Optimal random search using limited spatial memory.

Area of Science:

• Computational modeling of foraging behavior
• Statistical physics of random walks
• Animal movement ecology

Background:

• Lévy walks are efficient search strategies due to broad area coverage and minimal revisits.
• Self-avoiding walks (SAWs) offer effective search patterns by preventing revisits.
• Real-world foragers have limited memory, posing challenges for sustained optimal search.

Purpose of the Study:

• To investigate optimal foraging strategies for agents with limited memory.
• To model how memory loss affects search efficiency and movement patterns.

Main Methods:

• Developed an agent-based model incorporating self-avoiding walks (SAWs) with decaying memory.
• Simulated agent behavior where memory loss triggers adaptive responses to visited locations.
• Analyzed movement patterns, including the emergence of ballistic walks and power-law distributions.

Main Results:

• Agents initially followed SAWs, avoiding previously visited sites.
• Upon memory loss, agents adapted by using local information to infer global trail patterns.
• This adaptive behavior resulted in intermittent ballistic walks, characteristic of Lévy-like movements.

Conclusions:

• Limited memory does not preclude efficient foraging; adaptation is key.
• Agents can generate Lévy-like movement patterns through adaptive responses to memory decay.
• The model demonstrates a plausible mechanism for efficient searching in natural systems with memory constraints.