The universal decay of collective memory and attention

Our last article in Nature Human Behaviour shows that the temporal dimension of the attention received by cultural products, including scientific papers, patents, songs, movies, and biographies, decays following a universal bi-exponential function that uncovers the communicative and cultural nature of collective memory.

Thumb yo defense
Dec 10, 2018
3
0

The Decay of Collective Memory and Attention from Collective Learning group at MIT on Vimeo.

For decades scholars have studied the adoption and decay of cultural content. Yet, despite numerous efforts, we know little about how to model the decay of attention.

This article makes use of data on papers and patents citations, and on the present and past online attention of different cultural domains, namely:  songs, movies, and biographies, to study the decay of human collective memory and attention.

The paper has two significant contributions. First, it shows that the attention decays across all of these cultural domains follows a universal bi-exponential form. Second, it is built on the concepts of communicative and cultural memory—sustained by oral and “written” communication, respectively—to derive a model that accurately reproduces this bi-exponential decay. For all the considered datasets, this model is better at explaining the empirical decay than previous efforts, such as log-normal and exponential models. 

While the shape of the decay function is universal, its parameters are informative of the decay properties of each cultural domain. For instance, our model suggests that the memory of award-winning athletes is sustained mainly via communicative memory, while the memory of songs relies on the translation from communicative to cultural memory.

The article provides a fundamental advance in the understanding of human collective memory using a big-data approach. Not only it shows universality in the decay of attention, but it proposes a model for this decay based on the mechanisms known in the theoretical literature.

For more details please watch the video.


Medium yo defense

Cristian Candia

Postdoctoral fellow, Universidad del Desarrollo

Computational Social Science, Network Science, Complex Systems, Collective Learning, Learning Analytic, Knowledge Diffusion, Experimental Game Theory.

No comments yet.