Center       for   Science in Society Bryn Mawr College http://serendipstudio.org/local/scisoc

The plan for the semester is to alternate between theoretical, abstract discussion of emergence and concrete, practical discussions of modeling methods and applications of our emergence knowledge. Today we began a very abstract discussion on what generalizations we can make about emergence.

We started with a list of phenomena and conjectures, listed on the Wiki page
http://emergent.brynmawr.edu/index.cgi/WhatIsEmergenceWhatIsEmergent.

Doug: First on the list: snowflake form and tree form. Are they emergent phenomena? What differences are there between them? Paul noted that unlike snowflakes, trees exist in an infinite but bounded variety of forms, all recognizable as tree forms. Bounded variation, as he called it in a paper (REF?), and it describes variation in other biological systems and in grammatical sentences. A redwood is a tree like a birch is a tree, and somehow human perception places them in the same category.

Rob: it's like Wittgenstein's notion of the family resemblance.

Karen: but to what extent are our categories just in our minds, and not in nature?

Ted: there might be discontinuities in the variation in plant form such that trees to cluster together, and separate from other kinds of plants.

Paul: it's often remarkable how our perceived category boundaries match cluster boundaries in the world.

Rob: or how perceived categories have no basis in the world. Consider color perception. There are no natural discontinuities in variaion in wavelengths. The discontinuities come from how our retinas work.

Panama: at the 2003 Joint Mathematics Meetings in Baltimore, the Gibbs Lecture was given by David Mumford. He's trying to develop a mathematical way to characterize the similarity between two shapes.

Mark: a lot seems to come down to what one calls a pattern. In A New Kind of Science, Wolfram frequently just declares one pattern to be ordered and another pattern to be random. It's subjective, and his arguments depend on his categories.

Rob: it always seems to me that models that are claimed to have emergent behaviors frequently have a global selection criterion built in.

Deepak: in computer-science research, there are models that explicitly exclude global criteria. Unsupervised learning,

Ted: snowflake form isn't subject to selection, but tree form is. Tree form results from some kind of process with feedback.

Paul: tree form is emergent, and any particular instances of tree form -- particular generative rules, for example -- are themselves the result of an emergent process, evolution by natural selection.

Doug: Marching bands -- emergent?

Mark and Al: No! They're made to look emergent, but really they're carefully choreographied by a central choreographing mind.

Doug: Maybe phenomena aren't emergent independent of the processes that give rise to them. The phenomena are emergent?

Paul: Any phenomenon that arises from an emergent process can also be created by a nonemergent process. You can't look at the result and tell whether it's emergent.

Ted: And Conway's Game of Life? It's not impressive. Even less impressive is Langton's Ant. There's nothing so amazing about the fact that a particular configuration -- which the ant happens upon after searching the space of configurations -- has a behavior that jumps out to our human minds as qualitatively different from the other configurations' behaviors. My subjective dismissal of these phenomena is relevant to our discussion, because I'm claiming that much of what we call emergent is only emergent because someone finds it amazing.

Anne: Someone smarter, like the computer, would not necessarily find it amazing. Emergence might only result from our limited cognitive capacity.

Paul: Conway searched hard for a set of rules that would make Life interesting. Rules that caused the game to converge quickly to an equilibrium state -- he dismissed these as uninteresting. All these are interesting in that you can't predict what comes out from what goes in.

Ted: Maybe emergence is defined by the impossibility of predicting the outcome except through simulation.

Rob: I still think that the only interesting emergent phenomena are the ones that can solve problems. Optimizers, etc.

Next week: Ted demonstrates the implementation of cellular automata in NetLogo.

Home | Calendar | About | Getting Involved | Groups | Initiatives | Bryn Mawr Home | Serendip Home

Director: Liz McCormack - emccorma@brynmawr.edu | Faculty Steering Committee | Secretary: Lisa Kolonay
© 1994- , by Center for Science in Society, Bryn Mawr College and Serendip
Last Modified: Wednesday, 02-May-2018 11:57:04 CDT