A Dynamic State-Based Framework for Quantifying Triadic Autonomic Synchrony During Group Decision-Making

16/10/2026

AI Summary

State-based framework quantifies time-varying triadic autonomic synchrony using EDA and HRV, PCA reduction, and k-means clustering into five discrete synchrony states.
Empirical triads showed a graded synchrony continuum with bias towards the highest synchrony state; occupancy and stability exceeded chance by surrogate IAAFT (p < 0.001).
State 0 showed lower dynamic time warping distances and higher lag-optimised physiological linkage for both EDA and HRV; anxiety measures did not predict HSS prevalence.

TL;DR (AI Generated)

We propose a state-based framework to quantify time-varying interpersonal synchrony during a laboratory group decision-making task (Desert Survival Task) using concurrently recorded electrodermal activity (EDA) and ECG-derived heart rate variability (HRV).

Convergent validity of the proposed state representation was further supported by continuous signal-based benchmarks: compared with State 4, State 0 showed lower dynamic time warping distances and higher lag-optimized physiological linkage (max R2) for both EDA and HRV (all p < 0.001).

These findings suggest that discrete synchrony states capture interpretable, non-random coordination regimes in triads and provide a complementary framework for characterizing the temporal organization of autonomic synchrony beyond conventional continuous synchrony measures.

Basic summary

Psychophysiology. 2026 Oct;63(10):e70427. doi: 10.1111/psyp.70427.

ABSTRACT

In triadic social interaction, autonomic coordination is dynamic and may be poorly captured by session-level synchrony estimates. We propose a state-based framework to quantify time-varying interpersonal synchrony during a laboratory group decision-making task (Desert Survival Task) using concurrently recorded electrodermal activity (EDA) and ECG-derived heart rate variability (HRV). Twelve triads (36 participants) with complete physiological recordings were analyzed. Signals were segmented into overlapping 60 s windows (1 s step). For each participant, window-level features were extracted (EDA: 4 features; HRV: 74 features), standardized within individuals, and converted to dyadic divergence vectors (absolute differences) for the three dyads in each triad. Divergence vectors (combined EDA plus HRV) were reduced with PCA (95% variance retained) and clustered using a single global k-means model trained on pooled windows to yield five discrete synchrony states shared across triads. States were ordered along the first principal component (PC1) of the divergence space, with its orientation chosen so that State 0 corresponded to the highest-synchrony state (HSS). Triad-level state metrics quantified both prevalence and temporal stability: fractional occupancy rate (FOR), mean dwell time (MDT), and conditional long-run fraction (CLRF; ≥ 10 s runs). Empirical triads showed a graded continuum of synchrony across states and a strong bias toward the HSS. Surrogate validation using a leave-one-fixed Iterative Amplitude Adjusted Fourier Transform (IAAFT) null applied at the signal level demonstrated that HSS occupancy and stability exceeded chance alignment (FOR; p < 0.001; MDT and CLRF; p < 0.001). Convergent validity of the proposed state representation was further supported by continuous signal-based benchmarks: compared with State 4, State 0 showed lower dynamic time warping distances and higher lag-optimized physiological linkage (max R²) for both EDA and HRV (all p < 0.001). In contrast, group-mean anxiety measures (SPIN, STAI) were not reliably associated with HSS prevalence or stability in this sample. These findings suggest that discrete synchrony states capture interpretable, non-random coordination regimes in triads and provide a complementary framework for characterizing the temporal organization of autonomic synchrony beyond conventional continuous synchrony measures.

PMID:42885436 | DOI:10.1111/psyp.70427

Document this CPD