Mechanosensory Coupling Between Human Cardiac Activity and Plant Bioelectric Potentials: A Naturalistic Self-Study with Spectral Validation

We report a single-subject proof-of-concept naturalistic longitudinal within-office study (n ≈ 93,000 one-Hz observations from a single participant across seven daytime sessions; n here refers to time-series samples, not biological replicates) testing whether human emotional states couple to the bioelectric potentials of a co-located Kalanchoe dai-gremontiana plant through measurable physical intermediaries. Using a custom IoT platform comprising two complementary AD8232 bioelectric plant sensors (Plant1: RC lowpass, 5-second averages; Plant2: 0.1–20 Hz bandpass at 1 Hz), a Sensirion SCD41 CO₂/temperature/humidity sensor, an SGP40 volatile organic compound (VOC) sensor, a Polar H10 heart rate monitor, and HSEmotion facial expression recognition, we recorded simultaneous human emotion (valence, arousal), physiology (heart rate HR, heart rate variability RMSSD), and environmental chemistry (CO₂, VOC index) at 1 Hz alongside plant bioelectric voltage. Lagged mediation analysis confirmed partial mediation of the valence→plant relationship through heart rate (Δa = 8 s for valence→HR, r = +0.067, p < 0.001; 30% partial mediation with Plant1). Critically, a novel spectral mediation analy-sis—using short-time Fourier transform (STFT) band powers of Plant2 as outcome vari-ables—reveals that the valence→HR coupling (Δa = 8 s) is invariant across all eight plant frequency bands, while the HR→plant coupling (path b) is frequency-specific: full sta-tistical mediation occurs only in the 44–75 second oscillation band (Δb = 34 s), with partial mediation across five other bands. This frequency specificity rules out broadband me-chanical coupling and points toward a frequency-selective biological transduction mechanism analogous to Venus flytrap mechanosensory integration. HRV-derived emotion (RMSSD-based valence, independent of the camera system) provides convergent validation. VOC index is the strongest cross-session predictor of emotional state in random forest models.