I really just don’t get it.
The way we identify faces is by looking at certain landmarks on the face, allowing us to recognise a face we’ve seen before and distinguish between different faces. Dogs have actually learnt to do this with human faces, using the correct landmarks to identify humans faces, and even learning how to read out emotions from our facial expressions. My point being that this is an easy thing that our brains do without any conscious input from us (at least for neurotypical people, neurodivergent children may need coaching or special programs to learn to identify faces better) and social animals that we have domesticated have learnt this skill as well. 
This is relevant, because the only underlying (evo-psych) explanation for racism that I’ve heard is that seeing a face that isn’t what we’re used to seeing (i.e. is different to the faces we learnt facial recognition on – someone who doesn’t look like our friends and family) might trigger some kind of “oh noes, danger” alarm in the back of our heads, as back when we lived in family groups or tribes, people who looked different would be from other groups/tribes, and might be trying to attack us.