Have you ever woken up in a cold sweat after a particularly feverish dream, your body worn out and your mind racing?

Some dreams seem to deplete us rather than restore us, which goes against common sense. Recently, a team of neuroscientists from Japan set out to understand why this happens by imaging the brains of dreaming mice.

What they found was a second paradox related to how energy is consumed in the brain. They also found that the circulation of the blood plays a larger role in determining how the brain behaves than they had thought. The findings tell us something about the energy costs of dreaming—and about how closely the brain and the heart are connected, says Ko Matsui, one of the authors of the study, which was published in Nature Communications.

In ancient Greece, Aristotle believed the mind, soul, and intellect originated in the heart. The brain was simply a cooling station for the blood, he proposed. Over the ensuing centuries, as the science of the brain advanced, we came to understand the importance of the brain for processing sensation, perception, movement, emotion, and consciousness. But now, says Matsui, we’re in some ways confirming what Aristotle had intuited: That the heart and the blood play a starring role in the way the mind operates, whether dreaming or awake.

I spoke with Matsui, a professor of brain physiology at Tohoku University in Sendai, Japan, about what dreaming is for, how the animal brain is different from a computer, and what his findings tell us about the nature of certain brain disorders.

You set out to investigate this paradox about dreaming: Rest is supposed to be restful, but people sometimes feel physically exhausted after a particularly vivid dream. Then you found another paradox. Can you tell me how these two puzzles relate?

REM sleep is when we’re supposed to be dreaming. If you look at a person’s brain waves during REM sleep, it looks similar to the awake state. The brain is quite actively working, even though your muscles are completely relaxed. And this phenomenon has been studied by scientists for a long time.

It was called a paradox because you’re supposed to be resting to recharge your energy, yet the brain seems to be working quite busily, taking in a lot of energy through the blood pumping through it, supplying oxygen and glucose. But when we did our study, we found another paradox: Even though the body is supplying a lot of energy to the brain through the blood, the battery of the brain is rapidly decreasing.

How does this work?

If you look into the anatomy of the brain, it’s quite interesting. The only supply of energy is through blood. Blood carries oxygen and glucose. That’s a source of energy. But the neurons, which use a lot of energy, aren’t in direct contact with the blood vessels. They’re connected via the astrocytes, which are intermediaries in energy delivery. And there are several places where you can change the efficiency of the energy supply.

The blood vessels are surrounded by smooth muscle. Depending on activity in the brain, the blood vessels can dilate or shrink, which determines the volume of blood available to any particular region. But the activity of astrocytes also determines how efficiently energy is delivered to the neurons. What we found was that in REM sleep, the blood vessels expand, so more energy is being provided. And the concentration of the molecule that helps transfer energy to the neurons, called pyruvate, also goes up. But the concentration of energy molecules that neurons use to function, called ATP, goes down.

We have three hypotheses about why this happens. One is that the neurons are consuming so much energy in dreaming that the total energy in the system decreases. The second is that maybe the delivery of energy from astrocytes to neurons is suspended for some reason. The third is that even though the astrocytes are supplying energy to the neurons, the neurons’ mitochondria, the engine of the cell, stop converting this energy into ATP.

Read more: “Dreaming Is Like Taking LSD”

How do the energy supply needs of REM sleep compare to those of waking-life activities, like reading a book or watching a movie?

What we need to appreciate is that the brain constantly uses energy. And the amount of energy used, by the way, is quite a lot. If you stop blood circulation to the brain, a few seconds is enough, and you become unconscious. That shows that the constant flow of energy to the brain is important to keep up, whatever it’s doing. Interestingly, our brain is actually quite small—it occupies only 2 percent of body weight, but it consumes about 20 percent of the energy used in the whole body.

You found a pretty dramatic inflection point, a 50-second period right before REM sleep starts, when the blood vessels start pumping more blood into the brain. Did this surprise you, and what does it tell you about how the brain operates in REM sleep?

What we often imagine is that if you’re awake and doing a lot of information processing, and your mind is full of images or plans, the brain requests more metabolic energy via the neurons, the information processing units in the brain. But what we found is that it goes both ways. Some 50 seconds before REM starts—and 50 seconds is quite a long time—the blood-flow surge begins. That suggests that the metabolic landscape is being prepared long before we actually enter the dreaming state. In other words, it seems like the metabolic energy is determining how our information processing works. It’s reciprocal. Information determines how metabolic energy works, and metabolic energy also determines how neurons work.

It’s like the brain is making its bed. We thought the neurons were calling the shots, but in fact, the blood vessels make the decision. It’s a consequence of the metabolic landscape that you go into. And that’s the direction of study in our lab—trying to show that neurons aren’t the central decision-makers in your mind. Decisions are shaped by non-neuronal cells, such as blood vessels and glia cells, astrocytes, and other effects from your body.

Modern science would suggest that your brain is the most important thing. That your consciousness or your mind is in the brain. In ancient Greece, some people thought that the mind was in your heart. And I think that to some extent, that’s true: your heart and blood vessels and non-neuronal cells. Your body is affecting how your mind works. If you just take the neurons, they alone wouldn’t form a complex mind.

In a bigger picture sense, do your findings or these hypotheses have any bearing on the purpose of dreaming? There is so much ongoing debate over what dreaming is really for.

The purpose of dreaming is an interesting subject, and it’s been studied and discussed a lot in the science community more generally. But in other work in mice, what we see is that if you record the activity in the neurons when they’re dreaming, it repeats the patterns of their experience the previous day. It seems to be repeating those patterns at a faster speed but deformed in some way. My hypothesis of what is going on within the dream is that it’s probably replaying their past experience with the aim of deciding which experiences to store as memories and which experiences to forget. This process is called plasticity, and it may require a certain amount of energy to produce.

How likely are your findings to translate to humans?

The surge in blood volume during REM sleep has been shown in humans as well in fMRI studies, but we use the specific genetically encoded sensors in mice to look at the ATP levels, for instance, and that couldn’t be done in humans. Functional MRI studies are quite important. What they show is an increase in blood flow during REM sleep, but we thought this meant the brain must be enriched with ATP. Using mice, we found it’s not that simple. The information processing, blood flow, and energy all behave differently. And you can’t easily infer just from the MRI studies how much energy is available for neurons to use.

Do your findings throw fMRI readings of brain activity into doubt?

No. In this study, we were able to look into another component that had been missed. fMRI can provide information about the blood flow, and EEG can measure electrical activity in the brain—in humans as well as in mice. But the energy dynamics of ATP molecules were never visualized before. So we were able to access one more component that hadn’t been previously available. On the other hand, the general public may think that functional MRI reports reflect neuronal activity, when what it’s actually revealing is just blood flow. And scientists actually don’t know the exact correlation between blood flow and neuronal activity. This relationship isn’t completely understood. We simply focus on another component that’s being missed by the functional MRI. That’s the energy dynamics.

Read more: “Does Dream Inception Work?”

You write that your findings can help us understand what makes the animal brain distinct from the computer: It reroutes energy depending on behavioral state or memory demand or needs. A computer, on the other hand, distributes energy uniformly. Is that true of all animals—from the humblest sea sponge on up?

I don’t know. What is distinct about the animal brain is that it receives information through sensory inputs. But it doesn’t necessarily give the same output, so it’s an unreliable electrical circuit. Also, one of the differences is that we don’t calculate all possibilities. If you’re being chased by a lion, and you have to decide whether you go right or left, the computer works through all the possibilities and potential outcomes. But what the humans or animals do is that they think, “Oh, probably heading right seems better,” and they make a decision with only about 70 percent of confidence. They will choose one way or the other. If you stop to consider all possibilities, the lion will catch up and eat you, so you have to decide quickly. We conserve energy and time by choosing only a part of the whole calculation. That’s probably what many animals are doing—even insects. Animals—we’re lazy. We don’t do all the hard work.

Are there any efforts, as far as you know, to make computers more like the biological brain in this way?

I think that’s being explored, especially recently. You hear that Bitcoin or AI consumes a lot of energy—that it’s consuming a huge amount of power, and that is affecting the environment. Making this more efficient and more human-like, or more biology-like, I think it’s been explored a lot. With new AI technology, maybe there will be a way to conserve energy rather than getting all the answers correct.

At the end of the paper, you write that understanding how the energy system of the brain works in sleep might help us better understand or even treat sleep disorders or neurodegenerative diseases and psychiatric conditions. Are there any specific research questions you’re pursuing in this area right now?

We published a paper last year showing how energy flows change during epileptic seizures. In epilepsy, the neurons are hyperactive, and they consume a lot of energy. If you have multiple seizures, then you get more severe seizures. But we also found that the metabolic circuit adapts and changes, which leads to more severe seizures. That is an extreme case, but understanding how the energy circuit in the brain affects neuronal activity would be directly applicable to situations such as epilepsy.

We’re also looking more closely at neuropsychiatric disorders, like depression. We don’t have much data yet, but we believe they may be influenced by the metabolic energy dynamics as well.

Lead image: fran_kie / Adobe Stock