One argument for infra-Bayesianism is that Bayesianism has some disadvantages (like performing badly in nonrealizable settings). The existing examples in this post are decision theory examples: Bayesianism + CDT performing worse than infra-Bayesianism.
(And if Bayesianism + CDT perform badly, why not choose Bayesianism + UDT? Or is the advantage just that infra-Bayesianism is more beautiful?)
Are there non-decision-theory examples that highlight the advantage of infra-Bayesianism over Bayesianism? That is, cases in which infra-Bayesianism leads to “better beliefs” than Bayesianism, without necessarily taking actions? Or is the yardstick for beliefs here that the agent receives higher reward, and everything else is irrelevant?
I think realizability is the big one. Some others are:
Infra-Bayesianism lets you avoid bridge rules, for more detail, see the AXRP podcast.
The classical Bayesian simplicity prior, the solomonoff prior, might have problems with acausal attacks, though I am not sure how much I buy this. Infra-Bayesian physicalism still uses this, but also allows us to classify and discard these malign hypotheses.
One argument for infra-Bayesianism is that Bayesianism has some disadvantages (like performing badly in nonrealizable settings). The existing examples in this post are decision theory examples: Bayesianism + CDT performing worse than infra-Bayesianism.
(And if Bayesianism + CDT perform badly, why not choose Bayesianism + UDT? Or is the advantage just that infra-Bayesianism is more beautiful?)
Are there non-decision-theory examples that highlight the advantage of infra-Bayesianism over Bayesianism? That is, cases in which infra-Bayesianism leads to “better beliefs” than Bayesianism, without necessarily taking actions? Or is the yardstick for beliefs here that the agent receives higher reward, and everything else is irrelevant?
I think realizability is the big one. Some others are:
Infra-Bayesianism lets you avoid bridge rules, for more detail, see the AXRP podcast.
The classical Bayesian simplicity prior, the solomonoff prior, might have problems with acausal attacks, though I am not sure how much I buy this. Infra-Bayesian physicalism still uses this, but also allows us to classify and discard these malign hypotheses.