When models interact with their subjects: the dynamics of model aware systems

Area of Science:

• Complex systems dynamics
• Scientific modeling
• Behavioral economics

Background:

• Scientific models are often static, but subjects can influence models.
• Model-aware systems necessitate dynamic, adaptive modeling approaches.
• Understanding feedback loops between models and their subjects is crucial.

Purpose of the Study:

• To present two models for the dynamics of model-aware systems.
• To explore population behaviors influenced by self-descriptive models.
• To investigate experimentalists' publishing behavior influenced by confirmation bias and models.

Main Methods:

• Developing a model for prediction-seeking and prediction-avoiding populations.
• Creating a model for experimentalists' publishing behavior with confirmation bias.
• Performing numerical simulations and comparing them with real-world data.

Main Results:

• Model-aware systems can display convergent random or oscillatory behavior.
• Universal 1/f noise was observed in model-aware systems.
• Simulations of experimentalists' behavior quantitatively matched neutron lifetime and mass measurements.

Conclusions:

• Model-aware systems exhibit complex dynamics influenced by their models.
• The findings support the quantitative agreement between simulated and real-world scientific publishing behavior.
• Dynamic modeling is essential for understanding systems that interact with their own descriptions.