...his first inputs seem to be an adjustment for something the ap had trimmed out.
My understanding (which may well be incorrect) is that, when the a/p drops out, alternate law will maintain the pitch/roll status current at a/p disconnect. The implication in the report is that the right roll began after a/p disconnect, presumably as a result of turbulence. My understanding (again, possibly incorrect) is that the computers would have corrected the roll to the previous (presumably straight-and-level) setting without pilot input. But that doesn't explain why the PF believed he needed to get the nose up.
The question left by the report is why the PF did anything at all. Obviously there was a reason for what the PF did, even if not doing anything might have produced a better result. If it was the airspeed displayed on the right side display, it wouldn't be recorded in the FDR, which does seem to be a design oversight.
A related question is why the crew didn't work the QRH for unreliable airspeed.