Intelligent systems in the context of surrounding environment

Area of Science:

• Computational neuroscience
• Artificial intelligence
• Agent-based modeling

Background:

• Biologically motivated neural networks are used to model agent behavior.
• The minority model by Challet and Zhang provides a competitive framework.
• Understanding emergent behaviors in multi-agent systems is crucial.

Purpose of the Study:

• To investigate behavioral patterns of agents in a competitive minority model.
• To explore how agent characteristics, like memory, influence emergent behaviors.
• To analyze the relationship between neural network architecture and agent analytic capability.

Main Methods:

• Simulating a population of agents controlled by neural networks.
• Implementing the minority model for agent competition.
• Varying agent memory and neural network intermediary layer size.

Main Results:

• Agents with similar memory exhibit crowding behavior.
• "Rogue" agents with higher memory exploit majority populations.
• Analytic capability is correlated with the size of the neural network's intermediary layer.

Conclusions:

• Intelligence is an emergent property dependent on environmental context (embodiment).
• Neural network architecture significantly impacts agent cognitive abilities.
• Competitive dynamics can reveal fundamental principles of natural and artificial intelligence.