Symbolic dynamics of animal interaction

Area of Science:

• Information theory
• Animal behavior
• Dynamical systems analysis

Background:

• Transfer entropy aids understanding of cause-and-effect in coupled systems using time series data.
• Its application in animal behavior, particularly for social interactions and predator responses, remains underexplored due to technical challenges.

Purpose of the Study:

• To explore an alternative method for computing transfer entropy using symbolic dynamics.
• To assess the utility of this symbolic approach for analyzing animal behavior and inferring causal relationships.

Main Methods:

• Developed a framework using symbolic dynamics to represent animal locomotion patterns as symbols.
• Applied permutation transfer entropy to quantify predictive relationships between symbolized movement patterns of individuals.
• Validated the method on a fish behavior dataset with known causal interactions.

Main Results:

• Symbolic dynamics provides an intuitive and robust framework for transfer entropy computation in animal behavior.
• The approach effectively quantifies the predictability of one individual's movement patterns based on another's.
• Demonstrated successful inference of causal relationships from noisy, limited-duration observational data.

Conclusions:

• Symbolic dynamics offers a promising avenue for advancing the application of transfer entropy in behavioral ecology.
• This method enhances the ability to uncover causal links in animal interactions from observational data.
• The approach has the potential to overcome limitations of traditional transfer entropy methods in complex behavioral studies.