>UL and UR think that that policy is optimal! Whatever UL and UR are they're introduced as "reward functions". They "see", they "prefer", they "think" etc. Is

>UL and UR think that that policy is optimal! Whatever UL and UR are they're introduced as "reward functions". They "see", they "prefer", they "think" etc. Is it normal to anthropomorphize reward functions now? I find it makes this article hard to read. Is this common in robotics and if so, why would you describe reward functions in such a roundabout way?