I think this a valid concept that i have been taking into account in Live poker games for the past couple months..( after reading FTR obv) .. I would rather fold on earlier streets with the example N2intheworld mentioned because the games are deep and the barreling frequencies are high ( esp if u show weakness).

However in OP's post I feel that having a perfect read is a bit tougher. First of all 9-10 and AK (or similar hands) are very exact hand ranges, and if this were the case folding is probably better. However, having 78-and similar hands in there raising range on the BTn might skew the equity calculations considerably. Esp since i would expect him to 3brl on these particular run-outs. If that were the case then there were more hands in their value range that we beat while simultaneously having his bluffing range in their as well. That might make calling down better.

Another point to take into account is if he's on a higher level and he knows that u will think that u expect him to have 2nd nut or nut hands in his range while u might have a strong bluff catcher in ur range so he could look polarized (skewed ofr value) to get us to fold out those better hands.

I feel that these decision really depend on a variety of factors and aggregating them into bunches to make playing these tough spots more robotically is probably worse than making the best rational decision u can at that moment. And if his range is wide enough for him to be doing this for a couple of reasons then we just have to chalk it up as a cooler and call it off.