Philosophical Foundations of Probability Theory Flashcards
What is a probability distribution
function mapping each proposition in the relevant representation language to a number between 0 and 1 conforming to the axioms K1 and K2
What questions are we trying to answer here?
Who says degrees of belief should be probability distributions?
Who says that they should be revised by conditionalization?
What is a Dutch Book?
Sequence of bets, each of which the agent is disposed to accept (given his degrees of belief) which taken together will cause the agent to lose money no matter what happens.
What is the price of this ticket:
$1 if a
0 otherwise
$degree of belief in a
How do we demonstrate this ticket is a dutch book?
p(a) = 0.6 bet 1: $1 if a– 60p
p(b) = 0.5 bet 2: $1 if b– 50p
p(! (a or b)) = 0.25 bet 3: $1 if !(a or b)– 25p
p(! (a and b)) = 0.7 bet 4: $1 if !(a and b)– 70p
create a table for ll possible combinations of a and b
Put the result of each bet for each case (+ or - based on what paid)
Calculate overall outcome for each row
If a dutch book occurs, what do we know about the agents degrees of belief?
The degrees of belief violate K1 and K2
What is the fair price of this ticket?
$1 if a and b
$0 if !a and b
money back if !b
$conditional belief in a given b
$1 if a and b $0 if !a and b money back if !b We have our price $x for this ticket. We have the following tickets: $1 if a and b | $x if !b 0 otherwise | 0 otherwise What is the price of these tickets? Equate these to prove conditionalisation
$x . p(!b)
x = p(a and b) + x(a) p(!b)
= p(a and b) + x(a) (1 - p(b))
x . p(b) = p(a and b)
x = p(a and b) / p(b)
What does a diachronic Dutch book argument show?
a justification for the policy of conditionalizing
Explain the Diachronic dutch book argument
we want to say that if E has happened, p(A) = p(A | E).
If this is violated then p(A) < p(A | E).
We can show that this can create a dutch book through the following 3 bets:
Price $p(A | E)
$1 if A and E
0 if (!A) and E
Money back if (!E)
Price $a . p(E)
$a if E
0 otherwise
Price $p(A)
$1 if A
0 otherwise
From this we get payoffs for E and !E that show if p(A | E) != p(A) we get a dutch book
How do we solve the monty hall problem?
Let A stand for the car being behind A P(A and !B) = p(A) = 1/3 p(!B) = 1 - p(B) = 2/3 P(A | !B) = p(A and !B)/p(!B) = 1/2 But Monty opens one of the doors....so now we can say we know the car is not behind B: p(K !B | A) = 1/2 p(K !B | !A) = 1/2 P(A | K !B) = p(K !B | A) P(A) / p(K !B | A)p(A) + p(K !B | !A)p(!A) = 1/3