Symbol

When Bohr says that something is ‘symbolic’, it’s tempting — especially for us analytic philosophers — to understand him in light of post-Hilbertian logic, i.e. as saying that the thing is just chicken-scratch that one manipulates on paper. But such an understanding is ahistorical. Bohr’s notion of a symbol and of symbolic representation is informed by another tradition.

Here is what Harald Høffding says in his psychology book (first published in 1882, and with several subsequent editions):

Da vi aldrig kunne komme udover Dobbeltheden Subjekt-Objekt, er selv det nøjagtigste og fuldstændigste Verdensbillede kun et Symbol, opstaar som et Produkt af Vexelvirkningen mellem det erkendende Subjekt og den øvrige Del af Tilværelsen, et Produkt, der ikke behøver at ligne nogen af de Faktorer, der have frembragt det, men hvis Gyldighed og Betydning heller ikke beror derpaa.

Since we can never get beyond the subject-object duality, even the most accurate and complete picture of the world is only a symbol, arising as a product of the interaction between the recognizing subject and the rest of existence, a product that need not resemble any of the factors that have produced it, but whose validity and meaning do not depend on it either.

Høffding covers a lot of ground in this passage. First, he insists — along with his teacher Rasmus Nielsen — that humans can’t get beyond the subject-object duality. In other words, in any act of knowledge, the world gets split into two, with the subject on one side and the object on the other side. Comprehensive knowledge, with all of the world as object, is impossible for human beings.

Second, Høffding claims that knowledge emerges out of an interaction. This claim harks back to Kierkegaard’s idea that knowledge is momentary (øjeblikkeligt), and it points forward to Bohr’s emphasis on the concrete interaction between subject and object.

Finally, Høffding explicitly denies that the correctness of a symbolic representation depends on its resembling the object. Here he suggests that correctness of representation is relational, rather than the result of two things having monadic properties in common.