(Reinforcement?) Learning to forage optimally.

Area of Science:

• Neuroscience
• Animal Behavior
• Cognitive Science

Background:

• Effective foraging is crucial for animal survival and brain evolution.
• The precise learning and decision mechanisms behind foraging strategies are not fully understood.
• Human laboratory foraging tasks show complex behaviors not always explained by standard reinforcement learning models.

Purpose of the Study:

• To explore the learning and decision mechanisms underlying foraging behavior.
• To investigate if a unified framework can explain divergent foraging strategies.
• To propose model-based average reward reinforcement learning as a potential unifying model.

Main Methods:

• Analysis of human behavior in laboratory foraging tasks.
• Comparison of observed behaviors with predictions from model-free reinforcement learning.
• Evaluation of heuristic strategies and prospective prediction models.
• Assessment of model-based average reward reinforcement learning as an explanatory framework.

Main Results:

• Model-free reinforcement learning inadequately explains some human foraging behaviors.
• Simple heuristic strategies and prospective prediction models offer partial explanations for observed behaviors.
• Model-based average reward reinforcement learning shows potential as a unifying framework.

Conclusions:

• Human foraging strategies appear diverse and are not fully captured by current models.
• Model-based average reinforcement learning may offer a more comprehensive framework for understanding foraging decision-making.
• Further research is needed to validate this framework across different species and tasks.