Modeling and quantifying resurgence in the Evolutionary Theory of Behavior Dynamics

Area of Science:

• Behavioral science
• Artificial intelligence
• Computational psychology

Background:

• McDowell's Evolutionary Theory of Behavior Dynamics (ETBD) accurately models live organism behavior.
• Artificial organisms (AOs) animated by ETBD have replicated response resurgence in nonhuman subjects.
• Response resurgence is a phenomenon where a previously suppressed behavior re-emerges.

Purpose of the Study:

• To replicate a resurgence study using human participants within a three-phase paradigm.
• To evaluate computational models of resurgence based on Resurgence as Choice (RaC) theory.
• To identify the most parsimonious and accurate model for describing resurgence data in AOs.

Main Methods:

• Utilized a traditional three-phase resurgence paradigm with human participants.
• Developed and fitted two computational models derived from Resurgence as Choice (RaC) theory to AO data.
• Employed an information-theoretic approach to compare model fit, accounting for model complexity.

Main Results:

• A Resurgence as Choice in Context model, integrating Contingency Discriminability, provided the best fit to AO resurgence data.
• The chosen model demonstrated superior descriptive accuracy for resurgence phenomena.
• The study successfully replicated key aspects of resurgence behavior in a computational framework.

Conclusions:

• The Resurgence as Choice in Context model offers a robust explanation for behavioral resurgence.
• Computational modeling, particularly with ETBD and RaC, is valuable for understanding complex behaviors.
• Further research should consider integrating diverse factors into quantitative models of resurgence.