The QOTD, echoing Quine, on the impossibility of bare empiricist epistemology:

Indeed, there is a connection between belief and perception of beauty (and similar qualities) that goes much deeper than Weinberg suggests. As Leibniz and many since have noted, there are ordinarily many different theories or beliefs compatible with our evidence. If we plot our data on Cartesian coordinates, we will be able to draw as many lines as we please through the points we plot, and we could project any of the appropriately related hypotheses. All emeralds so far examined have been green; if so, however, they have also all been grue, where an emerald is grue if either it is examined before 2050 a.d. (bringing Goodman up to date) and is found to be green, or is not so examined and is blue. [390] So (instead of projecting that all emeralds are green) we could project that all emeralds are grue, thus concluding that emeralds not observed before 2050 are blue. The sun has come up every morning so far; we form the belief that it comes up every day and will also come up tomorrow. We could have formed quite a different belief, however: where T is today, we could have formed the belief that the sun comes up every day prior to T and never after T. Why do we accept the hypotheses we do; why do we project green rather than grue, and the hypothesis that the sun will continue to come up rather than the one according to which it won’t? Why do we project simple hypotheses rather than complex ones? Not because we have evidence that simpler hypotheses are more likely to be correct than complex ones; for, for any alleged evidence for this conclusion, there will be a more complex inference from the same data for the denial of this conclusion. So why do we do it?

Because we find simple beliefs (whatever precisely simplicity is) more natural and more attractive than complex beliefs. Only a madman would project grue or its partner in crime, bleen. [391] Messy, complex beliefs are ugly, disgusting, weird, repellent: we dislike them and therefore reject them. We may hope that the world is in fact such that simplicity (at least simplicity of a certain sort and in certain areas) is a mark of truth; but we have no hope whatever of establishing that in a way that doesn’t already rely upon simplicity. For suppose we note that in the last one thousand cases the simplest hypothesis has turned out to be true. Where t is the present, say that a belief is simplex’ if it is formed before t and simple, or after t and complex; what we will have observed, so far, is that simplex beliefs tend to be true. But that means that from now on we should go for complex beliefs.

– Alvin Plantinga, Warranted Christian Belief (available online for free, here), pg. 309-10.