When he bets on the river, then you are getting 8:1 to call. From his perspective, the odds against him bluffing should be 8:1 too. This is assuming that he is bluffing with an unexploitable frequency.
One interesting aspect is this: Suppose that your hand can beat every bluffing hand, but no other hands whatsoever, and suppose too that your opponent bluffs exactly one time in nine here. Then what? The answer is that it doesn't matter what you do. It doesn't matter at all.
Another interesting aspect is that if he knew that you will fold more than one time in nine here, then he can bet profitably in this spot every time. Conversely, if you fold, say, one time in twelve here - and he knew of that, then he should never bluff.
In this case, if you use your cards to randomize your calldowns of possible bluffs (which is not a bad idea), then that overpair is nowhere near the bottom 11.11...% of your "showdownable" hands here. Would you fold AA? Probably not, but AA has only a tiny bit more showdown value than TT here.
One more thing. If the opponent is holding a pair here then betting and folding to a raise may not be that bad a play on his part. If his options are "check-call" and "bet-fold", then "bet-fold" may be better for him because the board isn't very inviting for a thin value bet in position on your part because you would need some 65% equity if called. He doesn't need much equity at all to make the bet. He needs only be sure that he wouldn't be folding to a bet from you, and be fairly sure that you wouldn't bet a worse hand. This perhaps sounds murky, but suppose that you have AK/AQ and the board is slightly less threatening and the pot is slightly bigger. Would you fold? Hardly. Would you bet if checked to? Hardly. Would you raise him as a bluff? Hardly.
A good reference is "The Theory of Poker" by David Sklansky. It's quite boring to read, but it is very very correct.
/Johan =