certainty
– decisions where no probabilities are involved regarding the consequences of alternatives
Decision under uncertainty (Risky/ambiguous =mehrdeutig)
– decisions where the consequences of alternatives do only occur with some probability
Risky decisions: probabilities of consequences of alternatives are known
ambiguous decisions: probabilities of consequences of alternatives are only vaguely known
– ambiguity effect: decision-makers prefer risky decisions to ambiguous decisions
Ellsbergparadox
decision-makers behave inconsistently between two formulations of the decision situation that have mathematically identical possible gains but differ in ambiguity. Demonstrates the ambiguity effect
Beispiele:
An urn contains 30 red balls and a total of 60 other balls that are either black or yellow.
First game:
1(a) If a red ball is drawn from the urn, the participant wins.
1(b) If a black ball is drawn from the urn, the participant wins.
Second game:
2(a) If a red or yellow ball is drawn, the participant wins.
2(b) If a black or yellow ball is drawn, the participant wins.
Most players inconsistently choose 1(a) but 2(b)
St. Petersburgparadox
illustrates that decision-makers do not base decisions on expected value
Subjective expected utility theory (SEU)
Besonders subjektiv. St. Petersburgparadoxon bezieht sich darauf, dass nicht die mathematisch sinnvollste Alternative gewählt wird, sondern die subjektive.
– subjective expected utility = sum of products of subjective probabilities and subjective utility of consequences
– choose option with highest subjective expected utility
Normative decision models
– specify how an (idealised) individual should make optimal decisions
Prescriptive decision models
– use decision theory to offer step-by-step suggestions on how to proceed in a decision situation in order to make an optimal decision
Descriptive decision models
– describe how individuals actually make decisions
Economic games
empirical study of correspondence between theoretical predictions of normative theories and actual behaviour
Ultimatum game
– decision-makers have to divide a good between themselves and another Person
– if the other person rejects the proposed division, both leave empty-handed
• Game-theoretic prediction
– A gives minimum positive amount possible
– B accepts because any positive amount is better than Zero
• Empirical findings
– higher offers than minimum are made
– variations occur, but even in very different societies, gametheoretic prediction is not supported
Dictator Game
– decision-makers have to divide a good between themselves and another person
– the other person has to accept the proposed division
• Game-theoretic prediction
– A gives nothing
Empirical findings
– many people give something to the other person; on average, 28% of the amount
Prisoner’s dilemma
– two decision-makers independently have to decide between cooperation and defection
– while from an individual perspective defection(Überlauf->Geständnis) is the rational solution, the overall outcome is better if both cooperate
• Game-theoretic prediction
– both players defect
• Empirical findings
– many people cooperate, trusting that the other person will also cooperate
Backward induction
– analysis of repeated decision problems
– starts from considering rational solution in last round
– from that solution, rational solution in previous round can be derived, and so on
Decision anomalies
Result from: – information processing (Speziell: misunderstanding from potential growth) – emotions – time – heuristics
Monty Hall Dilemma
– decision-makers have to decide whether to revise a previous decision when presented with additional information
– demonstrates that people have difficulties understanding probabilities
• Maximization strategy
– switch
• Empirical findings – very few people switch from their first choice – explanations: • misunderstanding of probabilities • regret
Emotions
Feeling and anticipated emotions are relevant for the decisions
Affective forecasting
– prediction of emotional reactions to future events
• Four components(Wilson & Gilbert, 2003) – Valence (Wertigkeit) – specific emotions – intensity – duration
Impact bias:
– duration and intensity of emotional reactions to future events often overestimated
projection bias:
tendency to assume that one’s future feelings and preferences will be similar to the current feelings and preferences
Delayed gratifications ( Time)
– people have trouble deferring rewards, even when future reward is significantly larger than present reward
– “Marshmallow experiments” (Mischel, 1974)
– particularly difficult when System 1 is dominant
Time discounting ( Time)
– preference for immediate utility vis-à-vis delayed utility in decisions where consequences occur at different points in time
Problems:
– short-term benefits vs long-term Costs
– discounting often inconsistent
Melioration (Time)
– people often choose the alternative that puts them in a better position for the time being
– in the long run, melioration can counteract maximisation
• Law of relative effect (‘matching law’)
– selection ratio of various behavioural alternatives is proportional to the subjective value of the reinforcements attached to these alternatives
– and inversely proportional to the time that passes between the behaviour and its reinforcement
Distorted Memories (Better-than-average effect )
– people believe they are better than others
– positive traits more developed, negative traits less developed than those of other people
- Often higher for low Performers
- Serves to enhance self-worth
Distorted Memories (Peak and End rule)
– in retrospect, an experience is often not judged by its duration and the sum of all its elements, but is strongly influenced by outstanding elements and by the elements at the end
Hindsight(Rückblick) bias
– after learning an outcome, people tend to claim to have known that the event would turn out that way
• Causes:
– adaption of memories to enhance self-worth
– not remembering, but new prediction processes with outcome information as anchor and “motivated” answers
Mood-as-Information heuristic
– judgments are based on the current mood
• Can lead to errors because relevant aspects are not considered and mood may be caused by unrelated events
Representativeness heuristic
judgments are based on the similarity between an outcome and a model
• Relevant for judgements of category, frequency, probability, as well as evaluations or decisions based on these
• Can be correct because objects within a category share Attributes
• Can lead to errors because similarity alone is not sufficient, and people neglect other important statistical rules
• Gambler’s fallacy
– mistaken belief that, after a sequence of one outcome, the other outcome becomes more likely
• Hot hand fallacy
– mistaken belief that, after a sequence of successes, another success becomes more likely
• Conjunction fallacy
– people do not take into account that combined events cannot be more likely than the constituting single events
Fast and frugal heuristic
decision-making aids that allow a complex problem to be solved in a short period of time, using only a few pieces of information
• Recognition heuristic
– decisions are based on familiarity with objects
– when only one of two objects is recognised, that object is seen as more important
• Less-is-more effect
– in some cases, less knowledge can lead to better decisions because it allows using the recognition heuristic
• Take-the-best heuristic
– decisions are based on the one characteristic of options that seems especially relevant
• Elimination heuristic
– decisions are based on sequential elimination of alternatives
