The copyist model of response emission

Area of Science:

• Behavioral psychology
• Computational modeling
• Reinforcement learning

Background:

• Modeling performances under various reinforcement schedules is complex.
• Variable reinforcement schedules maintain steady response rates, making them suitable for simpler models.

Purpose of the Study:

• To develop a computational model for predicting response rates under variable reinforcement schedules.
• To account for the influence of interresponse times on behavior.

Main Methods:

• The model calculates response rate based on a decaying average of recent interresponse times.
• It randomly selects interresponse times with equal probability from the last 300, defining an exponential distribution for response emission.

Main Results:

• The model's response rates approximated empirical findings on variable schedules.
• It successfully reproduced three key effects: VR > VI rates, inverted-U rate function with reinforcement density, and matching on concurrent schedules.

Conclusions:

• The model's success suggests that responding under single and concurrent schedules can be explained by reproducing reinforced interresponse times.
• Behavior is viewed as a form of copying previously reinforced response patterns.