Barycentric extension of generalized matching.

Area of Science:

• Behavioral science
• Animal behavior studies
• Reinforcement learning

Background:

• Traditional choice studies use limited, identical options.
• Real-world choices involve numerous alternatives and inherent biases.
• Existing models struggle with complex choice environments.

Purpose of the Study:

• To present a novel methodology for investigating choice behavior with multiple alternatives and substantial biases.
• To extend the Generalized Matching Model to accommodate complex choice scenarios.
• To quantify individual biases in operanda choice.

Main Methods:

• Utilized twenty rats in a choice experiment with five distinct operanda (levers and pigeon keys).
• Employed probabilistic and concurrent schedules of reinforcement.
• Systematically varied reinforcer frequencies across operanda to generate biases.

Main Results:

• The Barycentric Matching Model effectively described the rats' choices across varying reinforcer frequencies.
• The model accurately estimated individual bias values for each operanda.
• A single exponent captured the sensitivity to reinforcer ratios.

Conclusions:

• The Barycentric Matching Model offers a robust framework for analyzing choice under concurrent schedules with multiple, biased alternatives.
• This method provides a more realistic approach to studying decision-making in complex environments.
• The findings advance our understanding of choice behavior and reinforcement learning.