Foraging as sampling without replacement: A Bayesian statistical model for estimating biases in target selection.

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Behavioral Ecology

Background:

• Foraging behavior in humans and animals often involves searching for multiple targets sequentially.
• A common observation is the tendency to forage in 'runs' of the same target type, particularly when targets are difficult to distinguish from distractors.
• Existing measures like run statistics are limited as they are interdependent and constrained by target distribution, hindering analysis of cognitive processes.

Purpose of the Study:

• To address limitations in current measures of foraging behavior.
• To develop a novel modeling approach for analyzing target selection biases in human foraging.
• To facilitate direct comparison of foraging tendencies across different search environments.

Main Methods:

• Modeled foraging as a generative sampling without replacement procedure.
• Implemented this approach using a Bayesian multilevel model.
• Broke down foraging behavior into specific biases, including target proximity and run-selection bias, independent of target quantity.

Main Results:

• The developed model allows for the quantification of specific foraging biases.
• This approach is not dependent on the number or distribution of targets present in the search environment.
• The model facilitates direct comparisons of foraging tendencies across environments with varying characteristics.

Conclusions:

• The proposed Bayesian multilevel model offers a more precise and flexible method for analyzing human foraging behavior.
• This approach overcomes limitations of traditional run statistics, providing deeper insights into underlying cognitive processes.
• The model serves as a foundation for future research in visual foraging and decision-making.