Philosophy | Theory of Knowledge
P562 | 26600 | Kaplan


Topic: Decision Theorry

Suppose you have five eggs in the refrigerator, out of which you want
to make an omelet for lunch.  Suppose you have already broken four of
them into a bowl.  Should you break the last of them into this same
bowl or use a new one?  If the egg is good, you’ll clearly be better
off if you break the egg into the same bowl: you’ll have the makings
of your five-egg omelet and only one bowl to clean.  On the other
hand, if the egg is bad, you’ll be better off if you break the egg
into a new bowl; you’ll have an extra bowl to clean, but at least you
will have the makings of a four-egg omelet.  Your decision problem
can be rendered thus:
			Egg is good		Egg is bad

Use same bowl		5-egg omelet,		no omelet,
			1 bowl to clean		1 bowl to clean

Use new bowl		5-egg omelet,		4-egg omelet,
			2 bowls to clean 		2 bowls to
clean

A natural thought is that what you should choose depends on how
confident you are that the egg is good, how confident you are that
the egg is bad, and what values you attach to each of the four
outcomes your choice may bring about.  The claim of Bayesian decision
theory is to have discovered, in considerable detail, how this
natural thought can be worked out:  how you can measure how confident
you are that the egg is good and how confident you are that it is
bad; how you can measure how valuable each of the possible
consequences is to you; how those measurements, once made, tell you
which of the two options is preferable for you.  The purpose of this
course will be to evaluate this claim.  This will involve examining
the foundations of Bayesian decision theory, drawing the (surprising)
consequences the theory has for our understanding of practical
reason, and looking at some of the (even more surprising)
consequences the theory has for epistemology.