Abstract # 25:

Scheduled for Thursday, June 18, 2015 12:30 PM-12:50 PM: (Cascade AJBCD) Oral Presentation


NETWORK DYNAMICS OF SOCIAL STABILITY

B. A. Beisner1,2, H. Fushing3 and B. McCowan1,2
1Dept of Population Health and Reproduction, University of Calfornia, Davis, Davis, CA 95616, USA, 2California National Primate Research Center, Univesity of California Davis, Davis, CA 95616, 3Dept of Statistics, University of California Davis, Davis, CA 95616
line
     Social systems are composed of multiple constituent networks that interact, and stability is a system-level property that emerges from this interaction. Network theory is designed to reveal emergent structure in complex systems and identify potential sources of network instability. We describe here a data-driven network approach, called Joint Network Modeling (JNM), which identifies changes in social stability by quantifying the relationships amongst these multiple constituent networks. To illustrate how JNM can detect instability in empirical networks of a complex system, we present analyses of the joint relationship between aggression and subordination networks in four stable and three unstable groups of rhesus macaques. All stable groups showed the same pattern of interdependence: 58-76% of the interdependence between aggression and subordination networks came from having far more dyads than expected (under the null hypothesis of independence) with opposite direction aggression and subordination, i.e., aggression always goes from A to B, and peaceful subordination signaling always goes from B to A. In unstable groups, however, less of the aggression-subordination network interdependence came from dyads with opposite direction aggression and subordination (38-51%). Furthermore, unstable groups had more frequent than expected dyads with ambiguous/contested dominance relationships (e.g. bidirectional aggression with either no subordination signals or bidirectional subordination). By revealing the hidden architecture of inter-network relationships, JNM readily distinguished stable from unstable phases across multiple replicates of a complex social system.