I am really glad to read this interesting work. I just express my agreement with all the praising comments that have already been stated among the discussion. As Martin Giurfa says, this work may open new research avenues. I add that not only on associative or social learning, but also on nestmate recognition.
I raise the question on why should it be adaptive for the bees to accept heterospecific workers and why, in a natural context, to understand the dance coding by another species.
Apis cerana accepts the A. mellifera workers under certain conditions regarding resources availability, but they reject them if, for instance, food becomes scarce. Is there any high cost on aggressing the exogenous workers, that prevent the A. cerana individuals from aggressing the A. mellifera, when there is enough food for all the individuals in the hive.
Maybe, A. cerana takes profit from A mellifera's ability to locate food sources, thus there is no advantage in killing the A. mellifera individuals (from A. cerana point of view). If such is the case, why should then all the A. mellifera individuals be rejected when the hive is urged for food?
Another underlying question is whether individuals from different species are always recognised as non-nestmates, despite not rejected when much food is available, or if there if accuracy for nestmate recognition vaies according to resources being available or scarce. How these different scenarios would affect perception and physiology within the hive and among the bee society members remains the source of insparation for much more future work!