A further application of the active time model to multiple concurrent variable-interval schedules

Area of Science:

• Behavioral Psychology
• Animal Cognition
• Reinforcement Learning

Background:

• Concurrent variable interval (VI) schedules are fundamental in understanding choice behavior.
• Predictive models are crucial for explaining and validating experimental findings in operant conditioning.

Purpose of the Study:

• To evaluate the predictive accuracy of the active time model (ATM) for probe data under concurrent VI VI schedules.
• To compare the performance of ATM against a variant of scalar expectancy theory.

Main Methods:

• Subjects (birds) were trained on concurrent VI 30-s VI 60-s and VI 60-s VI 120-s schedules.
• Unreinforced probes were conducted, pairing stimuli with equivalent absolute but different relative reinforcement rates, and vice versa.
• Behavioral data from probes were analyzed using ATM and scalar expectancy theory.

Main Results:

• Birds showed preferences in probe trials that aligned with reinforcement schedules.
• The active time model (ATM) accurately fit individual subject data for both probe types.
• A variant of scalar expectancy theory failed to fit the data at individual or group levels.

Conclusions:

• The active time model (ATM) provides an accurate account of behavior under concurrent variable interval schedules.
• ATM demonstrates superior predictive power compared to scalar expectancy theory in this context.
• Findings support the utility of ATM in modeling complex reinforcement schedules.