By the way, in googling something else I happened on this volume of a journal of High Ability Studies that appears to have a focus on just this question:
High Ability Studies, vol 18 issue 1, 2007
This is behind a subscription wall, but at least some of the papers are findable individually by googling their titles. In particular, the Ericsson et al. paper at the beginning looks interesting (and it starts by discussing the trainability of memory, as it happens). I haven't read it yet. The presence of another paper in the same issue by the same authors responding to criticism of their paper suggests that there is considerable controversy!


Email: my username, followed by 2, at google's mail