Robots are increasingly explored in mental healthcare, showing promise in autism, dementia, and pediatric contexts. Unlike other technologies, robots possess physical bodies, enabling embodied communication such as affective touch.
Prior human-robot interaction (HRI) research shows that robot-initiated touch can elicit positive emotions and prosocial responses. However, whether affective touch specifically influences mood, self-compassion, and linguistic expression during disclosure of personal concerns remains unexplored.
This study addressed this gap by examining how robot-delivered affective touch affects psychological states and conversational characteristics when participants discuss their concerns with RoBoHoN.
Experimental Approach to Robot Touch
Thirty participants were randomly assigned to either the Affective Touch group (n=16) or the Control group (n=14). The sample was underpowered for detecting medium effects, so null results should be
interpreted cautiously.
Exclusion criteria included heart conditions, regular yoga or mindfulness practice, non-native Japanese speakers, and high alexithymia scores (a trait involving difficulty identifying emotions). Personality traits and verbal intelligence were also assessed to ensure group equivalence. Ethical screening confirmed no participants met criteria for major depressive episodes.
The robot used was RoBoHoN, a humanoid robot phone controlled via a programming application. Affective touch was operationalized using three criteria, namely, velvet material, stroking speed of approximately six centimeters per second (within the optimal range for activating C-tactile fibers), and a subjectively pleasant sensation confirmed through pilot testing.
Participants completed questionnaires measuring mood states, negative attitudes toward robots, anxiety toward robots, self-compassion, distress related to their concerns, and perceived rapport.
After a pre-task survey, participants discussed personal concerns with RoBoHoN for 20 minutes. In the Affective Touch group, the robot gently stroked participants' hands during conversation. In the Control group, no physical contact occurred. Post-task questionnaires were then administered.
Analysis involved comparing changes in scale scores between and within groups, plus text mining of recorded conversations to examine emotional word usage and conversational topic diversity.
Changes in Mood and Conversation
No significant differences between groups were found for verbal intelligence, personality traits, or baseline distress, though the Control group showed numerically higher initial distress scores, warranting caution in interpreting distress changes.
For mood states, no significant between-group differences emerged. However, within the Affective Touch group, total mood disturbance, anger, fatigue, and anxiety all decreased significantly after the robot task. The Control group showed a significant reduction only in anger.
Self-compassion increased significantly only in the Control group, not in the Affective Touch group. Distress decreased significantly in both groups following the interaction. Both groups showed significant reductions in negative attitudes toward robots and anxiety toward robots after the task. Rapport perceptions did not differ significantly between groups.
Text-mining analysis revealed that the Affective Touch group used significantly more emotion-related words than the Control group. Co-occurrence network diagrams indicated more diverse conversational topics in the Affective Touch group, with discussions spanning daily activities, academic concerns, and social interactions.
The Control group's conversation was more narrowly focused on academic aspects and general self-expression of concerns. Topic-modeling analysis corroborated these findings, showing three distinct thematic clusters in the Affective Touch group compared to two less differentiated topics in the Control group.
Interpreting the Mixed Effects of Robot Touch
Within-group analyses showed significant reductions in anger, fatigue, and anxiety in the Affective Touch group, though no between-group differences emerged, so these changes cannot be specifically attributed to touch.
Self-compassion increased only in the Control group, possibly because receiving comfort from the robot may have subtly affected self-esteem, though this requires further investigation.
Both groups showed reduced distress, suggesting a cathartic effect of disclosure itself. Similarly, both groups showed reduced negative attitudes and anxiety toward robots, indicating that familiarity through actual robot use may be beneficial regardless of touch.
Text mining revealed significantly more emotion-related words and broader conversational topics in the Affective Touch group, tentatively suggesting enhanced emotional engagement. However, major limitations exist, such as small sample size, underpowered analyses, no physiological indicators, no manipulation check, physical distance confound, baseline distress imbalance, and no pre-registration. These findings provide preliminary foundational knowledge for future clinical applications.
Cautious Steps Toward Robot-Assisted Therapy
This preliminary study suggests that affective touch from a humanoid robot may be associated with within-group reductions in negative moods and enhanced emotional expression during disclosure of concerns.
However, the absence of significant between-group differences, combined with important methodological limitations, means these findings cannot be confidently attributed to touch alone. The observed reductions in distress and negative attitudes across both groups indicate that simply engaging with a robot may contribute to psychological changes.
While these results provide foundational knowledge for future robot therapy applications, replication in larger, well-controlled studies with clinical populations is essential before drawing firm conclusions.
Journal Reference
Suzuki, N., and Yamamoto, T. (2026). The impact of affective touch from robots on mood states and emotional expression during disclosure of concerns. Scientific Reports. DOI:10.1038/s41598-026-67539-1.https://www.nature.com/articles/s41598-026-67539-1.
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.