Elucidating simulated equivalence responding through dynamic visualization of structural connectivity and relational density

This article presents Affinity, a visual analytics tool that enhances the simulation of the emergence of derived relations between stimuli in humans. Built on the foundations of a reinforcement learning model called Enhanced Equivalence Projective Simulation, Affinity provides both real-time visualizations of the agent's relational memory and enables the simulation of Relational Density Theory, a novel approach to understanding relational responding through the modeling of higher-order properties of density, volume, and mass. We demonstrate these features in a simulation of a recent study into the quantification of relational volume. We also use this as an opportunity to examine the effect of the underlying model's consolidation mechanism, Network Enhancement, on the agent's relational network. Our results highlight Affinity's innovation as an explainable modeling interface for relational formation and a testbed for new experiments. We discuss the limitations of Affinity in its current state, underline future work on the software and computational modeling of Stimulus Equivalence and locate this contribution in the broader scope of integrations of Contextual Behavioral Science and Artificial Intelligence.