Daniel sent us this one — he wants to know why we dream at all, what the leading theories actually are, and then the really interesting part: why certain medications produce dreams that are so vivid or disturbing they practically have their own genre. SSRIs, beta-blockers, varenicline, melatonin, Wellbutrin, even withdrawal states. He wants us to tie the pharmacology back to the underlying mechanics of dreaming itself.
The second layer is where the real insight lives, because you can't understand why a drug wrecks someone's dream life unless you understand what dreaming is doing in the first place. The two questions are the same question, basically.
Let's start there. What is the brain actually up to? Daniel specifically asked about the scientific theories — memory consolidation, threat simulation, emotional processing, the activation-synthesis thing, and whatever REM neurochemistry tells us.
And I should say upfront — today's episode is being written by DeepSeek V four Pro.
Back to dreams.
The first thing to understand is that we have about five or six major theories still in play, and none of them fully explains dreaming. The field doesn't have its Darwin yet. We have a bunch of partial models that each capture something real but none of them closes the case.
Let me lay them out in order of how seriously the field takes them right now. Number one is the memory consolidation hypothesis. The idea is that during sleep, especially REM, the brain replays and reorganizes information from the day, strengthening some connections and pruning others. The dreams you experience are basically the subjective side of that process — you're experiencing the replay.
Dreaming is what memory consolidation feels like from the inside.
That's the argument. Matt Wilson's lab at MIT recorded hippocampal place cells in rats running a maze, then recorded those same cells firing in the same sequence during REM sleep. The rats were literally replaying the maze in their dreams. And the replay wasn't just a straight rerun — it was compressed, sometimes running backward, sometimes with novel combinations. Exactly what you'd expect if the brain isn't just recording but actually reprocessing and integrating.
That explains why you dream about stuff that happened during the day, but not why the dreams are often completely unhinged.
That's where the second theory comes in. Threat simulation — mostly associated with Antti Revonsuo, a Finnish cognitive neuroscientist. He argued that dreaming is essentially a virtual reality training system for dealing with threats. His lab analyzed thousands of dream reports and found that threatening events are massively overrepresented compared to waking life. About sixty to seventy percent of dreams contain some kind of threat, and the dreamer typically responds with realistic defensive behavior.
Which would make evolutionary sense — if you're a primate trying not to get eaten, rehearsing predator escapes while you're unconscious and safe is a pretty good strategy.
The threats in dreams aren't random — they tend to be ancestrally relevant. Being chased, being attacked, social humiliation, falling. Not modern threats like missing a credit card payment. The dream system seems tuned to ancient dangers, which is a point against memory consolidation being the whole story. If dreams were just replaying the day, you'd dream about spreadsheets and traffic lights.
You do sometimes, but not proportionally.
The third theory is emotional processing, and this one overlaps with both but has its own distinct emphasis. The idea is that REM sleep specifically helps process emotional memories — not just store them, but actually reduce their emotional charge. Matthew Walker at Berkeley showed that during REM sleep, norepinephrine levels in the brain drop to nearly zero. Norepinephrine is the stress chemical — it's what makes emotional memories feel raw and immediate. So you're replaying emotional experiences in a neurochemical environment where the stress response is offline, which lets you integrate the memory without the emotional sting.
That's why people say "sleep on it" after something upsetting, and why things often feel less devastating in the morning.
That's the hypothesis, and it's got decent empirical support. People deprived of REM sleep after an emotionally charged experience show stronger emotional reactivity the next day. The brain seems to use REM as a kind of emotional reset.
Okay, so we've got memory consolidation, threat simulation, emotional processing. What about activation-synthesis?
This is the one often presented as the "dreams are meaningless" theory, which is unfair. Activation-synthesis was proposed by Hobson and McCarley at Harvard in the late seventies. Their argument was that during REM sleep, the brainstem generates random bursts of neural activity — the pons fires these PGO waves, ponto-geniculo-occipital waves, which propagate up into the cortex. The cortex, being a meaning-making machine, tries to synthesize a coherent narrative out of this random input. So dreams are the cortex's best attempt to make sense of internally generated noise.
It's like the brain is a storyteller that can't help itself. You give it random static and it spins a plot.
The reason dreams have that bizarre, disjointed quality is that the input really is random, and the prefrontal cortex — the part that does logic and reality-checking — is largely offline during REM. So you get narratives that feel coherent in the moment but collapse into absurdity the second you wake up. That maps well onto the phenomenology: locations shift without transition, people morph into other people, you accept completely impossible things as normal.
Hobson later revised the model, right?
Yes, into what he called the AIM model — Activation, Input-output gating, Modulation. He acknowledged the activation isn't entirely random, that there's some influence from memory systems and emotional state. So modern versions are less dismissive of meaning. But the core insight holds — a lot of dream content is driven by bottom-up neural noise that the cortex is trying to narrativize.
Where does REM neurochemistry fit as its own theory?
It's less a standalone theory of why we dream and more the mechanistic layer underlying all the others. The neurochemical environment of REM sleep is genuinely weird. Acetylcholine is through the roof — the brain is more cholinergically active during REM than when you're awake. But serotonin and norepinephrine are both suppressed. Dopamine is present but in a specific pattern. And the prefrontal cortex is deactivated while the limbic system — amygdala, hippocampus, anterior cingulate — is highly active.
You've got high emotion, high memory activation, and the logic circuits are asleep.
That's basically the recipe for a dream. High acetylcholine drives the cortical activation and vivid imagery. Suppressed norepinephrine means you don't get the stress response that would normally accompany those images. Suppressed serotonin — we'll come back to that, it's crucial for the medication question. And the deactivated prefrontal cortex means you don't question the reality of what you're experiencing.
The dream state is essentially: the brain is highly active, flooded with acetylcholine, emotionally charged, memory systems churning, and the part that says "this doesn't make sense" is offline. The cortex tries to weave all of that into a story.
That's a good summary. And that brings us directly to the pharmacology question, because every single medication Daniel mentioned messes with one or more of those systems.
Let's go through them. Start with SSRIs.
SSRIs are fascinating for dreams because they're so widely prescribed and the dream effects are so consistent. Most people on SSRIs report more vivid, more memorable, and often more bizarre dreams. Some get outright nightmares. And the mechanism is directly tied to what we just discussed.
Serotonin suppression during REM.
SSRIs increase serotonin availability — that's their job. But serotonin normally suppresses REM sleep. The firing of serotonergic neurons in the raphe nuclei is one of the things that keeps REM from initiating. When you flood the system with serotonin via an SSRI, you suppress REM sleep. People on SSRIs spend less time in REM, they have fewer REM episodes, and the REM they do get is fragmented.
The dreams get more intense, not less. So what's happening?
First, REM pressure builds up. The brain has a homeostatic drive for REM sleep — if you suppress it, the need accumulates. So when REM does break through, it's more intense. The cholinergic system is pushing harder. Second, serotonin itself modulates dream content. Serotonin normally dampens the emotional intensity of dream imagery. When you're chronically altering serotonin signaling, you may be changing the emotional tone of the dream generation process.
The dreams get weirder and more intense because the system is both suppressed and dysregulated.
There's a third factor — SSRIs also fragment sleep more broadly. They suppress slow-wave sleep in some people and can cause more awakenings during the night. More awakenings mean more dream recall. You don't remember most of your dreams because you don't wake up during or right after them. If a drug fragments your sleep, you're catching more dreams on the way out.
It might not even be that the dreams are more vivid — you're just remembering more of them.
It's probably both. The dreams are objectively more intense due to REM pressure, and you're remembering more of them due to sleep fragmentation. The combination creates that classic SSRI dream experience — long, elaborate, emotionally charged narratives that you remember in unusual detail.
What about beta-blockers? Those are famous for nightmares.
Beta-blockers are a completely different mechanism. They don't primarily affect REM sleep architecture the way SSRIs do. They block beta-adrenergic receptors — they suppress the effects of norepinephrine and epinephrine. Which sounds like it should reduce nightmares, not cause them, because norepinephrine is a stress chemical.
Why do people on propranolol or metoprolol report such disturbing dreams?
This one is not fully settled, but the leading explanation involves what happens when you suppress the sympathetic nervous system. Beta-blockers cross the blood-brain barrier to varying degrees — the lipophilic ones like propranolol get into the brain more than the hydrophilic ones like atenolol. And the lipophilic beta-blockers are the ones most associated with dream disturbances, which tells you it's a central nervous system effect.
It's not about REM architecture, it's about something else.
The best hypothesis involves melatonin and the pineal gland. Beta-blockers suppress melatonin production — that's well established. Melatonin is synthesized from serotonin in the pineal gland, and the synthesis is regulated by beta-adrenergic receptors. Block those receptors, melatonin drops. And melatonin isn't just about sleep onset — it modulates the timing and quality of REM sleep. Low melatonin can lead to more fragmented REM and more dream recall.
That's a pretty indirect pathway. Are there other mechanisms?
There's also evidence that beta-blockers alter the balance of cholinergic and aminergic neurotransmission during sleep. REM sleep is characterized by high acetylcholine and low norepinephrine. If you're pharmacologically suppressing norepinephrine signaling even further, you might be pushing the system into a more extreme version of the REM state. Some researchers think beta-blocker nightmares are essentially a hyper-cholinergic dream state — too much acetylcholine relative to the aminergic tone.
That connects back to activation-synthesis. If acetylcholine is the main driver of the random cortical activation, and you're tipping the balance even further toward acetylcholine, you get more intense and chaotic dream generation.
That might also explain why beta-blocker dreams have a particular quality — they're often described as more fragmented, more threatening, more "real-feeling" than normal dreams. The emotional processing system is being pushed off-kilter.
Let's talk about varenicline. That's Chantix, the smoking cessation drug.
Varenicline is the cleanest story of all of these, because its mechanism directly targets the acetylcholine system. Varenicline is a partial agonist at nicotinic acetylcholine receptors. It's designed to substitute for nicotine — it binds to the same receptors, gives a mild stimulation, and blocks nicotine from binding. But those nicotinic receptors are all over the brain, including in the circuits that regulate REM sleep.
You're directly stimulating the acetylcholine system that drives dreaming.
Acetylcholine is the primary neurotransmitter driving REM sleep and dream generation. Varenicline is literally turning up the gain on the acetylcholine system. The result is extremely vivid, often lucid dreams. The dream reports from varenicline users are remarkable — people describe dreams that are more colorful, more detailed, more emotionally intense than anything they've experienced before. Some people actually enjoy it.
Some people get nightmares so bad they stop taking the drug.
It's a direct pharmacological boost to the dream-generating system. It's probably the closest thing we have to a drug that just turns up the "dream intensity" dial, and the mechanism makes perfect sense given everything we know about REM neurochemistry.
What about Wellbutrin? That's bupropion.
Wellbutrin is interesting because it's an antidepressant but it doesn't work like SSRIs. It's a norepinephrine-dopamine reuptake inhibitor. It doesn't touch serotonin directly. And its dream effects are different from SSRIs — people report vivid dreams, but they're often described as more positive, more energetic, less nightmarish than SSRI dreams.
Which makes sense if serotonin suppression isn't part of the mechanism.
Wellbutrin actually increases norepinephrine and dopamine availability, and it tends to suppress REM sleep less than SSRIs do. Some studies suggest it can actually increase REM sleep in certain patients. So you're getting a different neurochemical recipe. The dopamine boost might be particularly relevant — dopamine is involved in the reward and salience systems, and it's present in REM sleep. Higher dopamine tone during REM might make dreams feel more engaging and less threatening.
The emotional valence of the dream shifts with the neurotransmitter that's being manipulated.
That's the hypothesis. And it fits with the broader picture — dreams aren't just random noise, they're generated by specific neurochemical systems. Tweak one system, you change the character of the dream. Tweak another, you change something else.
What about melatonin? People take it for sleep, and some get really intense dreams.
Melatonin is complicated because it's not directly driving dream generation the way acetylcholine is. Melatonin is primarily a chronobiotic — it regulates the timing of sleep, not the architecture of sleep stages. But there are a few things going on. First, melatonin can increase REM sleep duration in some people, especially if they were REM-deprived before. More REM means more dream time. Second, melatonin affects body temperature regulation during sleep, and core body temperature drops during REM — if melatonin is altering that thermoregulatory pattern, it might change the quality of REM sleep.
There's the simple fact that if melatonin helps you sleep more deeply and consistently, you're getting more REM overall.
Part of the melatonin dream effect is probably just better sleep architecture. But there's also evidence that melatonin receptors are present in brain regions involved in emotional processing — the amygdala, the hippocampus. So melatonin might be directly modulating the emotional tone of dream content, not just the quantity of REM.
The one that really interests me is drug withdrawal. Daniel mentioned withdrawal from various drugs producing especially disturbing dreams. What's happening there?
This is REM rebound, and it's one of the most robust phenomena in sleep science. When you suppress REM sleep chronically — with alcohol, benzodiazepines, SSRIs, many recreational drugs — the brain doesn't just accept the suppression. It builds up REM pressure. The homeostatic drive for REM accumulates. And when you withdraw the suppressing agent, the brain goes into REM overdrive.
You get a flood of REM sleep, and with it, a flood of dreams.
Not just more REM — more intense REM. The REM episodes are longer, the eye movements are denser, and the dreams are more vivid and more nightmarish. Alcohol withdrawal is the classic example. Alcohol is a potent REM suppressant. Chronic drinkers spend very little time in REM. When they stop drinking, REM rebounds dramatically. The dreams during alcohol withdrawal are notoriously disturbing — people describe them as hyperrealistic nightmares, often with themes of pursuit and danger.
Which is the threat simulation system in overdrive.
It's not just alcohol. Benzodiazepine withdrawal does the same thing. SSRI discontinuation can cause REM rebound with intense dreams. Even cannabis withdrawal — cannabis suppresses REM, and when heavy users quit, they often report extremely vivid and bizarre dreams for a week or two.
The withdrawal dreams are essentially the brain catching up on all the REM it missed, and doing it all at once.
That's the core mechanism. And it tells you something important about dreaming — the brain treats REM sleep as a biological necessity. It will compensate for lost REM. You can't permanently suppress it. The pressure just builds until it breaks through.
That has implications for the theories we started with. If the brain is that insistent on getting REM sleep, dreaming must be doing something important.
That's one of the strongest arguments against the idea that dreams are just meaningless noise. If dreams were epiphenomenal — just random static the cortex spins into stories — why would the brain fight so hard to get REM sleep? Why would there be such a strong homeostatic drive? The brain doesn't have a homeostatic drive for random noise.
The counterargument would be that the brain needs REM sleep for something physiological — cellular repair, receptor resetting, whatever — and the dreams are just a side effect of that process.
That's possible. But the fact that the content of dreams seems to be doing specific kinds of emotional and cognitive work — memory reprocessing, threat rehearsal, emotional desensitization — makes the side-effect argument harder to sustain. If dreams were just noise, you wouldn't expect them to systematically target emotionally relevant memories. You wouldn't expect the threat simulation pattern. You wouldn't expect dream content to shift predictably with emotional state.
What's your synthesis? Which theory or combination do you find most convincing?
I think the evidence supports a multi-function model. Memory consolidation is almost certainly happening during sleep, and dreaming is probably the experiential correlate of that process for certain types of memory — especially emotional and episodic memories. Threat simulation is an evolutionary overlay that biases which memories get replayed and how they're transformed. Emotional processing — the norepinephrine-offline desensitization — is a specific neurochemical mechanism that makes emotional memory processing possible without retraumatizing the dreamer. And activation-synthesis captures the mechanistic reality that a lot of dream content is driven by bottom-up processes that the cortex is trying to make sense of.
All the theories are partly right, and they operate at different levels of explanation.
Memory consolidation and emotional processing are functional theories — they're about what dreaming is for. Activation-synthesis is a mechanistic theory — it's about how dreaming happens. Threat simulation is an evolutionary theory — it's about why the system evolved the way it did. They're not competing explanations, they're different layers of the same phenomenon.
The pharmacology backs this up, because every drug that affects dreaming is messing with one of these layers. SSRIs alter serotonin, which changes REM architecture and emotional tone. Beta-blockers alter norepinephrine and melatonin, which changes the cholinergic-aminergic balance. Varenicline directly stimulates the acetylcholine system that drives dream generation. Withdrawal unleashes REM rebound.
The varenicline case is almost too perfect. Here's a drug designed for smoking cessation, and one of its most notable side effects is producing dreams so vivid that people write essays about them online. And it works by stimulating the exact neurotransmitter system that we know drives REM sleep and dream generation. It's like a controlled experiment confirming the acetylcholine hypothesis.
I want to push on something. If all these drugs produce vivid dreams through different mechanisms, is there a common pathway? Or are "vivid dreams" actually different things depending on the drug?
I think "vivid dreams" is a catch-all term that actually describes several different phenomena. An SSRI dream might be vivid because it's emotionally intense and you're remembering more of it. A varenicline dream might be vivid because the sensory imagery is literally more detailed — higher resolution, more color, more texture. A withdrawal dream might be vivid because it's nightmarish and frightening. A melatonin dream might be vivid simply because you're spending more time in REM.
The subjective experience of "vividness" isn't one thing.
That's actually important for people taking these medications. If your doctor says "vivid dreams are a side effect," that doesn't tell you what kind of vivid dreams you're going to get. The SSRI dreams might be bizarre and emotionally flat. The beta-blocker dreams might be threatening and fragmented. The varenicline dreams might be beautiful and lucid. The experience is different because the mechanism is different.
Which brings us to a practical question. If someone is on one of these medications and the dreams are distressing, what can they do?
First, never stop a medication without talking to your doctor — that's not just boilerplate, it's actually important here because sudden discontinuation can trigger REM rebound and make the dreams even worse. But there are some evidence-based strategies. For SSRI-related dream disturbances, sometimes adding a low dose of trazodone or mirtazapine at night can help — these drugs have different receptor profiles and can normalize sleep architecture. For beta-blockers, switching from a lipophilic one like propranolol to a hydrophilic one like atenolol often reduces the dream effects. For varenicline, the dreams usually diminish after the first few weeks as the brain adapts.
Some people actually like the dreams.
There are people who take varenicline and are disappointed when the vivid dreams stop. There are people who take melatonin specifically hoping for more interesting dreams. The dream experience isn't inherently negative — it depends on the content and the person.
One last thing I want to circle back to. You mentioned earlier that suppressed prefrontal cortex function during REM is part of why dreams feel real despite being absurd. Does any of these medications affect that specifically?
That's an underexplored question. Most of the medications we've discussed don't directly target prefrontal function during sleep. But there's some evidence that drugs affecting dopamine — like Wellbutrin — might alter prefrontal activation patterns during REM. Dopamine is important for prefrontal function. If you're boosting dopamine, you might be partially reactivating the reality-checking circuits. Which could explain why Wellbutrin dreams are sometimes described as more lucid or more controllable.
You might actually be more aware that you're dreaming.
And that connects to the whole lucid dreaming phenomenon, which is a separate fascinating topic. But the short version is that lucid dreaming involves partial reactivation of prefrontal areas that are normally offline during REM. If a drug partially reactivates those areas, you might get more lucidity.
The pharmacology of dreaming is really a window into the fundamental architecture of the dreaming brain. Each drug is a probe that tells you something about which systems matter for which aspects of the experience.
That's exactly how I think about it. The dream theories give you the framework — memory consolidation, threat simulation, emotional processing, activation-synthesis, the neurochemical environment. And the drugs are natural experiments. When you see that a cholinergic agonist like varenicline produces hyper-vivid dreams, that confirms the central role of acetylcholine. When you see that serotonin reuptake inhibitors produce bizarre and emotionally altered dreams, that confirms serotonin is modulating dream content, not just sleep architecture. When you see REM rebound after withdrawal, that confirms the homeostatic drive for REM sleep.
It's rare to get such a clean alignment between theory and pharmacology. Usually biology is messier than this.
It is messy. There's plenty we don't understand. But the broad strokes are surprisingly coherent. Dreaming is a cholinergically driven state with suppressed aminergic tone, reduced prefrontal function, and heightened limbic activity. It serves multiple functions — memory reprocessing, emotional regulation, threat rehearsal. And you can manipulate almost every aspect of it by targeting the relevant neurotransmitter systems.
The one thing I keep coming back to is how insistent the brain is about dreaming. You can suppress it with drugs, with alcohol, with sleep deprivation, but the pressure just builds. The brain will get its REM sleep eventually. That doesn't feel like a system that's just generating random noise.
The homeostatic drive is one of the strongest arguments that dreaming serves an essential function. The brain doesn't care that much about things that don't matter. If REM sleep were optional, we wouldn't see REM rebound after suppression. We wouldn't see rats die after prolonged REM deprivation. The system is built to ensure that dreaming happens, and that tells you it's doing something irreplaceable.
And now: Hilbert's daily fun fact.
Hilbert: The rules of real tennis — the original indoor racket sport from which modern lawn tennis descended — survive in their purest interwar form at a single court in the Faroe Islands, built in 1928 inside a former fish warehouse in Tórshavn. The court features a penthouse roof, three asymmetrical walls, and a scoring system derived from medieval French currency, where fifteen, thirty, and forty represent multiples of a coin called the gros denier tournois.
...right.
The question I'm left with — and maybe this is where we should leave it — is what happens when we get better at manipulating these systems. We've already got drugs that accidentally make dreams more vivid. What happens when someone designs a drug specifically to enhance dreaming? Or to eliminate nightmares?
That's already happening, actually. There's active research on targeted dream engineering — using sensory stimulation during REM to shape dream content, using drugs to suppress nightmares in PTSD. Prazosin, an alpha-one blocker, is already used off-label for PTSD nightmares, though the evidence is mixed. The field is moving toward much more precise interventions.
Then we'll probably be revisiting this.
I'd count on it.
This has been My Weird Prompts, produced by Hilbert Flumingtop. If you want more episodes, you can find us at myweirdprompts.com or wherever you get your podcasts. We'll be back soon.
