The logic is that you are looking to maximize an advantage. If you can be disproportionately better by continuing to train beyond what is basically only a "perceived" limit, then I think you have to do so. Think of the work that it takes to bring a player of atrocious skill to pitiful, now think of the work involved in taking that player to mediocre. Even if you one position train, you're either going to reach a point of diminishing returns (due to training the same setting for too long), or it is going to take you two seasons just to bring a player's skill to a level that is still comparatively low. However, if you were to bring a prodigious attribute to colossal (and beyond), you are going to be statistically in the top 0.5% of all players. To me this is much more valuable. Additionally the category in question is driving, which is covered under multiple training sets, so you could (in theory) target this indirectly.
Two schools of thought, and I'm in D4 so I'm trying to say I know everything thing here.