Finishing Tim Ingold's piece and the book Linked by Albert-Laszlo Barabasi in the same day was more circumstance than anything, but the overlap was unavoidable. Barabasi, a Hungarian physicist at the University of Notre Dame, has been forging a new line of research known as Applied Network Theory. Seminal work in the field includes The Strength of Weak Ties, The Small World Problem, and more recently on power-laws: Power laws, Pareto distributions and Zipf's law.
Inspired partly by computer networks, applied network theory takes the basic data structure of a network of interconnected nodes and applies it as a model to various naturally occurring phenomenon. Some particularly good examples revealed in Barabasi's book include social networks, cell-biology and genetics, international financial markets, and everyone's favorite network, the world wide web*. Interesting properties, of varying degrees of complexity, fall out of simple network structure like "hubs" of proportionally large inter-connectivity, "islands" of relatively segregated sub-networks, and of course weak links such as those found abundantly in social networks.
The connection to Tim Ingold's call for a organism-centric biology is not hard to see. Network theory offers a simple and scalable model for organism-culture interaction. A directed graph (one in which nodes' connectivity distribution is said to follow a power-law) could explain mimetic / culturgen heritability and expansion. Networks have been used to show how companies rise to monopolies, how youtube videos go viral, and why child-naming patterns exhibit momentum. Networks offer what could prove to be an elegant reconciliation of implicate organismal traits and their relationship to culture.
*One emergent feature of directed networks are sub-structures referred to as "tubes" which are segments in which elongated, one-directional flow occurs. Thus, U.S. Alaskan Senator from, Ted Stevens, wasn't completely full of crap when he so elequently characterised the internet as a series of tubes.
Sunday, March 21, 2010
Tuesday, March 2, 2010
Simulating the Brain
In reading von Uexkull's piece on umwelten, it got me thinking about the ramifications of a popular strategy in brain simulation: simulate simpler animals' brains first, then we can move up to humans. Here is a small list of some popular (if not successful) attempts at brain simulation.
The Blue Brain Project has focused on simulating a neocortical column of a rat. Bi-products of the project, however, include a simulation framework and a number of noteworthy genetic algorithms for large-scale simulation. A number of popular articles have been written on the project.
On a smaller scale we have NEURON project from Yale, which seeks to simulate single brain neurons for use in larger network-based simulations.
On a bigger scale we have a DARPA funded group at IBM who recently claimed the "largest" cortical simulation to-date.
IBM's success on the company's Blue-architecture super-computers has spawned an even larger project: SyNAPSE. SyNAPSE is more in line with the chip manufacturer's agenda in that the goal is to instantiate brain-like redundancy and generality in a marketable hardware platform.
In reading about some of these projects and thinking of our friend von Uexkull, I'm wondering at what level of complexity umwelten may emerge from any kind of simulation. Clearly brains aren't the whole story in our engagement with the word. Might simulations of cortical tissue shed light on the organisation of our larger cognitive capacities? What really intrigues me is the idea that our more general abilities (like visual, tactile, and lingual systems) may have co-evolved with our exceptionally generalised brain. Will simulations, once computationally feasible, shed light on cognition? Or, will embodiment provide the next hurdle for technologists to jump? Might simulations replace imaging in neuroscience? And the bigger question: once we understand the brain, what's left?
The Blue Brain Project has focused on simulating a neocortical column of a rat. Bi-products of the project, however, include a simulation framework and a number of noteworthy genetic algorithms for large-scale simulation. A number of popular articles have been written on the project.
On a smaller scale we have NEURON project from Yale, which seeks to simulate single brain neurons for use in larger network-based simulations.
On a bigger scale we have a DARPA funded group at IBM who recently claimed the "largest" cortical simulation to-date.
IBM's success on the company's Blue-architecture super-computers has spawned an even larger project: SyNAPSE. SyNAPSE is more in line with the chip manufacturer's agenda in that the goal is to instantiate brain-like redundancy and generality in a marketable hardware platform.
In reading about some of these projects and thinking of our friend von Uexkull, I'm wondering at what level of complexity umwelten may emerge from any kind of simulation. Clearly brains aren't the whole story in our engagement with the word. Might simulations of cortical tissue shed light on the organisation of our larger cognitive capacities? What really intrigues me is the idea that our more general abilities (like visual, tactile, and lingual systems) may have co-evolved with our exceptionally generalised brain. Will simulations, once computationally feasible, shed light on cognition? Or, will embodiment provide the next hurdle for technologists to jump? Might simulations replace imaging in neuroscience? And the bigger question: once we understand the brain, what's left?
Subscribe to:
Posts (Atom)