The analysis of transient RR interval correlations can track consistency of autonomic response to a shared stimulus or activity. In this preliminary study, we examine a case of music performance and its impact on three players and a listener. We evaluate the usability of time delay stability (TDS) for analysing mutual changes in RR intervals and loudness and tempo while playing and listening to music.
Three musicians performed Schubert Trio Op. 100, Andante con moto, nine times on five different days. Electrocardiographic traces and extracted RR intervals were collected from each player and one listener (36 series total) during baseline (5 min) and each performance (~10 min). Loudness and tempo were extracted from recorded music audio.
TDS obtains the optimal time delay τ0v that maximises absolute cross-correlation between two signals. Stability is defined as optimal delay changing by ≤1 sec in ≥4 in 5 consecutive windows. The average percentage of TDS segments (%TDS) during music were compared with that of shuffled surrogate data (30-sec window, 15-sec hop size). We computed probabilities of observed stability; for score-aligned data, we tracked the number of performances where TDS was observed.
We observed significant difference between %TDS in original and surrogate data for the pianist-cellist (0.174, 95%CI=[0.104,0.243] vs 0.111, [0.086,0.136], p=.036) and violinist-cellist (0.232, [0.185,0.278] vs 0.097, [0.058,0.136], p=.001) pairs. Difference between baseline and music was observed for the pianist-cellist pair (0.174, [0.104,0.243] vs 0.085, [0.063,0.107], p=.016). In 8 out of 9 performances, we observed pianist-violinist and violinist-cellist TDS in the recovery period after the most strenuous sections. Thus, TDS offers a promising way to track dynamic RR interval interactions between people engaged in a shared activity.