Performance on the rapidly changing procedure, according to an associative learner

Area of Science:

• Behavioral psychology
• Computational modeling

Background:

• Associative Learner models are used to predict behavior based on event records.
• Understanding choice behavior under concurrent schedules is crucial in behavioral science.

Purpose of the Study:

• To evaluate an implementation of Shimp's (1978) Associative Learner for concurrent variable-interval schedules.
• To identify and rectify weaknesses in the base Associative Learner model's predictive capabilities.

Main Methods:

• Implementing Shimp's (1978) Associative Learner model.
• Testing the model's predictions against empirical data for two-alternative concurrent variable-interval schedules.
• Developing an extended Associative Learner implementation with modifications for reinforced and bout-like responses.

Main Results:

• The base Associative Learner implementation showed limitations in predicting inter-response times, reinforcement acquisition, and response rates.
• The extended Associative Learner implementation demonstrated improved accuracy in accounting for these behavioral measures.
• The modifications successfully addressed weaknesses related to post-reinforcement responding and response organization.

Conclusions:

• The extended Associative Learner model provides a more robust account of behavior under concurrent variable-interval schedules.
• Modifications enhancing the model's sensitivity to reinforcement history and response patterns are critical for accurate behavioral prediction.
• This research contributes to the refinement of computational models in understanding animal and human decision-making.