New Work on CGSs: ritual and incorporative modeling

Here are two short features (something I am calling social media posters) that I debuted on Tumbld Thoughts. Think of this as posting entries from your laboratory notebook (one of several, in my case) to social media. In this case, each entry is further development of the Contextual Geometric Structures idea. The first poster is on the exaptation of rituals from a mathematical modeling perspective, using Halloween as a seasonally-appropriate example. The second poster is on something called incorporative modeling, using my "return from Robotistan" [1] as an example.

I. Seasonally-appropriate Cultural Exaptation

Why do rituals change over time? Why do they resemble weird things? And why do people believe even weirder things about these weird things? Here is a demonstration of something I am calling ritual modeling, using Halloween as an example. 

Some elements (particularly the forgetting/ decay measurements) of this are drawn from [2]. Additional process modeling and graph with pseudo-data were done to demonstrate this idea.

II. Returning from Robotistan

Here is a counterpart to my recent post on ritual modeling. To place this in context, here are a few readings on the highly predictive parts of human behavior. In [3], the analysis of easily- tracked human behaviors (such as mobility) can lead to highly predictable patterns. In [4], Sandy Pentland from MIT discusses how big data (databases of internet behavior) allow us to predict human behavior independently of morals and values. This has the effect of uncovering behaviors (using machine learning techniques) that people are usually not straightforward about in their language or public persona.

These types of analyses are useful both in terms of understanding aggregate cultural trends and the construction of crude behavioral models. So while such approaches are highly successful at characterizing well-known behaviors, they do not consider how new information is incorporated into behavioral schemes, especially those which are not highly predictive to begin with. This requires a more purely computational (e.g. simulation-based) approach.

To address this, I introduce something called an incorporative model, which draws from earlier work on Contextual Geometric Structures (CGSs). CGSs [2] are a hybrid soft classifier/fluid dynamics-inspired computational model of cultural behavior (e.g. culturally-conditioned collective behaviors). The slides show how observations are incorporated into such models, and how this new information is shared across a population of agents.

Before we end, the video for my lecture to the BEACON Center in May is now available on the BEACON Center YouTube channel. The second part discusses the application of CGSs to understanding the evolution and dynamics of economic value.

NOTES:

[1] For more musings from my trip, please see: Alicea, B.   Fear and Loathing in Robotistan. Synthetic Daisies blog, August 20 (2013).

[2] Alicea, B.   Contextual Geometric Structures: modeling the fundamental components of cultural behavior. Proceedings of Artificial Life, 13, 147-154 (2012).

[3] Song, C., Qu, Z., Blumm, N., Barabasi, A-L.   Limits of Predictability in Human Mobility. Science, 327, 1218-1021 (2010).

[4] Pentland, S.   Predicting Customers' (Unedited) Behavior. Harvard Business Review, September 19 (2012).