I don't think it's useful to say you can have a credence of 1.0 in a belief (love in this case), admit that 1.0 means it can't rationally change at all, and then say that it can be irrationally changed. It sounds like a roundabout way to say the credence is actually some number less than 1.0, and it takes a very strong update (which the author mislabels as "irrational") to affect the credence much.