2015 Graduate Workshop in Complexity and Computational Social Science

Student Projects

Each student began a research project during the two-week workshop. Below are brief descriptions of these various projects. These projects will form the basis for dissertation chapters and/or journal articles.

Luis Alejandro Lee, Economics, University of Nottingham (alejandro.lee@nottingham.ac.uk).

Alejandro is looking at coordination in games. He explores a battle of the sexes game and tests whether the standard solutions provide insights into systems of adaptive agents modeled by learning automata. He finds that the automata learn to coordinate on one of the pure Nash equilibria for long periods with occasional shifts to the alternate equilibrium and, on rarer occasions, periods arise where the machines learn to alternate equilibrias each round. Contrary to standard intuition, the introduction of a coordination signal does not alter the above outcome.

Ilker Aslantepe, Economics, New School for Social Research (aslai552@newschool.edu).

Ilker is intrigued by the role of social interactions on individual decision making. Actions can either substitute or complement the acts of others. In a model with linear social influence and a representative agent, one can easily tie the model's outcome to equilibrium predictions. Under nonlinear social influences, critical bifurcation points arise in the system and multiple equilibria arise. He is currently deploying maximum entropy ideas to understand the system more fully.

Daniel Barkoczi, Cognitive Science, Max Planck Institute for Human Development (barkoczi@mpib-berlin.mpg.de).

Daniel is considering the tradeoff between exploration (finding new solutions) and exploitation (using known ones) in social learning. Social learning takes the form of sampling peers and using one of three possible decision rules (copying the best, conforming to the most frequent, and random sampling). These different decision rules lead to remarkably different patterns of intermediate and ultimate success. The current literature admits contradictory hypotheses about the impact of network structure on search, and Daniel's work shows that the inclusion of decision rules resolves this apparent contradiction.

Lawrence De Geest, Resource Economics, University of Massachusetts (ldegeest@resecon.umass.edu).

Lawrence is refining ideas about common goods and the influence of outside forces. In particular, he considers a model of a common pool resource that is subject to poaching from outsiders. Using a modified repeated prisoner's dilemma game, he accounts for changes in the common pool resource and how these can alter the incentives and impacts of poachers. Initial results indicate that the joint model admits interesting new behavioral regimes. Ongoing work is introducing more complex resource dynamics and adaptive behavior on the part of the agents.

Erika Frydenlund, International Studies, Old Dominion University (efrydenl@odu.edu).

Erika is analyzing the flow of refugees. She is paying particular attention to differences in mobility across the refugees in family groups. She finds that even small levels of family support for less mobile individuals, makes a large, positive difference in the system. She is currently elaborating key features of the model and applying the general insights to important policy and logistic considerations.

Amit Goldenberg, Psychology, Stanford University (amitgold@stanford.edu).

Amit is modeling the emergence of collective emotions based on the idea that interacting individuals influence, and are influenced by, group-level emotions. Individuals may want to conform (or not) to the group's emotional state. Experimental data suggest that individuals alter their individual emotional ratings when exposed to group ratings, and these data allow Amit to calibrate his model. Preliminary results suggest that the model can capture key aspects of how a network of individuals creates a collective emotion.

Kotrina Kajokaite, Biological Anthropology, University of California, Los Angeles (kotrina@ucla.com).

Kotrina wants to understand philopatry: the choice of whether an individual should stay in its current group or leave and seek another. Based on a stark mathematical model of this process, Kotrina developed a simple computational model of the system that produces similar results. She then modified her model to account for key behavioral and ecological elements found in groups of capuchin monkeys, such as alternative routes to becoming an alpha and the group's age structure. She finds that her new model accurately captures key features of the capuchin system that did not arise in the mathematical model.

Maxim Kuklin, Strategic Management, Carlson School University of Minnesota (kukli014@umn.edu).

Maxim is interested in how the success of one firm influences the opportunities of other firms. Using NK fitness landscapes as a base, he explored alternative models of transforming the landscape as firms adapt to local conditions. Depending on the degree of exploration, the system shows very different regimes, each of which may have interesting implications for firm strategy.

Abigail Sullivan, Environmental Social Science, Arizona State University (avsulliv@asu.edu).

Abie is exploring how institutions mediate ecological outcomes. While the work is general, she ties the model to data generated in Chitwan National Park in Nepal which is being harmed by an invasive weed. The local inhabitants can remove the weed using various actions, that differ in terms of the direct cost to the inhabitant, impacts on social reputation, and the likelihood of the weed returning and spreading. Without intervention, inhabitants choose the lowest-cost action (burning) which also results in large negative externalities as it exacerbates the spread of the weed. Abie is currently using her model to investigate potential policies that would improve the situation.

Toban Wiebe, Economics, University of Pennsylvania (tobanw@gmail.com).

Toban is developing a network model of social change. The agents in the model have underlying convictions but also have a desire to conform (or not) to the views of those agents with whom they interact. He finds that a small-world system shows a strong attraction to neutral beliefs, though polarized outcomes are possible depending on the parameters. He is currently developing evidence about the link between network configuration and behavioral outcomes.

John H. Miller , miller@santafe.edu.