The Influence of Robot Facial Expression and Inclusion Behavior on Rapport and Trust in Older Adults: Mixed-Factorial Experimental Study.

BACKGROUND: The World Health Organization framework for healthy aging emphasizes that the capacity to establish and sustain relationships is a vital component of functional ability. Social robots offer valuable support for this relational aspect, although their effectiveness is contingent upon the quality of the interactions. While it is established that facial expressions and inclusive behaviors can influence rapport building, the combined effects of these elements on older adults remain unexamined. OBJECTIVE: This study aimed to investigate the main and interaction effects of robot facial expression (happy vs unhappy) and interaction style (inclusive vs exclusive) on the development of rapport between social robots and older adults residing in the community. METHODS: A 2 × 2 mixed-factorial experimental design was used in this study, involving 53 community-dwelling older adults (mean age 70.57, SD 4.27 years; n=45, 84.9% female). Participants engaged in a gamified interaction known as Radish Squat featuring 2 humanoid robots that exhibited various combinations of facial expressions and interaction styles. The inclusive robot designated participants within the game 75% of the time, whereas the exclusive robot did so approximately 25% of the time. Dependent measures included robot acceptance, rapport, rapport expectation, and trust. Behavioral designation frequencies were recorded throughout the gameplay. A mixed-design repeated-measure ANOVA was conducted to examine both main and interaction effects. RESULTS: Facial expression exhibited significant main effects on acceptance (F1,51=14.27; P<.001; η2=0.22), rapport (F1,51=6.69; P=.01; η2=0.12), conversation partner expectations (F1,51=11.34; P=.001; η2=0.18), togetherness expectations (F1,51=10.28; P=.002; η2=0.17), and trust (F1,51=4.36; P=.04; η2=0.08). Interaction style significantly affected rapport (F1,51=6.22; P=.02; η2=0.11), as well as both rapport expectation subscales. Significant interaction effects were also observed for rapport (F1,51=8.44; P=.005; η2=0.14) and trust (F1,51=16.62; P=.002; η2=0.25) such that the positive effect of happy facial expressions was substantially larger under exclusive interaction conditions. Older adults designated the exclusive robot within the game more frequently than the inclusive robot (mean 6.42, SD 1.84 vs 5.32, SD 2.12; t52=2.18; P=.03), and qualitative data pointed to the use of fairness-based turn-taking strategies. CONCLUSIONS: Positive facial expressions and inclusive interaction behaviors significantly enhanced rapport and trust, with synergistic interaction effects primarily emerging under exclusive interaction conditions. The observed patterns, whereby older adults compensated for exclusive robot behavior, suggest that they perceive robots as social entities deserving of equitable treatment. Collectively, these findings offer evidence-based guidelines for designing social robots that foster the relational dimensions of healthy aging.