23 January 2006

Coordination Class Concepts and Scale-Free Networks

Ok ... so I just dumped my first post in as a copy and paste and haven't gone back to edit formatting - I'll get there.

Yet another aspect of the coordination class model for concepts that I like is that it exhibits scale-free properties. One of the central ideas of the coordination class is that the 'causal network' - the aspect of a concept that processes information that is read out from the environment - has components known as "resources". Essentially, this means that the coordination class model follows a "knowledge in pieces" philosophy, in which concepts are constructed of finer-grained cognitive units that tend to be naive and intuitive. Resources can come in a number of flavors - diSessa's phenomenological primitives, Minstrel's facets, and Tuminaro's epistemic frames. It seems reasonable, though, that the resources for some concepts may in fact be reasonably judged as concepts into and of themselves. Of course, evolution jumps out as a prime applicaiton for this idea, since the fully developed concept involves genetics, natural selection, homology, and a multitude of other knowledge that is finer-grained than the concept of evolution, but traditionally thought of as concepts themselves.

So why is this a good thing?

The idea for this refined model of the coordination class, and the application of the concept of evolution, is to develop a model for concepts that shows phenomenological overlap with the supporting layers from which it develops. I think of concepts as properties of the mind, which is itself an emergent property arising from the brain. Neuroscientific research has demonstrated that neural networks demonstrate scale-free properties (it's late for me right now and I'm feeling a tad lazy to do the fancy links...). Additionally, remembering that nerves are cells, researchers are also beginning to find that biochemical pathways that give rise to the emergent properties of cellular function also demonstrate scale-free properties. So, although we can hand-wave and reliably claim that mental objects - concepts - are emergent properties that can escape the limitations of the components from which they arise, I find it easier to support a model for concepts that does show some degree of overlap in properties - patterns - with its foundation. So, just as the scale-free networks of biochemical pathways can give rise to cellular-level properties, and scale-free networks of cells can give rise to mental properties, so can scale-free networks of cognitive resources give rise to concepts.

Of course, it's also interesting the the phenomenon of evolution itself seems to be scale free (in that it does not appear to be a pattern of relationships that is limited only to the biosphere). But more on that later.

No comments: