User Tag List

Page 2 of 7 FirstFirst 1234567 LastLast
Results 11 to 20 of 62

Thread: How the brain creates the experience of time

  1. #11
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    This is the presently prevailing view on time in neural networks.

    Understanding the computation of time using neural network models | PNAS

    As you can see, it is primitive and plainly inadequate.

    Much of this turf was covered by early experiments with the Boltzmann machine. In a nutshell, it stops learning after a while - for this exact reason, because frames are being associated and it clutters up the memory space with unnecessary detail.

    The brain has a whole interesting system that determines what exactly to store in memory and how to store it. Generally speaking, time and the experience of time are not part of that storage - although longitudinally they can be considered as indexing the frames.

  2. #12
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    Hm. This is one day old, I cant read it but it looks interesting.

    Estimating time with neural networks | Nature Machine Intelligence

  3. #13
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    In this article, you want to understand what they mean by "temporal basis sets".

    Computational Principles of Supervised Learning in the Cerebellum

    This is the cerebellum, it has a scaffold like network structure that looks like this:


  4. #14
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    This the view advanced by the receptive field types when they look at the hippocampus.

    Time cells in the hippocampus: a new dimension for mapping memories

    I'll tell you in a nutshell what they didn't tell you in all those thousands of words:

    The visual system is like every other brain pathway, it has a static channel (X) and a dynamic channel (Y). The presence and integration of the start/stop information is automatic.
    Last edited by nonsqtr; 09-15-2020 at 06:33 PM.

  5. #15
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    In the hippocampus, the memory "trace" (reflected as a pattern of synaptic weights) is only updated when neurons fire. The action potential is conducted backwards up the dendritic tree and causes a calcium influx in those synapses that are currently active.

    So you could say, a new sensory pattern is only "accepted" as a trace, when the neuron fires.

    Well, there bursting neurons in the septum that cause theta waves in the hippocampus. Theta waves are population activity, where the entire group of neurons becomes active at once, in synchrony.

    So the effect of a theta wave is to cause a hippocampal neuron to "repeatedly" update the information that's feeding into its memory traces.

    Why would it need to do this? Well, one reason is if it needed to "burn in" the information somehow, but the simpler explanation is that the sensory landscape is constantly changing, and you want to wait with storing the trace until you get a "good representation" so to speak - so you keep updating till you get it, and then you stop.

    Theta waves in the hippocampus are associated with alertness and attention, and generalized desynchrony in the cerebral cortex.

  6. #16
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    Anyone want to tackle the big enchilada?

    "Expectation".

    Which is not the same as anticipation.

    "Introduction to Probability 101"

    Anyone like to tell me what's wrong with the following definition of "prior probability"?

    Posterior Probability Definition

  7. #17
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    Here -

    prior probability represents what is originally believed before new evidence is introduced, and posterior probability takes this new evidence into account
    What's wrong with this statement?

    It should leap right off the page at you.

  8. #18
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    No one?

    How about that word "believed", it doesn't raise your eyebrows?

    Here, let's recap. Draw the visual system this way:

    hippocampus <= V3 <= V2 <= V1 <= retina

    (Not saying this is correct, it's just conceptual).

    I've drawn it this way for a reason. Pick a point in time t, call it "now", and for ease of discussion we can arbitrarily say "now" is at the origin t=0. Thus, in the diagram, "now" is on the far right. Light hits the retina "now", and then as the signal travels through the visual system it leaves a "trace" in the neural network in the form of an "evoked potential", which we discussed.

    We can draw a similar diagram on the motor side

    muscle <= M1 <= M2 <= ... <= Mn

    but in this case "now" is on the far left, at the muscle - and the network trace appears ahead of time in the form of a "pre-motor potential". So then, combining the two mappings, we get

    sensory <= 0 <= motor

    where 0 is the origin, the point in time defined as "now".

    This concept is called "anchor", it's defined as the intersection of the network representation with the environment. In the example we had retina and muscle, so consider that it's an eye muscle. Then "the anchors are co-located".

    But there's a problem with this mapping, isn't there? We clearly see that the motor activity is supposed to happen "ahead of" the anchor event, but that's not what happens in physical time. In physical time ("now"), the anchor event at the origin happened 100 msec "AGO" in M3. So our direction of time is reversed, we need to find a coordinate system that goes the other way.

    Well, if you're a point at the origin and your reference frame is such that you're looking in the direction of physical time, what we're talking about is a reflection through the origin, which results in the coordinate system becoming a mirror image.

    The neural network doesn't care about this, it's time-agnostic and there's no clock. If we flip the direction of physical time, we have correct representations of future and past IN AN EGOCENTRIC REFERENCE FRAME. The organism is very literally "looking into the future".

    So... belief... I have no idea what those assholes are talking about - that's a subjective word and there are two horrible assumptions underlying its use, one of which is provably untrue and the other so dubious as to be laughable.

    Probability, used in the sense of the quote, is a MEASURE, it's what the physicists call an "observable". It has nothing to do with "belief", it has to do with the representation of expectation and conditional expectation.

  9. #19
    Alumni Member V.I.P Achievements:
    50000 Experience PointsSocialCreated Blog entryVeteranRecommendation Second Class
    Overall activity: 49.0%

    Northern Rivers's Avatar
    Join Date
    May 2015
    Location
    Northern Rivers AUSTRALIA
    Posts
    25,330
    Thanks
    23,765
    Thanked: 23,521
    Rep Power
    21474867
    As a working and earning musician...timing is everything. To be accurate about it...took me decades. I can (finally) lay off and chime in exactly on the money...no matter how long of a rest the music has presented me. Right smack dab out of the blue...dead on...not even a 32nd, out.

    Saying that, an awful lot of what we do is innate timing. How we stride...climb steps...swim, chew...more than people realise. 4/4 timing is bred in the bone.
    I used to really like ducks.

  10. The Following 2 Users Say Thank You to Northern Rivers For This Useful Post:

    Kris P Bacon (09-18-2020),nonsqtr (09-18-2020)

  11. #20
    Alumni Member & VIP V.I.P Achievements:
    Social50000 Experience PointsVeteranTagger First Class
    Overall activity: 99.4%

    nonsqtr's Avatar
    Join Date
    Feb 2014
    Posts
    30,747
    Thanks
    12,380
    Thanked: 32,078
    Rep Power
    21474873
    Quote Originally Posted by Northern Rivers View Post
    As a working and earning musician...timing is everything. To be accurate about it...took me decades. I can (finally) lay off and chime in exactly on the money...no matter how long of a rest the music has presented me. Right smack dab out of the blue...dead on...not even a 32nd, out.

    Saying that, an awful lot of what we do is innate timing. How we stride...climb steps...swim, chew...more than people realise. 4/4 timing is bred in the bone.
    K - so as a musician then, you're already clear on the difference between listening and playing.

    When you hear a rhythm, and you "allow" your motor system to match it, that's called entrainment.

    https://en.wikipedia.org/wiki/Entrai...(biomusicology)

    Brainwave entrainment - Wikipedia

    Neural Entrainment to the Beat: The “Missing-Pulse” Phenomenon

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •