My question relates to lines which in general are covered by Avrukh (2015). However, Avrukh doesn't explain the phenomenon I noticed in the following lines. These lines are classed as Bogo-Indians, but I've kept the Catalan move order preferred by Avrukh.
[FEN "..."]
1. d4 Nf6 2. c4 e6 3. g3 Bb4+ 4. Bd2 Qe7 5. Nf3 Nc6 6. Nc3 Bxc3 7. Bxc3 Ne4 8. Rc1 d6 9. d5 Nxc3 10. Rxc3
Here out of 271 games in my database there are 160 with 10...Nb8 and 106 with 10...Nd8.
[FEN "..."]
1. d4 Nf6 2. c4 e6 3. g3 Bb4+ 4. Bd2 Qe7 5. Nf3 Nc6 6. Nc3 Bxc3 7. Bxc3 Ne4 8. Rc1 O-O 9. Bg2 d6 10. d5
Here out of 768 games there are 264 with 10...Nd8 and 69 with 10...Nb8. However, the most popular move (434 games) is 10...Nxc3, which leads us to the following position
[FEN "..."]
1. d4 Nf6 2. c4 e6 3. g3 Bb4+ 4. Bd2 Qe7 5. Nf3 Nc6 6. Nc3 Bxc3 7. Bxc3 Ne4 8. Rc1 O-O 9. Bg2 d6 10. d5 Nxc3 11. Rxc3
Now there are 535 games, of which 370 go 11...Nb8 and 146 go 11...Nd8.
Why is there a distinct preference for ...Nb8 in the first and third positions, and a distinct preference for ...Nd8 in the second position?
Avrukh, Boris (2015). Grandmaster Repertoire A1 – The Catalan. Quality Chess.