Why is this the best move and is it human?

Hi folks, 800 elo here, I just have a simple question that I've been trying to figure out.
On move 22 white plays Na5, and I was considering playing Rxa5, as it felt like it would help my position, but I ended up not playing that because I did not want to give a rook for a knight.
Upon analyzing the game, stockfish prefers to take the knight with 22. Rxa5 because it can see in the future.

Am I human for not taking the knight, or would you, in my position take the knight?

I mean... I don't see Rxa5 as being something about helping your position (if anything it hurts your position by giving up material) I also don't think of it as seeing into the future, it's about reducing white's play (not letting white play Nxc6 or Nxb7).

Since you're up a piece you can do whatever you want. What's important is that you keep your pieces active while not allowing a move like 24.Nxb7 (which happened in the game) which gave white a far advanced protected passed pawn.

For example a simple idea would be 22...Qd7 and double rooks on the a file.
For an example of a sacrifice that helps your position, I like 22...Nxe3 (aggressively activating your pieces).
22...Rxa5 is fine, but not if you were playing it randomly based on intuition (because giving away material randomly is bad). There are many ways to win this position, but winning it with 22...Rxa5 involves understanding the move reduces white's play, and if white can only make passive defensive moves, then your extra material will win eventually.

By the way, without the idea of capturing one of your pawns, your intuition should not be to remove the pin. Not only is the knight passive while pinned, but the pieces stuck defending the knight are also passive. So unless you saw the idea of the knight capturing one of your pawns, your intuition should not be to capture the knight.

So yeah, it could be a human way to win I guess... at least it's not a crazy engine move.

Thank you, you provided some very helpful insight and thoughts.
I could not really see how the game would turn out if I played 22. Rxa5, so I shelfed that idea.
I hindsight, I should have played 22. Nxe2 (it turned out to be a brilliant move), as I did calculate a line where I could pressure the king in his corner and simplify the position, but I ended up doing neither and screwed myself.

One thing you made me realize was that I did not consider the pieces defending the knight as passive, but you're completely right.

Thank you again for your time!