Interpreting Recovery in Human–Robot Affinity
Year: 2026
Type: preprint
Abstract: A response to a robot can return above its initial level even while an adverse contribution continues to grow. Interpreting such recovery requires distinguishing what a trajectory can exclude from what it can identify. We examine models in which the same positive process contributes to the response and drives an opponent through fixed nonnegative filtering. With zero history, fixed positive observation weights and a log-concave positive driver, the opponent-to-driver ratio cannot decrease, so a return above baseline after a negative excursion is impossible. The argument applies to arbitrary nonnegative delay mixtures and includes a published isolated-dose opponent architecture under its actual decaying input. A saturating two-state construction illustrates one way to escape the restriction while both contributions increase. Fixed nondecreasing scalar observation maps preserve the strict crossing diagnostic, whereas arithmetic averaging need not: an exact example has a recovering mean although neither individual recovers. These conditional comparisons clarify how individual trajectories, exposure history and observation assumptions enter model selection. The examples are synthetic and do not validate a psychological interpretation of the latent states.
Author: Vaseghi, Sam
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225466682