R :: conscious-computation

In Defence of Representation

There exists an entire industry whose sole purpose seems to be to call the very idea of representation into doubt. I refer of course to cognitive philosophy. To the vast majority of software engineers, who use representations on a daily basis to model one or other aspect of a system's interaction with its environment or its userbase, this is plain evidence of the uselessness of philosophers and philosophy. Software IS in a very real sense nothing BUT representations of data. The reason that such a scandalous situation is not just allowed to exist, but is actively funded, is a side-effect of the 'silo' culture of contemporary science, where super-specialisation is not just the rule, but the ruler. Cross-disciplinary studies are started in good faith, but rapidly allowed to wither on the vine, as I can personally attest to (see section 0).

It does not require a detailed technical analysis to demonstrate both the truth and utility of common-or-garden representations. When a child draws a 'child-like' drawing of themselves, mum and dad, and their house, they are making a representation of a real situation, their everyday 'lived reality'. This is no intellectual abstraction, as shown in Figure R.1 below.

Figure R.1

Firstly, it is a remarkably unambiguous portrayal. Very few people would fail to recognise the situation that the picture is describing. Every part of the situation that the author deems important has its own counterpart in the representation. There is a 1:1 mapping between parts of the situation and parts of the representation. Minor details do not appear if they are not important, but are included if they are. She appears in much more detail than her mum and dad. In fact, her mum and dad are generic adults. They don't even have faces. They are only recognisable as her parents because of their proximity to her, and to the house. In another part of the picture (which is called 'firemen are brave') two generic male adults appear near a house-sized vehicle. They are firemen because they are sitting in a large vehicle with a ladder and a hose spurting water. They wear hats of some kind because they are at work, whereas the child, mum and dad are bare-headed, suggesting they were at home when the fire started.

The first point to note is the role of positional association, and to distinguish it from positional order. The meaning of the representation would not change if the child stood to the right of its parents, or to the left of the burning house, as long as the family group and the house are spatially associated, and separated from the grouping consisting of the vehicle and its occupants. Spatial association is a combinatorial code with permutational aspects, not a permutational code like symbol sequence in a string (spatial position). The permutations of a given combination are like the microstates for a given macrostates- a measure of the representation's entropy.

The second point to note is that it requires considerable prior knowledge to understand. It is by no means a self-contained code (a 'cypher'). It is a situational language for the purposes of communicating understanding. You need to know about families and homes and fires and emergency services.