Video-Derived Pulse-to-Pulse Interval Variability and Acoustic Correlates During the Chevreul Pendulum Task: A Multi-Modal Pilot Study with Sham Control

Video-Derived Pulse-to-Pulse Interval Variability and Acoustic Correlates During the Chevreul Pendulum Task: A Multi-Modal Pilot Study with Sham Control Abstract Background: The Chevreul pendulum illusion—the apparent spontaneous motion of a hand-held pendulum—is classically attributed to ideomotor phenomena, wherein unconscious micro-movements generate observable motion. While electromyographic studies have confirmed muscular activity during this illusion, the associated autonomic and acoustic correlates remain underexplored. Objective: To investigate whether the Chevreul pendulum task elicits measurable changes in video-derived pulse-to-pulse interval (PPI) variability and concurrent acoustic parameters, and to compare these changes against sham conditions. Methods: A single participant completed five experimental sessions. Each session comprised a 30-second baseline (resting state) followed by a 30-second exposure period (active pendulum observation with circular/elliptical motion). Video-PPG was processed at 30 frames per second using a whole-frame averaging approach to derive RMSSD, SDNN, and heart rate. Voice recordings were obtained during baseline, exposure, and two sham conditions (pendulum held stationary), extracting fundamental frequency (F0), shimmer, and harmonicity parameters. Results: Across three valid paired sessions, RMSSD consistently decreased during exposure compared to baseline (mean reduction: −11.7%, range: −6.3% to −18.3%), while heart rate remained stable (mean change: +1.3%). Voice analysis revealed a mean F0 increase of +19.8% during active pendulum exposure (199.7 Hz to 239.2 Hz), compared to +6.8% during sham conditions (199.7 Hz to 213.2 Hz). Shimmer and harmonicity parameters showed inconsistent changes across conditions. Conclusion: These preliminary findings suggest a consistent pattern of reduced video-derived PPI variability alongside increased fundamental frequency during the Chevreul pendulum task, with changes exceeding those observed in sham conditions. This pattern is consistent with a state of focused attention and motor preparation. While methodological limitations (single participant, 30 FPS temporal resolution, whole-frame signal extraction) preclude definitive conclusions regarding autonomic mechanisms, these multi-modal measurements provide a proof-of-concept for non-invasive biomarkers of ideomotor engagement. Replication with larger samples, synchronized ECG validation, and controlled illumination is warranted. Keywords: Chevreul pendulum; ideomotor effect; video-PPG; pulse-to-pulse interval variability; fundamental frequency; pilot study; sham control 1. Introduction For nearly two centuries, the Chevreul pendulum illusion has captivated scientists and laypersons alike. First described by the French chemist Michel Eugène Chevreul in the 1830s, this phenomenon occurs when a person holds a pendulum stationary while focusing on a specific point or concept; seemingly without conscious volition, the pendulum begins to oscillate in patterns that often appear meaningful (Chevreul, 1854). While historically attributed to supernatural forces or "animal magnetism," contemporary research has firmly established the ideomotor theory as the primary explanatory framework (Carpenter, 1852; James, 1890). The ideomotor principle posits that mental representations of actions can evoke corresponding motor outputs without conscious awareness of the intervening muscular activity (Shin, Proctor, & Capaldi, 2010). In the context of the Chevreul pendulum, subtle, unconscious micro-movements of the hand and fingers—driven by the participant's focused attention and expectation—generate sufficient force to initiate pendulum oscillation. Recent motion-capture studies have confirmed that successful illusion production requires participants to generate driving movements at frequencies approximating the pendulum's resonant frequency, with finger movement amplitude and pendulum length being critical determinants of success (Cantergi, Awasthi, & Friedman, 2021). Despite this mechanistic understanding of the motor component, the associated physiological correlates of the ideomotor state remain largely unexplored. The act of sustaining focused visual attention while unconsciously generating micro-movements represents a unique neurophysiological condition that may impose specific demands on the autonomic nervous system (ANS). Heart rate variability (HRV)—particularly the root mean square of successive differences (RMSSD)—provides a non-invasive window into cardiac parasympathetic modulation (Task Force, 1996). Selective reductions in RMSSD, in the absence of concurrent heart rate increases, have been associated with states of focused attention and motor preparation (Thayer, Åhs, Fredrikson, Sollers, & Wager, 2012), suggesting that the ideomotor phenomenon might be accompanied by characteristic changes in beat-to-beat interval variability. Furthermore, the psychophysiological state induced by focused attention and motor preparation may also affect laryngeal muscle tension, which is reflected in acoustic parameters such as fundamental frequency (F0) (Van Puyvelde, Vanfleteren, & Deschacht, 2018). Increased sympathetic activation and muscle tension typically elevate vocal pitch, providing a potential secondary marker of the ideomotor state. Concurrently, advances in remote photoplethysmography (video-PPG) have opened new possibilities for unobtrusive cardiovascular monitoring. By extracting subtle color variations in facial or cutaneous regions from standard video recordings, researchers can derive pulse-to-pulse intervals (PPI) without physical sensors (Verkruysse, Svaasand, & Nelson, 2008; McDuff, Gontarek, & Picard, 2014). While video-PPG faces methodological challenges—including temporal quantization effects at standard frame rates and sensitivity to illumination artifacts—recent developments in signal processing have improved its feasibility for exploratory research. This pilot study represents, to our knowledge, the first investigation of video-derived PPI variability and acoustic responses during the Chevreul pendulum illusion. We hypothesized that active engagement with the pendulum would elicit: (1) a measurable reduction in video-derived RMSSD reflecting increased attentional-motor coupling, (2) stable heart rate indicating selective rather than generalized autonomic activation, and (3) increased fundamental frequency reflecting altered vocal/laryngeal state. Given the exploratory nature of this work and the methodological constraints of video-PPG at 30 FPS, we present these findings as hypothesis-generating rather than confirmatory, with the aim of informing larger, preregistered studies with synchronized ECG validation. 2. Methods 2.1. Participant and Experimental Protocol Data were collected from a single healthy adult participant (male, 28 years) across five experimental sessions conducted on different days as part of a pilot feasibility study. Each session comprised a 30-second baseline period (resting state, eyes open, pendulum held stationary) followed by a 30-second exposure period, during which the participant actively observed and mentally attended to a hand-held Chevreul pendulum undergoing circular or elliptical motion. The pendulum was held above the open palm with the pointed end facing upward, and its length was gradually increased until spontaneous oscillation began at a specific length (approximately 30-40 cm). In two sessions, sham conditions were included where the pendulum was held stationary without induced motion, serving as control conditions to isolate ideomotor-specific effects from mere time-on-task or general attention effects. One baseline recording (Session 2) was contaminated by severe motion artifacts and was excluded from the primary paired analysis, leaving three complete valid baseline-exposure pairs for the core statistical evaluation. An additional exposure recording (Session 5) was retained for descriptive comparison despite its baseline being excluded. The experimental protocol received ethical approval from the Institutional Research Board, and the participant provided informed consent before starting the experiment. All methods were carried out in accordance with relevant guidelines and regulations. 2.2. Video Acquisition and Signal Processing Facial video was recorded at 30 frames per second (FPS) using a standard webcam under ambient lighting conditions. Due to the absence of precise skin-region isolation in this exploratory phase, a whole-frame averaging approach was employed, extracting the raw photoplethysmographic (PPG) signal from the entire frame average green channel intensity. While this approach reduces spatial specificity and increases susceptibility to motion artifacts, it allowed for consistent signal extraction across sessions despite variable lighting conditions. Pulse-to-pulse intervals (PPI) were derived from the green channel using an adaptive peak-detection algorithm with the following parameters: minimum peak distance = 0.30 seconds (corresponding to 200 BPM maximum), prominence threshold = 0.15 × signal standard deviation, and confidence guard frames = 0.50 seconds around each detected peak. Signal quality was assessed via a custom Signal Quality Index (SQI) incorporating periodicity, harmonic ratio, peak sharpness, and beat stability metrics. 2.3. Voice Recording and Acoustic Analysis Voice recordings were obtained using a standard microphone at 44.1 kHz sampling rate. The participant produced a sustained /a/ vowel for 3-5 seconds during each condition (baseline, exposure, and sham). Acoustic analysis was performed using open-source software (Praat; Boersma & Weenink, 2023) to extract the following parameters: · Fundamental frequency (F0): The average pitch in Hz, reflecting vocal fold vibration rate. · Local shimmer: Cycle-to-cycle amplitude variation (%), indicating vocal fold vibratio