BERL and rationality

Over at Offsetting Behaviour Eric has A final word on rationality, including a nice coloured graph! He addresses, once again, the argument BERL makes about the use of strict rationality in Crampton and Burgess. He points out that their critique doesn't require strict rationality, just that on average across all of the customers that BERL assumes to be "harmful drinkers", the net benefits to these people is approximately equal to the "excess costs" of irrational behaviour.

I have said previously that I don't think you can write down an utility function with the standard properties which would have the characteristics that BERL seem to want of "irrational consumers", that is, prior to the 40 grams of alcohol threshold, benefits at least equal costs; after the threshold, benefits don't just equal zero, they're sufficiently negative to precisely offset all of the gross benefits from any prior consumption. In particular I can't see how BERL's function can be continuous. So I can't see how exactly BERL are modeling the "irrational behaviour" of some drinkers.

I think I understand how Eric and Matt think about consumer behaviour but I still don't understand the BERL approach.