As Dan Bron observes in a comment above, the preference for "hungry for X" over "thirsty for X" depends greatly on the identity of X. Here is an Ngram chart of books in the Google Books database across the years 1800–2000 for "hungry for knowledge" (blue line versus "thirsty for knowledge" (red line) versus "hungering for knowledge" (green line) versus "thirsting for knowledge" (yellow line) shows nothing like a 250% preference for "hungry" over "thirsty":
On the other hand, if we rerun the Ngram chart with love in place of knowledge, we get a very different-looking chart:
And if we replace love with life, we get a third chart, whose results fall somewhere in between the other two:
Despite any reinforcement that these first three chart might give to such a theory, I am skeptical that a division based on "need" versus "desire" as a guide to choosing between "hungry" and "thirsty" could withstand serious scrutiny. For example, consider what happens when we replace life with information (surely a closer match to knowledge than either life or love is):
Although the four charts presented above show varying margins of preference for "hungry for" in recent years, they also show that "thirsty for," "thirsting for," and "hungering for" remain in use at a significant level. I would let the circumstances of the narrative guide me in choosing one wording or another; I would not adopt a rule that consistently preferred one wording over the others and didn't take into account the specific context.