Sustaining Engagement in Long-Term Social Human-Robot Interaction

You are reading version 1 of this direction, frozen 2026-10-04. The latest version is version 2.

Version 1 of 262 papers81 conceptsfrozen 2026-10-04

In the first few minutes of meeting a social robot, users are almost universally charmed. Whether it is a humanoid like Pepper or a specialized coach, the "novelty effect" creates an initial spike in trust, engagement, and compliance. However, for researchers like Cory D. Kidd in A Robotic Weight Loss Coach, this honeymoon phase is a significant hurdle. If a robot is meant to assist with chronic conditions—like weight loss, diabetes management, or mental health—it must remain effective long after the "wow factor" has evaporated.

The core challenge of this research direction is moving Human-Robot Interaction (HRI) out of the controlled, single-session lab environment and into the "messy reality" of the home. To do this, researchers are developing strategies to combat the inevitable decline in user interest through habituation modeling, strategic unpredictability, and socio-cognitive engineering.

The Habituation Paradox

One might assume that as a robot’s technical flaws (like speech latency or repetitive motions) become apparent, user frustration would lead to total abandonment. However, longitudinal studies suggest a more complex "habituation" process.

Micol Spitale, Hatice Gunes, and colleagues found in Longitudinal Evolution of Coachees' Behavioural Responses to Interaction Ruptures in Robotic Positive Psychology Coaching that users actually perceive fewer interaction ruptures over time. Rather than the robot getting better, the humans adapt. They learn the robot’s "social scripts" and technical limitations, lowering their expectations to meet the machine's capabilities. This suggests that long-term success depends less on perfect mimicry of a human and more on establishing a stable, predictable "therapeutic alliance" where the user knows how to navigate the robot's quirks.

Combatting Boredom with Strategic Unpredictability

If a robot is perfectly predictable, it becomes a piece of furniture. To maintain influence over months, researchers are exploring ways to keep the interaction "fresh" without being annoying.

Shahabedin Sagheb and Andrea Bajcsy argue in Towards Robots that Influence Humans over Long-Term Interaction that robots lose influence because humans learn to "game" or ignore them. Their solution is to optimize for "belief entropy"—essentially making the robot’s strategy just unpredictable enough that the human cannot easily bypass it.

Similarly, Byron Boots and Patrick E. Lancaster explored this through adversarial play in Motivating Physical Activity via Competitive Human-Robot Interaction. They found that a robot that thwarts a user's goals (like a fencing partner) can sustain higher engagement than a purely cooperative one, as the competition provides a continuous source of novelty and challenge.

Socio-Cognitive Engineering for Chronic Care

For long-term behavioral change, the robot must transition from a "gadget" to a "partner." This requires a methodology that Willeke van Vught and Joost Broekens call Socio-Cognitive Engineering (SCE).

In Socio-Cognitive Engineering of a Robotic Partner for Child's Diabetes Self-Management, the researchers demonstrated that sustaining engagement over months requires a "multi-modal" approach. This includes:

  1. Personalization: Adapting content to the user's specific progress.
  2. Memory: Referencing past interactions to build a sense of shared history.
  3. Co-regulation: Allowing the user (and parents) to have granular control over the robot’s behavior, as seen in the work of Hui-Ru Ho and Bilge Mutlu on In-Home Parental Math Talk.

Detecting the "Spark" in the Wild

To adapt, a robot must first know when it is losing its audience. In the lab, researchers can use high-end sensors, but in the home, the environment is "noisy."

Balasubramanian Thiagarajan and Caitlyn Clabaugh addressed this in Modeling Engagement in Long-Term, In-Home Socially Assistive Robot Interventions for Children with Autism Spectrum Disorders. By using machine learning to fuse audio, video, and game-performance data, they created models that could detect engagement with 90% accuracy. This allows the robot to autonomously "refresh" the interaction—perhaps by changing its tone or introducing a new activity—the moment it senses the novelty effect is bottoming out.

Ultimately, the goal of this research is to move beyond the "first date" of HRI. By modeling the temporal decay of interest and building systems that can adapt their social strategies, researchers are turning robots into enduring tools for health and education.

Papers in this version

Showing 60 of 62 papers.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 82 papers
  2. Version 1 · 2026-10-04 · 62 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_novelty_effect_in_human_robot_interactio_d8161d06,
  title = {novelty effect in human-robot interaction},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/novelty-effect-in-human-robot-interaction/rdr_d8161d06c235410ab1f7b3342b305288},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_sustaining_engagement_in_long_term_socia_d8161d06_v1,
  title = {Sustaining Engagement in Long-Term Social Human-Robot Interaction},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/sustaining-engagement-in-long-term-social-human-robot-interaction/rdr_d8161d06c235410ab1f7b3342b305288/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}