I apologize in advance for the wordiness and vagueness exhibited below. The latter is something which I believe is unavoidable given the topic, and the former something which I slip into when I talk about this stuff in an effort to be more precise.
So I was thinking.
I assume for the moment that the universe exists, and I know that I exist by Descartes.
Now consider the fairly abstract question of whether or not anything 'matters', or is 'important' in a general sense. I'm not sure how to elaborate on this really, except through example. Maybe it's important that everyone eat at least one cookie a week. Maybe some religion or other is correct, and therefore its tenets matter. For the purposes of this post, 'good' can be identified with 'doing the things that matter' where as 'bad' can be 'not doing the things that matter', for suitably general definitions of 'doing'.
Now some people, often referred to as nihilists, would say that nothing matters, or else that we can't empirically prove whether or not any particular thing matters, so we should act as though nothing matters. They generally do not put this into practice, as evidenced by the fact that some of them are still alive. The reason usually given for this is that biological urges act independently of any possible general 'meaning' (though I prefer not to use that term due to its connotation of having one who means, I cannot think of a suitable replacement) and that we are all essentially naturally arising non-conscious robots. I have a problem with this interpretation, namely that I exist, but that is irrelevant to this question for now.
To consider the question of whether or not anything matters, consider the form of Pascal's wager. There are very flawed implicit assumptions Pascal's original formulation, the most obvious being that he failed to take into account the possibility of multiple religions. But the logic behind the wager is in and of itself, to the best of my knowledge and reasoning, valid. If I assume that the set of all things which could matter is non-empty, I must consider the following two possibilities.
1. something matters
2. nothing matters.
(note here that I make no mention of what it is that matters in the case that something matters. That this is empirically indeterminable is assumed for the purpose of this post)
In the latter case, it doesn't matter what I do, so there is no difference whether or not I act as though something matters.
In the former case, something matters, so in almost every case if I act as though something matters I might be right while if I act as though nothing matters I am guaranteed to be wrong*. In every case, acting as though something matters is at least as good as acting as though nothing matters.
So assuming you agree with me up until now, that still leaves the question of what I should believe matters, since 'nothing matters' is out of the question. But since I am assuming that whether or not any particular thing matters is empirically indeterminable, the odds of any particular thing being what matters should be the same as of any other particular thing mattering. So whatever thing actually matters, or even if anything does, should not affect my decision to act as though some particular thing matters. The end result of all this is that it doesn't matter what I act as though matters, so long as I act as though something matters.
Now consider what happens if knowledge that something does not matter is empirically determinable, but not innately known. So I can find stuff out, but I start with nothing. In the beginning the same logic as above applies. But now I can find out over time somehow that I am wrong about thinking some particular thing matters, and change my beliefs accordingly. However, as noted in the asterisk below, the only case in which acting as though something matters is not preferable to acting as though nothing matters is the case in which an infinite number of things might matter. So it is impossible to empirically determine that I should act as though nothing matters under any circumstances (except those where I assume that the set of things which can matter is empty).
In the case where I don't assume that the set of all things which can matter is non-empty, I have to also consider the case where the set of things which can matter is empty. But that case is equivalent to the 'nothing matters' case I considered initially, so I get the same result, that acting as though something matters is just as good as acting as though nothing matters. So even in the case where that nothing matters is empirically determinable to be true it is not better to act as though nothing matters.
So the moral of all this I guess is... don't be a nihilist, unless the possibilities for what matters are infinite? Otherwise do whatever you want.
*There is a slight complication to the problem in the case where what matters is 'acting as though nothing matters'. However, the odds of this case occurring, being indeterminable, can be considered equal to the odds of the case where what matters is 'acting as though something matters'. This renders both sides equally desirable in exactly the case where there are infinitely many possibilities for what things matter, and 'acting as though something matters' more desirable in all other cases. So if I want 'acting as though nothing matters' to be the best action, then you must accept that there are infinitely many things that could matter, but not if you want some particular thing to be what matters. Logic is weird.