people do not realize that the [inferential] process is biased by their goals, that they are only accessing a subset of their relevant knowledge, that they would probably access different beliefs and [inferential] rules in the presence of different goals, and that they might even be capable of justifying opposite conclusions on different occasions.”
