The dynamics of conditioning and extinction.

Area of Science:

• Behavioral Psychology
• Animal Cognition
• Learning Theory

Background:

• Intermittently reinforced classical conditioning in pigeons often results in variable responding.
• Previous models struggled to capture the trial-by-trial nuances of pigeon conditioning and extinction.

Purpose of the Study:

• To develop and validate a computational model for analyzing pigeon responses in classical conditioning.
• To investigate the influence of prior trial events on current responding patterns.
• To explore the dynamics of response rates during protracted periods without reinforcement.

Main Methods:

• A linear persistence-learning model was employed to track pigeon response states.
• A Weibull distribution characterized the number of responses within trials.
• Experimental manipulations of trial and intertrial durations were used to assess model sensitivity.

Main Results:

• Responding patterns were significantly dependent on the reinforcement history of the preceding trial.
• The model successfully predicted a plateauing of response rates even without reinforcement.
• Trial-to-trial reinforcement data yielded more accurate, jagged learning curves than probability-based predictions.

Conclusions:

• The developed model provides a robust framework for detailed trial-by-trial analysis of conditioning and extinction.
• This approach enhances data interpretability, allowing for more definitive testing of contemporary learning models.
• The findings offer new perspectives on behavioral momentum and conditioning processes in pigeons.