A Dynamic Link between Respiration and Arousal

Figure 1.. A, Neuroanatomy of respiratory and pupillary rhythms. Respiratory pacemaker cells in the pre-Bötzinger complex (pBC) project to locus ceruleus (LC), which controls the global release of noradrenaline (NE), and to centromedial thalamus (CM). In a feedback loop, the incoming airstream of each breath triggers receptors reaching from the epithelium into the nasal cavity, propagating breathing-locked activity to the piriform cortex (PC) and mediodorsal thalamus (MD). The inset time series shows representative measurements of typical respiratory and pupil dynamics over the course of 60 s. The blue respiration trace shows the clear rhythmicity of breathing. A rhythmic component is not immediately visible in the red pupil trace. The raw unprocessed pupil trace, also depicted in orange, still contains the typical blinks (spikes) that have been interpolated for analysis (see Materials and Methods for details). B, Spectral composition (power spectra) of 5 min resting pupil (red) and respiration (blue) time series. The shaded area shows SEM. C, Magnitude-squared coherence spectrum between pupil and respiration time series. Observed coherence (purple) plotted against coherence computed on…

Figure 2.. Changes over time. A, Pupil power spectra for three consecutive 2 min time windows. The darker color indicates a later time window. The black straight line shows a frequency range with significant changes between time windows (p < 0.05, cluster-based permutation test). B, Individual power values per time window at the hippus peak frequency (0.19 Hz) in A. C, Respiration power spectra, otherwise same as A. D, Coupling (magnitude-squared coherence spectra) between pupil and respiration, otherwise same as A and C. No significant monotonic changes observed between time windows.

Figure 3.. A, Contrasted normal versus deep breathing conditions (dark vs light colors in A–C). A marked decrease in respiration frequency. The inset raincloud plot shows individual distributions of peak frequencies (in Hz) in both conditions. Offset along the y-axis noninformative and for visualization only. B, Same as in A but for pupillary dynamics. No significant differences in hippus peak frequencies between normal and deep breathing. C, Individual coupling peak frequencies (*, variance-corrected jackknife estimates) are markedly reduced during deep breathing. Straight lines below spectra signify systematic coherence differences when compared with surrogate data (data not shown). D, Contrasted rest versus task conditions (different sample than A–C). Respiration power spectrum during rest (dark blue) and task performance (light blue). The straight lines with asterisks at the bottom of D–F indicate frequency ranges with significant changes between both behavioral states (p < 0.05, pairwise t-statistic, cluster-based permutation test). E, Same for power spectra of pupil time series; dark red, rest; light red, task. F, Comparison of respiration–pupil coupling, measured as magnitude-squared…

Figure 4.. Differences between task blocks. A, Respiration power spectra across the six blocks including task performance. Light colors indicate later blocks. The shaded areas indicate SEM. A straight black line above the x-axis indicates a frequency cluster with a significant time-on-task effect (linear regression, p < 0.05, cluster-based permutation test). B, Same as A but for pupil data. C, Same as A and B but for respiration–pupil coupling (magnitude-squared coherence). No significant differences in coherence between blocks.

Figure 5.. Effect of pupil preprocessing. A, Pupil power spectra during rest before (black) and after (red) preprocessing. B, Coupling (magnitude-squared coherence) computed using raw, unprocessed pupil data. The purple trace shows the coupling observed in the data, and the gray trace is based on surrogate data. The dark straight lines show frequency ranges with significant differences observed versus surrogate (p < 0.05, cluster-based permutation test).