Is it silly to notice the superficial similarity between EMDR and REM (both involve moving the eyes) and effects on psychiatric conditions? Might the evidence for the mechanism of action for one be relevant to the other?
As a data point, I am sometimes depressed, but always feel least shitty in the morning, where I have the most energy and mental clarity, and feel most shitty at night. This seems to be one of those axes along which there's your pretty typical spread of variation; if depressives seem to disproportionately be more night owl than morning lark, that *could* be due to the effects speculated here, or it could just be an "everything correlates with everything else" thing.
On this blog I've heard a bunch of times about how sleep deprivation can help (temporarily) with depression, but out in the normieverse I only ever hear about how sleep hygeine is good for mental health. Granted that there are other reasons why "permanently sleep deprive yourself" (which depression might do anyway lololol) isn't a good idea, is this just the broader culture *assuming* that a generally healthy thing (consistent, full sleep) is good in every specific, or is it still best, including for depression? (It's possible this was answered somewhere in the technical details that I lack the brains and background to follow.)
If I understand this right, maybe the better sleep hygiene thing recommended for depression needs to be amended with restricting the amount of sleep you get, if you tend to oversleep like I do.
Would it make sense to "jumpstart" synapse density first thing every morning by reading/learning? Perhaps that habit would help keep depression at bay.
I didn't look too long but I'm wondering if to what extent they have tried extended sleep deprivation as a treatment. The link you provided seems focused on these 36 wake periods in the short term, not the long term. As in, rather than just see how symptoms are reduced over the course of one or five cycles of 36 hour wake periods or whatever number they've chosen, how are they reduced if this sleep deprivation is done as a consistent lifestyle? Obviously this would pose a problem with living your normal life (family or work obligations, etc.), but if you have a severe case of depression those may not be functioning well anyway and you might as well give this a try if a researcher is offering it.
I know the depression seems to come back as soon as they fall asleep again, but perhaps with enough persistence this effect could be reduced through some sort of positive feedback loop, knowing that at the end of the day you'll feel remarkably better and you only spend maybe half your time feeling like crap instead of all the time.
I always ask depressed patients about sleep. One strange thing in DSM and ICD criteria for major depressive disorder is the hypersomnia or hyposomnia. Too much sleep or not enough? Aren’t those completely different symptoms?
I like Tononi’s synaptic normalisation hypothesis and there seems to be good evidence for it. There’s also evidence that a lot of metabolic waste products and extra cellular debris are cleared from the brain via the ventricular system during sleep, so it’s possible there could be more than one purpose for the change in activity.
I seem to recall that the build up of substance P during sleep was proposed as being connected to the depressogenic nature of sleep. Conversely, the build up of adenosine during the day, to make us sleepy, was proposed as an antidepressive effect (I can’t remember how many years ago I read this), but this seems implausible as we’d therefore expect caffeine as an adenosine antagonist to worsen mood.
The more I learn about sleep, the weirder and more fascinating the links with mental health become. I have a patient who was diagnosed with a non-epileptic attack disorder by neurology, but who is describing many symptoms of narcolepsy with cataplexy, enough that we have referred to a sleep specialist who can do sleep and awake EEG and maybe a lumbar puncture to confirm hypocretin levels. The patient is also reporting frequent nightmares relating to past abuse and trauma, and I am wondering if their hypocretin levels are normal that perhaps psychotherapy for the past abuse might alleviate the nightmares, help the patient to sleep normally, and correct the sleep paralysis and possible cataplexy they are experiencing.
I’ve had patients with PTSD who were experiencing recurrent trauma related nightmares in *every sleep cycle* awaking about once every 90 minutes to 2 hours. Simple trauma focussed CBT fixed this, surprising even me. I’ve heard other therapists and patients describing massive improvements in sleep post successful PTSD psychotherapy.
On the seasonal affective disorder link, I note that melatonin is synthesised from serotonin, but have not been able to find any anatomist to tell me if the melatonin in the pineal gland is synthesised from the supply of serotonin manufactured in the serotonergic neurons of the raphe nucleus or not, so this may be a red herring. Still, we might except a hypersomnic depression to more closely resemble an SAD presentation and a hyposomnic depression. Of course you also know that mixed depression and anxiety or atypical depression is often noted to occur more frequently than depression alone.
There is also some suggestive evidence (https://pubmed.ncbi.nlm.nih.gov/11696070/) linking natural short sleepers to a kind of mild anti-depressed/hypomanic phenotype.
Possibly related: Narcoleptics supposedly enter REM state almost immediately upon falling asleep (talking like 5 minutes). I'm Narcoleptic and it definitely matches my personal experience in having wildly vivid and epic dreams even if I just doze off for 10 minutes. Does that have any further implications for these findings?
On the abnormally low confidence on predictions / overly strong or trapped priors theory of depression, it might not be as much a feature of the whole brain as of particular areas involved in reward, reward prediction, and reward prediction errors that are causally involved. Although the whole brain as a neural network could be using Bayesian prediction to predict inputs as per the Bayesian brain hypothesis, when it comes to depression (and perhaps anxiety) it could be regions such as the orbitofrontal cortex, amygdala, and the habenula that are affected in reward, reward prediction and reward prediction error.
First, I am curious whether you mean "batch normalization" of artifiacal neural networks, or you actually mean "renormaliztion" which is something different and much more esoteric and unusual.
Batch norm does something like regularize neural network inputs so that all the inputs are rescaled into the same domain of possible values. A loose analogy would be your perception that an object "looks the same" whether it is illuminated by 10 lumens or 10,000 lumens; your eyes and brain have done a normalization for you across a broad range of objective input values.
I'll not from a machine learning standpoint that learning is all about leveraging nonlinearities. If you are routinely going through your network and just kind of smoothing everything out, then algorithmically you are probably making 99% meaningless changes that do very little and 1% terrible changes that effectively destroy some precise, interdependent arrangement of synapse weights.
It might be fun to take those creepy amorphous dog-generating GANs from a few years ago and intentionally do such an operation on them, "averaging down" all the weights, and then observing the result. I am not even all too sure what to predict. It is possible that, instead of seeing a frothing sea of Lovecraftian doggoths, you would get a more vague and less coherent visual image. It's also possible that you would still see some dogs, but maybe fewer. Or maybe you would see a disorganized mass of eyes and muzzles with no structure. Or maybe it wouldn't work at all.
This, plus experiences of friends and loved ones, definitely makes me wonder if some depression-adjacent sleep deprivation is trying to hold on to/chase the feeling of increasing synapses? Especially because getting enough sleep is something I had always tried to help depressed people with. But Luke comments below, maybe the negative side of sleep deprivation is just the extra-long sleep phases that follow?
Anyway, we can help with the cabin in Alaska, if you're looking to put your six months in.
Definitely a very interesting theory, but I'd be very cautious about buying into it. Just because staying awake makes you feel better and sleep makes you feel worse, doesn't mean sleep is doing something bad for your brain.
There's an analogy with muscles and sleep: If you work out really hard and then get a good night's rest, the next day you'll be sore. If you work out really hard and only sleep a few hours, you often don't get sore the next day. (That sounds counterintuitive, but it's a widely experienced phenomena.) But skipping sleep to avoid soreness is absolutely the wrong thing to do! The soreness is part of the recovery process. If you keep skipping sleep, your muscles will get weaker and weaker as they never recover from your workouts. So, by analogy, I'd worry that skipping sleep to treat depression might be trading off important brain recovery mechanisms.
Somewhat relatedly, the book "Why We Sleep" by Matthew Walker discusses studies that found that REM sleep reduces trauma: in sleep, the person re-lives traumatic memories, but in a detached way due to the emotional system not being fully connected. This causes the memories to be less emotionally-laden and traumatic going forward (or, one might say... "downgraded and renormalized"...)
Those studies might be relevant to the REM / non-REM question.
Extremely speculative, wondering if others have the same experience:
I often find reading a few highly analytic longposts is the best and only consistent way I can start the day. It seems to actually get my brain "working" while I can't motivate myself to do anything else.
As per the post, the later the day gets, the more energy I have.
But I wonder if reading non-fic essays is more "stimulating" to neurons than say, making breakfast, or going on a walk?
Do anti-seizure medications sometimes called “mood stabilizers,” like lamotrigine and gabapentin for example, ever treat depression? If they are ever effective in treating depression, not just anxiety or bipolar mania, how does that relate with the parts of this post about seizures, if at all?
Is bipolar depression so different from unipolar depression that anti-seizure medication can treat it effectively whereas unipolar depression can be described as effectively treated in the opposite way?
If anxiety can be a symptom of depression, again, why are another kind of anti-seizure medications, benzodiazepines, so effective at treating it? Admittedly they are described as potentially causing or worsening depression. And often prescribed together with other medications.
This post made me wonder, why is it that anti-seizure medications are ever safe enough and even helpful to give to depressed patients?
But I think I can answer that for myself: there are thresholds, risks, sometimes opposing symptoms and conditions co-occur and need treating simultaneously.
What I am more curious about is: does any kind of depression ever look like a seizure disorder instead of something like the opposite?
It also seems relevant that sleep-disordered breathing is common and often goes undiagnosed, yet is often comorbid with depression. Is it possible that some of of the correlation is because people with sleep issues often feel worse when waking, and people with sleep issues are more likely to have depression?
It's kind of hard to find great numbers on this, since at-home sleep studies frequently miss sleep apnea and usually miss other issues, but according to random popular articles (Healthline and similar), about 40% of people with obstructive sleep apnea (OSA) have depressive symptoms, and possibly 15 - 20% of depressed people have undiagnosed OSA. Again, I'd expect these to be underestimates, since the one reference I clicked through (10.1016/j.jpsychires.2019.06.015) relied on really basic at home tests for half of their data.
To me it is crazy that testing for sleep issues is not standard practice for depression given that OSA and other sleep disordered breathing issues are common organic causes for depressive symptoms, and treatment (PAP, surgery, etc.) can improve these symptoms, but that's a bit of an aside.
Sleep deprivation leads to ROS accumulation in/on the luminal-side of the gut but not other organs (including the brain). sci-hub.se/10.1016/j.cell.2020.04.049 This is evident in experiments with mice and flies after even two days of deprivation. Graphically evident with photos as in Fig 4. With longer deprivation, death results. If the luminal-side ROS is discharged through the by-mouth use of various compounds such as NAC, melatonin and lipoic acid, the animal is rescued from death even with sleep deprivation. Recovery sleep dissipates ROS accumulation in the gut. The effect of sleep deprivation is not all (or ?even?) in the brain.
The gut microbiota influences the brain and behavior through, among other things, the vagal nerve. sci-hub.se/10.1016/j.neuron.2019.02.008 Preventing vagal communication between the GI tract and the brain increased occurrences of psychiatric-related disorders. Vagal nerve stimulation can substantially improve MDD. Gut bacteria also influence the brain and psychiatric disorders through the vagal nerve. The vagal nerve has about 80% afferent (gut to brain) and 20% efferent (brain to gut) fibers.
Hypothesis: If the depression is caused by a specific gut ecosystem signaling, suppressing the signal between the gut and the brain through ROS accumulation in the gut could suppress the depressive symptoms. As demonstrated in the first reference, sleep dissipates the deprivation induced ROS. A possible test of this hypothesis would be to have patients whose depression are relieved by sleep deprivation take one of the tested supplements (e.g. lipoic acid, NAC, et.al.) and test for continued depression relief with sleep deprivation.
I guess this theory can explain your problem of trapped priors: Whenever a small not-so-negative experience (eg a only slightly aggressive dog) happens it gets deleted by a faulty renormalization afterwards and only a clear positive experience (eg a cute puppy in a cage) can help. Apparently, a correct renormalization would keep experiences of the first kind, but for depressed people it is more likely to throw small updates away (since their brains have less connections overall).
Maybe I'm melancholic depressive, though I don't know what that means. Any time I wake up at maybe 6-8 am, I'm suicidally depressed. The first thing I think on waking is, "Dammit, I'm still alive." But I know the feeling will dissipate in an hour or two, so I try to ignore it.
This happens regardless of my sleep cycle. If I go to bed around 3am and get up between 10 and 11, I have no depression at all.
(I haven't noticed whether it varies with season. I don't have natural optimal sleep-wake times; I have a natural daily cycle of 25 hours, so my sleep cycle moves forward 1 hour each day until I force myself to get up early and suffer a sleep-deprived day to restart the cycle.)
This corresponds precisely to my experience. I never thought about this systemically before, but I have always struggled with getting out of bed in the morning due to an overwhelming feeling of melancholic depression, exacerbated by the fact that I stay up later than would be wise for my own good because I usually feel most at peace with the world and therefore energetic at night. The occasional time I manage to stay up all night, I get to work the next day with verve and a more positive outlook, partly because of a feeling of accomplishment I suppose. Unfortunately there does not see to be any apparent solution to this, seeing as eventually Morpheus does call for his due.
It looks like a few people here are relating this blog to the one on trapped priors. Eric Hoel has some work (e.g. https://arxiv.org/pdf/2007.09560.pdf) on "The Overfitted Brain: Dreams evolved to assist generalization" that seems relevant. His theory (which he presents some data in support of but I don't have enough knowledge of the field to know how convincing it is) is basically that the brain, like a neural net, is prone to overfitting based on the data set that it encounters. And in neural nets they often solve that problem by adding noisy or corrupted inputs, which make the fit worse on the existing data but make the system overall more capable of being generalizable and adaptable. And that dreams are the human equivalent of that. From the abstract: "Dreams are a biological mechanism for increasing generalizability via the creation of corrupted sensory inputs from stochastic activity across the hierarchy of neural structures. Sleep loss, specifically dream loss, leads to an overfitted brain that can still memorize and learn but fails to generalize appropriately." I don't think I can quite put my finger on the model that connects them (maybe because it doesn't exist) but there seems to be a strong common thread between depressed brains suppressing new input, and dreams purported function in preserving openness/flexibility, and sleep being messed up in depression.
I wonder if this explains my experience, which is that I feel like I undergo a steady uptick in anxiety the longer I stay awake, which I think is a fairly general phenomenon, ie "your brain at 2am suddenly bombarding you with restless thoughts."
So what does waking up around 3AM, often associated with depression, do in this model? Is some failsafe processing in the brain detecting "too much" synapse clearing and waking you up to stop it?
>unsurprisingly SSRI antidepressants decrease REM sleep a lot (not just in depressed people, in >everybody).
I found this pretty surprising and would be interested to hear more. How significant is the decrease? how dose-dependent is it? etc.
Background:
I have had insomnia for basically my entire life which could charitably be described as "severe" and uncharitably described as "Holy F****ing H*** how are you still alive?!?!" I take a few different meds and supplements for it, in addition to sleep hygiene and other lifestyle interventions, but the main medication is Trazadone, an SSRI, and when my sleep specialist first prescribed it after my sleep study (which was apparently pretty horrific looking, even to someone who dealt with severe insomnia cases every day) she said it had been shown to help produce deeper more restful sleep, rather than merely helping with sleep onset. This was important because a big part of my problem was not getting any rest even when I theoretically slept; my sleep study showed (IIRC, it was almost 20 years ago) something like 5 hours 'asleep' out of 9 hours in bed, only about 20 minutes of that REM, and zero slow-wave sleep, I was just staying in the first stage of sleep all night, when I was asleep at all. No one sleeps well during a sleep study, you are in a strange room with electrodes taped all over you, but this was bad even accounting for that. I guess I am rambling quite a bit now, but my point is I was very surprised to hear SSRIs would /reduce/ REM considering what my sleep specialist said back in the day, and how well it seems to help me given how I sleep when I am not on it.
Low confidence in predictions may be an accurate read of reality in the face of a stressful or unpredictable environment. In this way it could be an adaptation to stress which could also include social withdrawal, apathy, and pessimism, or seen differently a switch into a mode of conservation and survival as opposed to a mode of exploration and expansiveness.
The amount of sleep that people have goes down with age, until retirement when it starts going up again.[0] Definitely some of that is due to the amount that they're able to sleep (a person with a job and two kids is going to have very little sleep), but is any of it related to different needs?
Has anyone ever tried sleep reduction therapy to treat depression? If avoiding sleep entirely seems to work until you go back to sleep, what happens if you only sleep 4 hours a night? How long can a human go with artificially reduced sleep?
What about going the other way? If we make people take sleeping pills to sleep more, do they develop depression?
For both questions, does changing sleeping patterns eventually cause the brain to renormalize to that level of sleeping?
Perhaps it is an under stimulation problem? You sort of allude to this in the post, but do not really explore that possibility further.
I basically fit this description to a tee. I have some intermittent problems with depression. And sleep deprivation removes it completely. Periods of slowly slipping into depression followed by sleep deprivation to snap me out of it used to be a recurring pattern for me. But I notice that when I force myself to get seriously mentally stimulated (reading some difficult texts, doing math problems), the depression goes away. Or is less severe. Especially when done earlier in the day. Caveat though, it is not major depression for me, although suicide has occasionally occurred to me in the past.
And actually working too hard, spending too much time and mental energy during the day gets me into mildly manic moods where I have a hard time falling asleep. Usually followed by tiredness where I spend less time doing difficult tasks, where I slowly slip into depression again.
This is so effective for me, that when I get depressed I know I have been lazy and need to improve my work ethic. It is kind of annoying that this only works for tasks that I do not easily pick up, especially when depressed. So once I slip into a depression it can be hard to get out of it without sleep deprivation.
So each person might have different needs for mental stimulation and sleep. And perhaps sleep deprivation to snap out of it, followed by a constant diet of doing mentally taxing and rewarding tasks is the cure for people who suffer from this. It helps if you set an overarching and at least somewhat ambitious goal that you can get excited about.
It seems like a reasonable inference from this is that someone should try treating depressed people with small doses of melatonin, like it's a sleep phase disorder. Does anyone know if this has been tried? It's in no pharma company's interest to find out if melatonin can treat depression so probably not, but maybe that fancy patented melatonin that there was an SSC post about once.
Also, what about depression with atypical features? It's definitely not rare. So it seems like there are also a lot of depressed people who feel *better* after sleeping, and I wonder how that works with this theory.
1. "You can't renormalize while the network is running." Says who? This seems like it might be a case of unconsciously extrapolating from *human* designed machines to natural machines. Human beings, for reasons of design and construction efficiency, do indeed design machines that generally need to be shut down in order to be worked on -- we have to close the road to repave it, we have to shut down the computer to replace its graphics card.
But I would say as a rule biochemistry tends not to operate this way -- there tend to be many multiple overlapping and redundant paths for getting the same, or similar things, done, and it appears pretty common for the body to work on a system at the same time as it operates. Bones don't grow longer (in children) only when they are not bearing weight, we don't heal wounds and injuries better or faster when we don't use the injured area, and we don't always choose to run only catabolic or anabolic pathways for the same substrate.
So I would say this idea should actually need to be measured to be sure it is true, especially since it's a key assumption underlying the idea that retuning the synapses needs to be done during sleep.
2. Why is it said depression gets "better" when one is sleep deprived? Maybe sleep deprivation merely adds a temporary cover layer of mania that hides the depression, as if you gave the depressed person a massive dose of amphetamines. That is, the change may not be in the direction of cure, but in the direction of adding *another* layer of brain dysfunction that conceals the original layer somewhat, so, away from a cure. If so, the last thing we'd want to do is try to extend it.
3. If a *general* synapse realignment underlay depression, why wouldn't there be much broader associated brain dysfunction? That is, shouldn't depressed people also have noticeably lower IQ, poorer executive function, worse memory -- even compromised sensory abilities and downgraded motor skills? That is, shouldn't an across-the-board physical degradation of the brain have across-the-board degrading effect on brain function, from soup to nuts? Why would we see the primary symptom in mood (which seems a fairly high-level complex epiphenomenon of neural function) rather than in, say, memory impairment or crappy motor skills, something more basal?
There are some ML failure modes this brings to mind as potentially useful ideas here.
Sparse gradients and local minima effectively make improving a model impossible. In practice, this usually means you throw the model away in start over. But you can't do that with a human, so you need some way to escape from the local minima. Stronger gradients are one way to potentially do that - a steeper gradient will tend to push what would have been smaller updates into bigger ones. The downside of taking bigger steps like this is that you often end up in a worse place than you started, because bigger steps means that the local gradient is less predictive of how good your new state is going to be.
But if your problem is escaping a local minima, that can actually be a very good thing! Even if your immediate situation is worse, you *might* now have a path to improvement, since even if your new state sucks, *it might be next to a place that doesn't suck*, and that might be next to a place that sucks *even less*, and there might be a path through state space all the way to an actually decent model.
More generally, a lot of depression treatments looks awfully similar to the kind of desperate tricks you might use to get a model out of a local minima.
Scott's is an interesting theory but it does not hold. People with schizophrenia also have lower synaptic density (see https://www.nature.com/articles/s41467-019-14122-0 ) but so far nobody noticed that sleep deprivation helps with schizophrenia.
Secondly, my guess would that the negative correlation with the depression score will turn out to be spurious when we get a publication with a larger sample. The sample in the article Scott quotes is very small (n=40).
On a lighter note, psilocybin increases synaptic density after a single administration (in pigs) https://www.mdpi.com/1422-0067/22/2/835 - Looks like we have a new therapy for schizophrenia and depression. And also you do not need to sleep, if you take some psilocybin, if Scott is right.
The #1 most common side effect of TMS depression protocols is extremely vivid dreams, by a long shot. I have had one of these protocols and can attest, it's friggin crazy. If it's non-REM sleep that's doing harm, maybe TMS increases not only the intensity of dreams but the amount of REM sleep.
Gina actually described three stages, REM, deep sleep, and a "transition period".
Deep sleep is the physical repair time, when the brain is least active.
REM and the transition period both are connected to the rewiring of your synapses, but one strengthens new connections, the other clears out unneeded connections. Or one organises, the other files/dumps depending on the results of the organising. Figuring out exactly what happens is obviously an area of ongoing research.
Two interesting notes that might help along or that you may find extremely boring:
* Donepezil is a great way to cause more REM, other drugs that act in complex ways on the cholinergic system do too. Looking at trip report of these might be interesting. The big problem here is that most people taking these REM-increasing drugs do so during the day and have serious conditions (e.g. alzhimers)
* REM and memory consolidation are potentially inversely related (not sure if this is what you mean by "sleep renormalizes the hippocampus"), or rather, REM and short term memory consolidation may be, at least in that N2 and N3 seem to play a big role there and increase N2 and N3 seem to => short term memory consolidation. And I'd usually associate a longer REM with a longer N4.
* Pulse oximeters can kinda-maybe measure sleep stages well enough to be not noise and so can some comfortable 2-4 neuron EEGs. These will never be used in a study *but* trying to get feedback on your sleeping patterns and what helps might be a much better way than speculation here.
* (Maybe obviously) Short/Long sleep studies with depression may be hopelessly confounded by common drugs (caffeine, nicotine, alcohol, benzos) which affect sleep a lot, and which depressed people would be prone to take, and unless you've looked into the methodology closely it might be worth batting an eye over (though maybe you did and you think it controls for this properly => ignore me, but it seems like the kind of thing that's hard to deconfound for many reasons, one way or another you end up with weird samples)
So we've established in this post that depression is caused by sleep, particularly REM sleep. It's well known that REM sleep is the only kind of sleep where dreams occur. When dreaming, your brain generates some arbitrary thing-that-resembles-a-narrative-story, and this thing further reinforces the intuitive idea of being a self-controlled sapient character in a grand narrative of the universe. In Buddhist/ish thinking, the idea of being a self-controlled sapient character is a myth and being enlightened causes one to intuitively grok it as a myth. So chaining the logic together we should conclude that enlightenment is a cure to depression.
Stupid thought on REM vs non-REM section: depressed people actually have MORE synapses, not less. Depression is a brain wave "brownout": barely enough to run everything at minimum capacity assuming nothing else taxes the system. As more synapses form throughout the day, the brain has to upgrade to full "blackouts" in various regions to keep the brain from permanently damaging itself, which frees brain waves to run everything else better, which is why they seem to get better as the day goes on, and why sleep dep seems to make things even better: more blackouts free up more brain waves for the rest of the brain. They get just enough non-REM sleep to normalize back to brownout levels, which starts the cycle over again. I would expect this to be falsified if the entire brain gets reduced function instead of just isolated regions during depression, and I would expect brain frying drugs like marijuana to help. I'm probably totally wrong though and welcome correction.
Pushing back on the "number of neurons doesn't change" claim, neurogenesis (new neurons being born and integrating into the circuit) is not only a thing (and despite some claims pretty likely to be happening in adult humans for a limited number of cell types) but is also proposed by some as the main mechanism for the positive effects of SSRIs, exercise, enrichment, etc on mood, since all of these increase both neurogenesis and mood.
If seizures help treat depression, then why are people with epilepsy more likely to experience depression? Or do electroconvulsive seizures work differently than epileptic seizures?
In my view to consider sleep deprivation as a valid approach is nothing short of insane. Reminds me of lobotomy. As a side note, the underlying specific assumptions of a computer-like functioning of the brain seem to me a strange caricature of current knowledge. I guess I hadn’t been connected to the community for quite long and forgot the strange belief system at work..
If rem sleep clears out the short-term memory and a having full short-term memory bank is linked to the reward system that would make sense from a more intense morning depression and lesser in the evening standpoint. Would then refilling of the short-term memory banks upon waking by reading something particularly interesting/enlightening (such as this blog) create a large number of new memories right off the bat and kick-start the day with a more elevated mood due to the well topped up short-term memory? Are there more effective ways of filling the short-term memory than reading, such as cramming Trivial Pursuit cards or listening to an audiobook on double speed (doing that feels like an information download straight to the brain)?
So the overall result is to increase "equality" of synapse strengths throughout the affected regions? This sounds kind of like destroying information, but maybe if the information is the pattern of connections making you depressed that's good?
From an information theory perspective, increasing equality doesn't necessarily reduce information. Suppose the brain had 1 really really strong synapse. That isn't much information. It isn't hard to point out which synapse is the strong one. On the other hand, if all the synapses were equally weighted, that carries no info. (But might make the info about where the synapses are easier to access.) Much more info can be stored by giving half the synapses a weight of 2, and the other half no weight. The distribution of synapse weights that maximises info (given fixed total synapse strength) will conceptually resemble the distribution of kinetic energy in an ideal gas.
The actual neuron weights will follow a geometric distribution (in this simplified model, where we are maximizing info with a bounded total weight and weight is just the count of some molecule), with a parameter that depends on the total weight bound.
TL;DR. A long tail of really strong synapses doesn't hold much info. A more even spread of strength can be more informative.
When I was in college, I was giddy-drunk if I went without sleep for a while, and that was fun. (Although I didn't drink in college, so I can't really compare.) I think my inhibitions just went way down and I'm normally too cautious.
Looks like they understand depression about as well as digestion. I can tell you for a fact that living with an optimist who loves you but won't put up with the attitude after a point can cure depressive tendencies over time.
>Is it fair to say that "decreased synaptic strength" is sort of the same thing as "abnormally low confidence on predictions"?
1. Consider that what may be involved is something like a "ready to hand" instrumental confidence in the effects of the {depressed, manic} person's actions; we can try to gloss this as corresponding to the epistemic confidence they have in their predictions, but that abstraction may be misleading in a lot of cases. (Famously it's hard to translate the ability to tie your shoes or execute a tennis serve into "predictions" or "decisions", and a lot of quotidian executive-function tasks are likely like that.)
2. What may be going on is that the depressed person's brain is *too good* at making the synaptic connections. If this manifests over the course of months, we call it a bipolar disorder. If their brain is somehow wise to what's going on and adjusts daily (nightly), then they wake up despondent/hopeless/directionless, go to bed slightly manic, and are at their most average functioning mid-afternoon.
3. This meshes with a lot of other pathways/covarying cognitive deficits. This kind of depression is connected to anxious behaviors like compulsive procrastination; that would make perfect sense if ingrained compulsive habits like e.g. checking blogs (and writing blog comments?) provide extremely predictable feedback. (This is something I think makes more sense on a "ready to hand" model than a literal "epistemic confidence" model; it's not clear why doing something just to be doing it would result from reliable predictions, since anxious procrastinators can usually predict when prodded that nothing valuable will result and they will regret procrastinating.)
4. It explains some stylized facts like: (a) why melancholy tends to attack people in relatively independent careers (they have more options and are vulnerable to decision fatigue early in the morning, and have fewer forced actions to get them through their low-functioning part of the day); (b) why they tend to have self-perpetuating circadian rhythm problems at night (they are most overconfident in the value of doing one more thing when they need to go to sleep, and have very firm convictions about tomorrow's agenda that they meticulously plan out when they're lying in bed); (c) why some people are convinced lifting weights helped their depression (very simple practical cause-and-effect relations, usually early in the morning); (d) why mastering some new hobby with quantifiable success/failure/improvement metrics often brings people out of a funk for a while.
I myself was in a melancholic funk a long time ago, and it stopped at the same time I started a ketogenic intermittent fasting diet that involved going to bed hungry and not eating until I finished my morning workout. At the time I thought that my rising spirits gave me the willpower to follow the diet, but now I wonder if it wasn't the opposite; starving my hyper-activated synapses in the evening forced me to bed and decreased the amount of overnight synapse-correction I needed. — Meanwhile I also wonder if the problem didn't originally arise because for a long time I tried to save time and diet by skipping breakfast, habituating my brain to absence of sugar in the morning. (Consuming my morning caffeine with artificial sugar instead of real sugar probably didn't help.) — That said, I wasn't at my most productive while practicing intermittent fasting; my most productive years in recent memory were again a period where I woke up in a fog, got nearly everything done 1pm-5pm, sometimes felt like trying to get a lot done at midnight, and sometimes had sleepless nights. So it's possible the overall pattern of strengthening synaptic connections over the course of the day isn't purely dysfunctional or diseased, even if it carries risks.
As someone with lifelong depression, I can attest that the sense of being "fully activated" only late at night is very real. I do my best creative work between about 10pm and 2am. I used to intentionally plan all-nighters when I had difficult school work or papers to complete; I eventually just realized that if I wanted 100% productivity, I had to do it late at night.
At the ends of semesters, I'd sometimes take a 36 or 48-hour stretch of final paper writing during which my productivity would reach *extreme* heights. For example, producing multiple graduate-level papers which received high grades in just a few days - something I *cannot* do normally. Toward the end of these periods, I'd experience a sensation I'd describe as feeling "too awake" - extremely alert, quick thinking, rapid ideation, borderline manic. If you've ever taken ADHD medications in a somewhat too high dose, it was a very similar feeling.
So, this discussion seems extremely accurate to my personal experience.
Is it possible to self-medicate depression by restricting sleep or reorganizing your life around it somehow? I'm thinking about shifting your work and important stuff to the evenings/nights and instead taking it easy in the mornings. I was reminded about how as a student I tended to gravitate towards doing more studying in the evenings (except for bar nights I suppose). It seemed to work then, and aside for a preference for having evenings off I don't see why it wouldn't work now as well.
This would seem to support a more complex view of the role of REM sleep in depression. SSRIs reduce REM sleep and fight depression; alcohol reduces REM sleep and worsens it.
Throwing another anecdotal evidence on the pile, I've always been a night owl, I find mornings difficult even though I've had to get up early for years, and on top of that I tend to have insomnia - either it takes me forever to fall asleep, or when I do fall asleep I keep waking up throughout the night.
And then when I get the opportunity to catch up on sleep, I can sleep for hours upon hours. I don't know how much linked that is with depression, but it certainly *feels* like it.
The times I have been sleep-deprived, I have ended up bouncing off the walls, but I don't think it had much of an effect on my depression; instead of feeling unmotivated and 'I hate myself', it was "Whee! I can finish all these tasks! *And* I still hate myself! Oh look all the straight lines are all bendy..."
What of other animals? Bats sleep much more than people, but have much simpler brains. Wouldn't you expect the much greater number of synapses to lead to some notable difference in sleep patterns or the types of chemicals released in sleep or something?
I was also thinking that dementia stems from a sleep disorder. It’s just that my mother who had something going on (neurologist could not define it other than dementia) would have moments of clarity then drift off to what seemed like she was describing a dream when she spoke. Meaning everything she was saying was disconnected like what happens in a dream.
What does this imply about people who need way less sleep than the average person?
Back in college the seniors would have a rotating shift of "night watchmen" who had to wake up and patrol the dorm a couple times a night to make sure nothing crazy was going on. When it was my turn I remember be constantly unnerved that the same 5 or 6 people out of a thousand and change were always up and about at 3 a.m.
Could it be that the "normalization waves" of these folks are abnormally strong and accomplish the task in just a couple hours? Or maybe they are susceptible to depression effects and have just evolved a strategy of exposing themselves less to sleep to maintain better mental health?
Glymphatics anyone? Searched for this in the comments along with lymphatics and aquaporin and go not hits.
Recently(ish) the extracellular mass transport of fluid has been shown to double during sleep. This is thought to be protective for Alzheimer's disease for example. It is also highly dependent on aquaporin 4 expression.
I note that early on Scott writes "As if depression is caused by some injury during sleep". The opposite could be true.
Hypothesis: some semi-soluble compound in the extracellular space is beneficial to depression but sleeping allows it to be washed away. During the day it reaccumulated.
So this coincides with a weird issue I'm investigating in my own brain.
For background, I have atypical depression (diagnosed, used to be medicated, mostly in remission for now) and probably ADHD (diagnosed as something ADHD-adjacent as child when that was not a category in my country, unmedicated). Mornings are universally the worst part of the day, both in terms of mood and sleep latency. Coffee helps a little bit but is mostly drank for pleasure. Brain fog is present during the entire day, but I manage.
The thing I discovered by accident is that drinking coffee in the evening gives me superpowers. I never liked to work in the evening, instead reserving it for chilling out, but with a sufficient burst of caffeine I get an incredible amount of stuff done, concentrating effortlessly while in a hypomanic state. Distractions that normally completely destroy my productivity are no problem - I can easily sustain conversation via IMs and not lose track of work, for example. This effect works _only_ in the evening and late afternoon.
This makes absolutely no sense under the standard dopaminergic (no relation to circadian rhythm) / adenosine antagonist (related to circadian rhythm, but it just makes you feel non-sleepy) explanation of caffeine action. The circadian depression voodoo explanation is... tempting, to say the least.
"This sounds kind of like destroying information, but maybe if the information is the pattern of connections making you depressed that's good?"
If I've got a textbook in mostly nine-point type, but one random sentence on each page is in forty-point bold, I don't lose any information and I do gain more efficient access to information if I reformat those sentences to the same font as the rest.
If it's not random sentences but the most recent sentences (the stuff I experienced yesterday and was supposed to reformat down to standard-archival-format), that sounds like it might combine synergistically combine with some poorly-chosen trapped priors. If my recent experiences are of the "life sucks, there is no hope" variety, and those keep getting overweighted, then I won't be able to draw effectively on my previous experience of life not sucking.
In which case, excessive sleep alone won't always cause depression, but would aggravate other potential causes of depression and impede natural recovery.
What about a simple alternative explanation: patients are going through withdrawal in the morning, because they haven't taken their SSRIs or other meds in 8+ hours.
Huh! I've noticed that during depressive episodes, my dream content "lags" by a few days. For example, if I watch a movie, those characters will show up in my dreams three days later, where normally they'd show up the same night. Related, maybe?
I have an interesting perspective of this, having done night work for over 30 years. Not just 30 years but for much of it six nights a week. Anyone who has worked the third shift can relate to the unnatural sleeping patterns. My day off (Friday--today in fact) I spend up to enjoy life the way day-trippers do. I am just about at 24hrs and going strong. Now I would not want to take a long drive at the moment, but lets hope this has some clarity.
Regarding depression, my own personal experience has been that sleep, deep sleep, is much needed to ameliorate it. Now, there is merit in sleep deprivation. The longest run I've had is three days straight. I fainted eventually (at the breakfast table!) but here's the odd thing--on the way home, I felt like a million bucks. A few things played into this I believe. The weather was warm and sunny, I kept myself moving (stopping is the bane of the sleep deprived) and was relatively trouble free. A miserable, rainy day in the winter after my cat just died might have another outcome. Never-the-less, I still fainted. My body pulled the plug on my mind.
Over the years, I have noticed a curious pattern. While I can stay up for long periods of time, I find after deep sleep, my mind fires on all cylinders. Let me tell you about my old friend Frank. His father was in a rest home and I would visit with him from time to time. Frank would warn me his pop would need a good 30 minutes to get "his brain moving" as he described it. He said it was almost like the blood had to get to his brain and took that long. Slowly his pop would come out of the malaise he was in, know our names, engage in conversation. I said I often stay up all day Friday. Depression (when it hits me) is Saturday morning. Early. 5am maybe after a good 5 hour sleep perhaps. I could lie there turning over thoughts that somehow seem the have the better of me. So, I one morning thought of Frank's pop. Thought perhaps my own blood had to get flowing, and simply chose to let my mind worry later, at a time when my mind was better apt to take the trouble on. Just as I would not fancy a long drive now for safety issues, I applied the same to my mind in that state. Fast forward to Monday night now. I am driving in, bright and alert, same troubles and "I've got this', I can say to myself. All my mind needed was rest. I've become a big advocate of proper sleep to bring clarity and convinced lack of sleep causes shotty thinking. I don't deny the research here, to some degree have experienced it, but also the benefits of a rested mind to achieve the same. I've seen the depression "reset" Saturday morning but the exact opposite in direct proportion Monday night sleeping sound all day.
I've often felt like i get more tired and depressed towards the end of the day. I have a demanding job, two kids, and aging parents. Towards the end of the day, I often feel down about almost everything. That seems to be the opposite of what the article describes.
When i wake up, this generally isn't the case. I've often used the image of something like 'light' that illuminates the space of possibility nearby - both what's happening now, and what might happen. When I feel good, this light is bright, and when i don't feel good, it's as if the light is 'dim' - everything just seems more dismal and hopeless.
That might make sense of connection strength gets stronger over the course of the day, and strong connections require more energy to maintain. If the connection strengths all get downgraded evenly over the course of sleep, then it takes less energy for my brain to map out all the various connections between things in my immediate environment and, as a result, i can more easily perceive the subtle, positive things that make life seem better.
It often feels, when I'm tired or angry or sad, like my experience of the world reverts into a 'cartoon' state, with all of the unpleasant feelings highlighted. That might seem consistent with only the strongest connections getting enough energy to run?
If we follow this theory that (certain patterns of) increased synaptic connection treats depression, and evidently our activities over a day tend to produce such, do we know if there are particular activities that do so disproportionately? It might make sense to front-load those.
There are some candidates on how top-down theories like predictive processing and bottom-up theories like homeostatic disregulation can meet.
In predictive processing, there are two major candidates on how/where the downwards prediction and the upward sensory information might actually be implemented.
One is that different parts of the layered structure (https://en.wikipedia.org/wiki/Cerebral_cortex#Layers_of_neocortex) of the cortex are used. Details are complicated, but probably upwards stream mostly arrive in layer IV, while the comparison with prediction might be done in layer V, or possibly in layer II/III. We don't understand the details of this "canconical microcircuit" very well yet, but there is a lot of structure.
The other (complementary) candidate is that different neurotransmitter are involved. Upstream information uses mostly AMPA (which is a LOT faster and is responsible for the lion share of energy consumption), while predictions probably/perhaps involve a lot more NMDA receptors (slower and longer-lasting, strongly involved in plasticity and learning).
Homeostasis does not affect all these structures the same way. As far as I understand Tononi's theories, mostly the AMPA part is upregulated over day and downregulated at night. Also, different layers are affected differently.
Once you know such things, you can start to speculate. If AMPA is stronger in the evening, does that mean that predictions are relatively weakened compared to sensory information? I don't think it gives a coherent picture (yet), and I don't think that we are at a point where we can match top-down and bottom-up theories. Probably we won't be there in the next 10 years. But I don't think it is hopeless in the long run.
Maybe but I still out substantial credence on finding out the sleep result reflects affects to the endogenous opiate system/cycle. I'd love an experiment that measured levels of those endogenous opiates over the time period.
What about schizophrenia? Sleep deprivation often leads to psychosis and the negative symptoms are a lot like depression symptoms? What's the interaction there?
It's weird to me to hear that sleep deprivation can help with depression, because I'm pretty sure sleep deprivation makes me *more* depressed, at least in the long term. But possibly short-term sleep deprivation is helpful and long-term sleep deprivation is harmful?
During college, I did a research paper on synaptic scaling during NREM sleep, and ran into Tononi on a ludicrous number of articles. I started noticing his name popping up on all the interesting ones, so I just did a search by author. I don't know how the **** he does it, but his name is on all sorts of crazy papers that are clearly trying to become a groundbreaking paradigm shift in their own way. At this point I'm convinced that he has a network of contacts across the entire field that give him early-warning alerts on weird new papers so he can swoop in and coauthor them.
For what it's worth, btw, the consensus among researchers ca. 2018 was shifting towards Tononi's NREM model being less helpful than the other big model in the field. There were a couple of studies that looked really slam dunk until one of the big names on the other side pointed out that they were anesthetizing their mice with an agent that specifically interfered with the part of the brain they were looking at. I don't know what the fallout was from that or if Tononi's people recovered, but it was an oof moment for sure.
Huh that explains it for me, thanks! "It" being: I sometimes feel depressed early in the morning, like first 5-30 minutes after waking up, although I'm normally not and never been diagnosed. It happens without any apparent reason like sleep deprivation, hangover or shitty day ahead/behind. And it's not just low energy, I literally feel like I'm worthless, my life sucks, things I said and done yesterday are all failures etc, even though I remember it's not what I was thinking yesterday and know I'll probably feel differently in half an hour. I eventually assumed it's some sort of a brain glitch and just made a habit of not thinking of any such matters right after waking up, now I finally know what kind of glitch this (probably) is.
As someone who experiences depression, it personally feels more like “your emotional system only has half as much RAM as usual today, so you're not going to be able to trust anything you think”. I don’t feel sad per say, I just don’t feel good either. Whenever someone asks me what I would like to do/eat/see/whatever I say “I don’t know” because none of the options seem like they would be better than the others. I feel like maybe I would be better off if I were able to at least keep strong negative associations with things in mind, because then I could figure out what I want though process of elimination.
In case anyone still reads these comments, since I read this article when it first came out I've started viewing nights where I have to sleep less as medicinal. "Okay, I can only sleep 5 hours tonight. That's good! I've been sleeping too much and feeling sad. Maybe tomorrow I'll feel better. :)" It might be a placebo, but so far this mindset has made me 1. more of a morning person. 2. happier. :)
I have always wished I can offer myself for any well funded research into the connection between sleep and low grade chronic depression. I have always believed and told people that I feel there's a strong link between the two phenomena.
I sleep a lot so much so that I was nicknamed Jonah the sleeper in childhood. I once went to my university health center to complain about sleeping too much. I meet many of the criteria for Narcolepsy and all the criteria for hypersomnolence. I once took modafinil (which I first heard about on this blog) and it worked tremendously which made me stop for fear of dependence. But it didn't relieve my depression, only excessive sleepiness.
Much of what this article says about the complicated relationship between sleep and depression applies to me. I have long discovered that I experience a negative rebound from the kind of overactivity associated with having fun such that I almost always have a mood crash afterwards but I have tended (obviously wrongly) to interpret it in psychological terms: I feel guilty after having so much meaningless fun such that I feel the need to atone by returning to my true original self. Alas, if this article is anything to go by, it is just pure circadian chemistry!(?)
Relationship with ketamine treatment?
Typo thread
The parenthetical beginning "Giulio Tononi (equally famous" is never closed.
Is it silly to notice the superficial similarity between EMDR and REM (both involve moving the eyes) and effects on psychiatric conditions? Might the evidence for the mechanism of action for one be relevant to the other?
As a data point, I am sometimes depressed, but always feel least shitty in the morning, where I have the most energy and mental clarity, and feel most shitty at night. This seems to be one of those axes along which there's your pretty typical spread of variation; if depressives seem to disproportionately be more night owl than morning lark, that *could* be due to the effects speculated here, or it could just be an "everything correlates with everything else" thing.
On this blog I've heard a bunch of times about how sleep deprivation can help (temporarily) with depression, but out in the normieverse I only ever hear about how sleep hygeine is good for mental health. Granted that there are other reasons why "permanently sleep deprive yourself" (which depression might do anyway lololol) isn't a good idea, is this just the broader culture *assuming* that a generally healthy thing (consistent, full sleep) is good in every specific, or is it still best, including for depression? (It's possible this was answered somewhere in the technical details that I lack the brains and background to follow.)
If I understand this right, maybe the better sleep hygiene thing recommended for depression needs to be amended with restricting the amount of sleep you get, if you tend to oversleep like I do.
Would it make sense to "jumpstart" synapse density first thing every morning by reading/learning? Perhaps that habit would help keep depression at bay.
I didn't look too long but I'm wondering if to what extent they have tried extended sleep deprivation as a treatment. The link you provided seems focused on these 36 wake periods in the short term, not the long term. As in, rather than just see how symptoms are reduced over the course of one or five cycles of 36 hour wake periods or whatever number they've chosen, how are they reduced if this sleep deprivation is done as a consistent lifestyle? Obviously this would pose a problem with living your normal life (family or work obligations, etc.), but if you have a severe case of depression those may not be functioning well anyway and you might as well give this a try if a researcher is offering it.
I know the depression seems to come back as soon as they fall asleep again, but perhaps with enough persistence this effect could be reduced through some sort of positive feedback loop, knowing that at the end of the day you'll feel remarkably better and you only spend maybe half your time feeling like crap instead of all the time.
I always ask depressed patients about sleep. One strange thing in DSM and ICD criteria for major depressive disorder is the hypersomnia or hyposomnia. Too much sleep or not enough? Aren’t those completely different symptoms?
I like Tononi’s synaptic normalisation hypothesis and there seems to be good evidence for it. There’s also evidence that a lot of metabolic waste products and extra cellular debris are cleared from the brain via the ventricular system during sleep, so it’s possible there could be more than one purpose for the change in activity.
I seem to recall that the build up of substance P during sleep was proposed as being connected to the depressogenic nature of sleep. Conversely, the build up of adenosine during the day, to make us sleepy, was proposed as an antidepressive effect (I can’t remember how many years ago I read this), but this seems implausible as we’d therefore expect caffeine as an adenosine antagonist to worsen mood.
The more I learn about sleep, the weirder and more fascinating the links with mental health become. I have a patient who was diagnosed with a non-epileptic attack disorder by neurology, but who is describing many symptoms of narcolepsy with cataplexy, enough that we have referred to a sleep specialist who can do sleep and awake EEG and maybe a lumbar puncture to confirm hypocretin levels. The patient is also reporting frequent nightmares relating to past abuse and trauma, and I am wondering if their hypocretin levels are normal that perhaps psychotherapy for the past abuse might alleviate the nightmares, help the patient to sleep normally, and correct the sleep paralysis and possible cataplexy they are experiencing.
I’ve had patients with PTSD who were experiencing recurrent trauma related nightmares in *every sleep cycle* awaking about once every 90 minutes to 2 hours. Simple trauma focussed CBT fixed this, surprising even me. I’ve heard other therapists and patients describing massive improvements in sleep post successful PTSD psychotherapy.
On the seasonal affective disorder link, I note that melatonin is synthesised from serotonin, but have not been able to find any anatomist to tell me if the melatonin in the pineal gland is synthesised from the supply of serotonin manufactured in the serotonergic neurons of the raphe nucleus or not, so this may be a red herring. Still, we might except a hypersomnic depression to more closely resemble an SAD presentation and a hyposomnic depression. Of course you also know that mixed depression and anxiety or atypical depression is often noted to occur more frequently than depression alone.
It certainly is a huge puzzle
There is also some suggestive evidence (https://pubmed.ncbi.nlm.nih.gov/11696070/) linking natural short sleepers to a kind of mild anti-depressed/hypomanic phenotype.
Possibly related: Narcoleptics supposedly enter REM state almost immediately upon falling asleep (talking like 5 minutes). I'm Narcoleptic and it definitely matches my personal experience in having wildly vivid and epic dreams even if I just doze off for 10 minutes. Does that have any further implications for these findings?
>I think this is the position that Rantämaki and Kohtala are taking in their Encoding...
>Is this Rantämaki and Kohtala's theory?
You got the name wrong here. It's Rantamäki.
On the abnormally low confidence on predictions / overly strong or trapped priors theory of depression, it might not be as much a feature of the whole brain as of particular areas involved in reward, reward prediction, and reward prediction errors that are causally involved. Although the whole brain as a neural network could be using Bayesian prediction to predict inputs as per the Bayesian brain hypothesis, when it comes to depression (and perhaps anxiety) it could be regions such as the orbitofrontal cortex, amygdala, and the habenula that are affected in reward, reward prediction and reward prediction error.
First, I am curious whether you mean "batch normalization" of artifiacal neural networks, or you actually mean "renormaliztion" which is something different and much more esoteric and unusual.
Batch norm does something like regularize neural network inputs so that all the inputs are rescaled into the same domain of possible values. A loose analogy would be your perception that an object "looks the same" whether it is illuminated by 10 lumens or 10,000 lumens; your eyes and brain have done a normalization for you across a broad range of objective input values.
I'll not from a machine learning standpoint that learning is all about leveraging nonlinearities. If you are routinely going through your network and just kind of smoothing everything out, then algorithmically you are probably making 99% meaningless changes that do very little and 1% terrible changes that effectively destroy some precise, interdependent arrangement of synapse weights.
It might be fun to take those creepy amorphous dog-generating GANs from a few years ago and intentionally do such an operation on them, "averaging down" all the weights, and then observing the result. I am not even all too sure what to predict. It is possible that, instead of seeing a frothing sea of Lovecraftian doggoths, you would get a more vague and less coherent visual image. It's also possible that you would still see some dogs, but maybe fewer. Or maybe you would see a disorganized mass of eyes and muzzles with no structure. Or maybe it wouldn't work at all.
This, plus experiences of friends and loved ones, definitely makes me wonder if some depression-adjacent sleep deprivation is trying to hold on to/chase the feeling of increasing synapses? Especially because getting enough sleep is something I had always tried to help depressed people with. But Luke comments below, maybe the negative side of sleep deprivation is just the extra-long sleep phases that follow?
Anyway, we can help with the cabin in Alaska, if you're looking to put your six months in.
Definitely a very interesting theory, but I'd be very cautious about buying into it. Just because staying awake makes you feel better and sleep makes you feel worse, doesn't mean sleep is doing something bad for your brain.
There's an analogy with muscles and sleep: If you work out really hard and then get a good night's rest, the next day you'll be sore. If you work out really hard and only sleep a few hours, you often don't get sore the next day. (That sounds counterintuitive, but it's a widely experienced phenomena.) But skipping sleep to avoid soreness is absolutely the wrong thing to do! The soreness is part of the recovery process. If you keep skipping sleep, your muscles will get weaker and weaker as they never recover from your workouts. So, by analogy, I'd worry that skipping sleep to treat depression might be trading off important brain recovery mechanisms.
Somewhat relatedly, the book "Why We Sleep" by Matthew Walker discusses studies that found that REM sleep reduces trauma: in sleep, the person re-lives traumatic memories, but in a detached way due to the emotional system not being fully connected. This causes the memories to be less emotionally-laden and traumatic going forward (or, one might say... "downgraded and renormalized"...)
Those studies might be relevant to the REM / non-REM question.
This is fascinating. Thank you for these in-depth explorations of science!
Extremely speculative, wondering if others have the same experience:
I often find reading a few highly analytic longposts is the best and only consistent way I can start the day. It seems to actually get my brain "working" while I can't motivate myself to do anything else.
As per the post, the later the day gets, the more energy I have.
But I wonder if reading non-fic essays is more "stimulating" to neurons than say, making breakfast, or going on a walk?
Do anti-seizure medications sometimes called “mood stabilizers,” like lamotrigine and gabapentin for example, ever treat depression? If they are ever effective in treating depression, not just anxiety or bipolar mania, how does that relate with the parts of this post about seizures, if at all?
Is bipolar depression so different from unipolar depression that anti-seizure medication can treat it effectively whereas unipolar depression can be described as effectively treated in the opposite way?
If anxiety can be a symptom of depression, again, why are another kind of anti-seizure medications, benzodiazepines, so effective at treating it? Admittedly they are described as potentially causing or worsening depression. And often prescribed together with other medications.
This post made me wonder, why is it that anti-seizure medications are ever safe enough and even helpful to give to depressed patients?
But I think I can answer that for myself: there are thresholds, risks, sometimes opposing symptoms and conditions co-occur and need treating simultaneously.
What I am more curious about is: does any kind of depression ever look like a seizure disorder instead of something like the opposite?
It also seems relevant that sleep-disordered breathing is common and often goes undiagnosed, yet is often comorbid with depression. Is it possible that some of of the correlation is because people with sleep issues often feel worse when waking, and people with sleep issues are more likely to have depression?
It's kind of hard to find great numbers on this, since at-home sleep studies frequently miss sleep apnea and usually miss other issues, but according to random popular articles (Healthline and similar), about 40% of people with obstructive sleep apnea (OSA) have depressive symptoms, and possibly 15 - 20% of depressed people have undiagnosed OSA. Again, I'd expect these to be underestimates, since the one reference I clicked through (10.1016/j.jpsychires.2019.06.015) relied on really basic at home tests for half of their data.
To me it is crazy that testing for sleep issues is not standard practice for depression given that OSA and other sleep disordered breathing issues are common organic causes for depressive symptoms, and treatment (PAP, surgery, etc.) can improve these symptoms, but that's a bit of an aside.
Babies sleep a lot, culling and pruning the masses of their synapses. Did depressed patients grow up too fast, or took it too far?
Sleep deprivation leads to ROS accumulation in/on the luminal-side of the gut but not other organs (including the brain). sci-hub.se/10.1016/j.cell.2020.04.049 This is evident in experiments with mice and flies after even two days of deprivation. Graphically evident with photos as in Fig 4. With longer deprivation, death results. If the luminal-side ROS is discharged through the by-mouth use of various compounds such as NAC, melatonin and lipoic acid, the animal is rescued from death even with sleep deprivation. Recovery sleep dissipates ROS accumulation in the gut. The effect of sleep deprivation is not all (or ?even?) in the brain.
The gut microbiota influences the brain and behavior through, among other things, the vagal nerve. sci-hub.se/10.1016/j.neuron.2019.02.008 Preventing vagal communication between the GI tract and the brain increased occurrences of psychiatric-related disorders. Vagal nerve stimulation can substantially improve MDD. Gut bacteria also influence the brain and psychiatric disorders through the vagal nerve. The vagal nerve has about 80% afferent (gut to brain) and 20% efferent (brain to gut) fibers.
Hypothesis: If the depression is caused by a specific gut ecosystem signaling, suppressing the signal between the gut and the brain through ROS accumulation in the gut could suppress the depressive symptoms. As demonstrated in the first reference, sleep dissipates the deprivation induced ROS. A possible test of this hypothesis would be to have patients whose depression are relieved by sleep deprivation take one of the tested supplements (e.g. lipoic acid, NAC, et.al.) and test for continued depression relief with sleep deprivation.
I guess this theory can explain your problem of trapped priors: Whenever a small not-so-negative experience (eg a only slightly aggressive dog) happens it gets deleted by a faulty renormalization afterwards and only a clear positive experience (eg a cute puppy in a cage) can help. Apparently, a correct renormalization would keep experiences of the first kind, but for depressed people it is more likely to throw small updates away (since their brains have less connections overall).
Maybe I'm melancholic depressive, though I don't know what that means. Any time I wake up at maybe 6-8 am, I'm suicidally depressed. The first thing I think on waking is, "Dammit, I'm still alive." But I know the feeling will dissipate in an hour or two, so I try to ignore it.
This happens regardless of my sleep cycle. If I go to bed around 3am and get up between 10 and 11, I have no depression at all.
(I haven't noticed whether it varies with season. I don't have natural optimal sleep-wake times; I have a natural daily cycle of 25 hours, so my sleep cycle moves forward 1 hour each day until I force myself to get up early and suffer a sleep-deprived day to restart the cycle.)
This corresponds precisely to my experience. I never thought about this systemically before, but I have always struggled with getting out of bed in the morning due to an overwhelming feeling of melancholic depression, exacerbated by the fact that I stay up later than would be wise for my own good because I usually feel most at peace with the world and therefore energetic at night. The occasional time I manage to stay up all night, I get to work the next day with verve and a more positive outlook, partly because of a feeling of accomplishment I suppose. Unfortunately there does not see to be any apparent solution to this, seeing as eventually Morpheus does call for his due.
It looks like a few people here are relating this blog to the one on trapped priors. Eric Hoel has some work (e.g. https://arxiv.org/pdf/2007.09560.pdf) on "The Overfitted Brain: Dreams evolved to assist generalization" that seems relevant. His theory (which he presents some data in support of but I don't have enough knowledge of the field to know how convincing it is) is basically that the brain, like a neural net, is prone to overfitting based on the data set that it encounters. And in neural nets they often solve that problem by adding noisy or corrupted inputs, which make the fit worse on the existing data but make the system overall more capable of being generalizable and adaptable. And that dreams are the human equivalent of that. From the abstract: "Dreams are a biological mechanism for increasing generalizability via the creation of corrupted sensory inputs from stochastic activity across the hierarchy of neural structures. Sleep loss, specifically dream loss, leads to an overfitted brain that can still memorize and learn but fails to generalize appropriately." I don't think I can quite put my finger on the model that connects them (maybe because it doesn't exist) but there seems to be a strong common thread between depressed brains suppressing new input, and dreams purported function in preserving openness/flexibility, and sleep being messed up in depression.
I wonder if this explains my experience, which is that I feel like I undergo a steady uptick in anxiety the longer I stay awake, which I think is a fairly general phenomenon, ie "your brain at 2am suddenly bombarding you with restless thoughts."
I wanted to ask. I suffer from anxiety and find that my anxiety is worst in the morning, by far. Does anyone else experience this?
So what does waking up around 3AM, often associated with depression, do in this model? Is some failsafe processing in the brain detecting "too much" synapse clearing and waking you up to stop it?
>unsurprisingly SSRI antidepressants decrease REM sleep a lot (not just in depressed people, in >everybody).
I found this pretty surprising and would be interested to hear more. How significant is the decrease? how dose-dependent is it? etc.
Background:
I have had insomnia for basically my entire life which could charitably be described as "severe" and uncharitably described as "Holy F****ing H*** how are you still alive?!?!" I take a few different meds and supplements for it, in addition to sleep hygiene and other lifestyle interventions, but the main medication is Trazadone, an SSRI, and when my sleep specialist first prescribed it after my sleep study (which was apparently pretty horrific looking, even to someone who dealt with severe insomnia cases every day) she said it had been shown to help produce deeper more restful sleep, rather than merely helping with sleep onset. This was important because a big part of my problem was not getting any rest even when I theoretically slept; my sleep study showed (IIRC, it was almost 20 years ago) something like 5 hours 'asleep' out of 9 hours in bed, only about 20 minutes of that REM, and zero slow-wave sleep, I was just staying in the first stage of sleep all night, when I was asleep at all. No one sleeps well during a sleep study, you are in a strange room with electrodes taped all over you, but this was bad even accounting for that. I guess I am rambling quite a bit now, but my point is I was very surprised to hear SSRIs would /reduce/ REM considering what my sleep specialist said back in the day, and how well it seems to help me given how I sleep when I am not on it.
Low confidence in predictions may be an accurate read of reality in the face of a stressful or unpredictable environment. In this way it could be an adaptation to stress which could also include social withdrawal, apathy, and pessimism, or seen differently a switch into a mode of conservation and survival as opposed to a mode of exploration and expansiveness.
The amount of sleep that people have goes down with age, until retirement when it starts going up again.[0] Definitely some of that is due to the amount that they're able to sleep (a person with a job and two kids is going to have very little sleep), but is any of it related to different needs?
[0]: https://www.bls.gov/tus/charts/sleep.htm
Has anyone ever tried sleep reduction therapy to treat depression? If avoiding sleep entirely seems to work until you go back to sleep, what happens if you only sleep 4 hours a night? How long can a human go with artificially reduced sleep?
What about going the other way? If we make people take sleeping pills to sleep more, do they develop depression?
For both questions, does changing sleeping patterns eventually cause the brain to renormalize to that level of sleeping?
Perhaps it is an under stimulation problem? You sort of allude to this in the post, but do not really explore that possibility further.
I basically fit this description to a tee. I have some intermittent problems with depression. And sleep deprivation removes it completely. Periods of slowly slipping into depression followed by sleep deprivation to snap me out of it used to be a recurring pattern for me. But I notice that when I force myself to get seriously mentally stimulated (reading some difficult texts, doing math problems), the depression goes away. Or is less severe. Especially when done earlier in the day. Caveat though, it is not major depression for me, although suicide has occasionally occurred to me in the past.
And actually working too hard, spending too much time and mental energy during the day gets me into mildly manic moods where I have a hard time falling asleep. Usually followed by tiredness where I spend less time doing difficult tasks, where I slowly slip into depression again.
This is so effective for me, that when I get depressed I know I have been lazy and need to improve my work ethic. It is kind of annoying that this only works for tasks that I do not easily pick up, especially when depressed. So once I slip into a depression it can be hard to get out of it without sleep deprivation.
So each person might have different needs for mental stimulation and sleep. And perhaps sleep deprivation to snap out of it, followed by a constant diet of doing mentally taxing and rewarding tasks is the cure for people who suffer from this. It helps if you set an overarching and at least somewhat ambitious goal that you can get excited about.
It seems like a reasonable inference from this is that someone should try treating depressed people with small doses of melatonin, like it's a sleep phase disorder. Does anyone know if this has been tried? It's in no pharma company's interest to find out if melatonin can treat depression so probably not, but maybe that fancy patented melatonin that there was an SSC post about once.
Also, what about depression with atypical features? It's definitely not rare. So it seems like there are also a lot of depressed people who feel *better* after sleeping, and I wonder how that works with this theory.
Three observations come to mind:
1. "You can't renormalize while the network is running." Says who? This seems like it might be a case of unconsciously extrapolating from *human* designed machines to natural machines. Human beings, for reasons of design and construction efficiency, do indeed design machines that generally need to be shut down in order to be worked on -- we have to close the road to repave it, we have to shut down the computer to replace its graphics card.
But I would say as a rule biochemistry tends not to operate this way -- there tend to be many multiple overlapping and redundant paths for getting the same, or similar things, done, and it appears pretty common for the body to work on a system at the same time as it operates. Bones don't grow longer (in children) only when they are not bearing weight, we don't heal wounds and injuries better or faster when we don't use the injured area, and we don't always choose to run only catabolic or anabolic pathways for the same substrate.
So I would say this idea should actually need to be measured to be sure it is true, especially since it's a key assumption underlying the idea that retuning the synapses needs to be done during sleep.
2. Why is it said depression gets "better" when one is sleep deprived? Maybe sleep deprivation merely adds a temporary cover layer of mania that hides the depression, as if you gave the depressed person a massive dose of amphetamines. That is, the change may not be in the direction of cure, but in the direction of adding *another* layer of brain dysfunction that conceals the original layer somewhat, so, away from a cure. If so, the last thing we'd want to do is try to extend it.
3. If a *general* synapse realignment underlay depression, why wouldn't there be much broader associated brain dysfunction? That is, shouldn't depressed people also have noticeably lower IQ, poorer executive function, worse memory -- even compromised sensory abilities and downgraded motor skills? That is, shouldn't an across-the-board physical degradation of the brain have across-the-board degrading effect on brain function, from soup to nuts? Why would we see the primary symptom in mood (which seems a fairly high-level complex epiphenomenon of neural function) rather than in, say, memory impairment or crappy motor skills, something more basal?
waitaminute... one model of autism is that autistic people have an overabundance of synaptic connections (https://www.cell.com/neuron/fulltext/S0896-6273(14)00651-5). And autistic people get less REM sleep (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3111973/). Is Autism a SLEEP DISORDER!?
There are some ML failure modes this brings to mind as potentially useful ideas here.
Sparse gradients and local minima effectively make improving a model impossible. In practice, this usually means you throw the model away in start over. But you can't do that with a human, so you need some way to escape from the local minima. Stronger gradients are one way to potentially do that - a steeper gradient will tend to push what would have been smaller updates into bigger ones. The downside of taking bigger steps like this is that you often end up in a worse place than you started, because bigger steps means that the local gradient is less predictive of how good your new state is going to be.
But if your problem is escaping a local minima, that can actually be a very good thing! Even if your immediate situation is worse, you *might* now have a path to improvement, since even if your new state sucks, *it might be next to a place that doesn't suck*, and that might be next to a place that sucks *even less*, and there might be a path through state space all the way to an actually decent model.
More generally, a lot of depression treatments looks awfully similar to the kind of desperate tricks you might use to get a model out of a local minima.
Scott's is an interesting theory but it does not hold. People with schizophrenia also have lower synaptic density (see https://www.nature.com/articles/s41467-019-14122-0 ) but so far nobody noticed that sleep deprivation helps with schizophrenia.
Secondly, my guess would that the negative correlation with the depression score will turn out to be spurious when we get a publication with a larger sample. The sample in the article Scott quotes is very small (n=40).
On a lighter note, psilocybin increases synaptic density after a single administration (in pigs) https://www.mdpi.com/1422-0067/22/2/835 - Looks like we have a new therapy for schizophrenia and depression. And also you do not need to sleep, if you take some psilocybin, if Scott is right.
>How does this relate to TMS?
The #1 most common side effect of TMS depression protocols is extremely vivid dreams, by a long shot. I have had one of these protocols and can attest, it's friggin crazy. If it's non-REM sleep that's doing harm, maybe TMS increases not only the intensity of dreams but the amount of REM sleep.
Convenient timing - the Royal Institution in England recently hosted an online talk by Gina Poe on this subject:
https://www.eventbrite.co.uk/e/essential-functions-of-sleep-learning-memory-and-changing-your-mind-tickets-138758722073
These usually appear on their youtube channel (https://www.youtube.com/user/TheRoyalInstitution) in a month or so, so it might be worth keeping an eye out for that.
Gina actually described three stages, REM, deep sleep, and a "transition period".
Deep sleep is the physical repair time, when the brain is least active.
REM and the transition period both are connected to the rewiring of your synapses, but one strengthens new connections, the other clears out unneeded connections. Or one organises, the other files/dumps depending on the results of the organising. Figuring out exactly what happens is obviously an area of ongoing research.
A title reminded me of the exurb1a's video: https://www.youtube.com/watch?v=-mu780uB7mI
Two interesting notes that might help along or that you may find extremely boring:
* Donepezil is a great way to cause more REM, other drugs that act in complex ways on the cholinergic system do too. Looking at trip report of these might be interesting. The big problem here is that most people taking these REM-increasing drugs do so during the day and have serious conditions (e.g. alzhimers)
* REM and memory consolidation are potentially inversely related (not sure if this is what you mean by "sleep renormalizes the hippocampus"), or rather, REM and short term memory consolidation may be, at least in that N2 and N3 seem to play a big role there and increase N2 and N3 seem to => short term memory consolidation. And I'd usually associate a longer REM with a longer N4.
* Pulse oximeters can kinda-maybe measure sleep stages well enough to be not noise and so can some comfortable 2-4 neuron EEGs. These will never be used in a study *but* trying to get feedback on your sleeping patterns and what helps might be a much better way than speculation here.
* (Maybe obviously) Short/Long sleep studies with depression may be hopelessly confounded by common drugs (caffeine, nicotine, alcohol, benzos) which affect sleep a lot, and which depressed people would be prone to take, and unless you've looked into the methodology closely it might be worth batting an eye over (though maybe you did and you think it controls for this properly => ignore me, but it seems like the kind of thing that's hard to deconfound for many reasons, one way or another you end up with weird samples)
[warning: extremely speculative]
So we've established in this post that depression is caused by sleep, particularly REM sleep. It's well known that REM sleep is the only kind of sleep where dreams occur. When dreaming, your brain generates some arbitrary thing-that-resembles-a-narrative-story, and this thing further reinforces the intuitive idea of being a self-controlled sapient character in a grand narrative of the universe. In Buddhist/ish thinking, the idea of being a self-controlled sapient character is a myth and being enlightened causes one to intuitively grok it as a myth. So chaining the logic together we should conclude that enlightenment is a cure to depression.
Can anyone whose worked with or suffered from severe depression comment on the relative misery of depression versus sleep deprivation?
Not to be cliche but would you rather be depressed or sleep deprived?
Stupid thought on REM vs non-REM section: depressed people actually have MORE synapses, not less. Depression is a brain wave "brownout": barely enough to run everything at minimum capacity assuming nothing else taxes the system. As more synapses form throughout the day, the brain has to upgrade to full "blackouts" in various regions to keep the brain from permanently damaging itself, which frees brain waves to run everything else better, which is why they seem to get better as the day goes on, and why sleep dep seems to make things even better: more blackouts free up more brain waves for the rest of the brain. They get just enough non-REM sleep to normalize back to brownout levels, which starts the cycle over again. I would expect this to be falsified if the entire brain gets reduced function instead of just isolated regions during depression, and I would expect brain frying drugs like marijuana to help. I'm probably totally wrong though and welcome correction.
Back to reading
Pushing back on the "number of neurons doesn't change" claim, neurogenesis (new neurons being born and integrating into the circuit) is not only a thing (and despite some claims pretty likely to be happening in adult humans for a limited number of cell types) but is also proposed by some as the main mechanism for the positive effects of SSRIs, exercise, enrichment, etc on mood, since all of these increase both neurogenesis and mood.
Some of you may be interested in the new paper from the Latham lab arguing that all synaptic modification is just part of Bayesian inference: https://www.nature.com/articles/s41593-021-00809-5
If seizures help treat depression, then why are people with epilepsy more likely to experience depression? Or do electroconvulsive seizures work differently than epileptic seizures?
Thanks for an interesting article 🙂
In my view to consider sleep deprivation as a valid approach is nothing short of insane. Reminds me of lobotomy. As a side note, the underlying specific assumptions of a computer-like functioning of the brain seem to me a strange caricature of current knowledge. I guess I hadn’t been connected to the community for quite long and forgot the strange belief system at work..
If rem sleep clears out the short-term memory and a having full short-term memory bank is linked to the reward system that would make sense from a more intense morning depression and lesser in the evening standpoint. Would then refilling of the short-term memory banks upon waking by reading something particularly interesting/enlightening (such as this blog) create a large number of new memories right off the bat and kick-start the day with a more elevated mood due to the well topped up short-term memory? Are there more effective ways of filling the short-term memory than reading, such as cramming Trivial Pursuit cards or listening to an audiobook on double speed (doing that feels like an information download straight to the brain)?
So the overall result is to increase "equality" of synapse strengths throughout the affected regions? This sounds kind of like destroying information, but maybe if the information is the pattern of connections making you depressed that's good?
From an information theory perspective, increasing equality doesn't necessarily reduce information. Suppose the brain had 1 really really strong synapse. That isn't much information. It isn't hard to point out which synapse is the strong one. On the other hand, if all the synapses were equally weighted, that carries no info. (But might make the info about where the synapses are easier to access.) Much more info can be stored by giving half the synapses a weight of 2, and the other half no weight. The distribution of synapse weights that maximises info (given fixed total synapse strength) will conceptually resemble the distribution of kinetic energy in an ideal gas.
The actual neuron weights will follow a geometric distribution (in this simplified model, where we are maximizing info with a bounded total weight and weight is just the count of some molecule), with a parameter that depends on the total weight bound.
TL;DR. A long tail of really strong synapses doesn't hold much info. A more even spread of strength can be more informative.
When I was in college, I was giddy-drunk if I went without sleep for a while, and that was fun. (Although I didn't drink in college, so I can't really compare.) I think my inhibitions just went way down and I'm normally too cautious.
Looks like they understand depression about as well as digestion. I can tell you for a fact that living with an optimist who loves you but won't put up with the attitude after a point can cure depressive tendencies over time.
>Is it fair to say that "decreased synaptic strength" is sort of the same thing as "abnormally low confidence on predictions"?
1. Consider that what may be involved is something like a "ready to hand" instrumental confidence in the effects of the {depressed, manic} person's actions; we can try to gloss this as corresponding to the epistemic confidence they have in their predictions, but that abstraction may be misleading in a lot of cases. (Famously it's hard to translate the ability to tie your shoes or execute a tennis serve into "predictions" or "decisions", and a lot of quotidian executive-function tasks are likely like that.)
2. What may be going on is that the depressed person's brain is *too good* at making the synaptic connections. If this manifests over the course of months, we call it a bipolar disorder. If their brain is somehow wise to what's going on and adjusts daily (nightly), then they wake up despondent/hopeless/directionless, go to bed slightly manic, and are at their most average functioning mid-afternoon.
3. This meshes with a lot of other pathways/covarying cognitive deficits. This kind of depression is connected to anxious behaviors like compulsive procrastination; that would make perfect sense if ingrained compulsive habits like e.g. checking blogs (and writing blog comments?) provide extremely predictable feedback. (This is something I think makes more sense on a "ready to hand" model than a literal "epistemic confidence" model; it's not clear why doing something just to be doing it would result from reliable predictions, since anxious procrastinators can usually predict when prodded that nothing valuable will result and they will regret procrastinating.)
4. It explains some stylized facts like: (a) why melancholy tends to attack people in relatively independent careers (they have more options and are vulnerable to decision fatigue early in the morning, and have fewer forced actions to get them through their low-functioning part of the day); (b) why they tend to have self-perpetuating circadian rhythm problems at night (they are most overconfident in the value of doing one more thing when they need to go to sleep, and have very firm convictions about tomorrow's agenda that they meticulously plan out when they're lying in bed); (c) why some people are convinced lifting weights helped their depression (very simple practical cause-and-effect relations, usually early in the morning); (d) why mastering some new hobby with quantifiable success/failure/improvement metrics often brings people out of a funk for a while.
I myself was in a melancholic funk a long time ago, and it stopped at the same time I started a ketogenic intermittent fasting diet that involved going to bed hungry and not eating until I finished my morning workout. At the time I thought that my rising spirits gave me the willpower to follow the diet, but now I wonder if it wasn't the opposite; starving my hyper-activated synapses in the evening forced me to bed and decreased the amount of overnight synapse-correction I needed. — Meanwhile I also wonder if the problem didn't originally arise because for a long time I tried to save time and diet by skipping breakfast, habituating my brain to absence of sugar in the morning. (Consuming my morning caffeine with artificial sugar instead of real sugar probably didn't help.) — That said, I wasn't at my most productive while practicing intermittent fasting; my most productive years in recent memory were again a period where I woke up in a fog, got nearly everything done 1pm-5pm, sometimes felt like trying to get a lot done at midnight, and sometimes had sleepless nights. So it's possible the overall pattern of strengthening synaptic connections over the course of the day isn't purely dysfunctional or diseased, even if it carries risks.
As someone with lifelong depression, I can attest that the sense of being "fully activated" only late at night is very real. I do my best creative work between about 10pm and 2am. I used to intentionally plan all-nighters when I had difficult school work or papers to complete; I eventually just realized that if I wanted 100% productivity, I had to do it late at night.
At the ends of semesters, I'd sometimes take a 36 or 48-hour stretch of final paper writing during which my productivity would reach *extreme* heights. For example, producing multiple graduate-level papers which received high grades in just a few days - something I *cannot* do normally. Toward the end of these periods, I'd experience a sensation I'd describe as feeling "too awake" - extremely alert, quick thinking, rapid ideation, borderline manic. If you've ever taken ADHD medications in a somewhat too high dose, it was a very similar feeling.
So, this discussion seems extremely accurate to my personal experience.
Is it possible to self-medicate depression by restricting sleep or reorganizing your life around it somehow? I'm thinking about shifting your work and important stuff to the evenings/nights and instead taking it easy in the mornings. I was reminded about how as a student I tended to gravitate towards doing more studying in the evenings (except for bar nights I suppose). It seemed to work then, and aside for a preference for having evenings off I don't see why it wouldn't work now as well.
Alcohol, in high doses, decreases REM sleep: https://onlinelibrary.wiley.com/doi/full/10.1111/acer.12006
This would seem to support a more complex view of the role of REM sleep in depression. SSRIs reduce REM sleep and fight depression; alcohol reduces REM sleep and worsens it.
Throwing another anecdotal evidence on the pile, I've always been a night owl, I find mornings difficult even though I've had to get up early for years, and on top of that I tend to have insomnia - either it takes me forever to fall asleep, or when I do fall asleep I keep waking up throughout the night.
And then when I get the opportunity to catch up on sleep, I can sleep for hours upon hours. I don't know how much linked that is with depression, but it certainly *feels* like it.
The times I have been sleep-deprived, I have ended up bouncing off the walls, but I don't think it had much of an effect on my depression; instead of feeling unmotivated and 'I hate myself', it was "Whee! I can finish all these tasks! *And* I still hate myself! Oh look all the straight lines are all bendy..."
What of other animals? Bats sleep much more than people, but have much simpler brains. Wouldn't you expect the much greater number of synapses to lead to some notable difference in sleep patterns or the types of chemicals released in sleep or something?
At least in children, "wake up grumpy but gradually cheer up" seems to be a lot more common than the reverse.
I was also thinking that dementia stems from a sleep disorder. It’s just that my mother who had something going on (neurologist could not define it other than dementia) would have moments of clarity then drift off to what seemed like she was describing a dream when she spoke. Meaning everything she was saying was disconnected like what happens in a dream.
What does this imply about people who need way less sleep than the average person?
Back in college the seniors would have a rotating shift of "night watchmen" who had to wake up and patrol the dorm a couple times a night to make sure nothing crazy was going on. When it was my turn I remember be constantly unnerved that the same 5 or 6 people out of a thousand and change were always up and about at 3 a.m.
Could it be that the "normalization waves" of these folks are abnormally strong and accomplish the task in just a couple hours? Or maybe they are susceptible to depression effects and have just evolved a strategy of exposing themselves less to sleep to maintain better mental health?
Glymphatics anyone? Searched for this in the comments along with lymphatics and aquaporin and go not hits.
Recently(ish) the extracellular mass transport of fluid has been shown to double during sleep. This is thought to be protective for Alzheimer's disease for example. It is also highly dependent on aquaporin 4 expression.
I note that early on Scott writes "As if depression is caused by some injury during sleep". The opposite could be true.
Hypothesis: some semi-soluble compound in the extracellular space is beneficial to depression but sleeping allows it to be washed away. During the day it reaccumulated.
So this coincides with a weird issue I'm investigating in my own brain.
For background, I have atypical depression (diagnosed, used to be medicated, mostly in remission for now) and probably ADHD (diagnosed as something ADHD-adjacent as child when that was not a category in my country, unmedicated). Mornings are universally the worst part of the day, both in terms of mood and sleep latency. Coffee helps a little bit but is mostly drank for pleasure. Brain fog is present during the entire day, but I manage.
The thing I discovered by accident is that drinking coffee in the evening gives me superpowers. I never liked to work in the evening, instead reserving it for chilling out, but with a sufficient burst of caffeine I get an incredible amount of stuff done, concentrating effortlessly while in a hypomanic state. Distractions that normally completely destroy my productivity are no problem - I can easily sustain conversation via IMs and not lose track of work, for example. This effect works _only_ in the evening and late afternoon.
This makes absolutely no sense under the standard dopaminergic (no relation to circadian rhythm) / adenosine antagonist (related to circadian rhythm, but it just makes you feel non-sleepy) explanation of caffeine action. The circadian depression voodoo explanation is... tempting, to say the least.
"This sounds kind of like destroying information, but maybe if the information is the pattern of connections making you depressed that's good?"
If I've got a textbook in mostly nine-point type, but one random sentence on each page is in forty-point bold, I don't lose any information and I do gain more efficient access to information if I reformat those sentences to the same font as the rest.
If it's not random sentences but the most recent sentences (the stuff I experienced yesterday and was supposed to reformat down to standard-archival-format), that sounds like it might combine synergistically combine with some poorly-chosen trapped priors. If my recent experiences are of the "life sucks, there is no hope" variety, and those keep getting overweighted, then I won't be able to draw effectively on my previous experience of life not sucking.
In which case, excessive sleep alone won't always cause depression, but would aggravate other potential causes of depression and impede natural recovery.
What about a simple alternative explanation: patients are going through withdrawal in the morning, because they haven't taken their SSRIs or other meds in 8+ hours.
Huh! I've noticed that during depressive episodes, my dream content "lags" by a few days. For example, if I watch a movie, those characters will show up in my dreams three days later, where normally they'd show up the same night. Related, maybe?
I have an interesting perspective of this, having done night work for over 30 years. Not just 30 years but for much of it six nights a week. Anyone who has worked the third shift can relate to the unnatural sleeping patterns. My day off (Friday--today in fact) I spend up to enjoy life the way day-trippers do. I am just about at 24hrs and going strong. Now I would not want to take a long drive at the moment, but lets hope this has some clarity.
Regarding depression, my own personal experience has been that sleep, deep sleep, is much needed to ameliorate it. Now, there is merit in sleep deprivation. The longest run I've had is three days straight. I fainted eventually (at the breakfast table!) but here's the odd thing--on the way home, I felt like a million bucks. A few things played into this I believe. The weather was warm and sunny, I kept myself moving (stopping is the bane of the sleep deprived) and was relatively trouble free. A miserable, rainy day in the winter after my cat just died might have another outcome. Never-the-less, I still fainted. My body pulled the plug on my mind.
Over the years, I have noticed a curious pattern. While I can stay up for long periods of time, I find after deep sleep, my mind fires on all cylinders. Let me tell you about my old friend Frank. His father was in a rest home and I would visit with him from time to time. Frank would warn me his pop would need a good 30 minutes to get "his brain moving" as he described it. He said it was almost like the blood had to get to his brain and took that long. Slowly his pop would come out of the malaise he was in, know our names, engage in conversation. I said I often stay up all day Friday. Depression (when it hits me) is Saturday morning. Early. 5am maybe after a good 5 hour sleep perhaps. I could lie there turning over thoughts that somehow seem the have the better of me. So, I one morning thought of Frank's pop. Thought perhaps my own blood had to get flowing, and simply chose to let my mind worry later, at a time when my mind was better apt to take the trouble on. Just as I would not fancy a long drive now for safety issues, I applied the same to my mind in that state. Fast forward to Monday night now. I am driving in, bright and alert, same troubles and "I've got this', I can say to myself. All my mind needed was rest. I've become a big advocate of proper sleep to bring clarity and convinced lack of sleep causes shotty thinking. I don't deny the research here, to some degree have experienced it, but also the benefits of a rested mind to achieve the same. I've seen the depression "reset" Saturday morning but the exact opposite in direct proportion Monday night sleeping sound all day.
Tim Shaw
I've often felt like i get more tired and depressed towards the end of the day. I have a demanding job, two kids, and aging parents. Towards the end of the day, I often feel down about almost everything. That seems to be the opposite of what the article describes.
When i wake up, this generally isn't the case. I've often used the image of something like 'light' that illuminates the space of possibility nearby - both what's happening now, and what might happen. When I feel good, this light is bright, and when i don't feel good, it's as if the light is 'dim' - everything just seems more dismal and hopeless.
That might make sense of connection strength gets stronger over the course of the day, and strong connections require more energy to maintain. If the connection strengths all get downgraded evenly over the course of sleep, then it takes less energy for my brain to map out all the various connections between things in my immediate environment and, as a result, i can more easily perceive the subtle, positive things that make life seem better.
It often feels, when I'm tired or angry or sad, like my experience of the world reverts into a 'cartoon' state, with all of the unpleasant feelings highlighted. That might seem consistent with only the strongest connections getting enough energy to run?
If we follow this theory that (certain patterns of) increased synaptic connection treats depression, and evidently our activities over a day tend to produce such, do we know if there are particular activities that do so disproportionately? It might make sense to front-load those.
There are some candidates on how top-down theories like predictive processing and bottom-up theories like homeostatic disregulation can meet.
In predictive processing, there are two major candidates on how/where the downwards prediction and the upward sensory information might actually be implemented.
One is that different parts of the layered structure (https://en.wikipedia.org/wiki/Cerebral_cortex#Layers_of_neocortex) of the cortex are used. Details are complicated, but probably upwards stream mostly arrive in layer IV, while the comparison with prediction might be done in layer V, or possibly in layer II/III. We don't understand the details of this "canconical microcircuit" very well yet, but there is a lot of structure.
The other (complementary) candidate is that different neurotransmitter are involved. Upstream information uses mostly AMPA (which is a LOT faster and is responsible for the lion share of energy consumption), while predictions probably/perhaps involve a lot more NMDA receptors (slower and longer-lasting, strongly involved in plasticity and learning).
Homeostasis does not affect all these structures the same way. As far as I understand Tononi's theories, mostly the AMPA part is upregulated over day and downregulated at night. Also, different layers are affected differently.
Once you know such things, you can start to speculate. If AMPA is stronger in the evening, does that mean that predictions are relatively weakened compared to sensory information? I don't think it gives a coherent picture (yet), and I don't think that we are at a point where we can match top-down and bottom-up theories. Probably we won't be there in the next 10 years. But I don't think it is hopeless in the long run.
Maybe but I still out substantial credence on finding out the sleep result reflects affects to the endogenous opiate system/cycle. I'd love an experiment that measured levels of those endogenous opiates over the time period.
What about schizophrenia? Sleep deprivation often leads to psychosis and the negative symptoms are a lot like depression symptoms? What's the interaction there?
It's weird to me to hear that sleep deprivation can help with depression, because I'm pretty sure sleep deprivation makes me *more* depressed, at least in the long term. But possibly short-term sleep deprivation is helpful and long-term sleep deprivation is harmful?
I'm wondering if what you wrote here can be used to explain why SSRIs take a week or two before they start working.
I really don't appreciate that you used a Sci-Hub link without warning
During college, I did a research paper on synaptic scaling during NREM sleep, and ran into Tononi on a ludicrous number of articles. I started noticing his name popping up on all the interesting ones, so I just did a search by author. I don't know how the **** he does it, but his name is on all sorts of crazy papers that are clearly trying to become a groundbreaking paradigm shift in their own way. At this point I'm convinced that he has a network of contacts across the entire field that give him early-warning alerts on weird new papers so he can swoop in and coauthor them.
For what it's worth, btw, the consensus among researchers ca. 2018 was shifting towards Tononi's NREM model being less helpful than the other big model in the field. There were a couple of studies that looked really slam dunk until one of the big names on the other side pointed out that they were anesthetizing their mice with an agent that specifically interfered with the part of the brain they were looking at. I don't know what the fallout was from that or if Tononi's people recovered, but it was an oof moment for sure.
Huh that explains it for me, thanks! "It" being: I sometimes feel depressed early in the morning, like first 5-30 minutes after waking up, although I'm normally not and never been diagnosed. It happens without any apparent reason like sleep deprivation, hangover or shitty day ahead/behind. And it's not just low energy, I literally feel like I'm worthless, my life sucks, things I said and done yesterday are all failures etc, even though I remember it's not what I was thinking yesterday and know I'll probably feel differently in half an hour. I eventually assumed it's some sort of a brain glitch and just made a habit of not thinking of any such matters right after waking up, now I finally know what kind of glitch this (probably) is.
Double curse for people with epilepsy, who have to maintain a rigid sleep schedule, whether they like it or not, to ward off seizures.
As someone who experiences depression, it personally feels more like “your emotional system only has half as much RAM as usual today, so you're not going to be able to trust anything you think”. I don’t feel sad per say, I just don’t feel good either. Whenever someone asks me what I would like to do/eat/see/whatever I say “I don’t know” because none of the options seem like they would be better than the others. I feel like maybe I would be better off if I were able to at least keep strong negative associations with things in mind, because then I could figure out what I want though process of elimination.
I know that sleep lowers seizure threshold / increases likelihood of seizures. seems relevant here
Staying up for ages takes you out of yourself. Same as exercise does.
In case anyone still reads these comments, since I read this article when it first came out I've started viewing nights where I have to sleep less as medicinal. "Okay, I can only sleep 5 hours tonight. That's good! I've been sleeping too much and feeling sad. Maybe tomorrow I'll feel better. :)" It might be a placebo, but so far this mindset has made me 1. more of a morning person. 2. happier. :)
I have always wished I can offer myself for any well funded research into the connection between sleep and low grade chronic depression. I have always believed and told people that I feel there's a strong link between the two phenomena.
I sleep a lot so much so that I was nicknamed Jonah the sleeper in childhood. I once went to my university health center to complain about sleeping too much. I meet many of the criteria for Narcolepsy and all the criteria for hypersomnolence. I once took modafinil (which I first heard about on this blog) and it worked tremendously which made me stop for fear of dependence. But it didn't relieve my depression, only excessive sleepiness.
Much of what this article says about the complicated relationship between sleep and depression applies to me. I have long discovered that I experience a negative rebound from the kind of overactivity associated with having fun such that I almost always have a mood crash afterwards but I have tended (obviously wrongly) to interpret it in psychological terms: I feel guilty after having so much meaningless fun such that I feel the need to atone by returning to my true original self. Alas, if this article is anything to go by, it is just pure circadian chemistry!(?)