Well, what are you supposed to do when a user who is asking for clarification is really just trolling you?
107 karma · joined June 12, 2015
Edit: SPWs, Sharp Waves, SWR, etc. are just labels for sharp waves. I can't believe I am getting attacked for using jargon. The names don't matter. The physics and science matters. Richard Feynman.
Edit 2: The link I sent uses the same jargon that I use. So, honestly mate, what is your deal here? Are you trolling?
If you google "SPW neuroscience", you will find many resources.
https://en.wikipedia.org/wiki/Sharp_waves_and_ripples
Brain science is highly interdisciplinary. Each expert should spend time to familiarize themselves with the various approaches used by other experts. In particular, machine learning theorists should spend time to understand the latest results (their implications, at least) in neuroscience and neuron biophysics. At some point, an ML theorist is not too different than a biophysics theorist. They just use different abstractions. This is why I am particularly interested in Google's and Hinton's work. That work seems to be subtly motivated by neuroscience and natural science...
I believe you are correct. Memories are maintained during sleep. Some are believed to get strengthened, while others are apparently pruned.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2039924/
http://neuralcircuits.uwm.edu/sharp-wave-ripples/
In summary, the hippocampus will organize neurons into clumps which encode spatial dimension. As a rat runs around throughout a track, certain groups of neurons fire[1] depending on where the rat is located. Before the task even starts, hippocampal neurons in a rat actually fire in quick succession (e.g. ABCDEFG). When the task is completed, the same pattern fires in the reverse order (e.g. GFEDCBA). Scientists believe that these SPWs play an important role in memory consolidation.
I have always thought of this as "the rat is thinking about where it's going to run to". So I started learning machine learning and landed on things like RBMs. I was so excited because RBMs are reminiscent of this actual phenomenon that occurs in the brain. So, I asked Hinton about it because this is no mere coincidence, and he told me that he had a theory that required reverse order firing during sleep state.
If you get a ton of new memories and patterns during wake, sleep may be responsible for not only pruning less important memories, but strengthening more important memories. For example, sleep might be asking the brain for a ranked list of memories, and then strengthens the memories above some threshold, and prunes the rest (just an idea).
[1] Synchronized firing, see Kuramoto coupled differential equations: https://en.wikipedia.org/wiki/Kuramoto_model
This seems, at first, to be a problem in academia, but, at further glance, is really a problem with education.
Fake news now is less effective against well-educated, smart, skeptical, scientific people. It just so happens that fake news is dumb enough to only fool 80% of us. As the methods for generating false information ("fake news" is not a general term) get more advanced, people will need to be more and more educated to resist falling into machine dependency.
If you're gonna get on the fake news train, do it now, as fake news is easy right now. Bots are still dumb, etc. In the future it won't be so easy to combat fake news. It will be a much, much more grave problem. If there are people who are uneducated enough to be influenced by fake news now, they will be sheep in the future even more so, when they are told lies of events, audio clips, video clips, etc. that never even happened.
Fake news, or more generally, false information, is going to be a big problem in the 22nd century. As our machines get smarter, so do our deception methods. At some point, humans will begin to place trust in machines because we will no longer be smart enough to figure it out ourselves. Currently the non-savvy population (maybe, ~80%?) has begun to do this already. At what point will our scientists begin to fall on the same side of that partition, until only a few scientists in the top of their field actually get it. We have already observed this God-Effect in esoteric fields.
God-Effect: Where a piece of technology, mathematics, or code becomes so complicated, eventually only a few people (or zero) actually understand it. E.g. complicated edge-physics, library implementations, encrypted data, etc.
Edit: Well, I slightly disagree. Even if the product doesn't require mass-adoption to be useful, it could surely benefit from branding/marketing and viral adoption to make it the desired product, in which case, it does still depend on who does all of that better. Even if Starbucks or Macbooks are more expensive, they have won their audience and increased their effective marketing force.
Hey, this is imaginative, even if it's armchair philosophy. Seems like you are thinking that the holography principle is self-similar at many scales.
> They found that some of the simplest quantum field theories could explain nearly all cosmological observations of the early universe.
This seems to be very, very old news. Like all theories, holography sounds very interesting. This article implies that a prediction made by holography has been observed. This is not the case, apparently. The math may a lot more elegant, but what new predictions are there, and have we observed them? This is what the article claimed to be about, alas it wasn't.
http://community.wolfram.com/groups/-/m/t/995095
Also, here's a more casual/fun post I happened to read earlier today, coincidentally:
http://www.dailymail.co.uk/femail/article-3280816/What-photo...
> “I thought I was going crazy. When my air conditioner is on, I wake up and hear light conversations. I would go to the window to see if anyone was outside, or I would turn the air conditioner off [and] it would stop. Sometimes it sounds like a radio.”
I have been sleeping with a fan on for six years, to provide background noise so that any outdoor audial perturbations would not disturb me as much. When I have housemates, or I am visiting family, I will often be disturbed so much sometimes, that I will get up and silent my fan because I am very sure that a conversation or break-in is happening downstairs. Every time, I realize it was my imagination, followed by wondering if I am developing schizophrenia or if I am just a hypochondriac, and then I go back to sleep.
> “I would hear faint voices—whispering, conversing, singing, or chanting! It sounded like a crowded room, full of people at a party in a distant room somewhere in the building. After a while I came to enjoy the sound, as they seemed to be enjoying themselves at the ‘party,’ and it helped lull me to sleep at night.”
> If you hear a symphony in Manhattan traffic, thank your auditory pareidolia for relieving your stress
Sounds like the better part of a great LSD trip. Good to know I am not alone in experiencing pareidolia, I thought I was going crazy, too. I can't help but think the background fan is not perfectly consistent background noise. Jitter in the wires due to external forces would lead to variable rotation speed in the motor, which would lead to ups/downs in amplitude and pitch. Our ears are sensitive enough to capture these distortions, and our brains are powerful enough to amplify or attenuate our perception of these distortions. Even if our ears can't capture these perturbations, I would think that the mere innocent thought or knowledge (or, more ominously, paranoia or suspicion) of these distortions is enough to create such a perception by the brain.
Initial Reaction: If you form a vector of time-convolution of the output of the fan with respect to all of your internalized sounds (fragments, phonemes, words, etc.), over all of their possible phases (accents, volume, pitch, etc.), you could see which sounds the fan is most likely to resemble based on how big the corresponding matrix element is compared to the others. Even if the overlap is .01% with a certain sound, if the next leading overlap is 1,000 times smaller, I bet our brains would pick up on that relative difference. I know it does with visual field (e.g. optical illusions), and there is reason to believe that the brain has a grand unified learning algorithm for all data[0][1] (data is normalized as electromagnetic input anyways)
After reading this, that view is challenged:
http://www.haskins.yale.edu/featured/sws/swssentences/senten...
Researchers at Haskins lab have gotten rid of the Fourier components in regular speech, in which case the inner products mentioned above would be zero, and the synthetic sound should be orthogonal to your internalized sounds. Nevertheless, I listened to some of this SWS, and could understand it with 90% accuracy. So, this is very strange. But does it make sense? If you apply a high frequency shift (e.g. gerbils effect) to some function, you can still understand the speech, even though it's effectively orthogonal to the original function. Reading more into the OP paper:
> “spill-over from parts of my auditory system that recognize the air conditioner, or the shower or the sound of the kettle, into areas of my nervous system that are responsible for processing speech and language.” He went on, “And that spillover doesn’t require similarity. It just requires some kind of shared activity.”
I am starting to like this theory, and reminds me of an earlier conversation on HN about corpus callosotomy not getting rid of spillover from other parts of the brain [2]. But that "spillover" resulted in useful information transfer, whereas the quote here would suggest it is much more random. Also, it would suggest that the frequency might not be as important as dA_i(t)/dt, for all i, where A_i(t) is the amplitude for the ith frequency, like in Haskins lab.
> But much of our perception of the world also results from the interaction between what we might expect to happen in a given situation, given our beliefs, and the processing of incoming sensory information
This explains why I imagined the sounds as my family or friends talking or someone breaking-in.
> minds impose on random noise, influenced by our idiosyncratic beliefs and predispositions, can also be harmful
Yes, it is important to take action to ensure you are not going crazy. E.g. get up, turn off the fan, and listen: did the conversation/robbery stop? If it did, try it again tomorrow. After several times, one should realize that it is unlikely that the robberies and conversation are happening, and that it is more likely due to brain perception mechanics.
[0] http://newsroom.ucla.edu/releases/ucla-study-shows-brain-s-a...
[1] https://www.google.com/webhp?sourceid=chrome-instant&ion=1&e...
> Finally, the team used the revolutionary gene-editing tool, CRISPR-Cas9 to engineer E. coli that don’t register the X and Y molecules as a foreign invader.
> "This will blow open what we can do with proteins."
> [The base pairs] can't be read and processed into something of value by the bacteria - it's just a proof-of-concept that we can get a life form to take on 'alien' bases and keep them.
> ...have not been designed to work at all in complex organisms, and seeing as they're like nothing found in nature, there's little chance that this could get wildly out of hand.
To quote Jeff Goldblum, "life, uh... finds a way".
This is true for people who do not consume heroin a priori. For the others, suddenly stopping heroin use can be life threatening. Most people need assistance to wane off of things like that. Also, I think you are right that enabling will increase usage, but you must also consider the black markets which are enabled in response to disabling the regulated markets.
It seems like a majority viewpoint on this thread is that it's better for more people to be using real opioids safely than to have a smaller, effectively invisible/black-market population using "opioids" very, very dangerously.
Only a single order of magnitude is not enough of a safety margin, depending on the substance, and how easy it is to administer the substance. Is alcohol's LD50/dose ratio an apt comparison? Does alcohol require a higher administration energy than heroin? It seems like it would be easier to "accidentally" consume half a gram of heroin in a small amount of time, as opposed to "accidentally" consuming 100 beers (or 2 handles).
Drug users experiment across this mere 10x range effortlessly, according to your source:
> As usage increases, however, the doses get much higher
Apparently, they get to 100x. Crazy!:
> The Swiss found that users offered unlimited quantities would, on average, max out at between 300 and 500 mg
This seems to support my view. I can't help but imagine a dystopian future (past?) where they start selling a soda called "HeroHera" which has trace amounts of heroin in it.
Anyways, it seems pretty dangerous for brains to depend on the naive-LD50 of heroin. Forgetting to consume it after one has an established a dependency must surely be an issue. How do we know that a potential widespread dependency on heroin and a requirement for widespread heroin sanctuaries won't actually cost more than the war on drugs?