We've all done the same thing...check and recheck our calculations then play something else! There is also another question that comes up sometimes and that is, “Should you trust your opponent?” In the fourth game of their 1965 Candidate's Match, Larsen played the unusual 5...Nd7 which caused Tahl to spend 50 minutes calculating 6.Nxf7, but he ended up playing 6.Bc4 instead. Tahl wrote, “My intuition insistently kept telling me that the sacrifice had to be correct, but I decided to calculate everything as far as mate, spent some 50 minutes, but then in one of the innumerable variations I found something resembling a defense, and rejected the sacrifice. This was a betrayal of myself, I saved the game only by a miracle after the adjournment.”
I suspect part of Tahl's problem was that if a player of Larsen's caliber offers you a chance to sacrifice a piece for a winning attack on move 6, he must know what he's doing and the attack is unsound! Tahl decided to calculate everything out to the end, but when he couldn't, he decided to trust his opponent.
The following year in a blitz game played at the Piatigorsky Cup in Santa Monica, California, Larsen played 6...Nd7 against Fischer, who was clearly willing to take a draw, but Larsen declined the invitation and quickly lost. I found 5 or 6 games in online databases with the move and white won them all. Engine analysis with Stockfish seems to indicate that the position does favor white, but there is no forced win. For a more complete discussion of the variation I suggest the article by Gary Lane at Chesscafe HERE.