,
5 tweets,
2 min read
Read on Twitter
1/5 This deserves a thread, @hulme_oliver.
"Utility" can be a scientific concept, or it can be a just-so word.
"Utility" can be a scientific concept, or it can be a just-so word.
2/5 Science operates as follows
a) propose a hypothesis, a mathematical theory (or model) of some physical process, e.g. of people making choices.
b) find a testable prediction.
c) check that prediction against observation.
a) propose a hypothesis, a mathematical theory (or model) of some physical process, e.g. of people making choices.
b) find a testable prediction.
c) check that prediction against observation.
3/5 Science can't operate without prediction. Prediction connects theory to data. Like the clutch in a car connects engine to wheels.
Utility theory makes the following kind of prediction.
1) Observe someone's behavior, and infer u(x).
2) use u(x) to predict future behavior.
Utility theory makes the following kind of prediction.
1) Observe someone's behavior, and infer u(x).
2) use u(x) to predict future behavior.
4/5 It's a scientific testable hypothesis that behavior optimizes a person-specific expected stable utility function.
But if u(x) is allowed to change, then no prediction is possible, science can't operate, and utility becomes a non-scientific concept.
But if u(x) is allowed to change, then no prediction is possible, science can't operate, and utility becomes a non-scientific concept.
5/5 It's not a scientific hypothesis that someone's action now tells us his current utility. That's just a dictionary definition of a word (utility).
"Paul did X"
"Paul had utility to do X."
With changing u(x) the latter is unnecessarily complicated and scientifically equivalent.
"Paul did X"
"Paul had utility to do X."
With changing u(x) the latter is unnecessarily complicated and scientifically equivalent.
