It's interesting that, in a certain phase of the learning process, words in natural language already work this way. Before a language learner has enough knowledge to understand word etymology, they have to just form an association between a word and a certain set of sensory experiences that the word is meant to represent.

Seems like we have the same thing happening here. The input images are the sensory experiences. The dummy word "S*" is the linguistic symbol that is attached to them.