Can we train a machine to detect if another machine has understood a concept? In principle, this is possible by questioning the second machine on the subject of that concept. However, we want this procedure to be done by avoiding direct queries. In other words, we would like to isolate the absolute meaning of an abstract idea such as "having understood" by putting it into a class of equivalence, hence without adopting straight definitions or showing how the idea works in practice with the help of tests. We discuss the metaphysical implications hidden in the above question, with the aim of providing a plausible reference framework. This could also clarify how the mechanism of self-consciousness develops, which requires however an interplay between members of a community. Indeed, we claim that "having knowledge of a concept" becomes an act of consciousness only when opinions are confronted with other actors, belonging or not to the same class of equivalence.