Globalization and the rise and fall of cognitive control

Mohsen Mosleh, Katelynn Kyker, Jonathan D. Cohen, David G. Rand

TL;DR

This paper uses an evolutionary game theory model to explore how the scale of interaction (global vs. local) influences the evolution of cognitive control. They find that globalization can have complex effects, sometimes favoring control and sometimes disfavoring it, depending on the specific type of globalization (learning, contact, or environment).

What Problem Does This Paper Address?

The paper addresses the question of how increasing global interconnectedness and interaction might shape the evolution of human cognition, specifically the balance between automatic (fast, reflexive) and controlled (slow, deliberative) processing. While cognitive psychology has extensively studied this distinction, evolutionary game theory has largely overlooked it. The authors aim to bridge this gap by modeling how different scales of interaction—local versus global—affect the prevalence of cognitive control in a population. This is particularly relevant given current societal trends of increasing globalization and the potential for both its benefits and drawbacks.

Methods

The authors employ an agent-based evolutionary game theory model. Agents are characterized by a probability of using controlled processing ($x_i$). Their fitness is determined by their own processing strategy, the strategies of other agents they interact with, and the state of the environment. The model incorporates several key features:

  • Cognitive Processing: Agents can use either automatic or controlled processing, with a trade-off between efficiency (automatic) and flexibility (controlled), and a fixed cost associated with controlled processing.
  • Interaction Structures: The model simulates different scales of interaction:
  • Learning: Agents can learn from local neighbors or the entire global population.
  • Contact: Agents interact with local neighbors or the entire global population.
  • Environment: The environmental state, which influences payoffs, can be local (agent-specific) or global (population-wide).
  • Evolutionary Dynamics: Agents update their strategies using a death-birth Moran process, where individuals with higher fitness are more likely to reproduce. A mutation rate is also included.
  • Cognition-Environment Feedback: A crucial element is the feedback loop where the collective level of controlled processing in the population influences the environment, which in turn affects the payoffs. This feedback can be subject to a time lag.

The novelty lies in integrating these distinct cognitive and evolutionary frameworks and systematically exploring the impact of local versus global interaction structures across these three dimensions.

Datasets

Not discussed in the paper. The study relies on simulations of an agent-based model, not empirical datasets.

Key Insights and Ideas

  • Globalization is Not Monolithic: The impact of globalization on cognitive control is not uniform. Different forms of globalization (learning, contact, environment) have distinct and sometimes opposing effects.
  • Contact Matters: The Sign of $w$ is Crucial: The effect of globalized contact depends critically on whether automatic agents harm or benefit controlled agents. If automaticity is detrimental ($w < 0$), global contact reduces cognitive control. If automaticity is beneficial ($w > 0$), global contact can sometimes increase cognitive control.
  • Global Environment Favors Heterogeneity, Local Environment Favors Homogeneity: Counterintuitively, a global environment, where all agents experience the same environmental state, can lead to greater diversity in cognitive strategies (some agents being fully automatic, others fully controlled). Conversely, a local environment, where each agent's experience is influenced by its neighbors, tends to homogenize strategies.
  • Global Learning Destroys Communities: Global learning, where agents can learn from anyone in the population, leads to a well-mixed population and eliminates the emergence of mesoscale communities of similar cognitive strategies.
  • Cognition-Environment Feedback Cycles: The model reinforces previous findings that lagged cognition-environment feedback can lead to cyclical fluctuations between high and low levels of cognitive control, suggesting a potential for societal collapse. Local learning, however, can suppress these aggregate-level temporal oscillations by creating spatial decoupling.

Key Results

  • Average Levels of Control: When $w < 0$ (automaticity harms control), globalization of any kind substantially reduces the average level of cognitive control. When $w > 0$ (automaticity benefits control), global contact can slightly increase average control.
  • Mesoscale Communities: Local learning promotes the emergence of mesoscale communities with similar cognitive control levels. Global learning destroys these communities, leading to a more homogeneous population. Interestingly, global environment, rather than local, can strengthen mesoscale communities when learning is local.
  • Variance in Strategies: Global environments tend to increase the variance in cognitive control strategies within the population, leading to more heterogeneous populations. Local environments tend to homogenize strategies.
  • Time Lags and Oscillations: Increasing time lags in cognition-environment feedback leads to larger amplitude oscillations in cognitive control. Local learning, while promoting spatial oscillations, suppresses aggregate-level temporal oscillations.

Limitations and Future Work

  • Model Simplification: The authors acknowledge that their model is an abstract and simplified representation of complex real-world phenomena. It uses a single parameter ($x_i$) to represent cognitive control and a simplified network structure (ring, with extensions to small-world networks).
  • Quantitative Predictions: The model is intended to provide high-level conceptual predictions rather than precise quantitative forecasts.
  • Future Work: The authors suggest exploring these conceptual predictions in laboratory experiments and by examining observational data. They also highlight the importance of understanding the specific nature of interactions in globalization and how to promote beneficial interactions where control benefits from contact with automaticity. The implications for global environmental issues like climate change and the risk of global collapse due to interconnectedness are also noted as areas for further consideration.

Summary updated

Abstract

The scale of human interaction is larger than ever before-people regularly interact with and learn from others around the world, and everyone impacts the global environment. We develop an evolutionary game theory model to ask how the scale of interaction affects the evolution of cognition. Our agents make decisions using automatic (e.g., reflexive) versus controlled (e.g., deliberative) cognition, interact with each other, and influence the environment (i.e., game payoffs). We find that globalized direct contact between agents can either favor or disfavor control, depending on whether controlled agents are harmed or helped by contact with automatic agents; globalized environment disfavors cognitive control, while also promoting strategic diversity and fostering mesoscale communities of more versus less controlled agents; and globalized learning destroys mesoscale communities and homogenizes the population. These results emphasize the importance of the scale of interaction for the evolution of cognition, and help shed light on modern challenges.
BibTeX
@inproceedings{mosleh2020globalization,
  author = {Mohsen Mosleh and Katelynn Kyker and Jonathan D. Cohen and David G. Rand},
  title = {{Globalization and the rise and fall of cognitive control}},
  booktitle = {Nature Communications},
  year = {2020},
  doi = {10.1038/s41467-020-16850-0},
}
Figure