Human Autonomy and Sense of Agency in Human-Robot Interaction: A Systematic Literature Review

Synthesis: Glawe, Schmeckel, Brauner, and Ziefle (2025) systematically review empirical studies on human autonomy and sense of agency in human-robot interaction (HRI), aiming to bridge the gap between design frameworks and regulatory demands (e.g., the EU AI Act, IEEE Ethically Aligned Design) on one hand and available empirical evidence on the other. Using the PRISMA workflow, they queried five databases and identified 22 empirical studies published between 2011 and early 2024. Across both concepts, separate methodological strands emerged: Self-Determination Theory-based measures of autonomy via psychometric scales, and neuroscientific measures of agency (primarily intentional binding).

Key Findings

• Human autonomy and sense of agency are core human values foundational to well-being, intertwined in definition and central to current ethical and AI Governance discourses on AI and cyber-physical systems (EU AI Act, IEEE Ethically Aligned Design).
• A PRISMA-guided systematic review of five databases identified 22 empirical studies published 2011–early 2024 on autonomy and sense of agency in HRI.
• Two distinct methodological strands emerged: self-determination-theory-based psychometric measures of autonomy, and neuroscientific measures of agency (primarily intentional binding).
• The review bridges design frameworks and regulatory demands with empirical evidence, highlighting gaps and directions for HRI design and governance.

Study Design & Method

This is a systematic literature review following the PRISMA workflow. The authors queried five databases and screened the literature to identify 22 empirical studies published between 2011 and early 2024 examining human autonomy and sense of agency in human-robot interaction. They analyzed the conceptualization and measurement of autonomy and agency across the studies, identifying distinct methodological strands (psychometric/SDT-based measures of autonomy; neuroscientific/intentional-binding measures of agency) and synthesizing implications for design frameworks and regulatory compliance.

What this means for practice

• Designers. Design for the factors the evidence supports and treat dependence as a risk: non-controlling communication and meaningful choice are backed by the reviewed studies, with gaze direction and robot anthropomorphism flagged as additions to current educational robotics guidance, while over-attachment, disrupted routines, and growing dependence appear as life-sphere threats.
• Designers. Evaluate beyond the interface and task spheres: the 22 studies cluster there, and behavior- and life-sphere effects such as declining skills, increased dependence, and routine disruption surfaced mainly in qualitative work, so long-term, real-world evaluation is the missing piece.
• Researchers. Measure both constructs in the same study: the corpus splits into SDT-based psychometric measures of autonomy and intentional-binding measures of sense of agency, and none of the 22 articles linked sense of agency to autonomy satisfaction.
• Researchers. Standardize definitions and instruments so results become comparable — the review could not run a meta-analysis because conceptual and methodological heterogeneity across the 22 studies was too pronounced.
• Researchers. Search with widely adopted vocabulary and expect a thin literature: 728 records from five databases screened down to 22 articles (2011–2024, with 10 published in 2023), and one relevant study was missed because it used "basic needs" rather than "basic psychological needs".

Limitations

• Twenty-two studies, a fragmented field. The review rests on 22 studies, reflecting a young and methodologically fragmented literature.
• Two concepts, incompatible instruments. Autonomy and agency are measured with very different instruments, which makes cross-study synthesis difficult.
• A window that closes in early 2024. The temporal scope of 2011 to early 2024 predates the widespread use of Large Language Models (LLMs)-powered social robots, so recent developments may be underrepresented.
• Human–robot interaction broadly, education only in part. The review's scope is HRI rather than education specifically.

Citation

Glawe, F., Schmeckel, T., Brauner, P., & Ziefle, M. (2025). Human autonomy and sense of agency in human-robot interaction: A systematic literature review.