The same may account for the ease with which malapropisms are perpetrated. Someone very close to me (though she now denies it vigorously) once asked me: "What is the entomology of the word `malapropism'?" Once the word "entomology" has forced its way into your consciousness, it becomes very difficult for the word "etymology" to squeeze past it.
All of this, I have just realised, is what makes chess problems so problematic. Take the position in the diagram (composed by Dombrovsky in 1977). It's White to play and mate in two.
All White needs is a decent check and Black is mated. The queen isn't doing anything, so the first thing to try seems to be 1.Ra7, clearing the way for a threat of Qb7 mate. But Black plays 1...e5 and there's no immediate mate. 1...e5 is also the reply to 1.Rc7 (threatening Bc6) and 1.Rd7 (threatening Rxd6). In the first case, an escape is created on e6 for Black's king. In the second, the rook on h6 defends d6. It looks as though the rook must stay on e7 to keep e6 guarded.
The solution, however, turns all these thoughts on their head. The opening move is 1.Rxe6! not only freeing e6 for the king, but positively inviting it there. But 1...Kxe6 allows 2.Qf7 mate, while 1...Rxe6 (blocking the vital square) is met by 2.Qb7 mate. Finally 1...Ng6 (to counter the threat of 2.Nf4 mate) allows 2.Rxd6 mate.
The opening move, 1.Rxe6, is difficult to see because it disrupts so many of the patterns that exist in the initial position.