8 ms·

Dreams are the default for intelligence

82 points by Amorymeltzer 4y ago

sam8401 4y ago
Wrong
harpiaharpyja 4y ago
I think it's an interesting potential answer to the ongoing question of "why do we have dreams, what evolutionary purpose do they serve?" Maybe the answer is that they do not need to serve any purpose because dreaming is actually the default mode - "normal" consciousness is the adaptation. From that perspective the evolutionary purpose (of normal rational consciousness) is obvious. Just musing here but I wonder if that line of thought could shed light on how non-sapient animals experience consciousness as well.
- molly0 4y ago
  From an evolutionary perspective, letting animals think/dream is an optimal way for them to explore things without risk.
- bitwize 4y ago
  I can tell you from firsthand experience that dogs dream. Some dogs "talk" in their sleep, move their legs as if walking or running -- the sequence from Cinderella (1950) when Bruno is dreaming of chasing Lucifer is only a slight to moderate exaggeration. A dreaming dog is definitely in a place far removed from waking life -- perhaps one where steaks grow on trees and blood-soaked frisbees fly through the air for no reason.
  - not_the_fda 4y ago
    Dogs dreaming is one of my favorite things in life.
- quesera 4y ago
  My (uninformed!) theory is that dreams are just what brains/neurons appear to do when the driving signal (thought) is mostly disconnected, but the filter (interpretation) is still active. I.e. just neural noise, following pathway patterns by biological necessity, but shaped into conceptual patterns by our overactive pattern matchers which do a ton of work to arrange input into a limited set of "observations". If that's true, nothing we ever dream is truly novel, but just assembled or permuted from the giant set of our own known or imagined pieces. Which seems pretty consistent with my dreams, at least. And it explains the phenomenon be being unable to recognize the dream state without altering it. Heisendreams!
molly0 4y ago
I got completely lost in trying to understand the idea he is trying to convey.
- T-A 4y ago
  I liked it (unusually enough for some random blog) so I'll volunteer my take on it. Remember how Stable Diffusion works? It's essentially noise reduction cranked up to eleven. You train a neural network to clean up slightly noisy images. When it gets good at it you turn up the noise a notch and repeat. You keep doing that until the network will look at pure noise and "repair" it into an image of whatever. Animal brains have to do something similar, because our senses suck. Take our eyes; the part of your field of view which is actually in focus is tiny [1], so you are constantly updating an internal 3D model of the world, one small patch at a time [2], interpolating between patches and editing out things like your nose to replace it with a hallucination of whatever it happens to be occluding (except right now, because I reminded you of it). In this sense, what you "see" is not so much the real world as a hallucination guided by your sensory input. What happens when you reduce the information content of that sensory input, either by making it "unstructured" [3] or by removing it altogether [4]? Same thing that happens to Stable Diffusion when you present it with pure noise: it hallucinates freely. So it makes sense that lying down on a comfortable bed in a dark, quiet room and shutting your eyes - all steps toward sensory deprivation - should have a similar effect. [1] https://en.wikipedia.org/wiki/Vision_span https://en.wikipedia.org/wiki/Vision_span [2] https://en.wikipedia.org/wiki/Saccade https://en.wikipedia.org/wiki/Saccade [3] https://en.wikipedia.org/wiki/Ganzfeld_effect https://en.wikipedia.org/wiki/Ganzfeld_effect [4] https://en.wikipedia.org/wiki/Sensory_deprivation https://en.wikipedia.org/wiki/Sensory_deprivation
  - molly0 4y ago
    Thank you! Interesting.
irrational 4y ago
I hope this isn’t true. I’m in my 50s and I have never dreamed. By that I mean, I have never woken up from sleeping and remembered anything happening between the time I fell asleep and I awoke. It’s a complete blank. Maybe I do dream, but I have no proof or indication that I do. I never have. As a side note, it really bothers me when dreaming is used as a plot device in books. I can’t relate at all to the concept of dreaming and remembering the dream.
- nomel 4y ago
  There are devices made to detect and disturb you (with light, vibrations, etc) during REM sleep, to help with lucid dreaming. If you want to experience a dream, something like that might help, with the fairly safe assumption that you do enter REM sleep.
  - jhot 4y ago
    I didn't realize this until I had kids and my sleep became constantly disturbed, but the only times I'm consciously aware of my dreams are when I'm awoken during the dream. If I am given the chance to have a peaceful night, then it's just an instantaneous time shift from falling asleep to awake with nothing in-between. So I could see how one of these types of devices could be handy to gently bring the consciousness out of deep sleep.
- ThalesX 4y ago
  Interestingly enough, I could be considered the opposite. Having aphantasia, I've never had a mental image, but I have very vivid dreams. Talking to my dad, he only dreams in black and white. It's just awesome the variability we humans have.
  - pessimizer 4y ago
    > Childhood exposure to black-and-white television seems to be the common denominator. A study published this year, for example, found that people 25 and younger say they almost never dream in black and white. But people over 55 who grew up with little access to color television reported dreaming in black and white about a quarter of the time. Over all, 12 percent of people dream entirely in black and white. > Go back a half-century, and television’s impact on our closed-eye experiences becomes even clearer. In the 1940s, studies showed that three-quarters of Americans, including college students, reported “rarely” or “never” seeing any color in their dreams. Now, those numbers are reversed. https://www.nytimes.com/2008/12/02/health/02real.html https://www.nytimes.com/2008/12/02/health/02real.html
zak1726 4y ago
Given that this might be so, wouldn’t it be even more apt to adopt Geoffery Hintons Forward Forward Algo which emphasises teaching the neural network the difference between dream data and real data at the neurological level.
crooked-v 4y ago
This also reminds me of some theories I've seen that sleep (e.g. mostly sessile behavior) is the biological default for Earth life, and being "awake" and highly mobile is itself a specialized adaptation.
- kingstoned 4y ago
  Interesting. Could you please share some links about it?
- alpaca128 4y ago
  I see a few problems with that. For example that we can freely act and think in reality but cannot in (non-lucid) dreams. Things like meditation seem more convincing. Though it is an interesting thought experiment, which implies that we evolved much farther than necessary for dreaming simply because the environment outside dreams as an energy source is so competitive, to a point we're now able to create different kinds of alternate realities as well. Which as so often leads to questions like the purpose of our existence. To me the idea that dreams exist in order to let the brain filter out unimportant experiences from the day before - or highlight important ones - seems more plausible, especially considering sleep deprivation can lead to hallucinations and more, I don't think this can be explained purely with other processes like the self-cleaning mechanisms of the brain during sleep.
amelius 4y ago
Or perhaps dreaming is GAN-style training.
idleproc 4y ago
I read a book about lucid dreaming a while back, and it said something like: the easiest way to tell you are in a dream is to look at something mechanical or complicated. The idea seemed to be: if you could get your dreaming mind to look at your hand and then turn it over to the other side, you didn't have the mental 'bandwidth' to imagine the opposite side correctly, and you'd realize you were dreaming. Equally, if you were around a bicycle in a dream, and tried to look at the gearing mechanisms etc., it just wasn't possible for your brain to generate that level of detail; it would just change the bicycle into an elephant or what-have-you. I never got to lucid dreaming, but did notice a similar thing happening. So, I always found it interesting how the mind might switch from internal 'concepts' to external 'reality' in a way that isn't readily available to 'conscious' thought. Not really going anywhere with this, but if 'AI' can generate better bicycles than our dreaming minds, then…
- robot_no_421 4y ago
  Another way to tell is to try to read things or look at a clock/numbers in your dream. Often time the information will not make sense, and will change if you look away and back. I've done this myself once or twice in a dream and I actually was able to tell I was dreaming but the shock of my realization woke me up.
  - BulgarianIdiot 4y ago
    Unlike most I can see things in great detail, including text in my dreams. Consistency in time is absolute mess, for sure, but then again, I don't find this odd to begin with. In fact I kind of expect this. As in I realize I'm dreaming, and yet I don't.
  - UncleOxidant 4y ago
    I've had dreams where I found myself traveling and was presented with maps (in one my car had a HUD where I saw the map on the windshield as I was driving) that seemed to be quite accurate when I recalled them after waking. Even freeway signs with names of towns, roads and freeway numbers that matched reality.
  - 4y ago
tracerbulletx 4y ago
Dreaming is the condition where the parts of your brain required to form memories, and to have consciousness are active enough to participate and take part in what the other awake parts of your brain are doing while the sleeping parts like sensory input and motor control are shut down. We already know the brain creates an active predictive model of the world that is what we experience, so it makes sense the brain can continue creating that model when asleep and untethered from sense input.
allemagne 4y ago
>When I am dreaming, I am receiving images/stories that are produced for me, not really by me This can really only be partially true for lucid dreaming, though, right? Maybe all dreaming is really just lucid dreaming. We're all usually just in some kind of split-brain patient situation where around half our brain is "lucidly" controlling a dream and the other half just goes along for the ride. A lot of dreams are just kinda nonsense, but some I've had are pretty clearly well-thought-out metaphors for things that are happening in my life. Somebody or something pretty smart that knows a lot about me had to have come up with it.
- saghm 4y ago
  > This can really only be partially true for lucid dreaming, though, right? I've had a lot of lucid dreams over the years (starting when I first was prescribed medication to help me sleep when I was a kid, where I'd have them several times a week, and gradually becoming less common over the years to maybe once a month). At least in my experience, being aware that I'm dreaming gives me the ability to choose how to act, the ability to force myself to wake up, and occasionally the ability to break minor "physics" rules (I pretty commonly find in lucid dreams that I can hover a few inches above ground and float around that way instead of walking), but I've never experienced the ability to change my environment, and the "people" in my dreams I interact with continue to act independently. Obviously my dreams are entirely a creation of my own mind, so some part of my brain is in control of it, but being aware of that isn't sufficient to let me control it arbitrarily any more than I can stop myself from feeling physical pain or disliking the taste of some food in real life. Ultimately, we're not really in "full" control of how our mind and body react to real-life stimulus either; if we were, things like depression and phobias wouldn't exist.
Grieverheart 4y ago
I once trained myself to recognise if I was dreaming by looking at something in my dream, looking away and looking back again. If the thing I was looking at had changed in any way, I was dreaming. This allowed me to partly shape my dreams, so I can’t totally buy the conjecture that senses are what’s taking the dream machine. Another thing supporting this is the fact that I could let my body fall asleep while keeping my consciousness awake. Even though the senses are technically shut off (sleep paralysis), I was not dreaming. I think the fact that dreaming is similar to generative models has more to do with how we learn shapes, and that is by learning sub shapes and categories of shapes. My theory is that that’s also how long term memory works. Instead of storing the full senses senses at that moment, parts of those senses are replaced by concepts/sub shapes.
felipemnoa 4y ago
Hehe. I actually had a thought around the same lines a few days ago. https://news.ycombinator.com/item?id=35516757 https://news.ycombinator.com/item?id=35516757 And what probably gives us awareness or consciousness is the multiple GPT like structures running in parallel. It really does make me wonder if we are really close to true AGI. It does feel like it. Very exciting times! Just hope we don't end up doing a Terminator.
- gabereiser 4y ago
  Oddly enough, we've been saying that since... The Terminator. NN's and LLM's are cool but not even close to the power of the human brain (maybe quantum llm's?). I'm keen to see how we can tap into our own minds as CPU's or at least drivers thereof. Consciousness is awareness and perception so one could say we are close to a rudimentary consciousness but we have a long way to go before it's aware. Still, even though we've been saying it for 30 years, exciting times indeed! #skynet
  - MacsHeadroom 4y ago
    > LLM's are cool but not even close to the power of the human brain There are only 20 Billion Neurons in the Neocortex. GPT-3 has 175 Billion. GPT-4 scores 130 on IQ tests. They are absolutely in the realm of the power of human brains. They're just exposed to different stimuli and have very narrow input/output.
    - Teever 4y ago
      organic neurons are much more complex than simulated neurons. It's more accurate to look at an organic neuron as an entire computing device than a simple transistor.
    - felipemnoa 4y ago
      >>GPT-4 scores 130 on IQ tests Then there must be something really wrong with IQ tests. GPT-4 to me seems more like an idiot savant.
    - 4y ago
mrangle 4y ago
Possibly related: I have a pet theory that holds that primarily attentive ADHD (ADHD-PI) is actually a facet of the type of autism that used to be called Asperger's. And that ADHD, generally, is an extension just outside of the autism spectrum: possibly bridging into it with ADHD-PI if not the other ADHD classifications. One of ADHD-PI's most noticeable traits is virtually uncontrollable daydreaming. When auditory / cognitive fatigue or fast-arriving boredom sets in, for example. Asperger's theory is plagued by glut. Which makes the condition challenging to accurately comprehend, but at the same time doesn't necessarily imply that a lot of the theory is incorrect. Instead, it may imply complexity. Which makes sense given that this condition isn't acquired but instead is characterized by divergent brain architecture and talent / cognitive resource re-parsing from "normal". Similar in concept to an RPG game. One facet of this glut of Asperger's theory is the school of thought that hypothesizes that Asperger's is some kind of evolutionary forward expression of a type of intelligence. Which is both generous and an unusually positive take within the glut of theory, ignoring for a moment any negatives. My hypothesis adjusts this a bit, and proposes that Asperger's may be a type of genetically de-selected (low reproduction chance) contribution of a specific intelligence type. Which may have existed in human populations from the beginning. Implying that Aspergian intelligence, which may be associated with an increase rate in daydreaming, isn't so much a forward evolution but a population mainstay that may serve some type of function in the sense of low-rate genetic contribution. Say having to do with visual processing that implies advanced talent in pattern processing specifically. Some human intelligence academics, who are not involved in Asperger's research, place emphasis on pattern processing as the possible heart of it. I can think of obvious reasons as to why whatever group genetic "intelligence" governs genetic evolution, as well as the roughly equal reproduction ratio of men and women as another example, would "want" to bias human evolution as much as possible away from a high rate of human-specific-intelligence gain. Possibly smoothing it out away from purely Aspergian expression and blending it, over time, with our other natures and talents.
- emmelaich 4y ago
  > Which may have existed in human populations from the beginning Well, autism is said to related to percent of Neanderthal. e.g. (2016) https://newsroom.uw.edu/story/human-neanderthal-gene-variance-involved-autism https://newsroom.uw.edu/story/human-neanderthal-gene-varianc...
komain7 4y ago
Is intelligence even intelligent? Only thinking creature and you use your brain for murder machines. Sounds so intelligent. Everything here for everyone, yet some greedy pigs feel more deserving. Intelligence, more like a waste of human life. Intellectuals and computers have been a giant circle jerk. You couldn't support your own weight.
ddmma 4y ago
That’s exactly what I’ce been exploring with this dream journal that is generating images in order to match your brain visualisation during sleep https://driary.com/ https://driary.com/
southernplaces7 4y ago
All of the suggestions here for conducting certain "physical" or mental tests to know if you're dreaming or not presuppose that you're capable of enough self awareness to realize you might be dreaming, and act on your suspicions. For most of us, being inside a dream, no matter how surreal, feels like being in another concrete reality in which it doesn't even occur to you to test its falsehood, because you take its realness as a given.

82 points by Amorymeltzer 4y ago

Wrong

I think it's an interesting potential answer to the ongoing question of "why do we have dreams, what evolutionary purpose do they serve?" Maybe the answer is that they do not need to serve any purpose because dreaming is actually the default mode - "normal" consciousness is the adaptation. From that perspective the evolutionary purpose (of normal rational consciousness) is obvious. Just musing here but I wonder if that line of thought could shed light on how non-sapient animals experience consciousness as well.

molly0 4y ago
From an evolutionary perspective, letting animals think/dream is an optimal way for them to explore things without risk.
bitwize 4y ago
I can tell you from firsthand experience that dogs dream. Some dogs "talk" in their sleep, move their legs as if walking or running -- the sequence from Cinderella (1950) when Bruno is dreaming of chasing Lucifer is only a slight to moderate exaggeration. A dreaming dog is definitely in a place far removed from waking life -- perhaps one where steaks grow on trees and blood-soaked frisbees fly through the air for no reason.
- not_the_fda 4y ago
  Dogs dreaming is one of my favorite things in life.
quesera 4y ago
My (uninformed!) theory is that dreams are just what brains/neurons appear to do when the driving signal (thought) is mostly disconnected, but the filter (interpretation) is still active. I.e. just neural noise, following pathway patterns by biological necessity, but shaped into conceptual patterns by our overactive pattern matchers which do a ton of work to arrange input into a limited set of "observations". If that's true, nothing we ever dream is truly novel, but just assembled or permuted from the giant set of our own known or imagined pieces. Which seems pretty consistent with my dreams, at least. And it explains the phenomenon be being unable to recognize the dream state without altering it. Heisendreams!

molly0 4y ago

I got completely lost in trying to understand the idea he is trying to convey.

T-A 4y ago
I liked it (unusually enough for some random blog) so I'll volunteer my take on it. Remember how Stable Diffusion works? It's essentially noise reduction cranked up to eleven. You train a neural network to clean up slightly noisy images. When it gets good at it you turn up the noise a notch and repeat. You keep doing that until the network will look at pure noise and "repair" it into an image of whatever. Animal brains have to do something similar, because our senses suck. Take our eyes; the part of your field of view which is actually in focus is tiny [1], so you are constantly updating an internal 3D model of the world, one small patch at a time [2], interpolating between patches and editing out things like your nose to replace it with a hallucination of whatever it happens to be occluding (except right now, because I reminded you of it). In this sense, what you "see" is not so much the real world as a hallucination guided by your sensory input. What happens when you reduce the information content of that sensory input, either by making it "unstructured" [3] or by removing it altogether [4]? Same thing that happens to Stable Diffusion when you present it with pure noise: it hallucinates freely. So it makes sense that lying down on a comfortable bed in a dark, quiet room and shutting your eyes - all steps toward sensory deprivation - should have a similar effect. [1] https://en.wikipedia.org/wiki/Vision_span https://en.wikipedia.org/wiki/Vision_span [2] https://en.wikipedia.org/wiki/Saccade https://en.wikipedia.org/wiki/Saccade [3] https://en.wikipedia.org/wiki/Ganzfeld_effect https://en.wikipedia.org/wiki/Ganzfeld_effect [4] https://en.wikipedia.org/wiki/Sensory_deprivation https://en.wikipedia.org/wiki/Sensory_deprivation
- molly0 4y ago
  Thank you! Interesting.

irrational 4y ago

I hope this isn’t true. I’m in my 50s and I have never dreamed. By that I mean, I have never woken up from sleeping and remembered anything happening between the time I fell asleep and I awoke. It’s a complete blank. Maybe I do dream, but I have no proof or indication that I do. I never have. As a side note, it really bothers me when dreaming is used as a plot device in books. I can’t relate at all to the concept of dreaming and remembering the dream.

nomel 4y ago
There are devices made to detect and disturb you (with light, vibrations, etc) during REM sleep, to help with lucid dreaming. If you want to experience a dream, something like that might help, with the fairly safe assumption that you do enter REM sleep.
- jhot 4y ago
  I didn't realize this until I had kids and my sleep became constantly disturbed, but the only times I'm consciously aware of my dreams are when I'm awoken during the dream. If I am given the chance to have a peaceful night, then it's just an instantaneous time shift from falling asleep to awake with nothing in-between. So I could see how one of these types of devices could be handy to gently bring the consciousness out of deep sleep.
ThalesX 4y ago
Interestingly enough, I could be considered the opposite. Having aphantasia, I've never had a mental image, but I have very vivid dreams. Talking to my dad, he only dreams in black and white. It's just awesome the variability we humans have.
- pessimizer 4y ago
  > Childhood exposure to black-and-white television seems to be the common denominator. A study published this year, for example, found that people 25 and younger say they almost never dream in black and white. But people over 55 who grew up with little access to color television reported dreaming in black and white about a quarter of the time. Over all, 12 percent of people dream entirely in black and white. > Go back a half-century, and television’s impact on our closed-eye experiences becomes even clearer. In the 1940s, studies showed that three-quarters of Americans, including college students, reported “rarely” or “never” seeing any color in their dreams. Now, those numbers are reversed. https://www.nytimes.com/2008/12/02/health/02real.html https://www.nytimes.com/2008/12/02/health/02real.html

zak1726 4y ago

Given that this might be so, wouldn’t it be even more apt to adopt Geoffery Hintons Forward Forward Algo which emphasises teaching the neural network the difference between dream data and real data at the neurological level.

crooked-v 4y ago

This also reminds me of some theories I've seen that sleep (e.g. mostly sessile behavior) is the biological default for Earth life, and being "awake" and highly mobile is itself a specialized adaptation.

kingstoned 4y ago
Interesting. Could you please share some links about it?
alpaca128 4y ago
I see a few problems with that. For example that we can freely act and think in reality but cannot in (non-lucid) dreams. Things like meditation seem more convincing. Though it is an interesting thought experiment, which implies that we evolved much farther than necessary for dreaming simply because the environment outside dreams as an energy source is so competitive, to a point we're now able to create different kinds of alternate realities as well. Which as so often leads to questions like the purpose of our existence. To me the idea that dreams exist in order to let the brain filter out unimportant experiences from the day before - or highlight important ones - seems more plausible, especially considering sleep deprivation can lead to hallucinations and more, I don't think this can be explained purely with other processes like the self-cleaning mechanisms of the brain during sleep.

amelius 4y ago

Or perhaps dreaming is GAN-style training.

idleproc 4y ago

I read a book about lucid dreaming a while back, and it said something like: the easiest way to tell you are in a dream is to look at something mechanical or complicated. The idea seemed to be: if you could get your dreaming mind to look at your hand and then turn it over to the other side, you didn't have the mental 'bandwidth' to imagine the opposite side correctly, and you'd realize you were dreaming. Equally, if you were around a bicycle in a dream, and tried to look at the gearing mechanisms etc., it just wasn't possible for your brain to generate that level of detail; it would just change the bicycle into an elephant or what-have-you. I never got to lucid dreaming, but did notice a similar thing happening. So, I always found it interesting how the mind might switch from internal 'concepts' to external 'reality' in a way that isn't readily available to 'conscious' thought. Not really going anywhere with this, but if 'AI' can generate better bicycles than our dreaming minds, then…

robot_no_421 4y ago
Another way to tell is to try to read things or look at a clock/numbers in your dream. Often time the information will not make sense, and will change if you look away and back. I've done this myself once or twice in a dream and I actually was able to tell I was dreaming but the shock of my realization woke me up.
- BulgarianIdiot 4y ago
  Unlike most I can see things in great detail, including text in my dreams. Consistency in time is absolute mess, for sure, but then again, I don't find this odd to begin with. In fact I kind of expect this. As in I realize I'm dreaming, and yet I don't.
- UncleOxidant 4y ago
  I've had dreams where I found myself traveling and was presented with maps (in one my car had a HUD where I saw the map on the windshield as I was driving) that seemed to be quite accurate when I recalled them after waking. Even freeway signs with names of towns, roads and freeway numbers that matched reality.
- 4y ago

tracerbulletx 4y ago

Dreaming is the condition where the parts of your brain required to form memories, and to have consciousness are active enough to participate and take part in what the other awake parts of your brain are doing while the sleeping parts like sensory input and motor control are shut down. We already know the brain creates an active predictive model of the world that is what we experience, so it makes sense the brain can continue creating that model when asleep and untethered from sense input.

allemagne 4y ago

>When I am dreaming, I am receiving images/stories that are produced for me, not really by me This can really only be partially true for lucid dreaming, though, right? Maybe all dreaming is really just lucid dreaming. We're all usually just in some kind of split-brain patient situation where around half our brain is "lucidly" controlling a dream and the other half just goes along for the ride. A lot of dreams are just kinda nonsense, but some I've had are pretty clearly well-thought-out metaphors for things that are happening in my life. Somebody or something pretty smart that knows a lot about me had to have come up with it.

saghm 4y ago
> This can really only be partially true for lucid dreaming, though, right? I've had a lot of lucid dreams over the years (starting when I first was prescribed medication to help me sleep when I was a kid, where I'd have them several times a week, and gradually becoming less common over the years to maybe once a month). At least in my experience, being aware that I'm dreaming gives me the ability to choose how to act, the ability to force myself to wake up, and occasionally the ability to break minor "physics" rules (I pretty commonly find in lucid dreams that I can hover a few inches above ground and float around that way instead of walking), but I've never experienced the ability to change my environment, and the "people" in my dreams I interact with continue to act independently. Obviously my dreams are entirely a creation of my own mind, so some part of my brain is in control of it, but being aware of that isn't sufficient to let me control it arbitrarily any more than I can stop myself from feeling physical pain or disliking the taste of some food in real life. Ultimately, we're not really in "full" control of how our mind and body react to real-life stimulus either; if we were, things like depression and phobias wouldn't exist.

Grieverheart 4y ago

I once trained myself to recognise if I was dreaming by looking at something in my dream, looking away and looking back again. If the thing I was looking at had changed in any way, I was dreaming. This allowed me to partly shape my dreams, so I can’t totally buy the conjecture that senses are what’s taking the dream machine. Another thing supporting this is the fact that I could let my body fall asleep while keeping my consciousness awake. Even though the senses are technically shut off (sleep paralysis), I was not dreaming. I think the fact that dreaming is similar to generative models has more to do with how we learn shapes, and that is by learning sub shapes and categories of shapes. My theory is that that’s also how long term memory works. Instead of storing the full senses senses at that moment, parts of those senses are replaced by concepts/sub shapes.

felipemnoa 4y ago

Hehe. I actually had a thought around the same lines a few days ago. https://news.ycombinator.com/item?id=35516757 https://news.ycombinator.com/item?id=35516757 And what probably gives us awareness or consciousness is the multiple GPT like structures running in parallel. It really does make me wonder if we are really close to true AGI. It does feel like it. Very exciting times! Just hope we don't end up doing a Terminator.

gabereiser 4y ago
Oddly enough, we've been saying that since... The Terminator. NN's and LLM's are cool but not even close to the power of the human brain (maybe quantum llm's?). I'm keen to see how we can tap into our own minds as CPU's or at least drivers thereof. Consciousness is awareness and perception so one could say we are close to a rudimentary consciousness but we have a long way to go before it's aware. Still, even though we've been saying it for 30 years, exciting times indeed! #skynet
- MacsHeadroom 4y ago
  > LLM's are cool but not even close to the power of the human brain There are only 20 Billion Neurons in the Neocortex. GPT-3 has 175 Billion. GPT-4 scores 130 on IQ tests. They are absolutely in the realm of the power of human brains. They're just exposed to different stimuli and have very narrow input/output.
  - Teever 4y ago
    organic neurons are much more complex than simulated neurons. It's more accurate to look at an organic neuron as an entire computing device than a simple transistor.
  - felipemnoa 4y ago
    >>GPT-4 scores 130 on IQ tests Then there must be something really wrong with IQ tests. GPT-4 to me seems more like an idiot savant.
  - 4y ago

mrangle 4y ago

Possibly related: I have a pet theory that holds that primarily attentive ADHD (ADHD-PI) is actually a facet of the type of autism that used to be called Asperger's. And that ADHD, generally, is an extension just outside of the autism spectrum: possibly bridging into it with ADHD-PI if not the other ADHD classifications. One of ADHD-PI's most noticeable traits is virtually uncontrollable daydreaming. When auditory / cognitive fatigue or fast-arriving boredom sets in, for example. Asperger's theory is plagued by glut. Which makes the condition challenging to accurately comprehend, but at the same time doesn't necessarily imply that a lot of the theory is incorrect. Instead, it may imply complexity. Which makes sense given that this condition isn't acquired but instead is characterized by divergent brain architecture and talent / cognitive resource re-parsing from "normal". Similar in concept to an RPG game. One facet of this glut of Asperger's theory is the school of thought that hypothesizes that Asperger's is some kind of evolutionary forward expression of a type of intelligence. Which is both generous and an unusually positive take within the glut of theory, ignoring for a moment any negatives. My hypothesis adjusts this a bit, and proposes that Asperger's may be a type of genetically de-selected (low reproduction chance) contribution of a specific intelligence type. Which may have existed in human populations from the beginning. Implying that Aspergian intelligence, which may be associated with an increase rate in daydreaming, isn't so much a forward evolution but a population mainstay that may serve some type of function in the sense of low-rate genetic contribution. Say having to do with visual processing that implies advanced talent in pattern processing specifically. Some human intelligence academics, who are not involved in Asperger's research, place emphasis on pattern processing as the possible heart of it. I can think of obvious reasons as to why whatever group genetic "intelligence" governs genetic evolution, as well as the roughly equal reproduction ratio of men and women as another example, would "want" to bias human evolution as much as possible away from a high rate of human-specific-intelligence gain. Possibly smoothing it out away from purely Aspergian expression and blending it, over time, with our other natures and talents.

emmelaich 4y ago
> Which may have existed in human populations from the beginning Well, autism is said to related to percent of Neanderthal. e.g. (2016) https://newsroom.uw.edu/story/human-neanderthal-gene-variance-involved-autism https://newsroom.uw.edu/story/human-neanderthal-gene-varianc...

komain7 4y ago

Is intelligence even intelligent? Only thinking creature and you use your brain for murder machines. Sounds so intelligent. Everything here for everyone, yet some greedy pigs feel more deserving. Intelligence, more like a waste of human life. Intellectuals and computers have been a giant circle jerk. You couldn't support your own weight.

ddmma 4y ago

That’s exactly what I’ce been exploring with this dream journal that is generating images in order to match your brain visualisation during sleep https://driary.com/ https://driary.com/

southernplaces7 4y ago

All of the suggestions here for conducting certain "physical" or mental tests to know if you're dreaming or not presuppose that you're capable of enough self awareness to realize you might be dreaming, and act on your suspicions. For most of us, being inside a dream, no matter how surreal, feels like being in another concrete reality in which it doesn't even occur to you to test its falsehood, because you take its realness as a given.