The Frequency and State-Dependent Relationship between Pupil Size and Respiration

Arousal reflects our level of alertness and impacts our readiness to engage with our environment. When arousal is low, such as when we are drowsy, we are less likely to detect a knock at the door compared with when arousal is higher, such as after drinking coffee. In the literature, arousal is a broad term used to describe cognitive states that influence perception and behavior, bodily states that are associated with affective experience, and feedback loops that connect these processes. In the cognitive domain, arousal level has been heavily linked with the norepinephrine-producing locus ceruleus (LC) in the brainstem (Aston-Jones et al., 1999). Measuring neural activity directly from the LC to estimate arousal level is technically challenging, however, and it is not always practical. Instead, systemic physiology is often used as a marker of arousal since physiological processes have been found to covary with arousal and in some cases have been specifically linked with LC activity.

Although the pupil's main role is to regulate the amount of light that enters the eye, pupil diameter is a common surrogate of arousal because of the relationship between the pupil and LC. An increase in LC activity leads to pupil dilation by inhibiting the Edinger–Westphal nucleus (EWN), which provides parasympathetic inputs to the pupil, and by exciting the superior cervical ganglion (SCG), which provides sympathetic inputs to the pupil (Joshi et al., 2016; Liu et al., 2017). As a result, a larger pupil is interpreted as a sign of elevated LC-linked arousal. Periodic fluctuations in pupil diameter have also emerged as a potential indicator of arousal state. These fluctuations, called pupillary hippus, are most prominent at rest, so their presence is considered an indicator of low arousal level. The exact frequency of hippus is variable across lighting conditions, experimental contexts, and species (ranging from 0.2 to 3 Hz), which has made understanding the significance of this pupil oscillation challenging (Loewenfeld, 1993). The circuitry giving rise to pupillary hippus is also poorly understood. One study found hippus depends on parasympathetic inputs to the pupil (Turnbull et al., 2017); but whether these oscillations are also driven by LC is unknown.

An obstacle to using pupil size as a marker of arousal is that changes can also be driven by processes unrelated to arousal, as is the case of light-evoked pupil responses. A study in rodents found that activity in LC only predicted a minority of pupil fluctuations and the pupil response to the same LC stimulation in different experimental sessions varied (Megemont et al., 2022). This emphasizes that pupil size alone may be a muddled representation of LC-linked arousal processes because LC is only one of many possible modulators of pupil size. Identifying other physiology measures that are influenced by arousal and cofluctuate with pupil size could aid in isolating changes related to arousal.

Similar to pupil diameter, respiration has systemic physiological functions but also has ties to arousal. The main purpose of respiration is to ensure adequate intake of oxygen and elimination of carbon dioxide. However, the circuits that influence the automatic as well as conscious control of respiration are intricately tied to arousal circuitry in the brain and body (Krohn et al., 2023). Furthermore, respiration is known to change as a function of arousal level. For example, respiration rate increases when arousal level is high during stressful conditions (Dias et al., 2025). Intriguingly, resting respiration rate in humans is typically ∼0.2 Hz (Kluger et al., 2024), and this frequency is similar to that of pupillary hippus. This leads to a compelling hypothesis that pupil size and respiration are coupled at this frequency.

There is, however, mixed evidence in the literature on the direction and strength of the relationship between pupil size and respiration (Schaefer et al., 2023). One challenge in comparing across studies is the diversity of experimental conditions and physiological measurements that could influence the findings. A study recently published in The Journal of Neuroscience aimed to further test the relationship between pupil dynamics and respiration by studying how these two measures vary under different conditions (Kluger et al., 2024).

Kluger et al. (2024) expected the pupil–respiration relationship to be strongest near the hippus frequency because this frequency is similar to the respiration rate. They measured pupil diameter and respiration simultaneously in human participants during two different states of arousal: periods of rest, when arousal level was low, and while subjects performed a visual detection task, when arousal level was higher. To quantify the pupil–respiration relationship, the authors calculated oscillatory phase coherence, a method that can be used to estimate the extent to which two signals are coupled at different frequencies.

Kluger and colleagues found that pupil diameter and respiration were significantly coupled at frequencies near the hippus frequency (between 0.17 and 0.37 Hz). They also found the frequency at which pupil size and respiration were most coupled could be modulated. When participants were asked to take deeper, slower breaths, as is often the case when arousal is low, the coupling frequency decreased. These findings suggested the pupil–respiration relationship was frequency and state dependent.

Next, the authors compared the pupil–respiration relationship between resting and task-engaged states. Kluger et al. found that, consistent with prior work, the power of pupil fluctuations at the hippus frequency decreased when subjects performed a task. This reduced power may have occurred because other inputs to the pupil, such as those that control processes associated with changing pupil size to focus on different visual stimuli, have a more dominant influence during a visual task. Yet despite the reduction in hippus power while performing a task, pupil size and respiration were still significantly coupled near the hippus frequency. But the strength of this coupling was decreased during the task compared with rest. The authors suggested decreased coupling during the task may have arisen because the weaker hippus fluctuations led to a poor estimate of coupling at that frequency. An alternative explanation is that arousal, which is higher during task engagement compared with rest, influences the strength of the pupil–respiration relationship. In future work, subjects could perform a nonvisual task, where the pupil would not be influenced by visual input, to provide clarity on the effect of task engagement on the pupil–respiration relationship.

In addition to these findings of frequency and state-dependent coupling, the authors also found evidence that changes in respiration preceded pupil changes. An exciting implication of this result is the potential for respiration to drive arousal-linked brainstem circuits. The authors speculated that inputs to LC from the pre-Botzinger complex, a brainstem nucleus responsible for generating regular respiration patterns (Krohn et al., 2023), as a possible mechanism. An oscillatory input from the pre-Botzinger complex to LC could drive pupil changes via the Edinger–Westphal nucleus, giving rise to pupillary hippus. Kluger et al. speculated that modulation of LC at the hippus frequency might allow norepinephrine levels to undergo periodic fluctuations at rest, which would conserve metabolic resources compared with a more constant release of norepinephrine at higher arousal levels. This prediction would be consistent with the arousal-state dependence of the pupil–respiration coupling strength the authors observed.

Hypothesizing about pathways that underlie arousal processes is tricky, however, because of the interconnected system of arousal-related brainstem nuclei. While pre-Botzinger complex and LC have bilateral connections, they also have common inputs from many other upstream nuclei, including the nucleus of the solitary tract (NTS). The NTS receives input from throughout the body and plays a key role in brain–body homeostatic processes (Krohn et al., 2023). An alternative explanation for pupil–respiration coupling is that it arises from this or another shared input to the pre-Botzinger complex or LC.

To test these predictions and tease apart the neural origin of pupil–respiration coupling, we need studies in animal models. However, characterization of the pupillary hippus phenomenon in animal models is sparse. One study in rhesus macaques identified a pupil oscillation at ∼1.3 Hz that was correlated with LC activity (Joshi et al., 2016); however, this frequency is higher than the average hippus frequency identified in humans. It is unclear how consistent the coupling between pupil size and respiration will be in cases where there is less overlap between the respiration rate and hippus frequency. While there may still be pupil–respiration coupling near the respiration frequency in these cases, another possibility could be that there are cross-frequency interactions, such that respiration at one frequency could still modulate pupil fluctuations at another frequency. To distinguish between these two possibilities, we need to investigate cases in humans and animal models where the respiration and hippus frequencies are different. This can be done experimentally by altering environmental or screen luminance, which is known to alter the hippus frequency but is unlikely to affect respiration (Loewenfeld, 1993). Once the nature of pupil–respiration coupling is established in animal models, we can begin to probe the circuitry that underlies this interaction to better understand potential arousal dynamics.

Finally, Kluger and colleagues’ work demonstrates the utility of within-study comparisons at different arousal levels, which can reveal relationships with an arousal-state dependence like the pupil–respiration coupling. Arousal-state comparisons could be done using periods of rest and task engagement, as in Kluger et al., or via experimentally induced arousal manipulations, such as varying norepinephrine levels in the brain using a noradrenergic antagonist. This type of experimental design could shed light not only on arousal processes in the neurotypical brain but also on physiological relationships that may be altered in psychiatric disorders with abnormal arousal levels, such as anxiety. Future investigations into pupil–respiration coupling in such patients could provide both a diagnostic marker of pathological shifts in arousal level and a better understanding of certain treatment mechanisms, such as how reducing respiration rate may help lessen anxiety.