My friend, the whole "study" is based on false assumptions and consist of interpretation of the data in the way which confirms these assumptions. This is not even science, this is some blog posting kind of crap.

Every bit of observable behaviour could be explained in terms of habitual behaviour and physical pattern matching, exactly what a trained NN will do - it will perfectly accurately separate patterns which corresponds to different quantities, while posses no notion of any abstract concepts whatsoever.

Animals pattern match, but do not have abstract concepts. Well, enzymes do pattern matching but not counting.