Friday Feb 18 2011 Douglas R. White, Causality project team

Evolutionary Causality: Discussions, Demonstrations, Possibilities link: http://bit.ly/ekyBwt

Doug White; Giorgio Gosti and Tolga at UCI; and Zeev Maoz at UC Davis

Friday Oct 22 2010 Dwight Read

Dwight Read, Professor of Anthropology, UCLA. http://bit.ly/c3sIfe
"Mathematical Modeling of the Logical Structure of Kinship Terminologies"

Friday Oct 8 2010 Yen-Sheng Chiang

Yen-Sheng Chiang, Asst. Professor, Sociology, UCI. http://bit.ly/aHQg0y

Cooperation Dynamics in Networks When Cued by the Structural Attributes of Nodes
Recent research endeavors in science investigate how cooperation evolves in complex networks. In the models, the network is either exogenously given or is formed endogenously. In either case, actors are modeled to occupy the nodes, play the prisoner's dilemma game with their network neighbors and adapt behavior in reference to either their local network neighborhood or the whole population. Yet, in these models an actor's decision of cooperation or defection in the game is independent of the network structure. A cooperator, for example, would cooperate unconditionally with all his network neighbors despite their differences in network positions. In this talk, I introduce a new direction for modeling cooperation dynamics in networks where an actor's decision making can be cued by the structural attributes of nodes. In the new model, cooperators are conditional in the sense that they follow some rules that govern whether they would cooperate or defect dependent on the network structural properties, such as betweenness, clustering and degree, of the nodes they occupy as well as their neighbors'. Simulating the evolution of cooperation across a variety of networks shows that conditional cooperators who are selective in choosing the recipients of cooperation based on the cues of nodal attributes increases the pervasiveness of cooperation across the population.

Friday Sept 24 2010 Douglas White

http://bit.ly/b2vyFf

Abstract: Results and strategies are illustrated for using a newly completed R package for cross-cultural or cross-community samples, Rccs, which generates potential causal inferences from coded data for ethnographic studies. It factors out network "peer effects" of spatial and shared cultural history, imputes missing data, and generates networks of potential causal relations among variables. These can be evaluated by Pearl's causal graph theorems for identifiable causal structures. Illustrations are given for a large network of variables from the Standard Cross-Cultural Sample and for a small time-series sample of South Asian villages undergoing transformation from a Jajmani patron-client to market exchange. It will also be applied to Lewis Binford's (2000) forager sample. Multi-level peer effects can be accommodated in these analyses. *Rccs* is used in UCI undergraduate science writing classes and is open access under the Gnu license for use in university classes worldwide.