This is the 'not even wrong' scenario described by Elizier Yudkowsky over at Lesswrong.
My response is-
Think of them as constraints on observation/decision spaces
Suppose I have a Utility function for possible states of the world. Is it reasonable to believe this Utility function is independent of the mechanism by which those states of the world come into being? No. Otherwise 'the monkey's paw' wouldn't be a good story. Indeed, for any state of the world brought about in a way we thought incompossible with our Universe, something so drastic has happened to our Information Set that we must abandon the decision problem altogether and forget about our Utility function because we need to start thinking about the nature of this unexpected Universe we find ourselves in and what sort of things we ought to want in it (meta-preferences). One way to do this is to assign negative probabilities to states of the world arrived at by what we currently consider incompossible means. This immediately means that there is at least one new possible state of the world which hasn't yet been assigned a probability. So decision making halts till that new state or states have been specified or speculated about.
In the Pascal Muggle example, clearly the fact that a Matrix Lord exists completely changes what we want out of Life.
Job's agon, on the other hand, changes what we want out of death.
All Philosophy is practicing that death.
Thus Pascal's muggle isn't interesting- it's a wank for nerds who get off on words like 'gogoolplex' and random invocations of Kosmogorov complexity theory.