Active Inferants: An Active Inference Framework for Ant Colony Behavior.

Area of Science:

• Theoretical Biology
• Ethology
• Computational Neuroscience

Background:

• Ant colonies exhibit complex collective behaviors through decentralized decision-making and information sharing.
• Active inference offers a multiscale framework for modeling behavior in biological systems.
• Understanding ant foraging provides insights into distributed systems and emergent intelligence.

Purpose of the Study:

• To introduce and implement an active inference model for simulating ant colony foraging behavior.
• To utilize a Markov decision process (MDP) to represent ant foraging dynamics.
• To validate the model's ability to reproduce empirical observations from ant behavior experiments.

Main Methods:

• Development of an active inference model based on Markov decision processes (MDPs).
• Simulation of the model using *in silico* experiments.
• Testing the model against the alternating T-maze paradigm commonly used in ant behavioral studies.

Main Results:

• The active inference model successfully simulated ant colony foraging.
• The model reproduced key phenomena, including trail formation after food discovery.
• Demonstrated the model's capacity to capture basic ant colony behavioral patterns.

Conclusions:

• The developed active inference model provides a computational framework for studying ant foraging.
• The model can be extended within a broader multiscale and systems biology approach.
• This work contributes to understanding collective behavior and decision-making in biological systems.