CHAPTER 1THE SCAFFOLDS OF THE DREAMWORLD
Anyone who has ever dreamed will know that many of the dreams we have are built on the scaffolding of our everyday life. Faces, places, or actions that we’re familiar with pop up in our sleeping lives, along with concerns and affairs borrowed from waking reality: an argument or conversation, a regular commute, a movie we’ve just seen.
After many years of studying the neuroscience of dreams, reading thousands of dream reports collected in the laboratory or at home, through online surveys or in everyday conversation, one thing that has become clear to me is that there are certain patterns in dreaming—patterns underlying how our waking life is not only represented but misrepresented in the sleeping mind. The dream has a unique, purposeful design: reflecting, yet transforming, our daily lives.
What we pay attention to while awake permeates where we journey in our dreams, with personal memories and culture filling in the scaffolds of the dreamworld. The sleeping self maintains a sense of identity, too, preoccupied by work, romance, survival. And the external world seeps into dreaming; this will come as no surprise to anyone who has slept through their alarm clock before, the sound disguised as songbirds or disco-themed dreams. Indeed, the sleeping brain is not so cut off from the sensory world, the lights and sounds of the room around, as scientists once believed. All of these components together—our memories and preoccupations and the guise of the sleeping body—drive the design of dreams.
What are the idiosyncratic and, perhaps more telling, universal ways that we dream?
How dreams are designed
Over the course of my career, I’ve had the privilege of conducting dream science in a number of sleep laboratories around the world. The quintessential sleep lab is built on the same basic principles: a bed in a closed room, a two-way communication system, ideally a video monitor, and various recording equipment. This includes electrodes to measure brain activity and fluctuating heart rate, a small tube that splits into the nostrils to measure airflow, and various electrodes measuring twitches or tension in muscles or eye movements. Even more sensors can measure the position of the head and body, or the rising and falling of the chest, and a small microphone taped to the neck can record the slightest tickle of snoring in the throat. All of these subtle signals of the living body are collected in massive amounts of data over an eight-hour night of sleep.
Countless times I’ve invited subjects into the laboratory, and before they go to sleep I spend some time placing electrodes on their scalp and casually making small talk. I try keeping the lights low, offering some herbal tea, and putting the subject at ease: The human element is essential to a good night’s sleep.
Invariably, almost every subject asks me how I got into this work. It’s a fair question—dream scientist is hardly a typical job title. And while I have always been fascinated by dreams, I must admit that I never thought I would be able to pursue a career in the science of dreaming. In a way, I stumbled into the field as an undergraduate at the University of Rochester. I had the opportunity to work in the Sleep and Neurophysiology Research Laboratory, located in the medical school a short jaunt away from the main campus. The sleep lab was tucked away in an old basement corner of the hospital, a hidden gateway to the unconscious mind.
It was there that I first learned about the electrophysiology of the sleeping brain—the waxing and waning patterns of electrical activity that occur as the brain descends through deeper and more active stages of sleep. While the science of sleep is, in its own right, a fascinating area of study, I was especially interested in dreaming, with entering and understanding that dark basement corner of our unconscious mind.
In Rochester, the bedrooms in the sleep lab were near luxurious, with large queen mattresses, thick warm duvets, and art deco bedside lamps; artistic paintings of sleeping figures adorned the walls. The control room where we monitored subjects was set apart from the bedrooms, separated by a long hallway, a private bathroom, a kitchen, and a lounge area. The biggest challenge for me was always staying awake all night while observing hypnotic brain waves on monitors, watching infrared videos of subjects’ sleeping selves. Having the control room separated from the bedrooms meant we could easily keep the lights on and talk or listen to music, while down the hall the bedrooms were near silent and pitch-black day or night—an ideal environment designed to be as conducive as possible for sleep.
In my current lab in Montreal, we have a more modest setup with two dorm-style bedrooms spotted with Ikea-like furniture. This kind of setting is in many ways familiar to our subjects, who are often students from local universities. The lab is one in a collective of sleep research set at the end of a fifth-floor hospital wing. Through a connecting doorway sits a sister sleep laboratory with three enclosed bedrooms, each with a private bathroom, where experimenters can control the lighting in the rooms to mock day-to-night phases of different lengths. Subjects sometimes spend multiple days in these time capsules, while experimenters observe how varying rhythms of light and life impact their sleep, relevant to night shift workers, blind individuals, and more.
Across the hall, the clinic diagnoses and treats patients with sleep apnea, sleepwalking, narcolepsy, and insomnia, to name a few, and our collaborations are revealing the many ways that dreaming, too, may be disrupted or treated in sleep disorders. Overall, the sleep center is a lively setting, and the lights and sounds around sometimes leak into our subjects’ dreams: from the labyrinth of fluorescent corridors on one side, to the windows outside on the other (think industrial Canadian snow removal over many nights each winter). Still, with a warm blanket, a dark room, and muffled sounds from the outdoors, our subjects often drift readily into slumber.
Sometimes, as I sit in the control room monitoring another subject, perhaps with a fellow dream researcher, I think back to the many other labs I’ve worked in. During my first postdoctoral research position in the UK, for example, we had a simple lab space with two comfy bedrooms and in a makeshift manner we adapted baby monitors into a two-way communication system. The lab was set in the center of an accessible college campus, near a beautiful coastal shoreline where subjects’ dreams were sometimes startled awake by seagulls in the mornings.
In Chicago, Cambridge, and the Netherlands, the bedrooms were fashioned from preexisting labs devoted to waking neuroscience, refitted with pull-out futons or couches to enable sleep studies. One of the bedrooms I conducted research from was tucked inside an old faraday cage, a room with a heavy metal refrigerator-looking door that isolates the subject inside an experimental chamber. Other studies in sleep clinics featured hospital-grade plastic beds and thin woven blankets, detracting from the comforting feeling so precious to sleep.
Suffice it to say, dreaming in a sleep lab can be far from perfect, and yet in my experience it has been endlessly revealing.
* * *
The most striking thing I’ve noticed across so many different studies, in different laboratories and countries, is that, despite the varied surroundings and comfort levels and ambient noise, there have been unexpected similarities in my subjects’ dreams. Of course, while the basic sleeping environments are designed to be comparable to a regular sleeping experience at home, the unusual pressure of needing to sleep and dream well for the experiment is omnipresent. It’s perhaps not surprising, then, that almost half of our subjects dream about the experiment itself.
From back in the control room, as soon as I see a subject enter REM sleep—with the characteristic rapid eye movements that define the sleep stage—I try to imagine what they might be dreaming. I wait five to ten minutes for their brain and muscle activity to ramp up, then grab my microphone and carefully, gently, call their name to ask, “Can you tell me what was going through your mind before I called you?” I pause and wait, as they usually stir for a bit, and hmm and umm as they try to pick up the pieces of the dream that’s just fallen away. In regular life, it’s relatively unusual to be woken in this way, but the gentle interruption is precisely what gives us prime access to the dream’s contents. I listen carefully to their halting words, as they strive to tell me everything they can remember. A dream:
I see the wires on my face. There are hospital corridors that I hover backward through. The student researcher follows me and I want to escape her. Then I’m in the lab and my parents come into the room. I am annoyed, I only have minutes to fall asleep.
Or:
I dreamed that the experiment was going on too long. I felt like I wasn’t performing well enough on the sleep side. I remember wondering (in my dream) if you could possibly see my dreams through the electrodes and if you were going to find me weird.
Poring over a decade of studies set in Montreal, my team and I uncovered references to the lab and to the experiment scattered throughout hundreds of dream reports.1 Subjects dreamed of the study setting, of corridors of the hospital, and experimenters appearing as dream characters; they dreamed of their mission to sleep well and to remember a dream, of objects like electrodes or computers or clipboards; and finally general sleep-related themes such as wearing pajamas or seeing a bedframe. Though several of these study elements often appeared together, dreams never reenacted the presleep laboratory episode in full. And this is a fact of dreaming: Fragments of recent experience are often woven into dreams and combined with other memories to create a novel story, but single memories are almost never replayed in their entirety. One exception to this is the special case of nightmares, where traumatic memories are re-created in part or in full, but this represents a breakdown in the normal process of dream creation—and we’ll return to these cases later.
While the lab dreams we observe do vary considerably, there are several recurring features that appear across many different subjects, suggesting the presence of certain design pressures that shape the narratives of dreaming.
Dream scenes in the lab often included research personnel appearing alongside other characters, especially friends or family, who crop up as part of fictional social scenarios in the lab. In fact, there is nearly always a social element to dreaming. In home dreams, social situations occur in over 80 percent of reports, and up to 25 percent of characters are family members and 20 to 40 percent are friends. In studies conducted in the lab, research personnel appeared in over half of our lab-related dreams—these are virtual strangers who the subject has only just met. This brings us to our first key element of the scaffolding of dreams: Dreams are consistently social in nature, encouraging dreamers to freshly reengage with people, even strangers, from their waking lives.
This social design provides a helpful clue about one function of dreams themselves: that dreamed social simulations2 may have evolved to support our development as a social species. That dreams are selectively social, that is, they overrepresent social elements of our waking lives, could serve a purpose in reinforcing social skills overnight, perhaps allowing us to test out interactions and their potential consequences during sleep.
Of course, visiting the sleep lab is not your typical social affair. Sleeping in the lab is unusually intimate, with an experimenter watching as a subject sleeps and recording their dreams. It’s not entirely surprising, then, that subjects also dream of being observed and of wanting to perform well for the experimenter. This ties in to a second element of how dreams are designed: Dreams are often performative in nature, fashioned around a theme of skill rehearsal or achievement. Lab dreams often incorporate the basic task of trying to sleep well and remember dreams, along with other experimental tasks, too.
One subject dreams:
My mother was with me, we walked in a corridor into the testing room. [The experimenter] put electrolytes in my head to write the dreams directly on paper. I went to another corridor to do the second test … we wanted to finish to get out of the laboratory.
Another reports:
I was trying to fall asleep and not doing very well, because there was no real barrier between the bed and the researchers … and at one point my bed was in fact outside. The researchers referred to my brain patterns as a “Dante pattern.”
Though there’s no such thing as a Dante pattern, the latter subject’s dream of trying and trying to sleep and perform well, despite many obstacles, is very common in the lab. It seems that while awake we make note of certain challenges or tasks at hand, and once asleep we are again put in a simulation to try to achieve our goal. The dream is often repetitive or circular. This subject repeatedly tries to fall asleep, but one thing after another gets in the way; that subject is doing tests for the experiment, but after finishing one test a second comes up.
In dreams collected at home, we witness similar themes of perpetually attempting (and often failing) to achieve a goal, like trying interminably to catch a bus or to get to the airport, endlessly looking for an object or outfit, underpreparing or arriving too late for an exam. These dreams seem to be goal-oriented and, like in the lab, they seem to be more about trying rather than succeeding, dreams of process rather than completion.
Beyond simply reflecting desired goals, research and perhaps common sense tell us that dreaming enables nightly rehearsal in useful skills and that this repetition may be key to learning. While asleep we can practice using our dream bodies, to maneuver dream objects and navigate dreamworlds. For instance, athletes and musicians dream of practicing their particular sport or instrument, sometimes in high-pressure dreams of performance. If our lab dreams are any indication, another skill commonly enacted by the dreaming mind is that dreams are often exploratory—we are seeking, searching, probing through an unfolding map of the world.
Consider the following:
The researcher came into the room, turned on the light, and another person came to talk to me. She cared for conscious people in dreams. The bedroom had another door with a corridor, and there were people in white uniforms at the end of this corridor in a room.
Another example:
The experimenter comes into the bedroom. I see in front of me, where there should be a wall, a door. The door is opened by Jim from The Office. I see on the other side of the door a small corridor, and another open door. Jim tells me I can get up.
In these examples, dreamers imagine waking up in the lab and peering out into the expanding scene. The dream seems to reinstate a world that extends out from us spatially and temporally, like ripples of space and time. Some subjects even dream of making their way home from the study, weaving through hospital corridors and traversing city streets. These dreams build on our ability to remember the recent past, including where we are and what we’re doing, and project ourselves into possible near futures, a form of mental time travel with a hint of absurdity.
* * *
While some scientists quite fairly argue that dreams collected in the sleep lab might not represent natural dreams—that they’re not the same as dreams collected at home—in my opinion they are still informative and useful, and they help flesh out the key pieces of scaffolding for the dreamworld. In fact, dreams collected from home and survey studies reflect similar themes to those collected in the sleep lab, fashioned around social simulation, skill rehearsal, exploration, and reference to the current environment and ongoing concerns.
Another piece of the puzzle, so to speak, of how dreams are designed and built up comes in the form of “typical” dream themes.3 These are those dreams whose symbols or narratives are found repeatedly and with high prevalence across the population, and even across different cultures and ages. You have almost certainly experienced one or some of these typical dreams yourself. Upward of 70 percent of people report experiencing the most typical dream themes like being chased, or falling, or trying again and again to do something. Over forty themes have been uncovered in survey studies, and many of them expand on the designs we’ve identified in lab dreams.
The social design of dreams, for instance, is evident in typical themes of being chased or pursued, bad dreams of infidelity or pleasurable erotic dreams, and the semi-social dream of sensing a presence in the room. Other typical social dreams seem to cluster around concerns of self-consciousness or embarrassment. This seems to be the common thread to dreams of being inappropriately dressed or being nude, which resemble lab dreams of being observed by experimenters who are peering into the bedroom or recording subjects’ private thoughts.
Other typical dream themes are more positive and exploratory, such as finding money or—my personal favorite—discovering new rooms or passages in your home. Sometimes, this dream plays out fantastically, like opening magical doors to another world. Other times, it’s much simpler, like noticing a new closet in the hallway or a useful dishwasher in the kitchen (a good dream!). This theme is similar to lab dreams where unexpected windows, doorways, and corridors keep appearing as subjects wander through the hospital setting. It seems like, rather than starting with a predetermined map, the dreamworld is continually created wherever the dreamer ventures, unlocking new spaces along the way and encouraging further exploration.
While typical dreams are so named because of how often they occur in the general population, it seems like each of us dreams a careful selection of themes that cluster around personal concerns. One person might recall frequent social dreams of embarrassment, like being naked in public; another person might have typical dreams of failure, of attempting to achieve a goal again and again without success. These dreams can take on a recurrent nature, repeating many times over many nights, and reappearing more frequently during times of stress. Up to 75 percent of adults experience recurrent dreams like this, which can begin at a young age and persist for the rest of one’s life. To give one example, the typical dream of missing an exam often begins during school years, when academic stress is at an all-time high; but this theme can recur for years, perhaps reappearing before a big presentation at work or an interview. Although the circumstances are different, the stress of performing well seems to trigger the familiar dream scenario.
While recurrent dreams are common to a large number of people, they seem to be more pathological than typical dreams; we’ll return to recurrent dreams later, because they overlap quite a lot with nightmares.
Coming back to typical dreams, these universal themes reveal certain consistencies in dream formation across many people. Like lab dreams, they reveal a dreamworld that seems to be constructed around several purposeful designs—pressuring our dreams to be filled with characters to interact with, tasks to accomplish, avenues to explore. These themes point toward a sort of structural foundation to dream formation and provide clues to dream function itself, as we will continue to discover. At the same time, typical dreams are curious because they reappear in so many cultures and ages, and yet are often not experiences typical of waking life. When have you ever discovered a new room in your house, fallen or flown through space, or found yourself nude in public? And yet, a good deal of people will experience these dreams, possibly even recurrently.
What does this tell us about the design of the dreamworld? It’s hard to argue, for the more unusual themes that seem far removed from waking life, that they play a role in reinforcing specific skills, or even any relevant social practice. To understand a bit more about some of the more unusual and unexpected dream themes, things like teeth falling out or being unable to find a toilet, we’ll have to look at another source of dream formation: the body itself.
The body’s dreaming, too
It’s not unusual to realize on waking that elements from our surroundings have found their way into our dreams: a barking dog, a car alarm, a chorus of birds, all transformed into the sounds and setting of the dreamworld as our brains try valiantly not to wake up. Needless to say, if the stimulus is too intense it will provoke an awakening, which is of course the basic principle of an alarm clock. But besides these morning interruptions, most of us consider sleep to be a time when we are cut off from the physical world, encased in our minds and insulated from our sensing bodies.
In actuality, dreaming is not so isolated from either the sleeping body or its sensory access to the world. Instead, there continues to be a flow of information through our senses, which is at times fluidly incorporated into dreams.
The influence of the body is perhaps most noticed in dreams at times when it’s least desired: the urge to urinate, the inability to speak or move, grinding teeth or ringing ears invading the peaceful blanket of sleep. An extreme example of this comes in the form of sleep paralysis, where you are unable to speak or move in the moments before fully waking, often accompanied by frightening dreams. I remember the first time I had sleep paralysis: Early one morning before high school, I woke to the terrifying realization that my whole body was completely immovable. I strained to open my eyes and even tried to scream but was unable to lift a muscle or make a sound. I had no idea what was happening, and in a panic I noticed a shadowy presence near the door of my bedroom, slowly coming toward me. I thought I was going to die. Then suddenly my eyes lifted open, and it was daylight in my empty room. I gasped to catch my breath, confused and afraid.
I went on to experience sleep paralysis regularly for years, and it was a primary reason I became intrigued by sleep research in college. I was reassured to learn that sleep paralysis is neither dangerous nor uncommon (up to 40 percent of the population experiences it at least once); it’s simply a relic of the natural muscle paralysis that occurs during REM sleep. While it may seem strange that the body is physically paralyzed during REM sleep, this has a functional benefit: It inhibits us acting out our dreams in our real bodies. On the downside, this paralysis sometimes seeps into our dreams, especially in the liminal space between sleeping and waking, and experiencing this can range from alarming to downright terrifying.
While sleep paralysis is one extreme example, in fact, the body is often a part of dreaming; it’s just that most of the time, we don’t notice it in this way. In many ways, the body physically experiences what is happening in the dream, and vice versa, as the dream also responds to what occurs in the body. In other words, dreaming is not generated solely by the brain; it develops in communication with the sleeping body.
The first known physical correlate of dreaming was rapid eye movements, discovered in 1953 and thereafter defining the stage of REM sleep.4 When subjects were awakened in periods containing these “rapid, jerky, binocularly symmetrical” eye movements, 74 percent of the time they reported vivid and visual dreams. This discovery led to a surge in dream science studies (although we’ve since learned that dreaming can occur in other stages of sleep), and kick-started a tradition that continues to this day, of attempting to link objective indicators—like bodily and brain activity—with the mental experience of dreaming.
Could these eye movements be revealing actual visual activity in dreams? This possibility became known as the scanning hypothesis, and some studies found support for the idea, with rapid eye movements matching visual dream descriptions, such as a recorded tennis-match dream with repetitive left-right-left-right eye movements. Though mismatches between eye movements and dreamed visuals can occur, groundbreaking studies of lucid dreaming proved that a correspondence does exist. In the late 1970s, independent researchers Stephen LaBerge and Keith Hearne showed that lucid dreamers could make deliberate left-right-left-right eye movements in their dream that resulted in matching eye movements in the sleeping body.5 This proved that dreamers could control their physical body using movements in a dream. In the years since, many other studies have followed suit, revealing how lucid dreamers can control various parts of their physical body: They can flex their fist in a dream, which creates muscular twitches in the forearm at the same time; and lucid dreamers can control their breathing in a dream, which modulates respiration and abdominal movements in their sleeping body.
Even in nonlucid dreams, at times the content of dreams seems to physically manifest in the body. During REM sleep, for instance, the body undergoes erratic fluctuations in breathing, heart rate, and muscular activity, all of which could correspond with the mental experience of dreaming. In one study, a nonlucid subject dreamed of swimming the breaststroke, and measurements taken while the subject was asleep showed that they repeatedly held and released their breath in a way that mimics going above and below the water. Nightmares, too, appear to be strongly linked to the body. In the minutes before awakening from a nightmare, most sleepers have an increased heart rate and breathing rate, as if the body is physically experiencing the stress of the dream. Of course, this can have consequences, as we’ll see later, for those who repeatedly have nightmares, such as patients with post-traumatic stress disorder (PTSD), whose dreams can reenact both mental and physical aspects of a trauma while asleep.
There is clearly some relationship between what we imagine in our dreams and how our bodies respond to or mimic our dreamed experiences. But just how does the dreaming mind link to the sleeping body? And what are the implications of this connection?
In the first place, we can look at the “body map” that exists in the brain, meaning the organized mapping of sensory and motor neurons onto certain body parts. This neural map of the body seems to underlie the vivid sensorimotor imagery—the imagined sensations (sight, touch, sound, smell, taste) and bodily movements—of dreaming. For instance, functional magnetic resonance imaging (fMRI) of lucid dreams has shown that when lucid dreamers clench their fist, activation is observed in the same sensorimotor brain regions as when they do this during wakefulness, and this also corresponds with real physical twitches in the forearm. The mind, brain, and body are aligned. The same pattern is found for other types of perception like observing faces or hearing language or seeing spatial scenes—these are all associated with specific patterns of sensorimotor brain activity in waking life, and similar patterns are found during dreaming.
In general, the sensorimotor cortex, then, is providing vivid sensory and motor experience to dreams—experiences of walking, moving, flying, falling, fighting, foraging, and more. It’s worth noting that perceptually, dreams are most marked by visual imagery, followed by auditory and movement imagery, and less so by odors and tastes; although this varies in different populations—deaf and blind subjects, for instance, have more smell and taste in their dreams, and we’ll learn more about these variations later.
At times, the motor imagery of dreams is linked to real body movements, despite the intended paralysis of REM sleep. In the case of dream enacting behaviors, sleepers physically act out dream content, often during emotionally intense dreams, such as nightmares, grief dreams, or even erotic dreams. Common dream enacting behaviors include speaking, crying, smiling, laughing, and expressing bodily fear or defensive behaviors. Fascinatingly, these behaviors can reflect the speech or actions of the dreamed self or of another dreamed figure. For instance, if you dream of flying to escape a cackling witch, you could wake up shaking in fear, or you could wake up laughing, as if you were the witch. This highlights a phenomenon of dreaming that we rarely acknowledge, which is the fact that all of the characters within a dream are simultaneously created by the mind—including your own dreamed self, a friend or family member, a stranger or pet or spiritual guide. While witnessing the dream from a first-person perspective, the sleeping brain is also conjuring up these other characters with their own thoughts, feelings, actions, and voices. It even, at times, uses different languages or accents for each of these characters.
This is quite impressive, that the mind is able to orchestrate the actions and emotions of all of these dreamed figures. How is it possible for the mind to do this, to coordinate such a diverse cast of dream actors?
Some scientists have looked to the mirror neuron system as one potential generator and controller of dream characters. Brain imaging studies have revealed that the mirror neuron system is active while both performing a specific action, such as gripping a mug, and observing that action in another person, such as seeing a friend grip a mug. In fact, the same brain activity can be seen when merely imagining this action in oneself or another person. In general, mirroring the actions and expressions of other people helps us to model and understand their behaviors and emotions while we are awake. In a similar manner, the mirror neuron system could also be responsible for simulating the actions and emotions of other characters while dreaming.
Some support for this idea comes from the study of mirroring behaviors—these are behavioral manifestations of mirror neuron activity, such as contagious crying or laughing (i.e., crying or laughing when seeing others do the same) and other kinds of motor mimicry like taking the same postures or facial expressions as others. Dream science studies have revealed that people who express more mirroring behaviors while awake also express more dream enacting behaviors while asleep. As an example, people who frequently cry upon seeing others crying are also more likely to wake up from a dream crying. The same is true for smiling and laughing.
Altogether, then, the sensorimotor brain and the mirror neuron system may be puppeteering the actions and emotions of our dreamed selves and other characters, in a manner similar to how we act and interact with others in waking life.
* * *
While it seems clear and perhaps even intuitive that the body might mimic or display some of the actions and emotions of our dreams, there is a potentially more interesting flip side: the physical body as an actual source of dream imagery.
Ever marveled at a sleeping pet or a baby, watching as their limbs or whiskers twitch, like mini spurts of dream activity? For many years, scientists assumed that these twitches were simply a by-product of dreaming, leaking out through incomplete paralysis in REM sleep. But there is now evidence that twitches actually provide a source of dream activity. Twitches produce substantial activation in sensory areas of the brain during REM sleep. As an example, a twitch in the forearm causes strong sensory signals in the “forearm” part of the brain-body map, as if the forearm has just experienced an intense sensation. This activation is much higher than would be expected for such a small movement; this is likely because the body is mostly paralyzed in REM sleep, that these tiny twitches are in a sense amplified and can be clearly distinguished by the brain.6 So every night when we go to sleep, these twitches allow the brain to fine-tune and recalibrate its one-to-one mapping of different body parts, almost like a call and response—twitch to forearm? Ah! there’s the forearm.
This is thought to be a fundamental purpose of REM sleep, to develop and maintain our brain-to-body map over time, so that we can effectively use our bodies in the waking world. Dreaming, too, could support this function, calibrating the sensorimotor brain as we maneuver our dream bodies through immersive and multisensory dreamworlds.
If twitches give rise to clear sensory signals in the brain, could they also give rise to strong sensations in dreams? Indeed, a common example of how dreams are influenced by twitches occurs just at sleep onset, when brief hypnic jerks are accompanied by vivid imagery. A hypnic jerk is when your legs or body suddenly twitch as you’re falling asleep, and an exaggerated dream image often coincides with this sensation, such as an image of falling down a flight of stairs. The twitch seems to be magnified by the dreaming mind, perhaps because the signal is so strong compared to the stillness of the rest of the body. While speculative, twitches could also be responsible for the curious case of “exploding head syndrome.” While it sounds dramatic, this is actually a harmless experience, where a sudden deafening noise shocks your mind as you’re falling asleep. It may be that tiny twitches of the middle ear muscle (the ear’s version of rapid eye movements) are suddenly amplified in the dreaming mind as the body falls into sleep. These examples show how subtle movements and sensations in the sleeping body could protrude into a dream with a potentially magnified effect.
Similar explanations have been provided for REM sleep behavior disorder, a neurological disorder associated with loss of muscle paralysis in REM sleep. Patients often have vivid and violent dreams that are enacted in extreme and sometimes dangerous repetitive body movements, such as kicking or punching or flailing in bed. Ordinarily, the muscle paralysis that occurs during REM sleep is thought to prevent precisely this kind of dangerous acting out of dream scenarios in the physical body. While some scientists view the violent dreams as the cause of kicking and punching behaviors, the opposite may be true: that the loss of muscle paralysis in these patients leads to excessive and repetitive twitching movements, which then feeds back into intense and aggressive dreams. In the absence of this explanation, it is hard to understand why patients suddenly develop such violent dreams at the onset of the disease.
Besides twitches and bodily movements, other internal body states could also contribute to dreaming. For instance, several typical dream themes seem related to physical sensations, such as searching for a toilet or trying to find food or water in a dream. Subjects in the lab sometimes dream of looking for food in the cafeteria or trying to use the bathroom despite unusual obstacles (such as the bathroom having no doors, or personnel entering the room). Some dreams seem to incorporate the actual sensation of sleeping, such as dreams of suddenly fainting or falling asleep within a dream. And bodily states such as fever and thirst have long been associated with changes in dream quality, too, though the evidence is primarily anecdotal to date.
Even dreams of teeth falling out (a prevalent typical dream theme that almost 40 percent of us have experienced) are associated with the experience of dental irritation in the morning, suggesting that these common teeth dreams are a result of clenching or grinding teeth during the night. This provides an explanation for why so many people have this bizarre dream theme that bears little resemblance to waking life: It’s your dream’s way of making sense of the ongoing sensation.
In related examples, though the texture and resistance of objects and bodies in the dreamworld are often inconsistent with waking reality, their very source may be real physical sensations that are incorporated into an ongoing dream. For instance, dreamers experiencing the natural muscle paralysis that accompanies REM sleep sometimes feel unable to move their body freely within dreams; everything feels slower than normal. Other times the body feels very light and may start to float or fly, gravity seeming almost nonexistent. Indeed, falling is one of the most commonly reported typical dreams, and flying is a seemingly universal positive dream experience. The body’s position lying in bed, without any sense of vertical gravity, is quite unlike how the body feels when moving upright in waking life and may in part explain such frequent flying and falling themes (though other explanations are possible, too7).
In general, while real bodily sensations from paralysis to twitches continue to influence the sleeping brain and dreaming mind, these sensations are often different from normal waking life and could explain some of the more peculiar typical dream themes, like dreams of teeth falling out, being unable to find a toilet, or flying and falling.
* * *
Finally, in addition to processing bodily states, there is also continued awareness of the external environment during dreaming. Subjects in the lab report dreams of feeling the electrodes on their skin or hearing sounds from around the room. These sensations seem especially likely in dreams reported the first night of sleeping in the lab. The “first-night” effect describes how parts of the brain stay more awake while sleeping in a new environment such as the laboratory, resulting in a less restful sleep. Some scientists refer to this as a “night watch,” where the brain continues to monitor its environs while asleep. This effect dissipates after the first night in the lab; in fact, many sleep studies discard the first night of data because of the disruption to sleep.
Regardless of how comfortable we try to make the lab, unusual sensations like wires coiled around the body and electrodes pasted to the scalp make for an objectively strange sleep experience, and these sensations often filter into our subjects’ dreams. Even when people sleep at home, the sensory world often finds its way into dreams, albeit with a little less frequency. There are certain filters in place, gating mechanisms, designed to keep us from waking to habitual sensations, things like cars on the highway, talking neighbors, or the AC hum. In general, though, this filtering is incomplete, and there are gaps where sensory stimuli seep into dreams.
To better investigate just how and when the dreaming mind reacts to sensory stimuli during sleep, studies have been designed specifically to probe at these gaps. Experimenters have tried presenting all sorts of tactile, visual, auditory, and scent stimuli to sleeping subjects to observe their influence on dreams. In studies set in Montreal,8 a pressure cuff was inflated on subjects’ legs during REM sleep, leading one subject to report:
In my dream, the experimenter was there to wake me up. She turned on the lights and asked me about my dreams. I was answering her. I could feel the pressure pump on my leg. She asked me what does it feel like. I said it feels like a hug. She said, “Doesn’t it feel like someone pulling on your leg?”
In this example, the dreamer feels the pressure on their leg, and the incorporation is quite direct: The dream clearly displays the sensation and even references the ongoing experiment. Other dreamers reported sensations of tingling or discomfort in their leg, but within novel dream scenarios, like a dream of being unable to kick right while swimming. In some cases, the leg pressure penetrates dreams in a more elaborate way, at times even projecting onto another character, such as a horse trapping its leg. These cases show how sensory stimuli from the real world can be perceived while asleep and in different ways become part of an ongoing dream.
Studies of auditory and visual stimulation find similar results, where beeping sounds and flashing red lights frequently appear within dreams. Sometimes the resulting dream matches the stimulus directly—like an auditory cue generating sound imagery. But it can also be transformed, such as a sound triggering a rather abrupt movement in a dream. It seems that the transformation of sound into movement may be quite common; in one example, a clattering noise resulted in the dream of a clown suddenly springing into a somersault.9
It is fascinating how stimuli can change modality (e.g., from auditory to visual movement) or even “ownership” upon entering the dreamworld, projecting onto other characters and blending with the dream environment. Importantly, studies of sensory incorporation prove that real sensations are at times a source of dream content and that dreams are not only imagined sensations in the brain.
What has gone unsaid in the discussion so far is just how and why the dreaming mind accomplishes this feat, massaging and transforming perceptions into dream narratives, seemingly in real time. And what determines the broader content of dreams, where one person dreams of a horse hurting its leg while another dreams of a hug? It seems clear that sensory experience alone is not enough to construct dreams.
In the next section we will start to explore how the mind of each individual fills in the content matter of dreaming. In an illustrative case, one video of a patient with REM sleep behavior disorder shows a man who dreams of smoking his pulse oximeter (this is a small clip on the finger that measures blood flow): The sense of pressure on the finger leads to a dream of enjoying a cigarette, in this specific individual who has fond memories of smoking. The dream emerges both from the physical sensation and a personal association from memory. A more common example can be seen in the case of flying dreams: While this sensational dream is almost universal, the specifics vary from one person to another. Where one person who lives by the sea dreams of swimming like a dolphin, another person who reads comics dreams of jetting through the sky like Iron Man. These variations are based on personal fantasies and memories associated with flying.
In fact, as early as the 1800s scientists started to establish a sensory basis for dreaming10 and later began to unravel a formula,11 where dreams were seen to arise from a combination of factors including sensations in the body along with recent and long-term memories, habits, and concerns, too. In this view, how we perceive sensations during sleep depends not only on the stimulus itself but also on the mind and mental matters of the dreamer. Let’s see whether we can disentangle some of these factors now.
The mental matter of dreams
In order to unravel how real sensations can modulate dreams, one question to consider is whether dreaming is more akin to waking perception or imagination. In other words, are dreams like real perceptual experiences of the world or are they more like thoughts or imagined experiences about the world?
To get at this question, several studies have recruited lucid dreamers who can perform experimental tasks within their dreams, designed to test whether dreams are more like perception or imagination. In the realm of visual perception, we know that in waking, humans are capable of something called “smooth pursuit” of visual objects. If we follow a continuous moving target with our eyes, our eyes make smooth and continuous movements. But, if we try to imagine this movement with our eyes closed, it results in small jerky eye movements called “saccades.” To see whether dreaming is more similar to perception or imagination, scientists asked lucid dreamers to attempt smooth pursuit of visual objects in their dream. Lucid dreamers traced an infinity shape with their thumb and followed the movement with their eyes, which were being recorded via electrodes. The resulting electrooculogram revealed a smooth-pursuit pattern of eye movements as they continuously traced the infinity shape, suggesting dreamed vision is more akin to perception.12 In other words, the things you see in your dreams are registered in a way that is similar to how we see things in waking life rather than as figments of your imagination.
To further back up these findings, we can look at other forms of perception. It turns out that, just like vision, our perception of time is similar in dreaming and wakefulness. To investigate temporal perception in dreams, lucid dreamers were asked to indicate with eye movements the passing of ten-second blocks of time from within the dream (ten seconds, twenty seconds, thirty seconds), and their metrics were accurate to real time.13 This goes against the commonly held belief that large chunks of time pass within a single dream and suggests that our perception of time in lucid dreaming is relatively true to waking life. More recent landmark studies of lucid dreaming have shown that dreamers can even accurately understand speech—hence auditory perception—while asleep and are able to fully comprehend and respond to simple questions from within a dream (but more on that later). These studies amount to a view of dreaming that is different from what we once thought: Dreamed perceptions are experienced in a similar way as real perceptions in waking life.
Nevertheless, the unique neurobiology of sleep interferes with some perceptual processes. As we go about our waking lives, we are continuously processing and combining multiple streams of sensory information. This is called “binding,” where our brains collect features of the world and integrate them into the coherent objects we perceive. This allows us to experience a comprehensible world, rather than a series of disembodied shapes, colors, motions, and sizes.
In dreaming, things are a bit different. We are all familiar with confronting examples of imperfect or at least unusual bindings in our sleeping lives: Your best friend appears in your dream with glowing blue hair, or the book in your hand is suddenly a cell phone. In part, this is because there is a deactivation of parts of the brain important for sensibly integrating different strands of information (the inferior parietal cortices). Because your brain has partly lost some of its ability, this can lead to inappropriate or atypical object-feature bindings. Luckily, within dreaming we tend not to question these bizarre image pairings, and this is likely due to another unique pattern of brain activation during sleep: inhibited prefrontal brain activity. The prefrontal cortex is responsible for higher cognitive functions in the brain and so would play a part in recognizing when a collection of features that make up an object is, for lack of a better word, weird.
Although the combined features of dream objects can be bizarre, we nevertheless experience them as whole, as making sense within the dream, despite being unusual. So where do the features that bind into dream images come from?
In part, as we’ve explored, the dreaming brain draws from ongoing sensory experiences. But there’s definitely more to the story. Identifying the sources of dream imagery has been the focus of much research, and has been especially studied in microdreams—those short dream snippets that slip into mind as you drift off to sleep. Almost everyone has had the experience of nodding off while sitting in a bus or watching TV, and suddenly coming to as a bizarre thought or image erupts into mind. In our daily lives, and even in dream science, these little dreams are often disregarded. But carefully studying these microdreams has revealed that often these brief images are composed of three things: ongoing sensory stimuli, recent sensory impressions (a color or shape or touch observed), and broader memory stores for context, including memories from both the immediate and more distant past.
One technique for collecting microdreams is the “upright napping” procedure, basically allowing yourself to repeatedly nod off while sitting upright, like what happens when attempting to sleep on an airplane or in a bus. This is one technique that prominent dream scientist, and my personal mentor, Tore Nielsen, has relied on to unravel the makeup of microdreams.14 In one example, while at a conference and dozing off, Nielsen had the following microdream:
A heavy door made of wood suddenly swings open and slams against the corner of a countertop.
Just prior to this dream, Nielsen was observing a PowerPoint slide depicting a large, closed, brown wooden door; then, at the moment of sleep onset, the conference speaker made a thudding sound by hitting the microphone. Together, the sound and image merged into the door-slamming dream. But wait. How is it possible that the dream of a slamming door occurred at exactly the same time as the unexpected microphone thud? If we look to waking perception again, we find some clues to explain this phenomenon. Our waking brains continually modify our perceptions in peculiar ways, presenting us with a seemingly more coherent experience. As an example, if you see a video of two blocks hitting each other, and a sound is played slightly before this visual collision, your brain will actually delay perception of the sound in order to match the timing of the blocks hitting. In a more common example, if the audio on a TV program is slightly delayed compared to the picture, you will nevertheless perceive the speech as occurring in sync with the actors talking. In the “ventriloquist effect,” we perceive a dummy’s mouth producing a ventriloquist’s voice, meaning our brain shifts the perceived location and timing of a sound to match the visual scene.
This is known as multisensory integration, where our attention binds multiple streams of sensory information in a way that makes sense; this is at times an illusion, a trick of the mind that shifts and alters our perception to create a coherent image. In the microdream example above, perception of the “thud” was essentially paused until it could be explained by the new dream of a slamming door. In other microdream examples, a tennis ball “whop” led to a dream of an arm being slapped, or a sudden dip in airplane turbulence coincided with a dream of a passenger spilling a glass of wine. Our sleeping brain merges real sensations with recent perceptions (e.g., seeing an airplane passenger or a door) and broader memory, too, like the prior knowledge and experience that tells us that doors make a thudding sound when slamming.
This capacity is not limited to microdreams; it is also evident in studies of lucid dreams, when audio or visual stimuli become integrated into dream scenes. For instance, when I presented flashing red lights to lucid dreamers, one of my subjects reported that in the dream, “I could tell when the red light came on because it got hot and the sun got brighter.” In this case, the light cues were perceived as part of the dream image of the sun. And we see another level of integration here: When the sun gets brighter, the dream also becomes hotter. This is based on broader memory, on knowing that the brightness of sunlight is associated with warmth. In another example, when a series of beeping sounds were played while a subject slept, they later reported that in their dream, “I was shopping in a supermarket and I could hear the beeping, and it was like I was getting loads of messages on my phone telling me what to buy … things like, ‘buy some biscuits.’” Here, the beeping sound is perceived as coming from the phone, and it further creates actual semantic content based on prior knowledge that beeping phones usually contain messages. The brain is trying to make sense of its experience: “Where is this light from? What is this beeping? It must be a part of this world I’m dreaming.”
In this way, our memories and knowledge add context and form to the sensory features of dream imagery. This can be seen in sleep paralysis episodes, too, where our fearful experience of paralysis is projected into dreams of a malevolent presence, a threatening figure that is somehow responsible for our immobility. Our mind tries to make sense of the situation: “There must be some terrible reason I cannot move.” What’s more, the actual form that this shadow figure takes is unique to different cultures. In cultures that believe in witches or ghosts or aliens or devils, these are the figures that people perceive during sleep paralysis, as memories instilled by culture shape the content of their dreams. For instance, in the United States, it’s thought that many accounts of alien abduction could actually stem from sleep paralysis episodes; the mind tries to make sense of what’s causing the body’s paralysis, and in a country with more UFO sightings than anywhere else in the world, alien abduction seems most likely. In other cultures, like Egypt, for example, sleepers perceive the Jinn, which is a supernatural demon from Islamic mythology. In China it’s more common to perceive ghosts, whereas in Newfoundland there’s the Old Hag, a traveling spirit from folklore who sits on the sleeper’s chest. In all of these examples, the physical sensation of paralysis is projected into a form that makes sense, according to cultural memories and beliefs.15
In sum, even as we sleep our brains are constantly binding multiple streams of information in order to present us with a coherent experience—integrating sensations, together with recent memories and past experiences, and broader knowledge and expectations about the world, too. These are all combined sources of dream imagery.
* * *
As a final point, studies of dreaming also depend on one other factor, which is attention, specifically one’s ability to pay attention to, recall, and report their dreams. Everything we’ve learned about dreaming so far we know only from subjects’ reports; after all, we have no other way of recording dreams (though this may change in the future, as we’ll see later). For now, to study dreams we need subjects to first encode their dream experiences in the sleeping brain, then remember them after waking, and finally report them with some level of detail, capturing the dream’s emotions, sensory vividness, narrative structure, and more. All of these steps require skills of attention and memory.
From brain imaging studies we know that “high dream recallers”—those people who recall their dreams more often than most—have greater white matter density and brain activation in areas of the brain associated with attention and memory, such as the medial prefrontal cortex. And high dream recallers also perform better on tasks requiring visuospatial attention such as the “mirror-tracing” task, where you have to trace a figure on paper by observing its reflection in a mirror. Those who perform better on this task, and have better visuospatial skills, recall more of their dreams. Verbal fluency, a measure of how easily someone can produce coherent speech, is also key to reporting dream experience, and is correlated with dream recall. In other words, your ability to pay attention to, remember, and report on an experience factors into how much you can generate, recall, and report dreams.
Despite these individual differences and where you fall on the spectrum, training attention and memory can reliably increase almost anyone’s dream recall. This can be as simple as establishing the habit of recording dreams in the morning and setting the intention to remember dreams each night. Research has repeatedly shown that keeping a dream journal and cultivating positive attitudes toward dreaming increase dream recall.
The circumstances around an awakening also influence dream recall. Any distractions present on awakening will interfere with memory for a dream, so something as simple as remaining still with your eyes closed while recollecting a dream will increase how much you can remember. Subjects recall more dreams if immediately prompted for a report (rather than getting out of bed before reporting, for example). And while recall is highest in the morning, many people report spontaneous dream recall that seems to be triggered by events throughout the day—my dreams tend to resurface the moment I return to bed each night, as if the familiar proximity to sleep sparks the memory of a dream. Undoubtedly, many, if not most, of our dreams simply go unremembered.
This begs the question: Does increasing attention to dreams, such as by keeping a dream diary, simply reveal dreams that were already occurring but usually forgotten? Or do practices for improving dream recall actually change the content of our dreams, generating more vivid and perceptual dreams as a result? To start to answer this question we can look to the case of “white dreams,” where you have the feeling of having dreamed but cannot recall any specific content. Up to one third of awakenings are associated with white dreams and many people assume these are cases where they had a full dream but forgot it by the time they woke up. But there is another possibility: White dreams could be minimal dreams that lack much perceptual detail to remember. An example would be a dream of occasional thoughts or vague images going through the mind or a sense of time passing without any further story. Waking from these dreams would leave the feeling that we were experiencing something, but without remembering much content. This resembles what’s called a “weak glimpse” in waking perception: When subjects are shown visual objects that are almost completely masked by a filter, they often report the experience of having seen an object but cannot identify or remember what it is. They know they experienced something, but their perception was too minimal to comprehend or recall.
Studies of waking perception can help us to interpret white dreams and minimal dreams even further. In waking life, our perceptual experiences vary dramatically in how vivid, clear, and stable they are. To illustrate, if we observe a tree from a short distance on a clear day, the tree is clearly visible, solid, and unmoving. But when we view a tree outside a rain-splashed window in a moving car, the tree becomes blurry and fleeting due to the overlapping images and motion. As anyone who has ever dreamed can attest, these perceptual qualities diverge dramatically in dreaming. For instance, some dreams encompass our entire visual field and are saturated with color and brightness, whereas other dreams consist of small, faded images lacking focus or color or luminance. Dreams can also be markedly discontinuous, with images that change fluidly over time, and dreamed perception can lack clarity: We may see something that looks like a fox or a wolf, or is it a dog or some other animal? At the lowest end of the perceptual spectrum are dream images that lack any vividness, clarity, or stability. This could be the white dream, a minimal dream that lacks much detail to recall.
We also know that minimal dreams can occur in the form of sleep misperception, which is when subjects mistakenly feel like they have been awake despite actually being asleep. Subjects sometimes have the impression that they were lying awake in bed thinking when really they were dreaming of lying in bed thinking. This occurs more often for people with insomnia, who have restless sleep and are more likely to feel awake even while sleeping. We’ll return to these cases when we uncover how dreaming contributes to different sleep disorders and how dream interventions can help. (For instance, sleep misperception almost never occurs when subjects awaken from vivid perceptual dreams, which leave the strong impression of having slept).
To some extent, then, it seems like learning to pay attention to dreams not only increases their recall but also enhances the richness of dream content and improves sleep perception as well (and we will see later just how relevant this is to sleep medicine).
* * *
Altogether, dreaming is a form of perceptual experience that draws on memory and sensation to deliver a coherent, if bizarre, experience of the world—one that varies in intensity, relies on attention, and follows certain purposeful designs. In dreams we inhabit avatar bodies and are immersed in simulated realities: rehearsing skills and building social bonds, navigating the labyrinth of the mind. Our prior knowledge and memories fill in these scaffolds—the functional designs of dreams. Over time and experience, dreams serve as a fertile playground for developing skillful means of being in the world, and we’ll learn more about how dreams function in the next chapter.
On a broader scale, one boon of modern dream science is that dreams can be collected from many different laboratories and studies, so we can collectively examine their characteristics and patterns. This has led to a consensus that dreams are novel creations, constellations of related sensory and memory bits from waking life. Dreams sample from and build around our experiences of interacting with, performing in, and exploring the waking world. And yet, dreams are also unlike waking life in consistent ways, with unusual typical themes that reoccur across cultures and eras, perhaps in part having their basis in our kindred human bodies. Over time, we each grow accustomed to our own brew of bizarre and implausible dream themes (like flying or meeting the deceased) and realize in relating to others that these unlikeliest of scenarios are in fact nearly universal.
Understanding dreams as products of our bodies and memories is the first step toward an approach for designing dreams and nightmares and using them to achieve certain outcomes—the ultimate goal of a dream engineer such as myself. But we’ll need to explore a little further to see how this can be accomplished.
CHAPTER 2THE DREAMING BRAIN
Going into graduate school, I landed a research gig with Tore Nielsen at the Dream and Nightmare Laboratory in Montreal. Most of our research used polysomnography (PSG) to learn about how the electrical activity of the brain and body corresponds with dreaming, and we designed studies to uncover the types of memories and emotions most often experienced in dreams. In laboratory studies, one of the primary methods used to explore these links is to give participants simple learning or emotional tasks before sleeping—pared-down forms of real-world learning, like memorizing vocabulary or watching stressful videos—that we can then witness sprouting into dream content.
Today, science has established that almost every waking function is actively improved by a night of sleep and that sleep is especially important for our cognitive health and emotional well-being. The question of whether dreams are connected to these functions, however, is still an open one. At the very least, it is clear that the content of dreams is related to memories and emotions experienced during the day and that dreams then impact our mood and cognition the following day. These findings provide clues to where dreaming intersects with sleep and how both sustain healthy minds in waking life.
The architecture of sleep
On a basic level, our brains cycle through four stages of sleep, each characterized by unique patterns of brain activity. Faster brain waves are prominent in more alert and awake states, and slower brain waves are indicative of deeper sleep or even when we feel tired or sluggish in the day. As we move from wakefulness through the stages of sleep, our brain activity gradually becomes slower and more synchronous:
In the sleep lab, when participants first get into bed, we ask them to lie still so we can collect resting brain wave activity. When someone is in a relaxed state with their eyes closed, we can see fast brain waves in the alpha range (8–12 cycles per second), especially at the back of the brain, in the visual cortex, where closing the eyes is essentially turning off the lights in the mind.Stage 1 is a light sleep experienced in the first seconds or minutes as we fall asleep, with a mixture of wake-like alpha waves (8–12 cycles per second) and slower theta waves (5–7 cycles per second). In stage 1 sleep there are also slow rolling eye movements and occasional muscle twitches as the body shifts into sleep.Stage 2 is a slightly deeper sleep with primarily theta waves and short bursts of high-frequency activity called “spindles” (12–16 cycles per second), along with occasional slower delta waves that start to appear (0.5–4 cycles per second).Stage 3 is the deepest sleep and is marked by an abundance of slow and large-amplitude delta waves (0.5–4 cycles per second) and is also referred to as slow wave sleep.Stages 1, 2, and 3 are collectively considered non-REM (NREM) sleep. Over the course of the night, we cycle through these stages in a descending and then ascending order. From stage 1 at sleep onset, parts of the brain begin slowing down, then into stage 2 we see mostly theta and occasional delta waves, then stage 3 sleep has large delta waves throughout the brain; from this deepest stage of sleep we then shift back up to stage 2, a bit of a lighter sleep. Curiously, subjects in the sleep lab often wake up briefly at this point, at the end of their ascending stage 2 period, and they might shift body positions and mumble, before falling back into the fourth stage of sleep, the more wake-like rapid eye movement (REM) sleep:
REM sleep is a period marked by high-frequency brain activity mixed with theta waves. There are two types of REM sleep: phasic and tonic REM sleep. Phasic REM sleep is accompanied by rapid eye movements, muscle twitches, and heart rate variability, whereas tonic REM sleep is relatively quiescent, and both are accompanied by the now familiar body paralysis. Similarities between stage 1 and REM sleep, including brain activity, eye movements, and muscle movements, have led some authors to theorize that stage 1 is more akin to REM than to NREM sleep,1 but this is still controversial.The whole progression, of descending into deep sleep and ascending back up through REM sleep takes about ninety minutes and is called an “ultradian cycle.” Graphically it’s like a wave, as the brain fluctuates to deeper and then lighter stages of sleep. At the beginning of the night, this ninety-minute cycle leans more heavily into slow rhythms, having more stage 3 sleep and less REM sleep; later in the morning, the ninety-minute cycle favors faster rhythms, having more pronounced and longer REM periods but little stage 3 sleep (almost like a rolling tide across the night).
This shift over the night occurs because the ninety-minute cycle intersects with another rhythm, the twenty-four-hour cycle—or circadian rhythm. Basically, as human beings (and similar to most biological life), we are entrained to the twenty-four-hour alternation between day and night. Diurnal animals, like humans, as opposed to nocturnal ones, are driven by our evolutionary biology to stay awake when exposed to sunlight, to eat and to act and accomplish all of our goals in the external world during this period, and then to rest and undergo the many necessary functions of sleep during the night. Everything from photoreceptors in our eyes, to how our metabolism syncs with mealtimes, works like a clock to time our bodies to this twenty-four-hour cycle. Sleep, too, follows a circadian rhythm: Our brains fluctuate in levels of alertness across the day, and at night they sink into slower activity and deeper sleep. Early in the night we have more deep sleep, with larger and more abundant slow waves throughout the entire brain, and we stay in deep sleep for long periods of time, ascending only momentarily up into REM. As the night goes on, though, our sleep becomes lighter overall. We dip only slightly into deep sleep, before surfacing to REM sleep for long, pronounced periods: active, dreaming, ready to awaken into the day.
* * *
There are numerous associations between the rhythms of the sleeping brain and the qualities of dreaming. Both the frequency of recalling dreams and the content of dreams are modulated by these cycles across the night: the approximate ninety-minute ultradian rhythm, as well as the twenty-four-hour circadian rhythm.
While dreaming can be reported from any sleep stage and at any time of night, morning REM sleep has come to be known as the stage associated with the most vivid and elaborate dreams. The first clues that REM sleep was more conducive to dreaming, or at least to dream recall, came with the discovery by Eugene Aserinsky and Nathaniel Kleitman in 1953, that 74 percent of participants recalled dreams from REM sleep, whereas only 17 percent recalled dreams when awakened from NREM sleep.2 These results were reinforced in 1957, when Kleitman and new student William Dement (who would later go on to be a prominent figure in dream science) again found that far more subjects recalled dreams from REM sleep (80 percent), and almost none from NREM sleep (only 6.9 percent).3 These early studies established the belief that dreaming was a REM sleep phenomenon. In fact, for several years research continued to show similar findings, with high levels of REM dreaming and very little dream recall from NREM sleep. This assumption became so ingrained that REM sleep became referred to by some as Dreaming-sleep or D-sleep.
Initially, scientists presumed that the few dream reports obtained from NREM sleep were just memories of prior REM dreams that had occurred earlier in the night. Many believed dreaming was not physically possible in NREM sleep at all. The brain in deep sleep was thought to be incapable of producing any form of cognition; it must be unconscious. This belief was dispelled by researcher David Foulkes in 1962, who conducted a study collecting dreams from across the night and revealed that dreams could be reported even from an NREM period early in the night, prior to any REM sleep.4 It’s quite remarkable how recent, and even pervasive to this day, this belief can be, viewing most of sleep (or NREM sleep at least) as a form of unconsciousness, incapable of meaningful cognition.
Today, dreams are still more frequently recalled from REM sleep (80 percent of the time on average), suggesting that REM sleep physiology strongly enables dreaming and the necessary cognitive capacity for dream recall. However, there has been a steady increase in the amount of dreams recalled from NREM sleep over time. In part, this has been due to changing definitions of dreaming: The original concept of dreaming was reserved for the more hallucinatory and storylike dreams we often think of and see in films, rather than simpler types of mental content that can and do occur during other sleep stages. Nowadays, when we rouse a subject from sleep, we ask broadly, “Can you tell me what was going through your mind before I called you?” The open wording invites more frequent reports than the question used in early studies, “What were you dreaming about?” In fact, as early as 1962, Foulkes discovered that modifying the question in this way resulted in 70 percent of subjects recalling some content from NREM sleep. Although criticized at first, similar findings by other researchers have since confirmed that dream recall is present in a majority of subjects, from all sleep stages of the night, if sampled appropriately. Thus, it is now widely accepted that dreams, defined broadly as any form of mental content in sleep, can and do occur in any stage of sleep and are not exclusive to REM sleep.
This is quite remarkable. Although most of us have the illusion that dreams are merely fleeting things that occur in the mornings when we awaken, the reality is much more impressive, that we are thinking, dreaming, and feeling across the night. I remember participating in a sleep study where I was awakened twelve times in one night to report my dreams. It was astonishing to realize just how much I was experiencing, how far my mind was traveling as I slept for those eight hours. Of course, we forget most of these experiences come morning and are left with the misperception that dreaming is reserved for only those last few moments of sleep. In actuality, dreaming can occur across the night, though, as we’ll see next, its contents change along the way. Later on we’ll explore further what this means: If dreams do occur throughout the night, how could this impact our sleep and mental health, even if we don’t remember them? And what does this mean for those of us with nightmares or other forms of disturbing dreams?
* * *
To return to some more of the basics: Across all stages of sleep, the content of dreams is largely created by the cortex, where we store the many memories of our personal lives: Lovers become dream characters, childhood homes become dream settings. Sensory cortices give rise to vivid perceptual details, especially visual and auditory imagery, though dreams can also contain smells and tastes (usually to a much lesser extent). And as we learned earlier, brain imaging studies show that the motor cortex is active during dreamed actions, similar to waking actions.
While these correlates of dream imagery seem constant across sleep stages, it is also the case that there are reliable differences in the content of dreams collected from different stages of sleep, which likely correspond to changes in the brain across each stage:
Stage 1 dreams are brief and occur in transitory states as you fall asleep or wake up, termed the hypnagogic and hypnopompic states. If collected systematically, up to 75 percent of stage 1 sleep periods contain some kind of imagery. While often occurring in mere seconds, stage 1 dreams can have vivid visual or visceral content. These dreams often incorporate real-world stimuli, such as an alarm clock or a meowing cat.Stage 2 dreams are reported in 60 to 70 percent of awakenings. While they are often immersive and storylike, they are also consistently shorter, having fewer scenes and shorter word length than REM dream reports. They also have less perceptual and emotional content, and fewer characters, places, and actions than REM dreams, and are less bizarre, include more logical thought similar to waking, and regularly feature events from recent waking life. However, stage 2 dreams also become longer, more vivid, and more narrative over the course of the night, and by morning some are even indistinguishable from typical REM dreams.Stage 3 dreams are recalled around 60 percent of the time, and reports can be very short, consisting of simple perceptual or emotional qualities without narrative. Stage 3 sleep can also include minimal forms of dreaming, such as having an awareness of being asleep without any content, or simply thinking during sleep. In the 40 percent of awakenings where no dreams are recalled, subjects sometimes report feeling as if they haven’t slept at all, or they have a gap in consciousness since falling asleep. As a researcher, it’s quite fascinating to rouse a participant from a long, deep sleep and have them report feeling they haven’t slept a wink. This line of research is important because of its clinical implications: The feeling of sleeping deeply is key to how satisfied we feel with our sleep in the morning.REM sleep dreams are widely understood to be the most immersive, bizarre, emotional, and narrative of dreams—the most dreamlike, for lack of a better term. Given that REM is more predominant in the morning, these are the dreams people most often remember in day-to-day life. REM dreams also commonly reference recent waking life concerns, though often through metaphorical links to more distant memories, which distinguishes them from stage 2 dreams. The sensory and motor cortices are more active in REM than in NREM sleep, giving rise to the more vivid and embodied dreams characteristic of this stage. The limbic brain—including the amygdala, which is responsible for emotions in waking life—is also more active and may contribute to the especially emotional content of REM dreams, and the occurrence of nightmares, too.Transitions between these sleep stages are fluid, ebbing and flowing from light to deep sleep and back again. Brain activity does not abruptly change from one sleep stage to the next—it shifts gradually, including some periods where the boundaries between sleep stages begin to blur. This is actually a huge hurdle when trying to categorize or score someone’s sleep. The American Academy of Sleep Medicine established specific rules for how and when to classify sleep stages, and their guidelines try to account for every messy possibility: how to score sleep when someone is in stage 2 but then starts to show signs of REM activity, like an eye movement or muscle paralysis; what to do when someone in REM sleep shows signs of wakefulness, like fifteen seconds of alpha activity in the occipital cortex. These rules were first established as early as 1968, back when polysomnographic recordings were printed in real time with ink on paper, and technicians would observe each thirty-second page as an “epoch” to determine a subject’s sleep stage. To date, we still follow nearly the same rules and stick to the thirty-second window out of tradition (though new technology and algorithms are beginning to change this).
One job of dream research is to map how dreaming varies (how emotional or bizarre dreams are, for example) across the ninety-minute sleep cycle. For instance, although on average NREM dream reports are shorter than REM dreams, we can observe that NREM dreams actually become longer, more vivid, and more filled with content when collected at the end of an NREM period, closer to an oncoming REM period. In fact, the brain itself becomes more REM-like as it approaches REM sleep, with more fast-wave activity and more dreamlike dreams. Along similar lines, dream reports become more dreamlike with increasing time in REM sleep; they are described as more active, emotional, vivid, narrative, and immersive, the longer someone has been in REM sleep.
On another scale, dreaming evolves over the course of a night. For instance, the number of dreams and length of dream reports increase across the night, and especially NREM dreams become longer with each successive sleep cycle. This is likely due to the circadian rhythm and the fact that sleep overall becomes lighter, with REM sleep more abundant in the morning. Other qualities that amplify across the night include perceptual details, vividness, bizarreness, and emotional intensity of both NREM and REM dreams. In other words, dreams become more dreamlike as the night goes on, with more realistic, immersive, and socially interactive dreams occurring in later sleep cycles and in the morning. Morning naps are especially conducive to dream research: In a simple one-hour morning nap, we found very high dream recall rates for both NREM and REM sleep (89 percent and 96 percent, respectively), with vivid sensory content in each.
Altogether, the frequency and content of dreaming fluctuates in sync with the oscillating ninety-minute sleep cycle and amplifies across a night of sleep, peaking in frequency and intensity in REM periods and in the morning.
* * *
So far, we have limited our exploration of dreaming exclusively to the sleeping portion of our lives, considering only how dreaming varies during the night. However, advances in technology now allow us to record a much vaster array of data than previously possible. In particular, modern high-density EEGs (electroencephalograms) allow us to monitor the brain in high resolution, to fine-tune our understanding of what is happening in specific areas of the brain at specific times. In this way, dream research has shown something unexpected: Dreaming processes may not be exclusive to sleep.
While traditional sleep scoring uses 6 electrodes placed on the frontal, central, and posterior regions of the scalp, modern high-density EEG can record up to 256 different electrodes at a time. This has led to the discovery of “local sleep,” where slow delta waves—those waves that are typical of a brain in deep stage 3 sleep—appear over one region of the brain even when the rest of the brain is awake. This means that dreaming can happen not only in any sleep stage but also during waking mind-wandering episodes, or what we might more commonly recognize as daydreaming.
Instances of local sleep usually occur when slow waves start to appear over the medial prefrontal cortex. These slow waves are indicative of essentially falling asleep in parts of the brain while remaining awake in others. In this case, the part falling asleep is the part associated with metacognition, that is, the ability to direct and control our thoughts. When we are fully awake, we enjoy a certain level of metacognition, but as we let our minds wander, we start to lose this ability; our thinking becomes less focused and directed. Of course, as slow waves become even more prominent as we drift into sleep, the metacognition available in dreams decreases even further.
To illustrate this progression: As we move from waking thought to daydreams, our thoughts might shift from arranging plans for the day to reminiscing or fantasizing about a romantic date. The former requires more focused attention while the latter is more imaginative, though we retain the capacity to refocus attention when desired. As we travel into sleep, in stage 2 dreams, for instance, we become more immersed in imagery, but dreams still feature some instances of “thinking.” An example might include a dream of playing hide-and-seek, where the dreamer is immersed but also thinking about where to hide and what to do next. REM dreams have the least metacognition and display the most active and involved imagery: The dreamer does not reflect too much on their experience (e.g., a microphone materialized so I started singing; an animal chased me so I ran away). REM dreams seem to be more about “doing” than “thinking,” about acting without reflecting much in response to the demands of the dream.5 This temporary release from metacognition is in part what allows us to become fully absorbed into the universe of dreaming, but it’s also why we get carried away into nightmares, as we’ll see later.
Overall, it seems like dreaming spans a continuum from waking thought through sleeping rather than being relegated to one state or another. In this view, dreaming is like an intensified form of mind wandering,6 and both are derived from the same neural substrates. In fact, our lab studies have shown that dream qualities like emotion, bizarreness, and sensory vividness vary from waking imagery to dreaming, with daydreams being the least vivid, REM dreams the most vivid, and NREM dreams somewhere in between. These patterns provide a window into processes occurring across sleep stages and the whole twenty-four-hour cycle. But we’ll have to look a bit deeper at the sources of dream content in order to understand why these variations occur.
Dreaming and memory
When I decided to go to graduate school at the University of Montreal, a not-insignificant obstacle was that all of my coursework would be in French. With naive confidence I decided to take an immersive French course in the summer prior to starting my degree. Over the weeks, I noticed that several of my dream characters began to speak in French, some brokenly, others seemingly fluently. I could observe the wheels of language slowly reinventing my social dreamscape.
Unbeknownst to me, I was living through the design of a classic study done by Canadian dream scientist Joseph De Koninck, where immersion in a French language course was gradually associated with dreaming in the new language and with improvement in that language, too.
In the realm of mental function, we now know that sleep is especially important for learning, for both strengthening individual memory traces and supporting long-term memory networks. It should come as no surprise, then, that dream researchers have for some time been investigating whether dreaming, too, unlocks a door to learning.
Before exploring further, first, a quick lesson in memory. It’s important to know that memory can be broken up into different types. Some memories can be consciously retrieved, such as “semantic” memory, which includes all sorts of general knowledge, like knowing the sky is blue and remembering the name of your favorite song. “Episodic” memory refers to memory for specific events, such as remembering your sixteenth birthday party or what you had for breakfast last Monday. As you might guess, many of our episodic memories get lost over time, dissolved into more general semantic memory. To illustrate, though I can’t remember exactly what I ate for breakfast last Monday (an episodic memory), I do know that typically on weekdays I eat toast with peanut butter before work (a semantic memory).
Both episodic and semantic memories, those that can be consciously retrieved, are called “declarative” memories more broadly. This is in contrast to those memories that are more unconsciously learned and recalled, termed “nondeclarative” memories. This includes things like learning to ride a bike (i.e., muscle memory), and implicit associations or involuntary recall, like when the smell of the ocean elicits the memory of a childhood vacation. With nondeclarative memory, we are not consciously aware of encoding or retrieving a memory—it seems to occur more automatically.
Without question, sleep plays a crucial role in all these types of learning, shaping how our brains store and retrieve all kinds of memories. Early sleep researchers thought that NREM and REM sleep were separately responsible for declarative and nondeclarative memory storage. In support of this, NREM sleep seemed to correlate most with improvements in declarative tasks: memorizing word pairs, recognizing face stimuli, and learning new vocabulary. REM sleep was associated more with nondeclarative memory: improving performance on implicit motor tasks, maze learning, and other visuospatial tasks. The case of semantic memory was a bit more ambiguous, as some research showed that REM sleep was important for improving in a second language, whereas NREM sleep was linked to vocabulary learning.
Over time, more complex learning tasks were found to depend on multiple sleep stages, and multiple memory types, too, so this model of attributing specific memory types to specific sleep stages was left behind. In its place, a “sequential” model emerged, where memory was thought to be strengthened over consecutive NREM and REM sleep periods, with both stages reworking different parts of a given memory. This is now widely thought to be a more accurate view of how sleep reinforces learning, with learning dependent on both NREM and REM sleep together and the succession of these stages overnight.
In sequential models, NREM sleep, which occurs earlier in the sleep cycle, is thought to be important for stabilizing recent memory traces, done especially early in the night when all the details of the day are fresh in mind. During NREM sleep, there are short bursts of fast activity launching through the brain, generated deep in the brain where memories are first encoded, and broadcasted out to more surface-level cortical neurons. Scientists think that a form of reactivation or replay of memory is occurring, that the brain is shuttling recent memory traces from their hippocampal origins, where they are first encoded, into distributed cortical regions for storage. In animal studies, we can see precise firing patterns when waking life memories are replayed exactly during NREM sleep (recorded one synapse at a time with tiny instruments), giving rise to the term “memory reactivation.” These reactivated memory traces are slated for long-term storage.
But not all memories undergo this stabilization. In fact, another core function of NREM sleep is in erasing irrelevant memory traces. Our brains accrue an enormous amount of new neural connections each day, presenting an increasing burden to the brain’s resources that is not sustainable in the long term. One function of slow wave sleep is to prune through these numerous newly encoded memories and abolish all the weak, unimportant traces to make space for new ones. In essence, each night during slow wave sleep we undergo a system reset, refreshing synaptic availability in the brain and erasing quite a bit of our waking life minutiae in the process. Taken together, NREM sleep helps to preserve important memory traces and save them from erasure on a nightly basis.
So what do we know about the role of dreaming in this process?
To study memory in dreams, we often give subjects specific learning tasks prior to sleep and then try to detect elements of the task in subsequent dream content. One early and influential study, conducted by Harvard sleep researcher Robert Stickgold in 2000, required subjects to play hours of Tetris during the day, which led to around 60 percent of subjects seeing those falling blocks behind closed eyes at sleep onset (even in patients with amnesia, who had no declarative recall of playing the game).7 As one subject reported, there were jumbled images of “tetris shapes floating around in my head.”
The constant visual and spatial attention required for Tetris makes it a perfect task for getting incorporated into dreams, better than simple cognitive tasks like word-pair learning, for instance (or at least, more easily identifiable for dream research). A range of studies have since used similarly engaging tasks to go one step farther: to demonstrate that dreaming about a task is also linked to learning, to better memory for the task following sleep.
One of the first studies to show this effect, by the same Robert Stickgold and Erin Wamsley, had subjects navigate a virtual maze before and after a daytime nap. In the first place, the simple fact of sleep improved subjects’ performance. But those who dreamed of the task improved tenfold more than others. A follow-up study extended these findings to a full night of sleep, and this time, performance on the maze was motivated by the promise of a monetary reward—known to augment the benefits of sleep on learning, since money “tags” the memory as important (we’ll delve into this more in a bit). To make use of the whole night, participants were prompted to give multiple dream reports: first before falling asleep, then several at sleep onset, a few in NREM sleep, and one in the morning.
In total, 71 percent of participants reported dreams that contained some reference to the maze. Some participants mentally rehearsed the task while awake, spurred on by the idea that they would get money if they solved the maze. This waking mental rehearsal closely resembled the task itself, whereas at sleep onset, things start to get funky: “I can see the maze, and I think if we could swim above it, we could see everything.” As sleep goes on, the dreams become less like the task, but some are still clearly related. A subject in stage 2 dreamed of “standing in the middle of a maze, waiting for my friend to find me. She just kept going around and around, calling my name.” And finally in morning REM sleep someone reported “walking through [a maze] but it was like a formal maze, one of those outdoor ones made out of hedges and bushes.”
Notably, these dreams do not resemble an exact replay of the task or even an attempt to solve the task. Even so, those who dreamed of the maze at least once improved more than those who didn’t (and more than those who mentally rehearsed while awake), reducing their time and distance to get through the maze and solving it more efficiently.8
Similar studies set in Montreal have also found dreaming linked to better learning. In one study, subjects who dreamed of a virtual-reality flying task—dreams with visual or kinesthetic content related to flying—improved more on the task than those who didn’t. The same was true for a Wii Fit balance task, where not only task-related dreams but also lab-related dreams (dreams of experimenters, the lab setting, or equipment) correlated with better performance after sleep.
In Rochester, too, we found that lab-related dreams were linked to better language learning. Already earlier studies had shown that dreams were tied to language learning: The more knowledge one has of a new language, the more that language appears in dreams, and the faster one progresses in learning a language, the sooner the language appears in dreams. In our study, we had subjects study video clips for fifty American Sign Language signs before a nap. Sign language, of course, is unique because it relies more on motor and visuospatial memory than other languages, using hand movements and facial expressions to communicate. We found that 50 percent of our subjects had dreams about the lab, and these subjects had better recognition for signs after sleep. We think that lab-related dreams reflect memory processing of the entire presleep experience, including the learning task itself.9
Time and again, these studies show the effectiveness of “offline” learning during sleep and dreaming. But it’s hard to know whether dreaming is actively contributing to strengthening memories during sleep or whether dreams simply reflect this function of sleep.
To this end, some studies have asked lucid dreamers, who can control their dream content to some extent, to intentionally practice a learning task in their dream. In one such study, subjects managed to practice throwing darts in a lucid dream.10 Some completed as many as thirty throws in a dream, while others faced many obstacles: One subject started using a mirror as a dartboard but got distracted trying to change it to a better target; in another dream, the dartboard was projected onto the head of a woman, deterring the dreamer from practicing; in yet another, dream figures demanded money from the dreamer to purchase more darts. In the end, only those lucid dreamers who were successful in practicing in their dream improved in darts throwing in the morning. Those who were unsuccessful threw darts with a worse skill level than they had before sleep. In this case, the actual content of the dream seems relevant to the impact of sleep and suggests that the dream itself could be critical to learning.
Overall, there is now substantial evidence for a link between dreaming and learning, though there are several interpretations of these findings. The most conservative view is that task-related dreams are simply a result of underlying differences between participants. Certain individuals might have better cognitive skills or more motivation, for example, and so they dream more about a task and improve more as well. A second—and not mutually exclusive—interpretation is that dreaming reflects or actively contributes to learning. Lucid dreaming studies seem to provide some evidence that dream content itself can boost learning. This could function in a manner similar to waking visualization, which is often used by athletes or musicians to prepare for and enhance their performance. Dreaming in this case would be a more immersive and embodied form of mental rehearsal, and potentially more helpful for learning.
* * *
Coming back to the sequence of sleep stages described earlier, a first step of selecting and storing recent memory traces seems to occur primarily during NREM sleep. A second step, during REM sleep, seems to involve augmenting and strengthening connections between memory traces. One unique feature of REM sleep is its hyper-associative nature: REM sleep provides access to multitudes of memory traces in the brain, looking far and wide into our autobiography to relate each new experience to many similar things we have experienced before. “Ah, this argument I had with my boss about filing reminds me of keeping inventory in my job as a cashier and also feels similar to when I got scolded by my teacher, and oh, the files are the color of old October leaves, and the slamming cabinet is like a slamming door, a slamming phone, a shouting stop.”
Why would REM sleep be hyper-associative like this? Well, one reason is that we need to form links between experiences to better integrate each new experience into memory, to reveal patterns in the world that will help us going forward. This seems to be a key function of REM sleep, to integrate new experiences into memory.
In waking life our experiences are first encoded as specific episodes that occur in a certain place and time (yesterday, I ate a grilled cheese at the Blue Wolf café). If this memory is selected for long-term storage, as part of the transfer process it becomes distributed—the spatial and temporal context (Blue Wolf café, yesterday) separates from other features of the memory (a warm sunny day, a conversation with the barista, a productive writing session). All of these specifics become distributed and over time become harder to retrieve and piece together into a single episodic memory. Now I can still recall going to the café, but I don’t remember exactly what I ate, or the conversation topic, or the focus of my work that day. In fact, most episodes in memory become degraded like this with time. Only highly emotional events, or those we repeatedly recall, stay bound to their context and persist as episodic memories. This will become particularly relevant when we explore recurring nightmares later.
In part it seems like REM sleep is responsible for distributing pieces of memory and recombining them with other similar memories from the past. In a sense, REM sleep ties together fragments of overlapping memories. This leads to more generalized, semantic, or even implicit memory with time—basically, what’s left of the overlap between several experiences. For instance, from all of the cafés I’ve visited, I now have a sense of what I like and what to look for when I go out into the city to work. In the long term, this generalization of memory is adaptive; it allows us to retain vast amounts of knowledge, holding on to the gist of things and transforming stacks of individual episodes into a holistic understanding of the world.
How does REM sleep make these links? Well, associations between memories are formed by what’s known as a Hebbian process: When two items in memory are active at the same time, the association between them is strengthened. Some items are very often activated together (like the words “bread-butter-knife”) and so they become closely associated. In waking life, this helps us to function more efficiently and decreases the brain’s processing time: If someone asks me for bread and butter, I automatically hand them a knife. Of course, there are other related concepts that could be relevant, depending on the context (e.g., “bread” as a term for money). In waking experiments, we know that subjects process more closely related concepts more quickly, and they take longer to process more distantly related concepts (differences in the range of milliseconds, mind you). What’s interesting is that it’s precisely these more distant or weak associations that are uniquely accessed and strengthened in REM sleep.
Before getting into the function of this hyper-associativity, let’s look at some behavioral evidence. I mentioned that words are processed more quickly when they are strongly related—this is called “priming”: “Bread” primes “butter,” “hot” primes “cold.” Typically, this priming effect is fastest for closely related words and then slows with increasing semantic distance. However, when subjects are awakened from REM sleep, they actually process more distantly related words faster than closely related words. This is highly unusual, and it is not the case for NREM awakenings, when subjects respond more like they do during wakefulness (faster access for closely related words). In another example, when subjects were required to find a link between three distantly related words (e.g., “falling,” “actor,” “dust”), they were better able to find the solution following awakenings from REM sleep rather than from NREM sleep or wake (Answer: “star”).11 As you can guess, this type of hyper-associativity is highly useful in creative pursuits, where we often need to think outside the box to find novel links between less obviously related concepts. It seems like REM sleep (and stage 1 sleep, but more on that later) is essential to uncovering and holding on to these distant links.
How does this happen? Well, the brain in REM sleep has high levels of activation all throughout the cortex, providing access to more (and more remote) memories, along with all of the interwoven connections between them. Dream scientists believe that the many unusual connections accessed by the brain in REM sleep can “help explain the bizarre and hyper-associative nature of REM-sleep dreaming.”12 Dreams often give rise to strangely associated images (my cat as a bat; the beach made of Play-Doh) and jump and transform from one scene to the next (my house turned into my office and then my old high school). But even across these seeming discontinuities, dreams seem to follow rules of semantic organization, flowing through loosely related concepts and bringing them together. Importantly, the bizarreness of dreams seems based in meaningful associations (something about my cat reminds me of a bat). The concepts seem to be measurably interrelated in some way; the connections are just a bit more obtuse than usual.
There may even be more to the story. There is some evidence that during REM sleep the brain actually inhibits access to closely associated memory items, in essence avoiding obvious links between memories, those that are more predominant in waking life. A similar possibility is that the brain turns off Hebbian weighting altogether, treating all connections between memories as equal so there is no longer any sense of strong or weak associations. If this is the case, it would allow a free flow of activation throughout memory networks, in any which way without preference. These possibilities support two important functions: (1) increasing access to distantly related concepts that could be useful or insightful in the future (even if not typical of daily life) and (2) making sure that memory networks remain flexible and able to evolve over time.
The latter is important because the brain needs to reorganize even core memory patterns from time to time. Think about the last time you had a big life change: Maybe you moved to a new city and needed to establish a whole different routine from one week to the next. Your brain, pretty efficiently, unlearns the old habits and forms new connections in its place. Overall, this is part of a key function of REM sleep: to integrate new information into memory, whether highly unusual or more of the same, and to update the programs running in waking life.
Dreams can offer insight into how this process is accomplished, how yesterday’s walk through the park becomes merged with like-minded strolls a week ago, a month ago, a year ago, or even more.
First off, dreams very frequently incorporate memories of the prior day into their content, what Freud aptly termed the “day residue.” Generations of research and common knowledge have since confirmed the phenomenon, with day residues present in 60 to 75 percent of dreams. More generally, in over 80 percent of dreams we can identify at least one of a dream’s “memory sources”—a waking life event from either the recent or remote past or even the anticipated future that can be detected in dreams.
To more fully explore when memories appear in dreams, scientists can ask a subject to identify a dream’s memory sources and to provide an approximate date for when the memory is from. In a study in Montreal, one subject was awakened thirty-one times during the night, repeatedly in stage 1 sleep just at the precipice of dreaming.13 The subject reported dreams at each awakening and was able to identify two or three memory sources for each fleeting image. As an example, one dream incorporated three pets from memory—a cat from recent life, a rabbit from twelve years ago, and a teddy bear from childhood. Dreams often combine memory sources like this from multiple time periods in life.
My team and I wanted to explore this phenomenon further, to see whether memory sources change across a night of sleep. In a study in Rochester, New York, we woke subjects up from each of the four sleep stages at the beginning, middle, and end of the night (twelve awakenings in total) and in the morning asked them to identify memory sources for all twelve of their dreams.14 To give an example, one subject, who happened to be an EMT, reported in the middle of the night, “There was a game of baseball with a bunch of members of my family, and two of us got really hurt.” The memory source for this dream was an experience from two days prior of helping a patient who was hurt playing baseball at a family gathering. What we found was that recent memories like this (from up to a week ago) were more prevalent in early-night dreams, and relatively more distant memories appeared in dreams later in the night.
There was also some evidence that the same memory source reappeared in multiple dreams over the course of the night. For instance, that same subject later dreamed of “walking home from a golf course with my dad. [I] got sidetracked playing baseball with some friends and kids that I knew.” The baseball memory is still evident, but it has transformed and become less directly related to the initial event. Other research, too, has shown that how memories appear in dreams evolves across the night. Dreams early in the night are more literally related to the recent past or anticipated future, such as “I was at work. We had orders coming in. I was cataloging,” whereas dreams later in the night incorporate more distant memories and display more metaphorical reference to waking life (in the following example, the memory source is an exam): “It’s a big party with exams, people were getting called into a room one by one. Everyone was in modern Victorian dress.”15
In general, memory sources seem to morph not only over the course of a single night but also over many nights, becoming less directly related to the original episode with time. And dreams often combine multiple memory sources together, linking recent memories with more distant ones, especially during REM sleep. This process follows a cyclical time pattern: After an initial spike in day residues, memories then reappear in our dreams after five to seven days, a phenomenon coined the “dream-lag” effect. Say you have an argument with your spouse—this memory is likely to pop up in dreams both immediately and again after a five-to-seven-day delay. Dreams of the sleep lab also crop up first as day residues (the night after being in the lab) and then reappear in dreams again about a week after the visit to the laboratory. The initial dreams are more true to source: “I dreamt of the laboratory bedroom. [It] was exactly the same as I saw it yesterday … somebody was taking the electrodes off my head,” whereas, in the week following, dreams about the lab are less direct: “I am being admitted to a hospital because I am unable to remember my dreams. A team of doctors stands over the gurney.”16 The latter excerpt is layered within a less obvious dream about a hospital, the lab episode relating to more distant and generalized knowledge over time.
Overall, studies of the memory sources of dreams reveal a time course of memory integration and expand on our scientific knowledge about how sleep is processing memory. A highlight reel of day residues is gradually digested and absorbed through dreams and into long-term memory. Altogether, sleep’s effects on memory are curious, complex, and continuous—preserving, strengthening, integrating, and generally sculpting the key moments of our waking lives.
Dreaming and emotion
A common mantra among dream scientists is that dreaming is a form of overnight therapy.
As is probably apparent by now, not all memories benefit from sleep. Emotional episodes, significant events, personal concerns: These are experiences important for memory that sleep is essential for safeguarding. Emotion is one key factor that determines whether an event is important enough to be remembered for the future. As an example, if emotion is experienced at the time of encoding, a memory will be tagged for storage. Our brains register emotion as a signifier that something is important. Other factors do this, too, like motivation and reward. When a memory is tagged in this way, it is likely to be preserved from the erasing effect of slow wave sleep and likely to be strengthened and integrated during both NREM and REM sleep.
While emotion is a helpful signal to the sleeping brain to record an event, it’s also important that over time the emotion associated with a memory dissipates. Just think how burdensome it would be if simply recalling a negative memory always stirred up the same level of emotion as the initial event. Sleep provides two emotional functions, then: soothing our distress and holding on to the essential lessons learned.
My favorite study of all time used fMRI to record amygdala reactivity during an induced state of self-conscious distress.17 The initial task was for participants to sing along to karaoke videos while wearing headphones. Later, while in the MRI scanner, experimenters induced self-conscious distress by playing back recordings of the participants’ own out-of-tune singing. Their brains screamed of shame and embarrassment, with significant blood flow responses in the amygdala and accordant subjective feelings of distress. After a night of sleep, though, amygdala reactivity decreased in direct proportion to the amount of uninterrupted REM sleep. If REM sleep was severely disrupted, this benefit was completely lost.
This demonstrates that a key function of REM sleep is forgetting the distress associated with a memory, or at least dissipating the emotional response over time.
The sleep to forget/sleep to remember18 model claims that the unique neurochemistry of REM sleep is designed both to consolidate emotional memories (sleep to remember) and to decrease the level of arousal associated with these memories (sleep to forget). As we’ve learned, there are high levels of activation in the amygdala uniquely in REM sleep, which supports the reactivation of emotional memories. However, the REM brain is also in a state of suppressed noradrenaline activity; noradrenaline is a neurotransmitter associated with arousal, so its inhibition during REM sleep means that we do not exhibit as much physiological arousal as we do in waking life. This unique combination of high amygdala activity combined with low noradrenaline means that the content of an emotional memory can be safely strengthened (and potentially reexperienced in dreams) in a low-arousal state. Subsequent recall of the memory then benefits from lessened arousal, too, a learned dissociation. This prevents an unwanted buildup of anxiety associated with emotional memories over time, but as we’ll see later, this benefit is lost for those who experience nightmares.
To give some experimental examples: In one study, subjects were asked to memorize negative pictures (such as a taxicab accident or a vicious snake)19 while their heart rate was measured as an index of arousal. The subjects then slept. After sleep, those subjects who had higher heart rate when initially memorizing the objects later had better recall for these objects (emotion tagged the memories for consolidation), and their heart rate was lower during recall. Sleep stored the memories while dissipating the associated arousal. Other studies induced negative mood states in subjects just prior to a learning task—such as memorizing word pairs—and sleep again improved memory but soothed the associated negative mood after a few nights. In this case, even though the task itself was not emotional, being in an aroused state tagged the memories for storage. (You can think of the implications here for patients with anxiety disorders, who are in a constant state of arousal, so an abundance of daily events are tagged as emotional and important.) Finally, when subjects are deprived of sleep, emotional memory suffers (such as having poor recall for negative pictures seen before), and amygdala reactivity stays high when reexposed to these stimuli later. Sleep is thus as important for forgetting as it is for remembering.
Is dreaming, then, a form of overnight therapy? We know that emotions are ubiquitous in dreams, occurring in up to 95 percent of dreams reported at home. Besides emotion, current concerns (like financial worries) are often found in REM dreams, as well as personally significant events (like an argument). Experimental studies have shown that exposing subjects to presleep stressors, such as watching aversive films or taking stressful intellectual tests, also influences dream content and amplifies dreamed emotions.
Dreams even stubbornly incorporate those concerns that we try to avoid thinking of during the day. The “dream rebound” refers to a phenomenon where thoughts that we attempt to suppress during the day, especially just prior to sleep, spring back into dreams later that night. When we asked subjects to attempt to suppress a thought just prior to sleep over the course of one week—either an unpleasant or pleasant thought depending on their assigned condition—we found that unpleasant thoughts more often rebounded into dreams. But these dream rebounds had a positive impact: Subjects who dreamed more of the thought rated it as more pleasant later on. This provides evidence that dreaming, too, is supporting sleep’s emotional functions.
Calling back to the hyper-associative nature of REM sleep, one way that dreams may help is in providing creative solutions to personal problems, solutions we might not think of in waking life. Indeed, in one study it was only when dreams presented possible solutions to a presleep problem that subjects felt relieved on awakening, even if the dream itself was negative. On the other hand, dreams containing no such resolution were associated with a more negative view of the problem on waking.
If this is part of how dreaming works, it’s likely a process that spans successive REM periods, and even multiple nights, as our ongoing concerns resurface and are reexamined in the fluid memory networks of the REM sleeping brain. This seems especially true for major life stresses, such as going through a divorce. Rosalind Cartwright’s seminal work in the 1980s and ’90s20 showed that divorced women who initially had more negative dreams involving their ex were less depressed one year later than women who did not experience stressful divorce dreams. She suggested that emotional memory is being reworked over many nights of sleep. The divorce calls for a major reorganization of memory networks, adapting to the negative experience and rewiring the brain over time. Cartwright was an early pioneer (and one of the only women) in sleep medicine and dream science in the ’60s, and played a prominent role in uncovering the role of sleep and dreaming in our emotional lives.21
Indeed it seems like both REM sleep and dreaming play a pivotal role in maintaining our emotional health: integrating salient experiences into memory and dampening the distress associated with these memories over time. We’ll explore these concepts further when we get to nightmares, where these benefits begin to unravel.
* * *
All in all, the cycles of brain activity we surf through each night have evolved for our adaptation and survival: to support our brain’s immense capacity for encoding information, to recover from the wealth of stimuli and stress accrued each day, and to prepare us to reenter the world each morning. Without question, sleep is instrumental for learning. Memory traces tagged as important are reactivated in the brain and strengthened during sleep. Those events of greatest concern and strongest emotion are most likely to be saved, becoming pieces of our personal story and teaching us how to engage with and what to expect from the world each day.
In dreams, we may be witnessing this process from within, as if accessing memory were a real experience (a reexperience). Though early views considered dreams to be random electrophysiological noise produced by the brain during sleep, today’s science is revealing that dreams offer a window into sleep’s mental functions, revealing which memories are undergoing consolidation and when stressful events are being processed. Dreaming is correlated with learning and with improved performance on various tasks. Some dreams unveil creative solutions to problems, a product of the hyper-associative REM state. (In fact, you can borrow from this state in the short period after waking from a dream—try using this time to reflect on a problem, for example).
Over many nights and many dreams we see an interweaving of close and distant, recent and remote memories. This resembles a sorting process, where each new experience is filed relative to similar experiences from the past, and the dream spirals out to link more remote memories into its web overnight. With this nightly process, our sense of self evolves and updates over time and throughout our lives.
NIGHTMARE OBSCURA: A DREAM ENGINEER’S GUIDE THROUGH THE SLEEPING MIND. Copyright © 2025 by Michelle Carr. All rights reserved. For information, address Henry Holt and Co., 120 Broadway, New York, N.Y. 10271.
