Most of what people believe about dreams is wrong. Dreams do not only happen in REM sleep. They are not random noise from a sleeping brain. They are not reliable windows into unconscious desires. And they are not something science has neatly explained away.

When Brains Dream by sleep researchers Antonio Zadra and Robert Stickgold is one of the most thorough and readable examinations of dream science published in recent years. Zadra is a psychologist at the University of Montreal who has studied dreams for three decades. Stickgold is a cognitive neuroscientist at Harvard Medical School whose research on sleep and memory is among the most cited in the field. Together, they bring both rigorous science and genuine intellectual honesty about what is still unknown.

The book proposes a new theoretical framework for what dreams are for, dismantles several popular myths about dreaming, and offers a genuinely satisfying account of phenomena like lucid dreams, nightmares, and the feeling that a dream meant something. Here is what you need to know.

Key Takeaways

• We dream in every stage of sleep, not just REM. Most people dream for two-thirds or more of the night, and NREM dreams are more common than most people realize.

• The NEXTUP model proposes that dreaming is the brain exploring weak, unexpected associations between memories and experiences, a process that supports creative insight and emotional processing.

• Sleep quality and dream quality are connected. Protecting consistent sleep schedules and sufficient sleep duration gives the brain more time to run the processes dreaming supports.

About the Authors

Antonio Zadra has spent his career studying the phenomenology of dreams, including nightmares, recurring dreams, and sleepwalking. His lab's work on dream content analysis is among the most detailed in the field.

Robert Stickgold runs the Center for Sleep and Cognition at Harvard Medical School. His research on sleep-dependent memory consolidation, particularly the role of REM sleep in emotional memory processing, has shaped how neuroscientists understand what sleep is doing beyond rest. Both authors approach the subject as scientists who find dreams genuinely fascinating rather than dismissible, which gives the book an unusual combination of rigor and enthusiasm.

The NEXTUP Model: What Dreams Are Actually For

The core intellectual contribution of When Brains Dream is the NEXTUP model, which stands for Network Exploration to Understand Possibilities. The idea is that dreaming is a specific type of brain activity in which the sleeping brain explores loosely connected memories and experiences, looking for non-obvious relationships that might be useful.

During waking life, the brain tends to activate strong, well-established associations. You think of "dog" and immediately go to common associations: pet, fur, bark. During sleep, the brain loosens its association filters and explores weaker connections: dog might link to loyalty, freedom, pack dynamics, a memory of a specific afternoon, an unrelated metaphor. NEXTUP proposes that this exploration is how dreaming produces the creative leaps and problem-solving breakthroughs that people sometimes experience upon waking.

This is not the same as saying dreams predict the future or decode hidden desires. It is a more modest and scientifically defensible claim: the dreaming brain is running a form of associative exploration that waking cognition does not typically engage in, and that exploration has functional value for creativity, memory integration, and emotional processing.

We Dream More Than We Think

One of the book's more surprising points for general readers is how much of the night involves dreaming. Zadra and Stickgold document evidence that dreaming occurs across all sleep stages, not just the REM periods most people have heard about. NREM sleep produces dreams that tend to be shorter, less narrative, and less emotionally vivid than REM dreams, but they are dreams nonetheless.

The implication is that most people are dreaming for significantly more of the night than they realize. The common experience of "I never dream" typically reflects poor dream recall rather than an absence of dreaming. Dream recall is unreliable and declines steeply as you move further from waking. People who are woken during NREM sleep report dream content with surprising frequency.

Understanding how much REM sleep you actually need is one practical takeaway from this section of the book: since both REM and NREM sleep contribute to the processes the authors describe, total sleep duration matters alongside sleep stage distribution.

Dreams, Memory, and Learning

Stickgold's memory research features prominently here. The evidence that sleep plays an active role in memory consolidation is now well-established: information encoded during the day is processed, strengthened, and integrated during sleep, particularly during slow-wave sleep (NREM) and REM. Dreaming appears to be part of this process rather than an epiphenomenon of it.

The authors describe research showing that people who dream about newly learned tasks subsequently perform better on those tasks than people who do not. This is not because the dream rehearsal itself matters, but because the dream is a marker of the underlying memory consolidation process running effectively. You dream about the maze because your brain is working on the maze. The work is what matters; the dream is a byproduct.

This connects directly to the practical question of what sleep is actually doing during the hours you spend in bed. Memory consolidation is one of the most important and least visible functions of sleep, and it is one of the first things to degrade when sleep is shortened or disrupted.

Lucid Dreams, Nightmares, and Prophetic Dreams

The book devotes substantial attention to the unusual categories of dream experience. On lucid dreaming, Zadra and Stickgold describe it as a genuine phenomenon with a documented neural signature: during lucid dreams, the prefrontal cortex shows unusual activity relative to regular REM sleep, which is consistent with the self-awareness lucid dreamers report. They treat lucid dreaming as interesting and real without overclaiming for its practical applications.

On nightmares, the authors are particularly good. Nightmares are not signs of psychological disturbance in most cases. They appear to serve a function in emotional memory processing, replaying emotionally significant events in a context where the physiological stress response is partially suppressed (because you are lying down, not in actual danger). When nightmares become chronic and distressing, as in PTSD-related nightmares, the process is dysregulated rather than absent.

On prophetic dreams, the authors are clear-eyed: confirmation bias drives most of the perceived accuracy. The brain generates thousands of dream scenarios. When one matches a subsequent event, it is remembered. The vast majority that do not match are forgotten. The statistical math does not support genuine predictive capacity in dreams.

The Emotional Function of Dreaming

One of the most compelling threads in the book is the argument that dreaming serves an emotional regulatory function. REM sleep in particular appears to be involved in processing emotionally charged memories in a way that reduces their acute emotional intensity over time. Stickgold's phrase "sleep to forget, sleep to remember" captures the idea: the emotional valence of a memory is reduced while the factual content is preserved.

This has clinical implications. Sleep disruption in people with anxiety and PTSD may partly explain why traumatic memories remain so emotionally raw: the normal REM-sleep process for attenuating emotional intensity is interrupted. The connection between sleep problems and emotional dysregulation has practical significance for anyone dealing with anxiety, ADHD, or trauma-related sleep difficulty.

What This Means for Your Sleep

The authors are careful not to overclaim practical implications, but several do follow from the science they present. Protecting sufficient sleep duration is the most important. Since both REM and NREM sleep contribute to the processes they describe, consistently cutting sleep short short-circuits all of them.

Sleep timing matters as well. REM sleep is concentrated in the later hours of the night. People who go to bed late and wake early disproportionately lose REM, which is where the emotional memory processing and associative exploration the book describes are most active. The connection between irregular schedules and poor dream content or emotional volatility is not a mystery once you understand the biology.

For more on the downstream effects of sleep quality, our guide on sleep tracking apps covers tools that measure REM and NREM stages over time, which can make these patterns visible rather than theoretical. And if you have ever wondered how long it should take to fall asleep, that piece addresses what sleep onset tells you about your overall sleep health.

How to Apply the Book's Insights

The most practical takeaway from When Brains Dream is that the processes dreaming supports, including memory consolidation, creative association, and emotional regulation, are not optional add-ons to sleep. They are core functions that require sufficient, well-timed sleep to operate.

Protecting consistent sleep and wake times is the single highest-impact behavioral change supported by the science in this book. A tool like Lifestack approaches this from the scheduling side: by mapping your day around your energy and health data, it helps you finish demanding work before it encroaches on sleep, maintain consistent wake times, and avoid the late-night schedule drift that disproportionately cuts into REM sleep. Plans start at $7/month or $50/year with a 7-day free trial.

Frequently Asked Questions

Who is When Brains Dream written for?

The book is written for educated general readers with no scientific background required. The authors explain technical concepts clearly without oversimplifying. It is suitable for anyone curious about the science of dreams, sleep researchers, therapists working with nightmares or trauma, and anyone who has wondered what dreaming is actually for.

What is the NEXTUP model of dreaming?

NEXTUP stands for Network Exploration to Understand Possibilities. The model proposes that dreaming is the brain's mechanism for exploring weak, non-obvious associations between memories and experiences. This exploration supports creative insight, problem-solving, and emotional memory integration. It is a functional account of dreaming that does not require attributing mystical significance to dream content.

Do we dream every night?

Yes. Research strongly supports that dreaming occurs on virtually every night of normal sleep across all sleep stages. What varies is recall, not dreaming itself. People who report not dreaming typically have poor dream recall rather than an absence of dream activity. Recall is highest for dreams that occur just before waking and declines rapidly with time.

How is When Brains Dream different from Why We Sleep?

Matthew Walker's Why We Sleep covers sleep broadly, including memory, health, and the consequences of sleep deprivation. When Brains Dream focuses specifically on dreaming: what it is, what it does, how the brain produces it, and what the various unusual forms of dreaming tell us about the underlying mechanisms. The two books complement each other well and cover different territory despite some overlap.

Does dreaming improve memory?

Dreaming appears to be a marker of effective memory consolidation rather than the cause of it. The underlying sleep-dependent processes, particularly those occurring during slow-wave sleep and REM, strengthen and integrate newly encoded memories. Dreaming about a recently learned task correlates with better subsequent performance, but the memory work is happening throughout sleep, not only during dreamed moments.