Organizations: Tsinghua University Beijing, China · University of Pennsylvania Philadelphia, USA · Syracuse University Syracuse, USA · National University of Singapore Singapore, Singapore · Royal College of Art London, United Kingdom
As social robots take on increasingly complex roles like game masters (GMs) in multi-party games, the expectation that physicality universally enhances user experience remains debated. This study challenges the "one-size-fits-all" view of tangible interaction by identifying a critical boundary condition: users' Negative Attitudes towards Robots (NARS). In a between-subjects experiment (N = 67), a custom-built robot GM facilitated a multi-party murder mystery game (MMG) by delivering clues either through direct tangible interaction or a digitally mediated interface. Baseline multivariate analysis (MANOVA) showed no significant main effect of delivery modality, confirming that tangibility alone does not guarantee superior engagement. However, primary analysis using multilevel linear models (MLM) revealed a reliable moderation: participants high in NARS experienced markedly lower narrative immersion under tangible delivery, whereas those with low NARS scores showed no such decrement. Qualitative findings further illuminate this divergence: tangibility provides novelty and engagement for some but imposes excessive proxemic friction for anxious users, for whom the digital interface acts as a protective social buffer. These results advance a conditional model of HRI and emphasize the necessity for adaptive systems that can tailor interaction modalities to user predispositions.
Figures & tables
Figure 1. A gameplay session with POIROT. A picture showing the experimental setup. A group of participants sits around a rectangular dining table in a living-room-style laboratory environment. The custom-built robot, POIROT, stands on the table surface at the head of the table, facing the participants who are engaged in the game.
b
SE
95% CI
p
Narrative Immersion
Intercept
1.72
1.19
[-0.61, 4.06]
0.147
Condition (App=0, Cards=1)
3.65
1.34
[1.01, 6.28]
0.007
NARS
1.03
0.38
[0.28, 1.78]
0.007
Condition × NARS
-1.27
0.43
[-2.10, -0.44]
0.003
ICC
0.130
Table 1. Multilevel Models Predicting Narrative Immersion, Game Experience, and Social Presence as a Function of Condition, NARS, and Their Interaction. Random Intercepts were Specified for Game Sessions.
Outcome
NARS level
b
SE
z
punc
pcorr
Narrative Immersion
Low ( −1 SD)
0.66
0.44
1.49
0.138
0.275
High ( +1 SD)
-1.24
0.46
-2.72
0.007
0.014
Game Experience
Low ( −1 SD)
0.52
0.45
1.14
0.254
0.508
High ( +1 SD)
-0.80
0.47
-1.72
0.085
0.170
Table 2. Simple Slopes of Condition (Physical-Card vs. In-App) Predicting Outcomes at Low ( −1 SD) and High ( +1 SD) Levels of NARS. P-values are Reported as Uncorrected ( punc ) and Bonferroni-Corrected ( pcorr ).
Figure 2. The interaction effect of experimental condition (Physical-Card vs. In-App) and participants’ Negative Attitudes towards Robots Scale (NARS) on: (a) Narrative Immersion, (b) Game Experience. Shaded areas represent 95% confidence intervals.
Appendix figures & tables11 assets
Supplementary material from the paper’s appendix.
Appendix
Role
Identity
Core Motivation
Fang Qi
Defendant
Maintains innocence
Tang Yali
Victim’s daughter
Justice for her father
Ye Caiyang
Prosecutor
Legal conviction
Yin Hong
Defence atty.
Overturn guilty verdict
Chu Mihe
Nun / Witness
Reveal complete truth
Chen Haotong
Police officer
Support prosecution
Appendix
Table 3. Six-Character Roster for Judgment Quartet
Table 4. Sample private clues provided to each role
Figure 3. Companion app setup inter face in sequence: (a) room creation interface, (b) room joining interface, (c) role selection interface, and (d) example of a script reading interface (see Section 3.2.3 for details). Note: In the Physical-Card condition, the "Clues" button was also present in the script reading interface but inactive, since all clues were provided via physical delivery. Four sequential screenshots of the mobile app interface. (a) shows a "Create Game" screen with a green button and a Room ID. (b) shows a "Join Game" screen with an input field. (c) displays a list of character names with radio buttons for selection. (d) shows a sample of detailed script page (featuring a photo of a male character in a suit named Ye Caiyang.)
Figure 4. Examples of physical clue cards (6.3 cm × 8.8 cm, laminated): (a) front side showing clue content, and (b) reverse side with neutral labeling. Private clues were dispensed individually to players, while public cards (eight per session) were dispensed at the table center before script reading started. A picture showing the front and back of a sample laminated clue card. The left image (front) contains text labeled "Exclusive Clue for the Police – Clue 3" followed by narrative text about a vehicle. The right image (back) shows the same label on a plain gray background without the narrative text.
Figure 5. Digital clues panel: (a) overview of multiple digital clues, and (b) detail view of a single clue, opened by tapping the corresponding item from (a). Two screenshots of the digital clue interfaces on a tablet. The left image shows a scrollable gallery view with two columns of text blocks representing different testimonies. The right image shows a detailed view where a single clue titled "Witness Testimony 4" is displayed in large text.
Figure 6. The POIROT social robot used in the study, shown in its default idle posture. To evoke the style of Agatha Christie’s detective Poirot, a symbolic mustache decal was added beneath its display screen. A front-facing picture of POIROT, which is a white, 3D-printed social robot. It has a smooth, rounded shape with small flipper arms. Its face is a black iPhone screen displaying two pixelated eyes. To evoke the style of Agatha Christie’s detective Poirot, a symbolic mustache decal was added beneath its display screen.
Figure 7. The custom-built clue dispensing module: (left) physical implementation with friction roller above laminated card stack, and (right) electrical diagram showing USB-powered GA12-N20 motor, friction roller, and inline SPST toggle switch. No MCU or feedback mechanism was used. A composite image showing the technical design of the dispenser. The left side shows a 3D physical implementation of the clue cards dispenser containing a stack of cards and a friction roller mechanism. The right side is a schematic diagram illustrating the circuit, connecting a USB power source, a push switch, and a motor to the friction roller.
Component
Specification
Qty
Motor
GA12-N20, 6V, 50rpm DC motor
1
Friction Roller
3D-printed + rubber O-rings
1
Card Housing
PLA, 3D-printed structure
1
Switch
SPST toggle switch
1
Power Supply
USB-A 5V (adapter or power bank)
1
Wiring
2-pin cable
1
Appendix
Table 5. Bill of Materials for Clue Cards Dispenser Module
Designing effective Human-Robot Interaction in task-oriented settings requires carefully balancing user engagement with socially acceptable levels of robot intrusiveness. In this paper, we examine how different robot intervention strategies shape user experience, interaction dynamics, perceived intrusiveness, and sense of support. We compare two approaches: a continuous engagement-seeking robot strategy, and a context-aware strategy that selectively intervenes based on the user's state and task context. Both approaches rely on multimodal behavioral cues, including body orientation and attentional signals, to guide robot actions. We evaluate these strategies in a user study with 32 participants performing a task in a simulated hospital environment. Our findings show that higher interaction frequency does not necessarily lead to better engagement. Instead, we observe a systematic trade-off between perceived support and intrusiveness, influenced by factors such as physical proximity and user effort. These results provide empirical evidence that effective engagement in HRI depends on adaptive, context-sensitive intervention policies.
Lavinia Hriscu, Valerio Bo, Alberto Sanfeliu +1
Institut de Robòtica i Informàtica Industrial (CSIC-UPC) · Universitat Politècnica de Catalunya (UPC), Barcelona, 08034, Spain.
Robot assistance is particularly crucial in unfamiliar tasks, where users must understand task requirements while coordinating with the robot. Previous research offers mixed evidence on the role of robot proxemics in user engagement: some studies suggest closer proximity enhances interaction, while others report it can feel intrusive. In this work, we argue that perceptions of intrusiveness depend not only on proxemics but also on the frequency of robot interventions, and are strongly influenced by individual traits such as personality and demographics. We conducted an experiment with 32 participants who interacted with two assistive robots that provided similar task support but differed in their intervention strategies. Results indicate that overall engagement remains stable across conditions, yet affective responses and perceived intrusiveness vary significantly with personality traits. Moreover, personality shapes interaction dynamics differently depending on the robot's behavior. These findings emphasize that effective human-robot interaction should account for individual differences, tailoring robot behavior to maintain engagement while respecting each user's unique affective and behavioral profile.
Valerio Bo, Lavinia Hriscu, Alberto Sanfeliu +1
Institut de Robòtica i Informàtica Industrial (CSIC-UPC) · Universitat Politècnica de Catalunya (UPC), Barcelona, 08034, Spain.
This work compares three interaction modalities for human-robot collaboration: passive, reactive, and proactive. We studied 18 participants assembling a seven-layer colored tower from memory while using nearby and distant blocks. In the passive modality participants worked alone; in the reactive modality a mobile robot helped only upon request; in the proactive modality it initiated brick delivery and error signaling without explicit requests. Although robot assistance increased completion time, most participants preferred collaboration: 67% preferred proactive behavior and 78% judged it most useful. These results suggest that timely proactive support can improve user experience in controlled collaborative tasks.
Dalle Molle Institute for Artificial Intelligence (IDSIA), USI-SUPSI, Lugano, Switzerland · Department of Surgery, Medicine, Dentistry and Morphological Sciences with interest in Transplant, Oncology and Regenerative Medicine, University of Modena and Reggio Emilia, Modena, Italy. · Department of Sciences and Methods for Engineering (DISMI), University of Modena and Reggio Emilia, Reggio Emilia, Italy.