PaperlayerView exact revision on arXiv

Source-linked AI summary

Designing Task-Induced Arousal: A Multimodal Stress Induction Method for Interactive Experiments

Morten Roed Frederiksen

arXiv:2609.26156v1cs.HCcs.RO

TL;DR

Interactive studies need arousal manipulations that remain compatible with ongoing device or robot interaction and sensing. This paper combines an escalating browser timing task with a concurrent hotwire challenge, and reports initial evidence of increased workload, arousal, GSR peak rate, and altered grip dynamics while valence remained above the scale midpoint.

  • Problem

    HCI and HRI studies need short, repeatable arousal manipulations that allow participants to continue interacting with devices or robots and producing sensor data.

    Source: §I · p. 1

  • Method

    The procedure combines a browser-based timing task with escalating demands and urgency cues and a concurrent physical hotwire challenge, evaluated in an A–B–A study using self-report, GSR, and grip dynamics.

    Source: §I · p. 1, G. Validation Criterion · p. 5

  • Results

    The induction increased workload and arousal, GSR peak rate, grip variability, and release speed, while valence remained above the scale midpoint.

    Source: §I · p. 2, D. Convergent Validation of the… · p. 6

  • Takeaways & Limitations

    The findings provide initial evidence for controlled, relatively high-valence task-related arousal during handheld interaction, with a reusable browser component for HCI and HRI studies.

    Source: §VI · p. 8

    The physical hotwire task increases setup complexity and may reduce cross-laboratory standardization because builds can vary in sensitivity and feedback intensity.

    Source: B. Scope and Limitations · pp. 7–8

Abstract

from arXiv · show

HCI and HRI studies often require short, repeatable arousal manipulations that can run while participants continue interacting with a device or robot. These experiments are often challenged by the need to induce arousal in settings that still resemble real interaction. Participants must continue using a device, touching a robot, or producing sensor data while the manipulation unfolds. We aimed to develop a compact and repeatable way to induce controlled task-related arousal during interactive experiments by combining a lightweight browser-based pacing task, escalating timing demands and urgency cues, and a concurrent physical hotwire-style challenge. In an A-B-A within-participant study, the induction condition significantly increased mental demand, temporal demand, effort, frustration, and SAM arousal (all p < .001), while perceived performance decreased (p < .001). GSR peak rate increased relative to both calm conditions (p = .030) and escalated over time (p < .001). Grip variability also increased (p = .036), as did release speed (both p < .001), while valence remained above the scale midpoint. These results provide initial evidence that the combined procedure induces controlled, relatively high-valence task-related arousal and may serve as a reusable experimental tool for future human-computer and human-robot interaction studies.

I. INTRODUCTION

Interactive studies need short, repeatable arousal manipulations compatible with ongoing interaction and sensing. The paper presents a browser-based timing task combined with a physical hotwire challenge and reports initial evidence of task-related arousal with valence remaining above the scale midpoint.

I. INTRODUCTION

  • Established stressors may be time-consuming, ethically demanding, socially evaluative, physically aversive, or tied to specific fears, unlike the constraints of many HCI and HRI studies.Interactive studies often require participants to continue touching a device, holding a robot, or producing behavioral data during the manipulation.Source: §I · p. 1
  • The method combines a standardized browser-based timing task with a concurrent physical hotwire challenge for short interactive studies.The browser controls timing, pacing, visual cues, and auditory escalation; the hotwire task adds embodied divided attention.Source: §I · p. 1
  • The induction increased mental demand, temporal demand, effort, frustration, and SAM arousal, while valence remained above the scale midpoint.GSR peak rate escalated across the two-minute window, consistent with the task’s escalating structure.Source: §I · pp. 1–2

II. RELATED WORK

  • Challenge-based arousal can increase workload and effort without strongly negative affect, a distinction relevant to interactive experiments.NASA-TLX and SAM together distinguish workload from valence and arousal.Source: §II · p. 2
  • The procedure is designed to run while participants continue interacting with a handheld device and producing continuous physiological and behavioral data.Its HRI relevance is experimental control and measurement during physical interaction, not a new autonomous robot capability.Source: §II · p. 2

A. Study Design

The study used an A–B–A within-participant design to test whether a multimodal induction increased task-related arousal relative to calm interaction periods. Participants used a handheld device that served as both interaction object and sensor platform.

A. Study Design

  • The A–B–A design used calm A1 and A2 periods around an arousal-induction B condition to examine whether responses returned toward baseline.Validation combined subjective workload and affect ratings, electrodermal activity, and grip-pressure dynamics.Source: A. Study Design · p. 2

B. Participants

  • Twenty-six people participated; 22 had complete A1, B, and A2 recordings, in a sample intended for an initial manipulation check rather than population-level estimates.Participants ranged from 10 to 44 years old and included children and university students.Source: B. Participants · pp. 2–3

C. Apparatus

  • The pocket-sized handheld tactile device combined two side-button pressure sensors with electrodes for GSR measurement and served as both interaction object and sensor platform.Participants held it in their dominant hand and pressed the side buttons in response to visual timing cues.Source: C. Apparatus · p. 3
  • The browser displayed the pacing task while responses were recorded through the handheld device, and sensor rows included timestamps, pressure, and GSR readings.The firmware intended to record approximately 20 Hz; the analyzed GSR was the device’s short-window average.Source: C. Apparatus · p. 3

D. Browser Task and Open-Source Design

The browser application provides calm and arousal-induction modes using the same basic timing task, with the latter adding escalating pace and urgency cues. Its inspectable, modifiable browser implementation is designed to pair with different interactive platforms.

D. Browser Task and Open-Source Design

  • The calm mode lasted 60 s and the stress/arousal mode 120 s, with the same basic visual timing task in both conditions.The B condition added time pressure, increasing pace, sensory escalation, and a concurrent physical task.Source: D. Browser Task and Open-Source Design · p. 3
  • In stress mode, the point’s speed ramped linearly to approximately 2.6 times baseline by the end of the condition.A visible countdown, timed auditory cues, flashing display, and target color shift added urgency during the two-minute procedure.Source: D. Browser Task and Open-Source Design · p. 3
  • The browser acts as a standardized pacing and induction environment, while the handheld device or other system handles interaction and measurement.The browser component can be paired with handheld robots, wearables, desktop interfaces, mobile devices, or other platforms.Source: D. Browser Task and Open-Source Design · p. 3
  • The inspectable browser implementation defines timing and cue parameters that researchers can reproduce or modify, including duration, speed ramp, auditory cues, and visual intensity.The system runs in a standard browser without specialized experimental software or a specific operating system or robot platform.Source: D. Browser Task and Open-Source Design · p. 3

E. Experimental Procedure and Conditions

Participants completed calm timing tasks before and after a two-minute induction that combined escalating demands with a concurrent hotwire challenge. The hotwire task repeatedly interrupted performance and was not completed successfully by any participant during B.

E. Experimental Procedure and Conditions

  • Participants completed 60-second calm timing tasks in A1 and A2, with A2 repeating the baseline task after induction.The same basic timing task was used in all three conditions.Source: E. Experimental Procedure and Conditions · p. 4
  • During the 120-second B condition, the moving point sped up over time while countdown, auditory, flashing, and color cues heightened urgency.These cues were designed to increase perceived urgency without social evaluation or physical discomfort.Source: E. Experimental Procedure and Conditions · p. 4
  • Participants divided attention between the timing task and a hotwire task guided with the non-dominant hand; touching the track triggered a buzzer and restart.The conductive loop navigated a track with approximately 6 mm clearance on each side.Source: E. Experimental Procedure and Conditions · p. 4
  • None of the participants completed the hotwire task successfully during the B condition.Source: E. Experimental Procedure and Conditions · p. 4

F. Validation Measures

The validation measures combine subjective workload and affect ratings with phasic electrodermal activity and grip-pressure dynamics. They assess activation and changing interaction behavior without treating absolute GSR level or mean grip pressure as the primary signal.

Subjective measures

  • NASA-TLX dimensions assess mental and temporal demand, perceived performance, effort, and frustration; lower performance scores indicate poorer perceived success.SAM arousal and valence assess activation and whether the experience is strongly negative or closer to challenge-related activation.Source: F. Validation Measures · p. 4

Physiological measures

  • Because active holding and pressing can affect absolute GSR level, physiological validation focuses on within-participant phasic peak structure and peak rate.Peak rate is detected using an adaptive prominence threshold and calculated as detected peaks per minute; electrodermal activity is the only autonomic measure available.Source: F. Validation Measures · p. 4

Behavioral measures

  • Grip validation emphasizes variability and release dynamics rather than mean pressure, testing whether gripping becomes more variable and releases faster.Grip variability is measured over time, while release dynamics are estimated from decreases in the combined pressure signal.Source: F. Validation Measures · p. 5

G. Validation Criterion

The induction was judged by converging evidence of workload and arousal, phasic electrodermal activity, and grip-dynamic change. Within-participant comparisons tested the induction against calm periods, with statistical significance evaluated at α = .05.

G. Validation Criterion

  • The B condition counted as successful if subjective workload and arousal, phasic electrodermal activity, and grip-pressure dynamics changed in converging directions.The criterion targets short-term task-related arousal during interaction, not purely fear-based, painful, or socially evaluative stress.Source: G. Validation Criterion · p. 5

H. Statistical Analysis

  • A1–B comparisons tested induction effects against initial calm, while B–A2 comparisons tested whether measures were also higher than postinduction calm.Comparisons used within-participant data.Source: H. Statistical Analysis · p. 5
  • Paired-samples t-tests were used for condition comparisons, with Cohen’s dz calculated from paired differences and significance evaluated at α = .05.Fig. 3 displays mean and individual GSR peak-rate trajectories across four 30-second windows of the induction.Source: H. Statistical Analysis · p. 5, Figure 3 · p. 5

IV. RESULTS

In the 22-participant complete-case dataset, the B condition increased subjective workload and arousal relative to calm conditions. Ratings indicate greater task demand and effort, with valence remaining above the scale midpoint during the induction.

IV. RESULTS

  • Analyses used complete recordings from 22 participants and were organized around subjective self-report, electrodermal activity, and grip-pressure dynamics.The central question was whether B produced a convergent increase in task-related arousal relative to calm baseline conditions.Source: §IV · p. 5

A. Subjective Validation

  • Mental demand, temporal demand, effort, and frustration increased significantly from A1 to B, while perceived performance decreased significantly.The performance decrease indicates participants experienced the task as more difficult to perform successfully.Source: A. Subjective Validation · p. 5
  • Temporal demand, effort, mental demand, and perceived performance showed the largest effects; frustration increased less, indicating workload and effort rose more than irritation alone.The pattern matches the procedure’s intended time pressure, cognitive demand, divided attention, and repeated performance disruption.Source: A. Subjective Validation · p. 5
  • SAM arousal increased during B and fell again in A2; valence decreased but stayed above the scale midpoint, then increased again in A2.The pattern is consistent with temporary activation and workload accompanied by a moderate reduction in pleasantness rather than a strongly negative affective state.Source: A. Subjective Validation · p. 6

B. Physiological Validation

Physiological validation found more frequent phasic GSR responses during induction, with peak rate rising as the task escalated. Individual peak amplitude did not significantly differ across conditions.

B. Physiological Validation

  • Relative GSR peak amplitude did not differ significantly across conditions, and none of the paired comparisons was significant.Mean peak amplitude was similar in A1, B, and A2.Source: B. Physiological Validation · p. 6
  • GSR peak rate was significantly higher in B than in both A1 and A2, while the calm conditions did not differ.The change reflects more frequent phasic responses rather than larger individual peaks.Source: B. Physiological Validation · p. 6
  • GSR peak rate increased across the four 30-second windows, with participant-level slopes significantly above zero and higher rates in the second than the first minute.The response increased as the task became faster, more urgent, and more difficult, rather than appearing only at induction onset.Source: B. Physiological Validation · p. 6

C. Behavioral Grip Validation

During the induction, grip behavior changed through greater variability and faster release rather than greater mean pressure. Across measures, subjective, physiological, and behavioral responses converged; exploratory associations linked perceived demand with GSR escalation and release speed.

C. Behavioral Grip Validation

  • Grip variability and release speed were higher during B than in both calm conditions, while mean grip pressure did not differ significantly.The calm conditions did not differ on variability or release speed; the behavioral signature was less stable, faster grip-release dynamics rather than stronger pressing.Source: C. Behavioral Grip Validation · p. 6

D. Convergent Validation of the Induction Method

  • During B, workload and arousal ratings increased, valence remained above the midpoint, GSR peak rate escalated, and grip variability and release speed increased without a change in mean grip pressure.The authors characterize this as convergent subjective, physiological, and behavioral validation, while cautioning that grip measures reflect this handheld device.Source: D. Convergent Validation of the… · p. 6, Figure 4 · p. 7

E. Exploratory Cross-Measure Associations

  • ρ = .54, p = .009: a larger A1-to-B increase in mental demand was associated with a steeper GSR peak-rate increase across B's four 30-second windows.The authors describe this as exploratory support for the task's escalating structure being reflected in both reported workload and physiological arousal.Source: E. Exploratory Cross-Measure… · p. 6
  • ρ = .50, p = .017: the exploratory workload-change composite was positively associated with release speed during B.The composite averaged changes in mental demand, temporal demand, effort, frustration, and reverse-scored perceived performance; it was not a validated NASA-TLX total score.Source: E. Exploratory Cross-Measure… · pp. 6–7

V. DISCUSSION

The procedure is a methodological tool for increasing workload, urgency, and physiological activation during ongoing interaction, not a broad stress induction or new robot capability. Its response pattern is best characterized as high-arousal challenge rather than low-valence distress.

V. DISCUSSION

  • The study offers an initial procedure for increasing workload, urgency, and physiological activation during physical interaction with a robotic device, rather than a new robot capability.The evaluation used a small handheld tactile robot; transfer to other robotic embodiments and interaction contexts remains to be established.Source: §V · p. 7

A. Characterizing the Induced State

  • Higher workload and arousal accompanied valence that remained above the midpoint, making the response pattern more consistent with challenge-based arousal than distress.The authors caution against using the paradigm to validate systems that claim to detect anxiety, fear, or distress.Source: A. Characterizing the Induced State · p. 7

B. Scope and Limitations

The procedure has initial support for handheld interaction, but physical implementation, grip measures, sample size, and physiological coverage constrain how broadly the findings can be interpreted. The conclusion calls for larger studies across devices and robotic embodiments before broad HRI validation.

B. Scope and Limitations

  • Grip results are device- and sample-specific behavioral indicators, not direct measures of stress.Hand size, finger placement, dominant-hand characteristics, baseline grip strength, age, motor control, and device fit were not systematically measured or normalized.Source: B. Scope and Limitations · p. 8
  • The complete-case sample of 22 limits precision and generalizability, while the study did not record heart rate or heart-rate variability.The authors recommend larger, more diverse evaluations and broader autonomic measures, including HR or HRV.Source: B. Scope and Limitations · p. 8

VI. CONCLUSION

  • The browser component is reusable, but replicating the complete procedure requires a comparable physical divided-attention task.Physical builds may vary in sensitivity and feedback intensity, increasing setup complexity and potentially reducing cross-laboratory standardization; an on-screen version is proposed.Source: B. Scope and Limitations · pp. 7–8, §VI · p. 8
  • The conclusion treats the findings as initial evidence and calls for larger studies across robotic embodiments and diverse interaction devices before broad HRI validation.It also recommends adding autonomic measures such as HR or HRV.Source: §VI · p. 8
Loading 2609.26156v1…