Some people are better than others at maintaining good situational awareness. I would not belittle training and experience. But even very experienced pilots can become disorientated, as indicated by many sad examples among the accident investigation reports I have studied over the years.
Therefore, IMHO, any instrument which might work counter-intuitively in unusual circumstances, significantly increases the risk of disorientation and its associated consequences if those unusual circumstances should arise.
It remains my tentative view that the stall warning (the one at 2h10min51, perhaps; but mainly the one at 2h1202-17) could well have
seemed to indicate that the nose-down inputs were for some reason causing the aeroplane to
enter a stall, so discouraging the PF from developing the very inputs which might have led to an
exit from the stall.
According to the preliminary report the angle of attack "n’est pas
présentée aux pilotes": ie, the instruments did not include an AoA indicator. No longer trusting the airspeed or other indicators ("je n’ai plus aucune indication", and "on n’a aucune indication qui soit valable") it may have seemed to the PF that he faced two risks: going too fast for the airframe, or going too slow for the AoA. He seems to have discounted the second of those risks
despite the stall warnings. So stated, this was an irrational response. Therefore, it is probable that something made it appear rational at the time. I suggest that he discounted it
partly because of the stall warnings or, rather, because of the timing of the stall warnings and their seeming linkage to his nose-down commands.