A two-phase model of collective memory decay with a dynamical switching point

Naoki Igarashi, Yukihiko Okada, Hiroki Sayama, Yukie Sano*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Public memories of significant events shared within societies and groups have been conceptualized and studied as collective memory since the 1920s. Thanks to the recent advancement in digitization of public-domain knowledge and online user behaviors, collective memory has now become a subject of rigorous quantitative investigation using large-scale empirical data. Earlier studies, however, typically considered only one dynamical process applied to data obtained in just one specific event category. Here we propose a two-phase mathematical model of collective memory decay that combines exponential and power-law phases, which represent fast (linear) and slow (nonlinear) decay dynamics, respectively. We applied the proposed model to the Wikipedia page view data for articles on significant events in five categories: earthquakes, deaths of notable persons, aviation accidents, mass murder incidents, and terrorist attacks. Results showed that the proposed two-phase model compared favorably with other existing models of collective memory decay in most of the event categories. The estimated model parameters were found to be similar across all the event categories. The proposed model also allowed for detection of a dynamical switching point when the dominant decay dynamics exhibit a phase shift from exponential to power-law. Such decay phase shifts typically occurred about 10 to 11 days after the peak in all of the five event categories.

Original languageEnglish
Article number21484
JournalScientific reports
Volume12
Issue number1
DOIs
Publication statusPublished - 2022 Dec
Externally publishedYes

ASJC Scopus subject areas

  • General

Fingerprint

Dive into the research topics of 'A two-phase model of collective memory decay with a dynamical switching point'. Together they form a unique fingerprint.

Cite this