Here's an observation which may or may not be relevant. The plane decelerated from horizontal Mach 0.82 to 107 knots over a period of 3 min 30s. By my estimate that is an average horizontal deceleration of around 0.1g.
It seems to me that, given no visual cues nor instrument readings, a pilot flying by the seat of his pants may interprete a pitch up attitude coupled with such a horizontal deceleration as level flight. Furthermore, lowering the nose would immediately decrease the horizontal component of deceleration, which the pilot may perceive confusingly as his nose pitching up (which is what it has been all along). Pulling back on the stick would return to the illusion of level flight, albeit with a stick operating "in reverse". (This is assuming the control sufaces are still capable of controlling the a/c.)
Note the report states "inputs made by the PF were mainly nose-up", (my italics) and in its 3D diagram the descent appears to consist of straight lines.