Moonwalking with Einstein by Joshua Foer. It is about a year that he spent honing his memory sufficiently to win the U.S. memory championships. Most of the book is about memory and its relationship, if any, to intelligence. He never really reaches a firm conclusion, but the book is delightful to read.
What I focused on most was the chapter on the "plateau" phenomenon. That is, when you try to master a skill, you may reach a point where you are no longer improving. Foer, citing Anders Ericcson and confirming with his own experience, says that what is happening at a plateau is that you are doing too much on auto-pilot. Instead, you have to jar yourself into engaging in the activity more consciously. Thus, if I were trying to get better at my Othello endgame, I could not just play more and more games the way I have been playing them. Instead, I would have to study the endgame, identify my weaknesses, and work to correct them. In fact, that is exactly what I did in 1987, when I had my best year. This is not a pleasant process, however. By reducing the role of your autopilot, you actually get worse for a while.
I wonder if there is an analogy with firms or even larger economic units. That is, a firm is bound to operate on "autopilot" to a large extent, but if it does so it will reach a plateau. And maybe firms or larger economic units sometimes have to cut back on autopilot and do worse for a while in order to escape a plateau.