Emergence of collective memories

THE EMERGENCE OF COLLECTIVE MEMORIES Sungmin Lee 李圣民 (Umeå University 于默奥⼤大学, 瑞典) Verónica Ramenzoni (MPI Psycholinguistics Nijmegen, Holland) Petter Holme 郝培德 (Umeå University 于默奥⼤大学, 瑞典, Sungkyunguan University 成韵馆⼤大学，韩国) PLoS ONE 5, e12522 (2010).

http://www.tp.umu.se/~holme/

全国复杂⺴⽹网络学术会议 October 17, 2010

培德在苏州

complex network boom of early 00’s

Narrative network 叙述⺴⽹网络 ≈ Memory web 记忆⺴⽹网络

research leader fieldtrip to 北京

Outline

- What is the topology of real narrative networks?

- How can we model the evolution of narrative networks / memory webs?

- How (much) can communication stabilize collective memories?

- What are the conditions for groups to form?

- How does the observations behave as a function of population size?

Empirical memory webSource material

Bearman, Moody, Faris, 2003. “Networks and History” Complexity 8:61–71.

Empirical memory webIndividual memory web 1

Bearman, Moody, Faris, 2003. “Networks and History” Complexity 8:61–71.

Empirical memory webIndividual memory web 2

Bearman, Moody, Faris, 2003. “Networks and History” Complexity 8:61–71.

Empirical memory webAggregate memory web

Bearman, Moody, Faris, 2003. “Networks and History” Complexity 8:61–71.

Empirical memory webDegree distribution

Empirical memory webClustering coefficient vs. degree

Empirical memory webDegree correlations

Empirical memory webCommunity structure

Model ingredients

- N agents with individual memory webs.

- Agents can learn about history from experience or communication.

- Closer friends are more likely to communicate & people who communicate often are closer friends.

- Events / associative arcs remembered more strongly are more frequently discussed, and vice versa.

- Events / associative arcs not talked about are gradually forgotten.

Communication modelLearning dynamics

Example outputMemory overlap

15  

Community structure dynamicsNumber of groups

Model community structureNumber of groups

Model community structureMemories and social ties

18  

Model community structureSize- & communication-rate scaling of the number of communities

Model community structureSize-scaling of the diversity regime

Model community structureSize-scaling of the diversity regime

Model community structureSize-scaling of the number of groups

Summary - Narrative networks represent memory webs. - Network modeling can be helpful to understand these quantitatively. - An empirical narrative network shows skewed (in- and out-) degree distribution. - Propose a communication model that shows the emergence of collective memories - … and the formation of communities with strong internal connections and similar memory webs. - The number of communities is limited as a function of N

Petter

谢谢⼤大家！

Sungmin Verónica

http://www.tp.umu.se/~holme/

Model community structureWhy a peak i?

Communication modelErrors & misinterpretations