Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
Reconstruction from brain activity
  • Vitaliy_KiselevVitaliy_Kiselev




    It is, of course, slightly different thing from that you think, as this is pattern matching and recognition combined with images averaging.
  • 21 Replies sorted by
  • cbrandincbrandin
    This reminds me of an experiment carried out years ago. People were put upside down and within days they started seeing right side up (this was done with children - which is kind of disturbing). When they were put right side up it took a few days again for them to also see right side up.

    The brain is not a literal device - it adapts to what it needs to perceive. The main requirement for the brain is to be able to differentiate what it needs to - the literal form that takes is malleable.

    A very long time ago I was involved in research like this - it's fascinating. People like to think the brain achieves what it does by massive capability. I've always favored the idea that it accomplishes more through architectural - rather than quantitative - superiority.

    Take the ability to sense temperature, for example. You have warm and cool sensors. The warm sensors are stimulated by a rise in temperature. Likewise cool sensors are stimulated by a drop in temperature. There is a wrinkle, however. Cool sensors are actually stimulated by both low and very high temperatures. Cool sensor stimulation = perception of cool. Warm sensor stimulation = perception of warm. Warm + Cool sensor stimulation = perception of hot. Have you ever been in the shower when suddenly there is a sensation of very hot water? What is really happening is that the water isn't thoroughly mixed and you get both cool and warm stimulation at the same time - and your brain interprets that as being hot.

    The brain is an amazing thing; not only because it is a massive and complex device, but also because it's an extremely clever design as well.
  • Vitaliy_KiselevVitaliy_Kiselev
    My views is very similar to ones described in Jeff Hawkins book on the subject:

    image

    http://www.amazon.com/Intelligence-Jeff-Hawkins/dp/0805074562#_
  • cbrandincbrandin
    Thanks for the link - I'll look into it.

    It's been a long time since I was involved in this sort of thing. One book that profoundly affected my work for many years was Julian Jaynes' "The Origin of Consciousness in the Breakdown of the Bicameral Mind". Much of my work in coding and information theory was inspired by it. I doubt that I would have ever gotten around to inventing Digital Pattern Processing had it not been for his book. What I learned from him was that it is not necessary to literally represent anything; all that is required is a mechanism that can differentiate everything you need to - and that's the basic principle behind symbolic processing.
  • Vitaliy_KiselevVitaliy_Kiselev
    @cbrandin

    I used Jeffs book ideas in my courses.
    And all the data and other books I know also suggest that it has more useful approach comparing to traditional "this part does this thing". He also underlines that such specialization could be not hardwired, but rather be made during normal early brain developing stages.
    At least even for my own cat I could say that they could almost full repair from major insult rebuilding brain functions.
    It is also very interesting to see in dynamics. From the cat who can't keep his head correctly oriented to fully healphy looking cat.
  • cbrandincbrandin
    Brain mapping is a very counter-intuitive thing. Most people assume that specific memories are localized and that personality is holographic. In fact, the opposite is true. There have been many cases of fairly massive brain injuries not resulting in loss of memories. On the other hand, there are many cases where very small, localized, injuries resulted in devastating changes in personality or capability. These discoveries led to many of the "holographic memory" theories popular today.
  • Vitaliy_KiselevVitaliy_Kiselev
    I think that no such thing as "memories" exist.
    During normal life brain work as very complex learning associative and predicting machine.
    It do not store any "memories" but use reality transformed to sensor inputs and internal inputs to change or add parts to associations machine :-) It also perfectly explain various techniques used to implant false memories.
  • Vitaliy_KiselevVitaliy_Kiselev
    About brain repairs for cats.
    It is also interesting that very long topic existed few years ago (now it is gone) with owners telling about such things. And cats are very good at it. Dogs are not. Most of them die after very short time.
  • cbrandincbrandin
    There is a surgical procedure called a Commissurotomy. This basically involves severing the corpus callosum - the connecting tissue between the brain halves, sometimes used to treat intractable epilepsy. The most extraordinary thing about such surgery is the degree to which people can recover from it. At first people will select clothing with one hand because one side of the brain likes the outfit, and put them back with the other because the other side of the brain doesn't. Given a surprisingly short amount of time, however, people will adapt away from this "split brain" problem - even though the corpus callosum does not regenerate.
  • cbrandincbrandin
    If you mean that there are no literal memories - I agree. I believe that memories are symbolic representations of experiences. We have no ability to perceive anything directly - what we see is a projection on the retna, what we hear is the sympathetic vibrations of cilia, etc... That makes us helpless but to remember reactions to reality rather than reality itself.
  • Vitaliy_KiselevVitaliy_Kiselev
    >I believe that memories are symbolic representations of experiences

    No I do not believe in symbolic representations.
    As I said, brain is just very complex associative machine thing.
    Jeff pushes this idea quite hard also.
    So, to get some memories you need to put into input something that will trigger such machine to get certain output. And most of the inputs are internal, not external.
    I also think this is why we need proper sleep to function normally.
    Brain needs to cut external inputs and make rearrangments feeding and refeeding internal inouts.
  • Vitaliy_KiselevVitaliy_Kiselev
    >There is a surgical procedure called a Commissurotomy

    Cats problems are much bigger.
    They lost ability to even get signals required to properly orient (can't get proper up down measurement).
    And part of the brain stops functioning.
    But they learn how to get this information from other sources. Brain ramaps and relearns.
  • cbrandincbrandin
    I think we agree - by symbols I mean association Icons, abstractions that facilitate association engines. I've written quite a bit about association Icons; basically they are representations of reality that possess the logical properties of data, but none of the data itself. The advantage of them is that you can represent everything with relatively short numbers, and process and manipulate them, yielding the same outcome you would get if you processed original literal data.

    Think about this: How many bits would it require to enumerate every atom in the known universe? It's not so many bits. Double the number of bits, and the probability of two atoms having the same serial number becomes very small, even if you assign the numbers randomly. In fact, odds are that only one collision of numbers will occur out of the entire set. Now, figure out how to generate the numbers according to logical properties, and these numbers can stand in for the reality they represent for the purpose of processing. This works because of the unlikelihood of statistical collisions because of the size of the numbers (symbols, or Icons) that represent reality - not the amount of data required to literally render the reality they represent.

    This is pretty esoteric, I know. You can Google me with "Digital Pattern Processing" and find several of these papers. In my defense I'll point out that I have actually used these theories to produce things that work - and much better than the alternatives.

    I'll read Jeff's book, maybe that will help to clarify some of the terminology.
  • cbrandincbrandin
    Speaking of sleep - which I believe allows the brain to convert short term literal data to long term spatially represented data (but that's a whole other subject), among other things - I'm going to get some now!
  • brianlucebrianluce
    Who the hell are you people???!!!!!
  • cbrandincbrandin
  • brianlucebrianluce
    @cbrandin

    Okay. It all makes sense now...
  • brianlucebrianluce
    @crandin
    "If you mean that there are no literal memories - I agree. I believe that memories are symbolic representations of experiences. We have no ability to perceive anything directly - what we see is a projection on the retna, what we hear is the sympathetic vibrations of cilia, etc... That makes us helpless but to remember reactions to reality rather than reality itself. "
    ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

    I don't really know what you mean. Of course perception is about receiving, filtering, processing and analyzing various aspects of reality. How would it ever be possible for the brain to reproduce actual reality as part of perception? A hacked GH2 can't even do that -- though Vitaliy's 9,000 mbps hack looks promising.


    Seriously, what am I missing here Chris?

  • Vitaliy_KiselevVitaliy_Kiselev
    >Of course perception is about receiving, filtering, processing and analyzing various aspects of reality

    Brian, get Jeff book and read it. It is very interesting. Brains is not about "filtering, processing and analyzing". It is how we like to think about it :-)
  • brianlucebrianluce
    You know, the notion that a brain could evolve to the point where it *could* in fact exactly duplicate reality is not a bad concept for a sci fi story.
  • Vitaliy_KiselevVitaliy_Kiselev
    Braind do not dublicate reality. Evolution also showed that this is not his goal in long term :-)
  • brianlucebrianluce
    I'm just saying if it did, it might make a cool short story or film. To the larger point, is that what Chris meant? That the version of the reality the brain gives us is not an accurate representation? I don't know what Chris meant.
    Thanks for the book tip. I might get it. I read a lot of the reviews on Amazon.

Add a Comment