Adaptive learning in large populations

Area of Science:

• Evolutionary Game Theory
• Behavioral Economics
• Computational Biology

Background:

• The adaptive learning rule by Harley (1981) models behavior selection using accumulated rewards.
• This rule is observed across diverse life forms, from bacteria to humans.
• Understanding learning in heterogeneous populations is crucial for evolutionary dynamics.

Purpose of the Study:

• To analyze Harley's adaptive learning rule within symmetric conflict games.
• To model the evolution of learning strategies in large, heterogeneous populations.
• To investigate the impact of memory factor on convergence to optimal strategies.

Main Methods:

• Derivation of a partial differential equation (PDE) for agent distribution in stimulus space.
• Analysis of PDE model solutions for symmetric games.
• Mathematical modeling of adaptive learning dynamics.

Main Results:

• The PDE model describes the evolution of learning in heterogeneous populations.
• In symmetric games with small residual stimuli, adaptive learning rules with larger memory factors exhibit faster convergence.
• The memory factor significantly influences the speed of reaching optimal outcomes.

Conclusions:

• Adaptive learning rules are effective in driving populations towards optimal strategies in conflict games.
• Memory factor is a key parameter in the efficiency of adaptive learning.
• The study provides insights into the evolutionary dynamics of learning in biological and economic systems.