Log in / Register
Home arrow Computer Science arrow Social Computing, Behavioral-Cultural Modeling and Prediction
< Prev   CONTENTS   Next >

3 A Human Development Dynamics Model

We maintain individual agent attribute relationships and postulated changes of RS, SE, D and Y in keeping with HD theory. These endogenously derived, individual agent attributes (RSi, SEi, Di and Yi) impact how economic transaction games occur, either increasing or decreasing individual wealth and, at increasing scales, determining societal productivity [8]. Geography and proximity are allowed to play a role by instantiating in random two-dimensional lattice worlds.

Social co-evolutionary systems allow each individual to either influence or be influenced by all other individuals as well as macro society [29] [35], perhaps eventually becoming coupled and quasi-path interdependent. Accordingly, we instantiate non-cooperative, socio-economic Prisoner's Dilemma (PD) transaction games given the similarity of agent i's attribute vector (Ai) of social, cultural, political and economic preference (RSi, SEi, Di and Yi) to agent j's attribute vector (Aj) for selected Aij pairs. Here, symmetric preference rankings and asymmetric neighborhood proximity distributions allows “talk-span,” a Euclidean radius measure, to proxy for communications reach, social connectivity and technology diffusion constraining the potential set of Aij game pairs. Low talk-span values restrict games to local neighborhoods among spatially proximate agents, while higher talk-span values expand potential Aij pairs globally, modeling socially compressed space.

Fig. 2. HDD Architecture (implemented in NetLogo [34])

Following Social Judgment Theory, the attribute positions of two agents are conceived as a Downsian continuum [11] [15] where distance between these positions symmetrically affects the likelihood of one accepting the other's position. Agent i evaluates the likelihood of conducting a transaction with agent j based on similarity of socio-cultural preferences |RSi-RSj| and |SEi-SEj| within the given neighborhood. This captures communications and technology diffusion for frequency and social tie formation [22].

After transaction counterparties are identified, similarity is measured against an exogenous threshold to gauge compatibility. If both parties are satisfied, compatible agents, endowed with RAP cognition, enter into an engagement and search their memory for prior transactions with their period t counterparty. In the case of no prior transaction experience, agents individually each select strategy Sij Î [Cooperate, Defect] probabilistically based on similarity of political preferences as expressed by |Di-Dj| [28].

In repeat transactions, agents have perfect memory of t-n and will predicate their strategy in period t transactions on their counterparties' t-1 behavior such that Sijit = Sji .

Agents are unaware of counterparties' strategy rule at any point in time. This can lead to the emergence of stable productive relationships, bad relationships featuring pure

defection strategies over repeated interactions, and tit-for-tat relationships, where agents alternate between strategies and never sync into a stable productive transactional relationship. This reflects recent work on the affects on co-evolution of both dynamic strategies and updating rules based on agent attributes [20] [23].

Following Nowak and Sigmund [24], we randomly assign game transaction values. However, we do not asymmetrically constrain such values; any particular game transaction value between pairs, Vij, lies in between [-.1, .1]. This instantiation allows for different potential deal sizes, costs, or benefits. We specifically model socio-economic transaction games as producing either positive or negative values as we want to capture behavioral outcomes from games with both upside gains or downside losses.

In our HDD framework, Ai strategies are adaptive, which affect Aij pairs locally within a proximate radius as first order effects. Other agents, within the system but outside the talk-span radius, are impacted through cascading higher orders. Agents simultaneously co-evolve as strategy pair outcomes CC, DC/CD or DD at t affect Yi at t+1, thus driving both positive and negative RS, SE and D feedback process through t+n iterations. These shape Ai attributes which spur adaptation to a changing environment, summing Yi, RSi, SEi and Di vector values. Feedback into subsequent Aij game selection networks and strategy choice yields a CAS representation across multiple scales.

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
Business & Finance
Computer Science
Language & Literature
Political science