When you're stumped on critical positions, how do you find better candidate moves?

Most games will reach a point where there is at least one critical position, sometimes I find the best move. Other times, no matter how long I look, I simply cannot find good candidate moves. For example, this problem stumped me in tactics trainer: 

(problem taken from https://www.chess.com/tactics/78969 )

Move 25 just gets by your tactical vision. I knew I wasn't finding the move, if I could have stopped the problem there, I would have. I spent 5 minutes 34 seconds on the problem. All the moves in the problem came instantly to me except for the critical move which I thought on for 5 full minutes, and could not find. Something was wrong with my candidate move-picking thought process that I could not find the move no matter how long I looked. I knew I wasn't finding it, but was at a loss for how to do a better job of spotting it.

I remember finding some sort of grandmaster video out there which mentions re-picking your candidate moves. Never really followed the video, but I recognize this as an area that I can improve in when I was stumped in the above puzzle.

When I know all my candidate moves suck, how do you go about finding new candidate moves? A passive-looking move like retreating a bishop isn't easy to spot. But when I know I haven't found the best move yet, and had all the time in the world, you'd think I'd eventually find or at least consider the best move. I might have considered all places the bishop could go and figured retreating the bishop accomplishes nothing.

Those of you which are able to find such moves,
what does your thought process look like that you're able to find them?

I mean, I solved this in 27 seconds. It seems pretty logical to me - you want to save the knight, so just create a safe square for it by making the bishop go to a3. I don't think this is any different from you missing any other tactic - you just didn't consider the tactical idea.

*However*, here's some general advice concerning candidate moves in general. 

Your intuition and logical skills will quickly come up with candidate moves in any position by themselves - they will look at the forcing moves and positionally motivated moves. However, if you do it just like this, you will probably miss some moves, especially in complicated positions. 

The key is:

*In complicated positions, or positions where there are tactics but you just don't know what to do (like in the above position), you should go through each and every one of your pieces, and consider all legal moves possible for a few seconds, one by one. Consider them all seriously for a bit.

This should only be used in very messy positions or as a last resort - typically you want to save time. However, if you did this, you would find Ba3 is reasonable (since you didn't find it through pattern recognition or logic, obviously), and you would consider it as a real candidate move.

But part of it is also your tactical skills. Black's threat with Rb7 is to play Bf5, winning the trapped knight. So I wanted to create a square for the knight to go to, and Ba3 is logical.

Even if you didn't think logically, but went through *all* legal moves, you would take no more than 30 seconds to find Ba3.

*King moves - all useless

*Rook moves - only safe move just forces black to win the knight

*pawn moves - obviously no use

*Bishop moves - and soon you see the only reasonable move in the position

Process of elimination, ah.

This method of thinking is just like that of a computer's. It makes sense that, since computers are so strong, we should model the parts of the computer's thinking algorithm that we have the capacity to carry out as our own. 

We can't calculate millions of variations or use a complicated evaluation algorithm at each position we consider, but we *can* consider all legal moves in some positions (when it's necessary), and get rid of the "human bias", which is a topic that many intermediate players will inevitably stumble across at some point. You missed Ba3 becaue you were biased.