>There was a series of studies a few years ago that put people in classic "shoot / no shoot" scenarios that police are trained on (i.e., you go through a scenario and have to shoot someone before they shoot you, but only if there's actually a gun). They randomly changed the color of the skin of the people involved. Civilians shot far more blacks than whites in "no-shoot" scenarios. Police shot about the same, but there was a longer delay: meaning, their impulse was to shoot blacks faster, but their training allowed a secondary impulse to come in and moderate the first one. (Sorry I can't find a link just now.)
That sounded interesting, so I tried to find some.
First one I found was this https://www.researchgate.net/publication/235256692_Results_f...
-- the conclusion seemed to be exactly counter to what the research you were talking about showed. "In addition, where errors were made, participants across experiments were more likely to shoot unarmed White suspects than unarmed Black or Hispanic suspects, and were more likely to fail to shoot armed Black suspects than armed White or Hispanic suspects."