Flashcards in The nature of rationality Deck (58):
What is the basic aim of studying the nature of rationality?
To examine the broad question of whether humans are rational.
What are the three kinds of models of rationality?
Descriptive, prescriptive and normative models.
Define descriptive models of rationality.
Models of how thought processes operate irrespective of whether the decision is good or bad.
Define prescriptive models of rationality.
Models that state how we ought to think in order to make the best decisions.
Define normative models of rationality.
Models that evaluate a decision in terms of the goals of the decision maker; a decision is good to the extent it reaches these goals.
What is a good (/rational) decision?
It's normative, it maximises utility, and rationality is subjective.
What is meant by the statement that 'rational thinking is normative'?
That a decision is good to the extent that it allows the decision maker to reach their goals.
What is meant by the statement that 'a rational decision maximises utility'?
It maximises utility to the decision-maker, where utility is the extent of goal achievement.
What is meant by the statement that 'rationality is subjective'?
We have different goals, thus making what is rational to one person, irrational to another
What did Todd and Gigerenzer (2000) state about model types (i.e. visions of rationality)?
There are four different types of model, which can be divided by whether they're 'demons' or 'bounded rationality' models.
What's the difference between a 'demons' and 'bounded rationality' model of rationality?
Whether our evolved cognitive processes (architecture) are capable of making a rational decision - 'demons' models believe we can, 'bounded rationality' that we can't.
What are the four types of rationality models outlined by Todd and Gigerenzer (2000) and which category do they belong to?
Demons: unbounded rationality + optimisation under constraints
Bounded rationality: satisficing + fast and frugal heuristics.
What type of rationality model refers to the normative benchmark?
Unbounded rationality. Believed by a subset of economists.
What is the main premise behind optimisation under constraints models of rationality?
That we are (i.e. our cognitive architecture is) capable of making rational decisions, but that we are lazy/uninformed.
What are the main ideas behind fast and frugal heuristics models of rationality?
That we have evolved specific decision-making apparatus that are domain-specific, forming rules of thumb.
What are the advantages and disadvantages of using rules of thumb to make decisions?
They're fast and involve little computational power, but they don't always result in perfectly rational decisions.
What kind of model is Rational Choice Theory an example of?
What is meant by the statement that RCT is social physics?
It's a search for universal mathematical laws to explain social and economic phenomena in terms of the behaviour of their “fundamental particles” (human decision makers).
What do individuals do in a conventional rational choice model?
They strive to satisfy their preferences for the consequences of their actions given their beliefs about events which are represented by utility functions and probability distributions and interactions among individuals are governed by equilibrium conditions. Normative benchmark!
What models does the conventional rational choice paradigm include?
Most standard models of mathematical economics, finance theory, business research (e.g. marketing) and statistical decision theory. Also rational-actor models that have become prominent and controversial in other fields such as political science, sociology, philosophy and law.
What is the normative benchmark?
The most rational decision possible against which decisions are compared - involves a perfect decision maker with perfect apparatus.
How far back does the modern RCT date?
50 years - three publications:
- von Neumann and Morgenstern’s Theory of Games and Economic Behaviour (1944/47)
- Kenneth Arrow’s Social Choice and Individual Values (1951)
- Savage’s Foundations of Statistics (1954)
What were von Neumann, Arrow and Savage's publications a part of?
A general ferment of cross-disciplinary operational research during the early post-war era which
laid the foundation for a dramatic escalation in the use of mathematical methods in the social sciences.
What mathematical methods exactly did cross-disciplinary operational research during the early post-war era encourage?
Axiomatic methods, concepts of measurable utility and personal probability, and tools of general equilibrium theory and game theory.
Where can the roots of RCT be found?
In the first formal use of equilibrium models by marginalist economists in the 1800s, Bernoulli's proposal of expected-utility-maximisation in the 1700s, probability theory in the 16th and 17th centuries, and even Euclid's development of the axiomatic method.
In what situations is choice under uncertainty used?
In judgement and decision making experiments.
What is the concept behind choice under uncertainty?
That given a choice between two or more alternatives the rational choice is the option with the highest value or utility.
What is the difference between value and utility?
Value is numeric and often monetary, whereas utility is abstract.
How do we measure expected value (EV)
We multiply the utility by the probability of receiving that outcome.
What is the EV of £100 with 50% probability?
According to RCT, what do all decisions boil down to?
An estimation of EV.
Why can any choice be represented as a gamble?
Because we cannot know the future.
What is the main premise of Expected Utility Theory?
All choices can be represented as a gamble, and we can apply EV to Expected Utility (EU).
Give an example of the application of Expected Utility Theory.
Pascal's wager - the comparison between living a holy or a secular life, considering the probability that God exists and the utilities of each outcome (i.e. heaven/hell, fun/less fun).
According to Expected Utility Theory, what is a rational choice?
The one with the highest EU.
What kind of theory is Expected Utility Theory?
A normative theory - it states what the rational choice is.
What arguments does Expected Utility Theory use?
The long-run argument and the argument from principles.
What is the long-run argument?
That because EU is based on probabilities and gambles, the rational choice is based on an average of what would happen if you were able to make the same decision on a number of successive occasions, or if a large number of people were to make the same decision.
What are the three choice axioms used in theories of rationality?
Weak ordering, transitivity and the sure-thing principle.
Define weak ordering.
The idea that for any set of choices we must always be able to say we prefer one over the other or neither. Because we can always express the utility of an option as an integer, we can never say that the two cannot be compared.
As utility is expressed as an integer our choices must be transitive, i.e. x > y, y > z, therefore x > z.
What did the Sure-thing principle arise from?
Multiplications of probability rather than the numerical status of utility.
What is the Sure-thing principle?
If there’s some state of the world that leads to a particular outcome irrespective of our choices, then this is cancelled out of the equation and we should not let it affect our choices.
What example was given by Bernoulli to show that the idea of EV computation is flawed?
The St. Petersburg Paradox.
Describe the St Petersburg Paradox.
Peter agrees to give Paul 1 ducat if Paul gets heads (toss of a coin) on the first go, 2 ducats if he gets it on the second, 4 for the 3rd, 8 for the 4th etc.)
The expected value of this gamble is infinite, but "any fairly reasonable man would sell his chance for 20 ducats"; the subjective utility is low, as indexed by people’s willingness to sell their chance to play for a small sum.
How did Bernoulli (1738) explain the utility of wealth?
By suggesting that EV doesn't equal EU. Instead the utility of wealth is proportional to its logarithm - the utility of each unit decreases as total wealth increases (marginal declining utility).
What are three violations of normative theory?
The Allais Paradox, risk seeking vs. risk averse behaviour, and that people don't integrate prospects with existing assets.
What is the Allais Paradox?
Maurice Allais (1953) showed that people don't maximise expected utility and violate Savage's choice axiom (sure-thing principle).
How do people violate the sure-thing principle and normative thinking on Allais' tasks?
According to Tversky & Kahneman (1974), most people prefer 1A (£100m for sure, rather than a chance of £100m, £500m and £0) and 2B (10% chance of £500m over 11% of £100m)
How do Allais' tasks show that people violate the sure-thing principle?
We should subtract common outcomes and outcomes with zero utility from the equation, as these shouldn’t affect our choice. In Choice 1, preference for £11m over £50m clearly doesn’t maximise EU, but in Choice 2, this results in a rational preference.
What did Kahneman & Tversky (1979; 1992) find about risk seeking vs. risk averse behaviour?
Most people choose a certain gain even if the EV is higher for an 80% chance of more money, but sure thing preferences are reversed when outcomes are losses. People are risk averse for gains and risk seeking for losses.
How does risk seeking vs. risk averse behaviour differ when the amounts of money involved are small?
Most people become risk seeking for gains and risk averse for losses - it switches round.
What does utility theory suggest about integration of decision outcomes and current assets?
That we should do it - we should choose the riskier option for both gains and losses when they have the same EV
What evidence is there that people don’t integrate prospects with existing assets?
Given the choice of two options with the same EV, people tend to pick the certain gain and risky loss options.
What did Gigerenzer & colleagues do?
Consider the origin of heuristics and re-defined rationality.
What did Gigerenzer & colleagues suggest about heuristics?
That they were evolved in the Environment of Evolutionary Ancestry (the Pleistocene period) on an ad hoc basis to satisfy particular problems.
What did Gigerenzer & colleagues re-define as genetic fitness?