Why are dreams so weird?

If you keep a dream journal, you know the feeling. You read back an entry from two days ago and it says something like "in my childhood kitchen but the fridge opened into an office building, coworker was there but she was also my high school teacher, late for something, no shoes." And you have zero memory of this being strange while it was happening.

I kept wondering why dreams are like that. Not the content specifically, but the weirdness itself. Why does the sleeping brain produce scenarios that would be immediately flagged as nonsense by the waking brain? Most people write it off as random noise. But I went looking into what neuroscience actually says, and the answer is more interesting than randomness. The weirdness might actually be doing something.

Your reality-checker shuts off

The main reason, and the one I find most satisfying, is that the part of your brain responsible for logic and critical thinking goes quiet during REM sleep.

The dorsolateral prefrontal cortex handles reasoning, planning, and noticing when something doesn't add up. During REM, it gets directly suppressed by acetylcholine, the same neurotransmitter that's ramping up sensory processing everywhere else in your brain. The chemical that makes dreams vivid is also the one silencing your ability to question them.

This is why you can fly through a grocery store, have a conversation with someone who died years ago, or show up to a meeting in your underwear and think nothing of it. The part of your brain that would normally flag those as impossible is offline.

Lucid dreamers are the exception that proves the rule. Brain imaging studies show that when someone becomes lucid during a dream, the prefrontal cortex partially reactivates. The moment your logic centers come back online, you notice how weird everything is.

Meanwhile, your emotional brain is running hot

While the prefrontal cortex powers down, the amygdala, your brain's emotional processing center, stays highly active during REM. Neuroimaging consistently shows strong amygdala activation during dreaming, at levels comparable to processing real emotional events while awake.

At the same time, norepinephrine drops to near zero. Matthew Walker, the UC Berkeley neuroscientist behind much of the modern sleep research, describes this as a kind of "overnight therapy": your brain replays emotionally charged experiences in an environment stripped of the stress chemical that normally accompanies them. The theory is that this helps defuse the emotional intensity of difficult memories.

So you've got a brain that's feeling things intensely but can't evaluate whether those feelings make sense. A mild worry about a work deadline can become a full-blown chase sequence. A passing thought about an old friend can become a vivid reunion in a place that doesn't exist. The emotional engine is generating content that the logical editor would normally reshape or reject, but the editor is asleep.

The weirdness might be the whole point

This is the part that got me. In 2021, Tufts University neuroscientist Erik Hoel published a paper in the journal Patterns that put dream weirdness in a completely different light. His "overfitted brain hypothesis" borrows a concept from machine learning: when a neural network trains too narrowly on the same data, it memorizes the specifics but fails to generalize. It gets good at recognizing the exact images it trained on but falls apart on anything slightly different.

Hoel's argument is that brains face the same problem. You spend your day in roughly similar environments, doing roughly similar things. If your brain only consolidated those exact patterns during sleep, you'd become overly specialized. Great at your routine, bad at anything new.

Dreams, in this framing, are corrupted training data. By generating bizarre, recombined, impossible versions of your waking experiences, the sleeping brain forces itself to stay flexible. The strangeness prevents your neural networks from becoming too rigid.

What makes this more than speculation: the most reliable way to trigger dreams about a specific topic is to perform a novel, repetitive task while awake. Your brain seems to target exactly the material that's at highest risk of overfitting.

Most dreams aren't actually that strange

Here's what I didn't expect. Despite how it feels, the majority of dream content is mundane. When researchers collect unselected dream samples from the general population, genuinely bizarre content shows up in only about 10% of reports.

Looking more granularly, more than half of all dream elements, including settings, people, and activities, are either direct copies or slight variations of real waking experiences. Only about 15% of physical surroundings in dreams are truly unlikely or impossible. For characters, that number drops to around 5%.

What gives dreams their strange feeling is how these elements get combined. Researchers categorize dream bizarreness into three types: incongruity (a dog breathing fire), vagueness (a location that's undefined but feels familiar), and discontinuity (sudden jumps in time or setting without transition). Dreams don't so much invent impossible things as shuffle real things into wrong combinations.

Many dream journalers notice this once they start paying attention. The individual pieces of a weird dream are usually recognizable. It's the arrangement that's off.

An older theory that still holds up (mostly)

The original neuroscience take on dream weirdness came from Harvard psychiatrists Allan Hobson and Robert McCarley in 1977. Their activation-synthesis hypothesis, published in the American Journal of Psychiatry, proposed that dreams start as random electrical signals from the brainstem during REM. The cortex receives these signals and tries to build a coherent narrative from them, like constructing a story from a handful of random words.

This was the first dream theory firmly grounded in neurobiology rather than Freudian interpretation. It explained why dreams are disjointed and bizarre without needing to invoke hidden meanings or repressed desires.

The theory has been refined since then. Hobson himself updated it over the years, acknowledging that dreams aren't purely random, that they're shaped by memory and emotion and recent experience. He and Robert Stickgold even ran an experiment at Harvard where they cut dream reports into pieces, spliced them back together in random order, and asked judges to tell the rearranged versions from the originals. The judges couldn't. Dreams already read like shuffled narratives, because that's basically what they are.

But the core insight holds: the brain is a meaning-making machine, and when you give it fragmented, emotionally charged input while its quality-control systems are offline, you get weird narratives.

What this means if you keep a dream journal

None of these theories say dreams are meaningless. Even the activation-synthesis model says the cortex is doing real cognitive work. The raw material might be chaotic, but the assembly isn't.

If you journal your dreams, a few things jumped out at me from the research.

The weird parts are worth writing down. If Hoel's overfitting theory holds, the most bizarre elements are where your brain is doing the most active work to stay flexible. Those might be the most interesting entries to revisit later.

I also found Walker's work changed how I think about journaling. His research suggests that the emotional content of dreams, not the specific events, is where the real processing happens. Noting how a dream made you feel is probably more useful than trying to reconstruct every plot twist.

And the patterns that matter tend to show up in the ordinary stuff. Since most dream content is actually drawn from waking life, the recurring elements across multiple entries tend to reflect what your brain is actively working on. The flying and the monsters are memorable, but it's the repeated appearance of your old apartment or a specific friend that usually points to something.

One last thing: don't overthink the narrative. Your sleeping brain is literally missing the hardware for logical sequencing. The jumps, the impossible combinations, the way one scene melts into another — that's not a code to crack. It's what cognition looks like without a prefrontal cortex keeping things in order.

The weirdness of dreams used to be treated as either mystical or meaningless. The neuroscience is landing somewhere in the middle. It's functional noise. Your brain is doing maintenance, emotional and cognitive and maybe creative, and the strange experiences you remember are a byproduct of that work. Or, if Hoel is right, they are the work.

I used to skip past the weird entries in my journal. Now they're the ones I go back to first.