Why Do We Dream? What Science Actually Knows

You spent roughly two hours last night dreaming. You probably don't remember most of it — perhaps a fleeting image, a half-formed scenario, an emotion that lingered after you woke. But while you were sleeping, your brain was running one of the most elaborate simulations it's capable of, and after decades of research, scientists are still debating exactly why.

Dreams are one of those phenomena that everyone experiences but almost nobody can explain. They've been interpreted as messages from gods, windows into the unconscious, and random neurological noise. The truth, as is usually the case in sleep science, is more interesting than any of those simple narratives. We now have several well-supported theories about why we dream, and none of them is the whole story. Let me walk you through what the evidence actually shows.

The Major Theories of Dreaming

There is no single, universally accepted theory of why we dream. What we have instead is a set of complementary frameworks, each supported by different lines of evidence. Think of them less as competing explanations and more as different lenses on the same phenomenon.

Memory Consolidation: Dreams as Overnight Filing

This is probably the most robust theory in the literature. The core idea is that dreaming — particularly during REM sleep — plays a role in consolidating memories, transferring information from short-term to long-term storage, and integrating new experiences with existing knowledge.

The evidence is substantial. Robert Stickgold and colleagues at Harvard showed that subjects who slept after learning a task performed significantly better the next day, and that the degree of improvement correlated with the amount of REM sleep obtained (Stickgold et al., 2000). More recently, a 2021 study in Science Advances demonstrated that during REM sleep, the hippocampus "replays" the day's experiences at roughly 20 times real speed, effectively rehearsing and strengthening neural connections (Boyce et al., 2016; updated findings, 2021).

But here's the nuance that often gets missed: it's not just about storing memories. Dreams seem to involve a process of extracting patterns — pulling out the gist, the emotional core, the useful abstractions — rather than literally replaying events. This is why dreams often feature distorted, fragmented versions of real experiences. Your brain isn't recording a documentary; it's writing abstracts.

Threat Simulation: Dreams as Evolutionary Rehearsal

The Finnish cognitive neuroscientist Antti Revonsuo proposed the threat simulation theory in 2000, and it remains one of the more provocative ideas in dream research. The argument goes like this: for most of human evolutionary history, the world was genuinely dangerous. Predators, rivals, and physical threats were constant. Dreams that simulate threatening scenarios — being chased, falling, facing danger — may have provided a safe virtual environment in which to rehearse threat responses (Revonsuo, 2000).

The supporting evidence is suggestive. Studies consistently find that negative emotions (particularly fear and anxiety) dominate dream content far more than positive ones — a pattern that persists across cultures and age groups. A 2019 meta-analysis of over 1,000 dream reports found that approximately 65–70% of dream emotions are negative (Murkar & Shenouda, 2019). From an evolutionary standpoint, a dream-rehearsal system calibrated for danger detection makes sense — it would have been more fitness-enhancing to over-prepare for threats than to under-prepare.

Critics point out that modern dreams often involve mundane or surreal content that has nothing to do with survival threats. Revonsuo's counter-argument is that the system was shaped in a very different environment, and the "bugs" in modern dream content are simply what happens when an ancient system runs in a modern world.

Emotional Processing: Dreams as Overnight Therapy

This theory, championed notably by Matthew Walker at UC Berkeley, proposes that dreams — especially REM dreams — serve as a form of emotional processing. The idea is that during REM sleep, the brain re-experiences emotionally charged events but in a neurochemical environment stripped of noradrenaline (the brain's stress chemical). This allows you to revisit painful or intense memories without reliving the physiological stress response, effectively draining the emotional charge from those experiences (Walker, 2017).

The evidence here comes partly from neuroimaging. During REM sleep, the amygdala (the brain's emotional centre) is highly active — more active than during waking, in fact — but the prefrontal cortex (the logical, rational part of the brain) is relatively quiet. This combination may allow emotional experiences to be processed without the usual cognitive filters and self-judgement getting in the way. (I sometimes describe it to patients as "free therapy — your brain working through your feelings without you getting in the way.")

Walker's lab also found that people who were sleep-deprived before being shown emotionally disturbing images showed a 60% increase in amygdala reactivity compared to those who slept normally — suggesting that without the overnight emotional processing dreams provide, we become disproportionately reactive to negative stimuli the next day (van der Helm et al., 2011).

What Actually Happens During REM Sleep

To understand why dreams look the way they do, it helps to know what's happening in the brain during REM. As I covered in my article on sleep cycles, REM is the stage where most vivid dreaming occurs, and it occupies roughly 20–25% of total sleep in healthy adults.

During REM, your brain is in a peculiar state. The prefrontal cortex — responsible for logic, self-awareness, and the sense of a continuous "self" — is largely offline. Meanwhile, the limbic system (emotional processing), the visual cortex (imagery generation), and the hippocampus (memory) are highly active. This combination explains several features of dreams:

• Vivid imagery with poor logic. Your visual system is generating scenes, but the part of your brain that would normally say "this doesn't make sense" is asleep.
• Emotional intensity. The limbic system is running hot, which is why dreams often feel overwhelmingly emotional — even bizarre scenarios can carry genuine terror, joy, or grief.
• Faulty time sense and memory. Without the prefrontal cortex online, you can't reliably track how long things have been happening, and your ability to form new episodic memories is severely impaired. This is why you forget most dreams within minutes of waking.
• Reduced body awareness. Motor cortex signals are generated but blocked by atonia (the body's temporary paralysis during REM), which is why you don't physically act out most dreams. (When this mechanism fails, you get REM sleep behaviour disorder — a condition that deserves its own article.)

The neurotransmitter environment during REM is also distinctive. Acetylcholine levels are high (promoting vivid imagery), while serotonin and noradrenaline are near zero. This chemical cocktail is essentially the opposite of the waking state, and it creates a neural landscape where internal experience dominates over external input. Your brain is, quite literally, creating its own reality.

Common Dream Myths, Debunked

Dreams have attracted more pseudoscience than almost any other sleep-related topic. Here are the most persistent myths and what the evidence says.

"You always dream — you just don't remember." This is largely true. Research using EEG during sleep shows that dreaming occurs in all stages, not just REM. NREM dreams tend to be less vivid and more thought-like, but they're real. Studies that wake subjects from REM sleep find dream recall about 80% of the time; from NREM sleep, it's closer to 50–60% (Fosse et al., 2003).

"Dreams only happen during REM sleep." As just noted, this is false. You dream during non-REM sleep too — the dreams are just less narrative, less emotional, and harder to recall. A landmark 2004 study at the University of Chicago used sophisticated monitoring and found that dreams reported during NREM stages were qualitatively different but no less real (Takeuchi et al., 2000).

"Dreams have hidden meanings that can be decoded." The psychoanalytic tradition, particularly Freud's work, proposed that dreams are symbolic representations of unconscious desires. There's no compelling scientific evidence for this. Dream content tends to be more straightforwardly connected to recent experiences and current concerns than to deep symbolic meanings. That said, dreams do reflect your waking preoccupations — not because they're encoded symbols, but because your brain is processing the material that matters to you.

"Falling in a dream means you're going to die." / Other predictive dreams. Dreams do not predict the future. The sensation of falling is extremely common and is likely related to muscle atonia during REM (your body interprets the loss of muscle tone as falling). Dream prediction studies have consistently failed to demonstrate any precognitive ability. Not even close.

"Blind people don't dream visually." People who became blind after early childhood typically do experience visual dream imagery. Those born blind tend to have dreams dominated by other senses — sound, touch, smell — but these are no less vivid or meaningful.

Dreams and Mental Health

The connection between dreams and mental health is one of the most active areas of current research, and it goes in both directions.

Nightmares and PTSD. Post-traumatic stress disorder is perhaps the clearest example. Recurrent nightmares are a hallmark symptom of PTSD, occurring in 70–80% of sufferers. These aren't random bad dreams — they're often direct replays of traumatic events, and they can prevent the very emotional processing that REM sleep is supposed to provide. This creates a vicious cycle: trauma disrupts the normal function of dreams, which prevents recovery, which maintains the nightmares. Prazosin, a medication that suppresses noradrenaline, has shown some success in reducing PTSD nightmares, supporting the neurochemical model of emotional processing during sleep (Raskind et al., 2013, though subsequent larger trials have produced mixed results).

Anxiety and depression. Research consistently shows that people with depression have altered dream patterns — more nightmares, more negative emotional content, and, interestingly, more vivid and memorable dreams in general. There's also evidence that the frequency of nightmares correlates with the severity of depressive symptoms (Schredl & Reinhard, 2011). Whether this is causal or merely correlational remains debated, but the direction of the relationship suggests that disrupted dreaming is both a symptom and a potential contributor to mood disorders.

Anxiety disorders. People with generalized anxiety tend to report more threatening dream content and more frequent awakenings from dreams — which may reflect the same hyperactive threat-detection system that characterizes their waking state. Interestingly, exposure-based therapy for anxiety has been shown to reduce nightmare frequency, suggesting that processing fears during the day can improve their processing during sleep.

Can You Control Your Dreams? Lucid Dreaming Basics

One of the most fascinating developments in dream research is the study of lucid dreams — dreams in which the dreamer becomes aware they're dreaming and, in some cases, gains the ability to influence the dream's content.

Lucid dreaming isn't new. References to it appear in texts going back centuries. But it became a subject of serious scientific inquiry in the 1970s, when Stephen LaBerge at Stanford developed techniques for signalling from within a dream (using pre-arranged eye movements that could be detected on polysomnography). This proved lucid dreaming was a real, verifiable phenomenon — not just dreamers misremembering their experiences.

What's actually happening in the brain during a lucid dream is interesting. Neuroimaging studies show increased activity in the prefrontal cortex — the very region that's normally quiet during REM. It's as if a piece of the rational, self-aware brain "wakes up" while the rest remains in dream mode (Dresler et al., 2012). This creates that distinctive combination: you know you're dreaming, but the dream imagery continues.

Lucid dreaming techniques tend to centre on a few key practices:

• Reality testing during the day. Regularly asking yourself "am I dreaming?" and performing a check (like trying to push a finger through your palm) builds the habit of self-awareness that can carry into dreams.
• Dream journaling. Writing down dreams immediately upon waking improves dream recall dramatically and helps you identify recurring patterns — which can serve as "dream signs" that alert you within a dream.
• MILD technique (Mnemonic Induction of Lucid Dreams). Developed by LaBerge, this involves waking during the night (typically after 5–6 hours of sleep), briefly recalling a recent dream, and then repeating an intention like "next time I'm dreaming, I will remember I'm dreaming" as you fall back asleep. It's the technique with the strongest empirical support.
• Wake-back-to-bed (WBTB). Setting an alarm for 5 hours after falling asleep, staying awake for 20–30 minutes, then returning to sleep. This increases the likelihood of entering REM with greater cortical activation.

A 2020 meta-analysis found that MILD combined with WBTB produced lucid dreams in approximately 40–50% of attempts for experienced practitioners — not every time, but far better than chance (Stumbrys et al., 2012; updated meta-analyses through 2020).

I should note a few cautions. Lucid dreaming is generally safe and doesn't appear to disrupt sleep quality when practiced moderately. However, people who are highly prone to dissociation or have certain psychiatric conditions should approach it carefully — the boundary between lucid dreaming and wakeful dissociation isn't always clear. And the popular idea that you can "solve problems" or "rehearse skills" while lucid dreaming has only limited evidence behind it. Some studies show modest benefits for motor skill rehearsal, but it's not a substitute for waking practice.

What We Still Don't Know

Despite enormous progress, dreaming remains one of the less-understood aspects of sleep. We still don't have a complete answer to the most basic question: why does the brain generate these elaborate, often bizarre narratives rather than simply remaining in a quieter state during REM?

The most honest answer is that dreaming is probably not one thing. It's likely that the brain uses the REM state for multiple functions simultaneously — memory consolidation, emotional regulation, threat rehearsal, and perhaps others we haven't identified — and the subjective experience of "dreaming" is what it feels like from the inside when all of these processes are running at once.

What I can say with confidence is that dreams are not meaningless, and they're not mystical. They're a product of a brain doing important work in a very particular neurochemical environment. Understanding that work — even imperfectly — can change how you relate to your own dreams. They're not prophecies, and they're not hidden codes. They're your brain processing your life. And that, I think, is remarkable enough.

Related reading: Understanding Your Sleep Cycles · How Many Hours Of Sleep Do I Need? · Sleep Hygiene Checklist