L2 Article Willingham 11 - decision making Flashcards
Three ways in which we can define inconsistency
- Logical benchmark
- Probabilistic benchmark:
- Rational Decision benchmark: consistency of preference & choice
What is logical benchmark?
Consistency of belief
↪ making deductively valid inferences
If you believe A, you also certainly believe B. If you believe A&B and they are related in that way, you’re being consistent
- Person who says they believe A & B but then believes A but not B is being inconsistent
E.g. flipping the cards example - people struggle with checking the logical implications of a rule
↪ because we’re humans and we use language flexibly
What is conversational implicature?
Meaning that isn’t stated but which one can reasonably take to be intended given the context in which the sentence is written
- a statement only implies a given proposition only if the listener who is aware of the context would take that proposition to have been intended
- can be misleading when doing logical assessment
What is the belief bias?
When logically assessing an argument, we need to check whether the conclusion must be true from the premises
Belief bias → existing convictions/beliefs influence the evaluation of the structure of an argument
E.g. P1: All dogs are animals
P2: Some animals are German Shepherd dogs
C: Some dogs are German Shepherd dogs
↪ invalid argument = conclusion doesn’t logically follow the premises but we are inclined to believe it as true since we know german shepards are dogs
What we should do instead of what we actually do according to the logical benchmark?
What we actually do → what we should do
- Confirmation bias → strive for falsification
- Rely on conversational implicature → consider only relevant info
What is the probability benchmark?
Consistency of degrees of belief
↪ Updating beliefs in accordance with the rules of probability
If that is true or I believe this to some degree, I should accept that this is true
↪ It’s a matter of degree instead of certainty
Answer to picture 1 - why? What is the name of this error?
Don’t look at the answer yet, try figuring it out
The answer is B, the smaller hospital
Because if you find a result that deviates from the norm (60% instead of 50%), it’s more likely to come from the smaller sample size because if sample size increases, the sample mean will approach the population mean.
Sample size neglect
Example:
P1) P(7Habits|Effective)=high
P2) 7Habits
C) Effective
What is P (Effective|7 habits)?
What is the name of this error?
All the effective people do the 7 habits.
Let’s say the probability of finding a person that does the 7 habits but is not effective is 60%.
The probability of being effective given that I have those 7 habits = ?
= P(7 habits | effective) / (P(not 7 habits|not effective) + P(7 habits | not effective)
Picture 2
Base rate neglect
What’s the difference between frequencies and probabilities in terms of decision-making?
People usually perform much better on choice problems when it’s presented in terms of frequencies instead of probabilities
- some theories say that we evolved in preliterate societies where info would be remembered in terms of frequencies
What we should do and what we are actually doing according to the probability benchmark?
What we are actually doing → What we should be doing?
- Sample size neglect → use sample sizes
- Base rate neglect → calculate conditional probabilities
What is Rational decision benchmark?
Consistency of preference & choice
↪ deciding in a manner that maximizes expected utility
If you have a preference for this, then logically should also have a preference for this
What is rationality according to the rational models?
Consistency across the choices made
What does it mean when choices show transitivity?
Relationship between three elements such that if it holds between the first and second and it also holds between the second and third → it must necessarily hold between the first and third
↪ logical relation
According to the rational model, how would you calculate this?
Ticket = 1eur
We have 1 chance in a million to win 1000000 eur
Should we buy the ticket?
- We multiply 1 in a milion → 1/1000000 = 0.000001
- Calculate the whether it’s worth it buying the ticket
Preference = Probability of winning * (objective value - the cost of the ticket)
0.000001 * (1000000 - 1) = 0
We should not buy this ticket, it’s not worth it
What is the normative model (normative theory)?
Some choices are considered better than others
↪ one choice is optimal from among the possibilities
What makes choice optimal varies with the theory