There have been numerous studies showing the inherent bias towards people of color. U of Chicago even did a first person shooter test with the same results. There were measured physical response differences when targets were black vs white.
If you're referring to the same study that I recall being reported, there was a mix of participants from civilians and police. And while there were indications of people making more mistakes while the target was black, the police participants were able to make the right call at a higher % than the civilian participants. There was actually a small (few ms) extra delay on the decision making of the officers. The suggestion was that training kicked in and overwrote the "gut" feeling that the civilians were reacting on.