Disentangling Respiratory Phase‐Dependent and Phase‐Independent Components of Anticipatory Cardiac Deceleration

Figures

• 

  FIGURE 1. Experimental paradigm and the RSA modeling approach. (A) Visual threshold‐level stimulus detection task. The tiles represent the visual display presented to the participants during each trial. Duration ranges of each state in a trial are given below the corresponding tile. The contrast of the target stimulus is increased in this figure to improve its visibility to the reader. QR codes at the corners of the screen are April Tags used for the eye‐tracking software to recognize the screen and compute gaze locations within this surface. (B) Physiological recordings. ECG (top), respiration data (middle), and EEG (bottom row) were continuously recorded throughout each experimental block. During data analysis, R‐peaks were organized according to their timing relative to the target stimulus onset. We refer to the last R‐peak before the target onset as “R:stim”, and one R‐peak before that as “R:stim‐1”, and the rest of the annotation pattern follows intuitively. (C) Beat‐by‐beat, trial‐averaged RSA estimation. The respiratory phase and the change in the IBI centered around a chosen R‐peak (e.g., “R:stim‐1”) are placed on the x‐ and y‐axis, respectively. Each trial yields one such data…

• 

  FIGURE 2. The role of RSA in ACD. (A) Anticipatory Cardiac Deceleration (ACD) during stimulus anticipation (top), and group‐level respiratory phase‐locking at the time of each pre‐stimulus heartbeat (bottom). Each point on the circular plots represents the mean respiratory phase direction of a participant at the corresponding time point. (Error bars correspond to SEM, ***: p < 0.001 on all sub‐figures; the doughnut‐shaped grating represents the stimulus onset, which comes after the R‐peak named “stim”). (B) Illustrative plot for the effect of RSA on heart rate at the group‐level. The plot was constructed from resting‐state R‐peaks (bars indicate minimum and maximum data values, boxes correspond to the interquartile range, and the median value is denoted with a black horizontal line within each box). (C) Single‐participant examples of resting‐state RSA curves. Sinusoidal functions were fitted to the cluster of data points corresponding to each R‐peak during the 10‐min resting‐state recordings (|RSA|: RSA amplitude estimate). (D) Comparison of RSA estimation methods by linear correlation. Our sinusoidal model is compared to the established Porges‐Bohrer method. (E) Correlation between the…

• 

  FIGURE 3. Quantifying respiratory phase‐dependent and phase‐independent heart rate modulations during stimulus anticipation. (A) ACD in stimulus anticipation during exhalation and inhalation. Numbers on the x‐axis labels represent the timing of the R‐peak with respect to the stimulus onset (Error bars indicate SEM). “exh‐0” and “inh‐0” correspond to “R:stim”, the last R‐peak before stimulus onset, while “‐1” and “‐2” connotations correspond to 1 and 2 R‐peaks before the last pre‐stimulus R‐peak, respectively (***: main effect of Time on ΔIBI from R:stim‐2 to R:stim). (B) Modeling the effects of respiratory phase‐dependent and phase‐independent contributions to the ACD. The amplitude (parameter A) and vertical offset (parameter B) of the fitted sinusoidal RSA curves are postulated as the respiratory phase‐dependent and phase‐independent components of ACD, respectively. The plots are solely for illustrative purposes. (C) Parameters A (i.e., RSA; left) and B (right) during resting‐state and stimulus anticipation. Progress of the parameter values starting from the resting state (Rest) and throughout the stimulus anticipation phase (from “R:stim‐2” to “R:stim”; ***: p < 0.001, ns: p > 0.05,…

• 

  FIGURE 4. RSA and respiratory phase‐independent cardiac deceleration in Hit and Miss trials. (A) Total cardiac deceleration (n.s.: p > 0.05 for the main effect of Detection), (B) Estimated RSA amplitude during stimulus anticipation in Hit (successful detection) and Miss (false rejection) trials (n = 21: two participants had to be excluded due to outlier values; *: p < 0.05 for the main effect of Detection). (C) Progress of the vertical offset parameter within stimulus anticipation. The time point of “R:stim‐2” heartbeat was taken as reference. Error bars in A, B, and C correspond to within‐participant 95% confidence intervals (Cousineau 2005; Morey 2008). (D) Comparison of shifts in vertical offset (ΔB) from the beginning to the end of stimulus anticipation in Hit and Miss trials (left), and the average vertical offsets during stimulus anticipation (right). (Wilcoxon signed‐rank test, *: p < 0.05, ns: p > 0.05; error bars indicate the range of data points).

• 

  FIGURE 5. Across‐participant correlations between cardiac deceleration parameters and Visual Awareness Negativity (VAN). (A) Left: Average visual evoked potentials at the occipital electrodes (Oz, O1, O2) in response to the target stimulus onset. Gray rectangle denotes the time window of 188–318 ms (with respect to stimulus onset, dashed vertical lines) in which the difference between the ERP in Hit and Miss trials was significant. Middle: Average scalp topography of the significant time window (unit of heatmap is in microvolts). Right: Exploratory analysis of significant time window clusters for the across‐participant correlation between the full visual evoked potential time window (0–1 s with respect to the stimulus onset) and ACD amplitude in Hit trials. The thick blue line denotes the significant cluster of Spearman's correlation coefficients between 244 and 302 ms with respect to the stimulus onset (dashed vertical lines). (B) Across‐participant linear correlation between the average VAN amplitude and ACD for each participant. (C) Across‐participant correlations between VAN amplitudes and the respiratory phase‐independent (vertical offset shift, ΔB) and (D) phase‐dependent (RSA)…