Gary Marcus

one example: a topographic map in vision is a representation of the world; if you filter it, you get a new representation, that represents a different aspect of the world (say likes or zero-crossings etc)

okay, for vision it's probably easiest - there is a pattern in the brain that is partly driven by the thing in the world. (It's clearly less obvious for olfaction, as Freeman describes)

Though even for vision, as we've just seen from @anne_churchland et al, much of the activity does NOT reflect the thing in the world, but what the animal is doing

In any case, if it's a "representation", then it must be re-presented to some other area - but again, does that imply semantic content or pragmatic consequence? (Does it matter?)

There are 'representations', but we use the word too casually. Often what we really mean are 'signs'.

No, we don’t. “Representations” is a great concept because it denotes both the process (hierarchical abstraction) and the result (symbol). “Signs” I don’t think captures the same meaning.

Signs or symbols have the same issue though - they *mean* something. How is that meaning communicated?

Contextually. Think of a graph data structure (entities and relationships) where each node is a graph itself (meaning each entity is a graph of entities) as well as one node in a higher level graph. That’s what a hypergraph is (universal data structure)