Your first dream of the night and your last one are nothing alike

Most dream journals are biased and the people writing them don't know it.

Here's what I mean. The dreams you write down in the morning, the ones vivid enough to survive waking up, reaching for your phone, and actually typing something out, those are almost exclusively late-night dreams. They're the long, strange ones that happen in the final hours of sleep. They're real, but they're not the whole picture. Your brain was dreaming hours before that. Those earlier dreams just look completely different.

I started reading the research on how dream content shifts across the night, and the gap between your first dream and your last is bigger than I expected.

Your night has two halves

Sleep isn't one continuous state. It cycles through stages roughly every 90 minutes, alternating between non-REM (lighter and deep sleep) and REM (where the most vivid dreaming happens). But those cycles aren't identical. The balance shifts as the night goes on.

In the first half of the night, deep slow-wave sleep dominates. More than 80% of slow-wave sleep is concentrated before the midpoint. REM periods are short. Your first one lasts about 10 minutes.

In the second half, the ratio flips. Deep sleep fades and REM takes over. Each REM period gets longer. By your fourth or fifth cycle, REM can run 45 to 60 minutes. The second half of the night contains roughly twice as much REM as the first.

So you're dreaming more in the back half of the night just by volume. But the content is changing too.

Early-night dreams replay your day

A 2021 study by Josie Malinowski and Caroline Horton in Consciousness and Cognition woke 68 participants during both early and late sleep and had them report their dreams alongside their recent waking experiences. The early-night dreams were recognizably connected to the day. They were less bizarre. More grounded. More "continuous with waking life," as the researchers put it.

This lines up with what sleep researchers have long thought about NREM dreaming. When people are woken from non-REM sleep and asked what was going through their minds, they report something, roughly 43–50% of the time. But the reports tend to be short, thought-like, and mundane. More "I was thinking about the meeting tomorrow" than "I was flying over a city made of bread."

These early dreams seem to be tied to memory consolidation. Your brain is filing away the day's events. The dreams, if you can call them that, are closer to echoes of recent experience being sorted and stored.

Late-night dreams get strange

Malinowski and Horton found the opposite pattern in late-night reports. Those dreams were more emotionally intense and more personally important. They were also more "hyperassociative," which is the researchers' term for the way late-night dreams mash together unrelated memories, shift settings without warning, and pile metaphor on top of metaphor.

This is the stuff most dream journals are full of. The reason is partly structural (longer REM periods mean longer, more elaborate dreams) and partly chemical.

Cortisol is part of the explanation

As the night goes on, cortisol climbs. It starts rising around the middle of the sleep period and peaks in the early morning hours, reaching 15–20 μg/dL. That's a range that, during wakefulness, is associated with memory impairment.

Jessica Payne and Lynn Nadel laid out the connection in a 2004 paper in Learning & Memory. Their argument: early in the night, when cortisol is low, the hippocampus functions normally. It can retrieve full episodic memories and replay them coherently. That's why early dreams feel like recognizable scenes from your day.

Later, as cortisol rises, it disrupts communication between the hippocampus and the neocortex. The hippocampus can't pull up complete memories anymore. Instead, you get fragments. Disconnected pieces of different experiences stitched together. Payne and Nadel describe the result as "fragmented and often bizarre."

So the weirdness of your morning dreams isn't random. It follows from a hormone curve that shifts what kind of memory retrieval your brain can do.

You're dreaming in NREM too, just differently

One thing that surprised me in the research: dreaming isn't exclusive to REM sleep. That was the assumption for decades after REM was discovered in the 1950s, but it's been revised.

A 2020 study using graph analysis of dream reports found clear structural differences between REM and NREM dreams. REM dreams had greater "connectedness," meaning more characters, more scene changes, more going on. NREM dreams were simpler and shorter, more like isolated thoughts or static images.

The recall rates tell the same story from a different angle. Wake someone from REM and they'll report a dream about 82% of the time. Wake them from NREM stage 2 and it drops to roughly 43%. Many researchers now think of it as a spectrum: NREM produces faint, thought-like mental activity; REM produces full immersive narratives. The boundary is blurrier than anyone assumed.

What this means for dream journals

If you keep a dream journal, the dreams you capture are almost all from the last stretch of the night. That's not a flaw. Those are the longest and most memorable dreams you have. But it's worth knowing that your journal is a skewed sample.

The quiet, day-replay dreams from early in the night almost never make it into a journal entry. They're too short, too forgettable, and they happen during sleep stages where memory encoding barely works. You'd have to set an alarm for 2 a.m. to catch them, which, outside of a sleep lab, nobody is going to do.

This matters for how you read your own journal. If every entry feels surreal and intense, that's not because all your dreams are like that. It's because writing down whatever you remember when you wake up naturally selects for the most vivid material.

A few things worth paying attention to:

If you ever do wake up in the middle of the night and jot something down, compare those notes to your morning entries. The middle-of-the-night fragments tend to be more mundane, more clearly tied to recent events. That's the early-night pattern showing up.

People who journal from multiple awakenings (new parents, light sleepers, anyone doing wake-back-to-bed) sometimes notice this directly: the first fragments are grounded, the later ones get progressively weirder. That's not just perception. It tracks with longer REM periods and rising cortisol as the night goes on.

Malinowski and Horton's finding about emotional intensity also matches what a lot of journalers already sense. The dreams that hit hardest are almost always the ones right before waking, from those final long REM windows.

You're only seeing the last act

None of this means your dream journal is inaccurate. The dreams it captures are real. But they're the end of a process that's been running all night, one that starts with mundane memory sorting and ends with the strange, recombined material that feels most "dreamlike."

Knowing that has changed how I read my own entries. When a dream seems disproportionately strange or emotional, it helps to remember it came from the part of the night where cortisol is high, REM periods are long, and the hippocampus has mostly checked out. The early-night version of that same concern was probably a straightforward replay of something that happened Tuesday. I just wasn't awake to catch it.

The full show runs every night. Your journal only covers the finale.