The OP has already cited an important counterexample to this generalization.

... I would disagree with that entirely. I'd be highly surprised if a statistically relevant random selection size of judges, after being informed of the 'meal break distance' bias, would believe that it applied to any particular case they had decided.

Unless you have additional information -- new postulates -- whatever your conclusion was originally will remain your conclusion thereafter. If you get "primed" into thinking that it was colder yesterday than you normally would have, you will still believe that it was cold yesterday -- even if "primed" into thinking that the same temperature is "hot" today.