Dynamic Human Trust Modeling of Autonomous Agents With Varying Capability and Strategy

TL;DR

The paper investigates how human trust in autonomous agents evolves over time in a cooperative screen-based grid-search task, focusing on how agent capability, strategy, and the order of exposure shape trust dynamics. It combines self-reported trust measures with ARIMAX time-series modeling to capture temporal dependencies and exogenous influences on trust, comparing against traditional OLS approaches. Findings show that ARIMAX models, which incorporate prior trust states and agent characteristics, provide better one-step-ahead predictions and reveal recency effects, especially when agent capabilities and strategies change across time scales. The work highlights the importance of representing autonomous agent characteristics over time to predict and manage human trust in evolving human-autonomy collaboration, with implications for designing adaptive, trustworthy autonomous systems.

Abstract

Objective We model the dynamic trust of human subjects in a human-autonomy-teaming screen-based task. Background Trust is an emerging area of study in human-robot collaboration. Many studies have looked at the issue of robot performance as a sole predictor of human trust, but this could underestimate the complexity of the interaction. Method Subjects were paired with autonomous agents to search an on-screen grid to determine the number of outlier objects. In each trial, a different autonomous agent with a preassigned capability used one of three search strategies and then reported the number of outliers it found as a fraction of its capability. Then, the subject reported their total outlier estimate. Human subjects then evaluated statements about the agent's behavior, reliability, and their trust in the agent. Results 80 subjects were recruited. Self-reported trust was modeled using Ordinary Least Squares, but the group that interacted with varying capability agents on a short time order produced a better performing ARIMAX model. Models were cross-validated between groups and found a moderate improvement in the next trial trust prediction. Conclusion A time series modeling approach reveals the effects of temporal ordering of agent performance on estimated trust. Recency bias may affect how subjects weigh the contribution of strategy or capability to trust. Understanding the connections between agent behavior, agent performance, and human trust is crucial to improving human-robot collaborative tasks. Application The modeling approach in this study demonstrates the need to represent autonomous agent characteristics over time to capture changes in human trust.

Figures (11)

• Figure 1: In the experimental task, the subject was instructed to search a hidden area for outlier circles: those with contrasting centers. The cutaway region on the right side shows a grid of possible outliers. Each trial, an Autonomous Searcher with differing capability and strategy searched alongside the subject. In this sample, the orange autonomous searcher used the "lawnmower" strategy.
• Figure 2: Sample AS paths from each strategy shown in orange. The paths shown are eroded to illustrate the behavior better. Outliers are blue circles. The Random and Omniscient paths are generated randomly but shared within subjects of the same group.
• Figure 3: Subjects were assigned to one of two groups according to the order in which they began the experiment. Both groups started with an identical tutorial and self-search block before starting the main experiment. Subjects in group 0 had a constant strategy in each block, while group 1 had a constant capability in each block. In this figure, capability is indicated by the shade of the box and strategy by the outline of the box.
• Figure 4: This study follows the modeling process outlined in this flowchart.
• Figure 5: Single subject representative trust trajectory.