Interesting post. It would seem to pretty much verify my point. Even in the case of children and higher animals recognizing themselves, there is no evidence of "knowledge," which requires conceptualization of the percepts that are recognized.
Perceptual recognition is pretty much automatized in most creatures, and to some degree in men.
Hank
Octopuses can learn how to take a desired food thing out of a bottle, simply by watching a neighboring octopus do it, and the process of removal can be somewhat complex, as in removing a stopper from a bottle to get at the food. If the animal can learn the action and repeat it without first trial and error discovery, that would appear to refute your assertion, also.
When Koko's --the Gorilla who learned amislan-- kitten was killed and the handlers brought another of very similar size and color to her to become her pet, she refused to call it 'allball', insisting that it was not allball, though she hadn't touch allball after it died out on the street outside her enclosure.