While eating some freeze-dried peas and corn I wondered how the nutritional content had changed compared to raw or boiled peas and corn. That opened the door to the whole gamut of considerations of how different ways to age, cut, process, cook, and mix foods affects their chemistry and thus their nutritional value. Complications are further exacerbated by the dynamics of food digestion, activity levels, body chemistry, and metabolism. Even the order, amount, and time in between eating two foods have a tremendous affect on how the body will absorb nutrients from them. The science of food chemistry and nutritional assessment is clearly a complex system akin to gene regulatory networks and should therefore be analyzed as such.
May 31: Searching for the Killer App
In the very near future I will begin constructing software to process and evaluate data in the form of Markov diagrams for my tipping points, robustness and path dependence analysis. These Markov diagrams are supposed to represent the state transitions of some actual system. Thus to perform the analysis I need data across time for some actual system. Since I aim to use this methodology for my methodology prelim exam for the political science part of my PhD, a political science-related data set is what Iím looking for. This would be some kind of longitudinal study or poll data or perhaps political economy data that tracks congressional behavior with economic features. Alternatively I just collect data from a generative model/simulation of a political process. The point is that I need some killer application of the analytical method so I can take something that people already care about and tell them something new about it that is both helpful and interesting. The problem is that I have no idea what system/data to use.
Let's make the plausible assumption that people's happiness depends upon the happiness of certain other people such as family members, friends, and public figures. Naturally some people care less about others and some people are cared about by more people than others; i.e. the distribution of attention is heterogeneous. Clearly it is also directional; shared happiness is usually reciprocated in familial groups, but the pop idol on magazine covers is cared about by fans he or she will never meet. The model presented here explores the idea that certain people, who are the object of many others' attention (celebrities), ought to receive a disproportionate amount of a society's resources because so doing will actually increase overall happiness.
Jun 23: Congress of Mind
Have you ever had an argument with yourself? Ever feel that different sides of an argument keep presenting themselves just as you think other considerations have won out? Ever wonder how to model this in a computer agent with a parsimonious mechanism that nonetheless usually produces quick, concerted, and correct outcomes most of the time? I have and here's something that I came up with based upon research is spatial voting models. It's a spin-off of Marvin Minskyís society of mind idea that human cognitive abilities are really the complicated combination of multiple simple, single-purpose problem-solving mechanisms. I call it the congress of mind model.
Jun 22: Evolution of Culture 2
Over the past two weeks I have been slaving away on a computer model (and paper) to test and expand claims in the culture and morality literature. The model was built to address questions regarding the role of punishment and reciprocity, learning rules, information constraints, scalability, and the possibility of group selection in models of the evolution of cooperation in the Prisoners' Dilemma game (among others). These simple games are purported to yield insight into the development of cooperation and coordination in common pool resource problems (n-person versions of PD); though I have my doubt in that regard, I believe that sophisticated models will be able to provide insight and every enterprise has to start from the beginning.
Prompted by Boyd and Richerson's formal models of the evolutionary origin of cultures (especially by the admitted limitations the replicator dynamics technique), I have embarked on a mission to build an agent-based network structure computer model to combine, test, and expand their claims and conclusions. Some of the extensions are to incorporate factors that they identify as important but could not include in their models while others are my own modifications to scale up the processes and broaden the application. Furthermore, the model pushes several "cutting edge" concepts in complexity and agent-based models as well as requires the development of new methodologies for operation and analysis.
Feb 21: Neutral Mutation Network
Walter Fontana's paper "The Topology of the Possible" presents the idea of a neutral mutation network. The idea is that since the fitness of a creature is determined by its phenotype, changes in the genotype that do not alter the phenotype have a neutral effect on fitness. However, these neutral mutations are important because they alter what future mutations the genes have access to. This neutral mutation framework is presented as sufficient to generate punctuated equilibria in phenotypes over time. I created a simple computational model to confirm (or disaffirm) this conjecture.