Referee 1's review:

This paper represents a highly original attempt to study a complex and difficult problem: the neural correlates of improvisation. Without going into details, let us just say that the authors have approached this complex issue in a reasonable way, given that they are starting essentially from scratch. Indeed, the main reason why this reviewer will recommend the paper for publication is not because the results are so clear per se, but rather because the authors have decided to tackle such an unknown question. So regardless of the validity of the conclusions, the study will spark interest in the question, and will give others a starting point from which to pursue the many outstanding issues.

My main suggestion for revision is that the discussion should be shortened, as it is quite long, given the necessarily tentative nature of many of the conclusions. Much of the discussion takes the form of mentioning each region that was found to be active or deactive, and proposing some ad-hoc reasons why it may have shown the pattern it did. I would recommend avoiding this approach, and focussing instead on the more global patterns of results that were obtained.

The other recommendation is that perhaps the authors could add as supplementary data the results alluded to on p 8 from a conjunction analysis, as this analysis is a good way to deflect criticism that only six subjects were scanned in all.

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.