In other words one optimization for living is to seek out and maximize surprise!
I definitely agree with the compression of information and simulation of potential futures as a default operation for brains, but I don't agree with the blanket "reduction in surprise" as the reward function, as it's much more nuanced.
In other words, cave-man could have been more successful minimizing lifetime surprise, if he sat in his cave all day, not playing with tools or meeting other tribes, and that's clearly not what happened.