@cshentrup No, not necessarily.

As an existence proof (not usually good decision rule!), consider random choice. However you model your choice set or space, one can always choose a random item or point. It's an entirely distinct method from attaching utilities to items or points and choosing a maximum. It involves no intransitivities (it doesn't involve any rankings that might be intransitive!) or self-contradictions.