Tackling the Scaffolding Paradox: A Person-Centered Adaptive Robotic Interview Coach

TL;DR

This work addresses how to balance psychological safety with instructional guidance in a robotic interview coach for anxious university students. It presents a three phase iterative design using Person-Centered Therapy and adaptive scaffolding, culminating in an Agency-Driven Interaction mode and the Adaptive Scaffolding Ecosystem. The Phase I results show a Safety-Guidance Gap, Phase II reveals a Scaffolding Paradox, and Phase III demonstrates that user agency buffers anxiety while preserving warmth and therapeutic alliance. The study proposes a framework for dynamically calibrating emotional support and guidance, with implications for deploying socially assistive robots in high stakes training and evaluation contexts.

Abstract

Job interview anxiety is a prevalent challenge among university students and can undermine both performance and confidence in high-stakes evaluative situations. Social robots have shown promise in reducing anxiety through emotional support, yet how such systems should balance psychological safety with effective instructional guidance remains an open question. In this work, we present a three-phase iterative design study of a robotic interview coach grounded in Person-Centered Therapy (PCT) and instructional scaffolding theory. Across three weekly sessions (N=8), we systematically explored how different interaction strategies shape users' emotional experience, cognitive load, and perceived utility. Phase I demonstrated that a PCT-based robot substantially increased perceived psychological safety but introduced a Safety-Guidance Gap, in which users felt supported yet insufficiently coached. Phase II revealed a Scaffolding Paradox: immediate feedback improved clarity but disrupted conversational flow and increased cognitive load, whereas delayed feedback preserved realism but lacked actionable specificity. To resolve this tension, Phase III introduced an Agency-Driven Interaction Mode that allowed users to opt in to feedback dynamically. Qualitative findings indicated that user control acted as an anxiety buffer, restoring trust, reducing overload, and reframing the interaction as collaborative rather than evaluative. Quantitative measures further showed significant reductions in interview-related social and communication anxiety, while maintaining high perceived warmth and therapeutic alliance. We synthesize these findings into an Adaptive Scaffolding Ecosystem framework, highlighting user agency as a key mechanism for balancing emotional support and instructional guidance in social robot coaching systems.

Figures (8)

• Figure 1: Three-phase developmental trajectory of the robotic coach, showing key findings and design changes in the Adaptive Scaffolding Ecosystem.
• Figure 2: System workflow illustrating real-time interaction between participants, the robot, the Node.js server, and the OpenAI Realtime API, with researcher monitoring via an experiment dashboard.
• Figure 3: Laboratory setup for the human–robot interview session
• Figure 4: Overview of the iterative interaction design cycle and study procedure. Across three phases, the system was iteratively deployed, evaluated through user testing and mixed-method assessments, and refined based on quantitative measures and qualitative feedback through the 5-step design cycle.
• Figure 5: Individual RoSAS scores across sessions and conditions for Warmth, Competence, and Comfort. Solid lines represent individual participants, and the dashed line represents the group average.