RobBB has made some recent posts about how an agent's representation of the world can interact with its perceptions and plans. That seems like a good topic, so let's discuss it.

Bonus question: If an agent represents its utility as a function over states of the world including itself, can it still wirehead?