Scholars and philosophers have linked the breath to thought for many thousands of years. Some of the early Upanishads, philosophical Sanskrit texts from Hinduism’s oldest scriptures, directly coupled stability of the mind to stability of the breath. And in the West, this connection was made explicit in Homeric epics, The Iliad and The Odyssey, as Ancient Greeks thought the lungs were the seat of thought, emotion, and deliberation.
Modern cognitive scientists have identified still more explicit links, finding that the breath can have profound impacts on memory, attention to sensory signals, and sensitivity to emotions such as anxiety. But new research suggests the relationship between breath and thought may be even more intimate and subtle than scientists had thought. A team of scientists found that the electrical signaling deep in the limbic and cortical parts of the brain may track the specific shape of each breath on a millisecond basis, like a fingerprint. They published their results in the Journal of Neuroscience.
Read more: “Does Consciousness Require a Body?”
The new work combines big questions from cognitive science, such as the nature of thought, with the methods of data science, says Bradley Voytek, coauthor of the study and chair of the University of California, San Diego, cognitive science department. “One of the cornerstones of cognitive science is that we’re not just brains in a vat,” says Voytek. “We’re not just transistors, like little binary signaling units. We are embodied. Our brains exist in a body that evolved in this complicated world that obeys certain laws of physics, and that embodiment affects our cognition.” What data science brings to the work is the perspective that the world is rich with signals, and everything is data that can be measured, including temporal variations in the breath.
Eena Kosik-Rose, a UC San Diego PhD student in the Department of Cognitive Science, led the study, together with Voytek, as well as scientists from Northwestern University, the George Washington University School of Medicine, the University of Iowa, and Children’s National Hospital.
I spoke with Voytek about what features of the breath they measured, why they worked with epilepsy patients, and how much of our breathing is under our conscious control.
There are so many ancient traditions that have treated breath and mind as one and the same. The Greek word pneuma and the Latin spiritus both mean breath and also carried the sense of life-force or soul. Your findings seem to suggest the ancients were on to something.
Breathing is linked even to the word inspiration. When you say, “I got inspired,” you’re talking about the act of breath. We’re one of the few animals that can consciously control our breath, as far as we can tell, which is fascinating. Animals can sniff and things like that, but we can slow and control our breathing to a degree that other animals cannot, which some evolutionary psychologists believe is how language was able to evolve. We could manage controlled passage of air over our vocal cords.
The central insight of your study seems deceptively simple: that breath is more than a number. Could you tell me what this means?
We talk about breathing as breath rate, breaths per second, but everybody breathes a little bit differently. You can think about it very crudely. We tend to breathe asymmetrically, where inhalation is faster and exhalation is longer. If you plot that breath, it looks kind of like a shark fin. But the respiration rate doesn’t care about shape. That number hides all of the richness of variability. What we found is that the voltage in the brain, deep inside these regions where we were recording, tracks the shape of each one of our breaths on a millisecond level basis.
What kinds of changes in the shape of the breath were you measuring?
Breathing is so wild when you look at these signals. There’s such richness of information. With inhalation, the signal goes up, and with exhalation, the signal goes back to a baseline of no airflow. But people flat line their breath all the time. We stop breathing for tens of seconds at a time, just randomly. We might do a weird little rapid kind of chittering breathing, like double or triple inhales. More commonly, if you have a young child, if they’re crying, they might gulp air.
There are, on average, some gender differences and age differences, but we also know that respiration rate changes in certain neurological disorders, dementias, and things like Alzheimer’s disease. There’s a ton of follow-up work that we want to do because those differences are usually measured just in respiration rate, how many breaths per second. Sometimes they’ll measure the change in respiration volume, how much air passes through the airways. But we’re interested whether these conditions change the shape or the nature of breathing.
We know that in the brain, you’ve got these 86 billion or so neurons that communicate with each other electrically. We have certain cells, deep in the brain that are responsible for maintaining that respiration rhythm. And then we have a feedback loop where we know that as I breathe in, I can feel that passage of air across my nostrils. You can feel the air cooling. You can feel that breathing. So we know that we have receptors in our nose and our body that can sense that breathing. We also know that cognition and emotion and things like that are related to breathing. My collaborator on the paper, Eena Kosik-Rose, was curious: How does the brain track each breath? Is it more than just inhalation versus exhalation, which is how people have been studying it?
We took a data science approach to cognitive science, looking at the squiggle of the airflow and marking exactly when the inhalation began and when the inhalation stopped and changed to exhalation, how much airflow moved during that time, how rapid was it, how long did it take to decay.
My understanding is that the breath is seen as a kind of bridge between our voluntary and involuntary nervous systems. How much of our breath is under our conscious control?
We normally don’t have to think about breathing. We just breathe. Now you get the semi-conscious kind of control when you’re speaking. I don’t have to think about how much air I’m moving across my vocal cords, but I am consciously controlling by speaking. My wife is a singer and my daughter’s a singer and they obviously have very fine breath control. You train breath control very carefully to precisely move a certain amount of air over your vocal cords to change the pitch and the duration and so on. But we can control it very acutely, right? I can take a deep breath if I know I’m going swimming and will need to hold it.
It’s conscious and not conscious. But when I control the breathing, like you mentioned, it is a bridge. For example, if I slow my breath, I can slow my heart rate. But I can’t directly slow down my heart. The breath gives us indirect conscious control of the heart.
Plenty of scientific evidence also supports breath control for anxiety reduction. It’s one of the first things that you learn if you have an anxiety disorder or PTSD: to be mindful of your breathing, especially if you’re starting to feel the onset of a panic attack. During an acute anxiety episode, we breathe more rapidly, more shallowly. People are taught to try and take these like long, slow breaths.
Some evidence suggests that the breath has specific impacts on memory and attention. Do your findings change our understanding of that relationship?
Not immediately or directly, but they lend themselves to the experiments that could explore that. Normally we treat these rhythms of the body as roughly binary kinds of events. Circadian rhythms: You’re asleep or you’re not asleep. But we know that there’s a lot more variability in the circadian rhythm. You wax and wane in tiredness over the day.
For breathing, we tend to measure whether you are inhaling or exhaling. We take these rich biological cycles, and we turn them into these simple categories. But the reality is that biology doesn’t work that way.
Our collaborator on this paper, neuroscientist Christina Zelano, has shown that if you give somebody a memory task and you present them with the thing that they’re supposed to remember during inhalation versus exhalation, their cognitive performance on this memory task changes.
Why did you choose to work with treatment-resistant epilepsy patients?
The reason that we worked with this group of people was twofold. One, normally in neuroscience, if I want to get what these 86 billion cells are doing in the brain, I can take some brain scanning measurements. One of the most common ones that people might be familiar with is fMRI, which measures oxygenation in the brain. If a couple million of these neurons are active because they’re using energy, the brain responds by providing a flood of oxygenated blood to that region to resupply the energy. But that takes many seconds after these brain cells are active. And these neurons, they signal on the order of one/1000th of a second, which means that fMRI is great for seeing where in the brain something happened, but terrible at understanding when.
The other measure that people are familiar with is almost 100 years old now: EEG, electroencephalography. You put electrodes on people’s heads to measure brainwaves. The neurons send signals and the electrical fields that they emanate hit the inside of the skull like a wave crashing into a rock wall and splashing everywhere. You have no idea where it happened. By analogy, it’s like trying to understand a conversation between a handful of people in a domed baseball stadium by having a microphone outside of the stadium. It doesn’t work. You can hear that everybody’s cheering and clapping and stuff, but you can’t hear the fine details.
The other way that people approach this in neuroscience is by doing surgeries on animals. They open up a part of the skull and implant electrodes inside of the animal’s brains. In this study, we got the best of both worlds. We worked with neurosurgeon collaborators here at UC San Diego and the University of Iowa, who are doing surgeries on people with epilepsy to treat their seizures.
For patients whose epilepsy is resistant to drugs, the option is to do a series of surgeries that take a couple of weeks, sometimes up to two weeks, where the goal is ultimately to identify this tiny little part of the brain that’s causing the seizures and just surgically remove it. To find this so-called epileptogenic zone, the surgeon has to open up the person’s skull and implant electrodes and wait for them to have a seizure. With these wires directly in their brain coming out of their heads, these patients are sitting in the epilepsy monitoring unit for up to two weeks waiting to have seizures. In the meantime, they’re just on their phones, they’re reading, they’re watching TV, playing video games, chatting with their friends and family, just waiting, kind of bored usually. During that time, we approach them and say, “This is a very unfortunate situation, but given the situation we’re in right now, you provide us with an unparalleled opportunity to understand the human brain at a level that we cannot otherwise do.”
Does the fact that the study participants were all treatment resistant epilepsy patients bias the findings in any way?
What we wanted to do in the study was understand this coupling between the brain and the body, specifically respiration and the brain. Our interest is understanding cognition, but a nice side effect of this data science approach could be to find some kind of predictive signal that will help us anticipate when someone might stop breathing.
One of our neurosurgery collaborators on the paper, Brian Dlouhy, from the University of Iowa, has a particular interest in what’s called SUDEP, sudden death in epilepsy. It’s very rare, but some people with epilepsy just die, and he thinks it’s because they stopped breathing. He has a particular interest in respiration, which is how we got connected. By analogy, it’s not proven, but some people think it’s somewhat similar to SIDS, sudden infant death syndrome, where infants just stop breathing in their sleep, which was terrifying to me when my kids were little.
This is follow-up work that we’re planning on doing. In clinical settings, we call that a yellow-light warning system, so instead of being reactive, you’re proactive. That’s where we’re going next. In other words, the short answer is, it does bias our signals, but it allows us to do something that we otherwise couldn’t. 
Enjoying Nautilus? Subscribe to our free newsletter.
Lead Image: Donkeyworx / Shutterstock
