How to Understand Behavioral Patterns in Big Data: The Case of Human Collective Memory

Area of Science:

• Behavioral Science
• Cognitive Science
• Data Science

Background:

• Simple patterns frequently emerge from complex systems, such as exponential decay in human perception and log-log relationships in word usage.
• Recent big data advancements allow for the characterization of common behavioral patterns, posing a challenge to understand their underlying mechanistic processes.

Purpose of the Study:

• To illustrate the challenge of understanding mechanistic processes behind observed behavioral patterns using big data analysis of collective memory.
• To emphasize the need for considering a broad set of alternative causal explanations in big data analyses within the behavioral sciences.

Main Methods:

• Analysis of collective memory decay patterns using big data.
• Application of a two-stage mechanistic model to fit the observed biexponential decay.
• Utilizing signal frequency analysis to generate alternative causal models.

Main Results:

• Collective memory exhibits a biexponential decay pattern over time, characterized by an initial rapid decay followed by a slower, longer-lasting decay.
• A two-stage mechanistic model was successfully fitted to the collective memory decay pattern.
• Signal frequency analysis provided several simple alternative models that precisely replicate the observed collective memory decay pattern.

Conclusions:

• While mechanistic models can fit observed patterns, big data analyses must incorporate a wide range of alternative causal explanations.
• Developing empirically testable alternative causal models is crucial for realizing the full potential of big data in the behavioral sciences.