I didn't think about whether these blocks were active responses, or passive omission of a justified inference (eg., disconnected beliefs).

It operates as a metaphor by suggesting co-evolutionary dynamics as a way of looking at the problem. It's not a valid metaphor for trying to figure out the exact mechanism.

As it stands now, it's all omitted inference. But I think the monk is the default--almost all inferences are omitted. If that's the default, I think drawing attention to them and calling them "antibodies" is a figure-ground error. (But maybe you don't think it's the default.)

I might talk about co-evolution, not between beliefs and blind spots, but between actions and excuses. The excuses can't be too incoherent, because some people pay some attention to them. What I took to be "antibodies" were elaborate excuses, excuses for not drawing inferences between the first-order excuses, but I think the race example was the only example you gave of this. Maybe these are rare and most people just use first-order excuses for what they do, not excuses for why they don't actually follow the first-order excuses.