Rule of the thumb: Closed Sicilian recaptures

During a training blitz game with a student, I realised that after this standard Anti-Closed Sicilian pawn block...

....f5, which stops the White pawn storm for some time, the subsequent play is not as 'automatic' as I would imagine. Although I personally had no concrete idea (probably it's just ingrained after many years of reading annotations by masters),my recollection is that, generally, after exf5, Black should respond gxf5 to keep the central pawn mass intact and and after gxf5, Black does well to recapture with exf5 to bolster the kingside. My student did not react as such and this allowed me to get the upper hand soon.

Out of curiosity, I asked GM Kevin Goh how he responds in such situations. he said, "I will retake the other way when they take with e-pawn,  I recapture with the g-pawn after exf5 too. Don't ask me why...I don't know actually". So even for a GM, it's more or less a standard instinctive option.

I rustled through my much neglected chess library (blame the laptop and pgn files) and dug out GM Danny King's take in his 'The Closed Sicilian (1997)'  book and his comments regarding gxf5 are 'I prefer the recapture with the e-pawn, Black's king keeps its pawn cover. The only drawback can be if the a2-g8 diagonal becomes a little sensitive or if White can put the d5-square to good use'. King also added that after gxf5 if Black responds with gxf5, 'I would be very glad to see Black recapture like this, it justifies White's blunt strategy'. 

There is a fairy tale ending to this story though as my student was faced with an analogous situation in a Filipino professional league game a week later against an NM and this time, he parried off the exchanges correctly and won clinically.

And so that was the end of the matter...and the coach feels that he has done his job...but just a few days ago....an automatic recapture by my opponent after exf5 with gxf5 basically gave my pieces full access to his kingside and he resigned soon enough.

Of course, these things are dependent on the position at hand so it's always better to put a disclaimer to chess concepts to indicate that they are general and not absolute rules.