Jim Fetzer (24 Jan 1996) brings to bear on the issue of representations an important set of distinctions between types of symbols. I want to draw people's attention to the use made of exactly these Peircean distinctions by Thierry Delmarcelle and Lambertus Hesselink in an article ``A Unified Framework for Flow Visualisation'' in Gallagher (1995). I mentioned in my starter paper that one of the more important uses of non-propositional representations was in computer simulations and visualisations. Delmarcelle and Hesselink (D&H) argue that in computer visualisations of scientific phenomena, icons (in Peirce's sense) are primary, because there has to be some structural similarity between the representation and the system being visualised, but that the icons should be indexical icons i.e. icons caused by data that are themselves causally produced by the system. There is thus a contrast between graphical images produced in, say, architectural renderings, which are non-indexical icons and the indexical icons D&H require.
This is an important point, not the least because it allows a way to respond to (the by now tedious) under-determination arguments of the kind peddled by Goodman and others, to which Fetzer draws our attention, and quite rightly criticises. In cases where we have indexical icons, you can't perform the sort of permutation operations that you can on symbolic symbols. (Of course we know the Goodman/Putnam/Quine response - ``resemblance'', ``causes'' etc., are just more parts of the worldview/language/ideology. This simply misses the point. What's going on in the system that produces indexical icons is going on in the world, it's not going on in a language. Symbolic languages are involved in simulations, but what causes the graphical output are concrete entities the interpretation of which is not part of the causal process.) Linguistically oriented philosophers of various kinds love symbolic symbols, and it is their obsession with symbolic systems, usually encoded into languages, that is one of the primary targets of my starter paper. Fetzer via Peirce has pointed us towards a way out of that linguistic trap and it's a promising route.