I have never seen an octopus "handling" a mirror to be sure about the quality of his own camouflage.
Seriously : I'm looking for an hypothesis (or a notion) about the "comparative" feedback mechanisms that assess if the octopus's skin is in "pendant" with the current background.