To extend on @BenCole comment, an interesting summary of different models of time perceptions can be found in this paper. These models are in a sense more descriptive than the fundamental biological hypothesis mentioned by caseyr547, so you might not be ready to call these "explanations", depending on what you mean by that.
The models attempt to give a precise (functional) form to the extent to which time perception decrease as the "stock" of experienced time increases.
A simple model which seem to have initially attracted some attention is the logarithmic time perception model (see http://www.kafalas.com/Logtime.html#LM) :
[perception of change from time $t$ to $t'$] = constant * $[\log(t') - \log(t)]$.
Apparently, this model was an attempt to link time perception with the so-called Weber-Fechner law relating the intensity of a perception to the magnitude of some initial stimulus of the same nature (e.g. the perceived increase in the weight of an object depending on the initial weight of the object).
However, researchers seem to have found little evidence of such relationship in the case of time perception (see the aforementioned article and http://link.springer.com/article/10.3758/BF03204158#page-1). So other model have been developed. The first paper describes some of these newer models.