Although utility fails as a description of human behavior, it does come closer to being a norm for human behavior (e.g., one can argue that individual preferences should be consistent). Of course it then collapses as a model if one attempts to build from there to collective preferences. Further, the collective-preference problems cast further doubt on the applicability of utility to "in-dividuals", since we seem more divisible when examined in psychological/neurological depth. And this, perhaps, undermines whatever claim it may have to defining a norm.

Words and ideas I offer here may be used freely and without attribution.