Proto-writing is still quite far along on the spectrum I'm talking about of contextually defined symbols. Lascaux and chauvet have plenty of examples generally agreed to be partially symbolic, just off the top of my head.