Jump to content

Welcome to FutureTimeline.forum
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. If you already have an account, login here - otherwise create an account for free today!
Photo

Memory Uploading... how will we validate what we have experienced?


  • Please log in to reply
7 replies to this topic

#1
pinkrose

pinkrose

    New Member

  • Members
  • Pip
  • 4 posts
This question has been on my mind lately- how will we “validate” our memories from the time we were born up until we have BCI as commonplace recording all neural activity as it occurs? We know now that the brain is not perfect at remembering every detail of our strongest memories, much less every event that has occurred in our lives so in the event of mind uploading, will we only be able to rely 100% on what we have experienced since recording began? How does Kurzweil think that with enough random data about his father (his own faulty/biased memories of him and other anecdotes or writings) he will be able to recreate him? Surely our inner monologues are a large part of who we are but if we have no recordings of these for our whole lives how would our uploads be 100% us? Any thoughts or comments on anything remotely related to this I’m interested in.

#2
funkervogt

funkervogt

    Member

  • Members
  • PipPipPipPipPipPipPip
  • 1,303 posts

There are colossal numbers of videos, photos, writings, and social media posts that have captured the daily actions and thoughts of hundreds of millions of people for years now. The troves of metadata (including location tracking) that tech companies and spy agencies have saved could also fill in a lot of blanks. 



#3
starspawn0

starspawn0

    Member

  • Members
  • PipPipPipPipPipPipPipPip
  • 2,263 posts

I decided to pop in to write about some new research; but then also saw this, so thought I would comment:
 
There's a now old dogma about human memory that says that the act of recalling a memory causes us to change it -- in fact, we can even overwrite and pervert it beyond recognition.  This dogma was especially strongly applied to the case of children who recalled being abused and sexually assaulted by daycare workers.  The claim was that the psychologists and police who interviewed them planted the suggestion of the abuse, and that is what they remembered.  
 
It's probably true that memories can be warped in this way; and we know from our own experience how easily we can forgot something from a long time ago.  And, obviously, since our own brain changes, and the mental representations used to anchor memory along with it, we're not going to remember things exactly as they were.  Perhaps this is why old movies look outdated when we see them again, even though we remember them as being so "new" and "relevant".
 
All of that is true.  BUT... the thing about psychology and neuroscientific research is that old dogmas are constantly being overturned.  I strongly suspect, for example, that the brain "remembers" far more about our experiences than we think; it's just that what it stores away isn't always accessible to the conscious mind.
 
One example is how Pokemon permanently alters the brain:

https://www.theverge...ain-information
 

By scanning the brains of adults who played Pokémon as kids, researchers learned that this group of people have a brain region that responds more to the cartoon characters than to other pictures. More importantly, this charming research method has given us new insight into how the brain organizes visual information.


One can find all sorts of relics of childhood like this by performing a brain scan.  As we grow and learn new things, we don't merely reshape synaptic weights, but alter the structure of the bran itself!

In addition to this, there are some memories that are known to be extremely stable over a long period of time, due to a neural population encoding that improves statistical power to recall them -- e.g. motor memories:

https://www.nature.c...1593-019-0555-4
 

We recorded from populations of neurons in premotor, primary motor and somatosensory cortices as monkeys performed a reaching task, for up to 2 years. Intriguingly, despite a steady turnover in the recorded neurons, the low-dimensional latent dynamics remained stable. The stability allowed reliable decoding of behavioral features for the entire timespan, while fixed decoders based directly on the recorded neural activity degraded substantially. We posit that stable latent cortical dynamics within the manifold are the fundamental building blocks underlying consistent behavioral execution.


Procedural and semantic knowledge is also probably stably encoded in the brain. This doesn't mean that the person always has reliable access to it; but that the information is "still there", long past after they have forgotten about it (like the Pokemon area).

And, probably, the stuff that we think we have overwriten -- e.g. like those children who had their memories altered through persuasion and suggestion -- is still there, inaccessible to the conscious mind -- but will be accessible to sufficiently powerful BCIs and decoding models.

One reason I say this, because it seems to happen with Alzheimer's patients (for anyone who has ever observed them): they often have episodes where they appear to think they are living in the past, and can recall experiences with perfect clarity from 50 years ago that they had forgotten -- and that almost nobody else remembers; but that has the ring of truth, so probably really happened. It's really freakish to see, for those who have never been around an Alzheimer's victim!

Another case study that comes to mind is how we can remember when things have changed, without being aware we even "recorded" what the thing was like. e.g. if you go walking down a familiar street, and someone changed the color of a building -- say, from gray to brown to match the other buildings -- you would probably notice that something was different; but might not be able to say exactly what it was.

My suspicion -- that runs against current dogma -- is that human memory acts more like a video and sound recorder than we think. Sure, it's not a perfect, and is more like a highly compressed recording; but it's far better than we think it is. And, sure, things get forgotten and warped due to suggestion and denial; but the truth is still in there, somewhere, and will be extract some day...



#4
pinkrose

pinkrose

    New Member

  • Members
  • Pip
  • 4 posts

And, probably, the stuff that we think we have overwriten -- e.g. like those children who had their memories altered through persuasion and suggestion -- is still there, inaccessible to the conscious mind -- but will be accessible to sufficiently powerful BCIs and decoding models.

One reason I say this, because it seems to happen with Alzheimer's patients (for anyone who has ever observed them): they often have episodes where they appear to think they are living in the past, and can recall experiences with perfect clarity from 50 years ago that they had forgotten -- and that almost nobody else remembers; but that has the ring of truth, so probably really happened. It's really freakish to see, for those who have never been around an Alzheimer's victim!

Another case study that comes to mind is how we can remember when things have changed, without being aware we even "recorded" what the thing was like. e.g. if you go walking down a familiar street, and someone changed the color of a building -- say, from gray to brown to match the other buildings -- you would probably notice that something was different; but might not be able to say exactly what it was.

My suspicion -- that runs against current dogma -- is that human memory acts more like a video and sound recorder than we think. Sure, it's not a perfect, and is more like a highly compressed recording; but it's far better than we think it is. And, sure, things get forgotten and warped due to suggestion and denial; but the truth is still in there, somewhere, and will be extract some day...


Interesting points... what leads to you to believe that memory behaves like a video/sound recorder? From my understandings we jog our memories through associated stimuli in “networks” of sorts rather than chronologically, so how would this all be extracted via BCIs and what does this mean for memory uploading or sharing? If we are able to share our memories how will the other person experience our subjective experiences to integrate with their own lives? I personally believe that we will have objective memories recorded as “truth” once we have these uploads happening in real time. Any thoughts on this?

#5
starspawn0

starspawn0

    Member

  • Members
  • PipPipPipPipPipPipPipPip
  • 2,263 posts
It can be "both / and", instead of "either / or" -- that is, it can store memories both as sequences and as random associations without regard to time.  This kind of thing is common.  A neuroscientist will say, "If there's one thing we know for sure, it's that hand movements are encoded in the motor cortex"; and then somebody will discover that, in fact, contrary to this, they are also encoded in the visual cortex -- and can be decoded from there.  And then another neuroscientist will say, "If there's one thing we know for sure about how the brain encodes place memories, it's that it uses place cells and grid cells"; and then someone will find a way to decode them from population responses. 
 
Nevertheless, there is recent work on sequence playback in the brain:
 
https://www.scienced...00505121711.htm
 

When we fall asleep, our brains are not merely offline, they're busy organizing new memories -- and now, scientists have gotten a glimpse of the process. Researchers report the first direct evidence that human brains replay waking experiences while asleep, seen in the brains of two participants who had been implanted with microelectrode arrays as part of a brain-computer interface pilot clinical trial.


And Deepmind has done some work on this. In fact, their original Deep Reinforcement Learning work on playing Atari made use of experience-replay.

I think there are much older neuroscientific theories around sequence learning, that motivated constructions like Echo State Networks, but will need to refresh my memory on this.

One could argue that "sequence prediction" that has been observed in the brain encodes "sequence memory". Being good at predicting the next stimulus response accurately will necessarily require a good memory of previously-seen patterns. There is a tension between being good at "generalizing" and "memorizing"; when memory is called-for, to be good at prediction means you better be good at memorizing the sequence.
 
Addendum:  More on unconscious memory:
 
https://www.sciencem...hout-knowing-it
 

Hannula and Ranganath found that when subjects' eyes focused on the correct match, the hippocampus and related memory areas in the medial temporal lobe, which the hippocampus is a part of, lit up. Even if subjects picked the wrong face, the hippocampus was still more active when they stared at the correct face. The scientists conclude today in Neuron that the hippocampus may be retrieving the link between face and scene even when the person doesn't seem to remember it.



#6
purpose

purpose

    Member

  • Members
  • PipPip
  • 11 posts

Memory recall is a complicated topic and one that tends to be often oversimplified in these discussions. 

 

It is important to differentiate between the various kinds of brain interfaces to the sensorium phenomena:

 

1) Realtime experience (i.e. the most direct and temporally present version of sensorium phenomena, i.e. the chemo-electrical wave patterns that are still oscillating in the brain)

2) The ability to immediately recall details of a very recent event (neutral tissue that is still primed to re-fire and thus recreate aspects of recent wave patterns)

3) The neutral tissue connections formed long after the wave patterns have changed, encoding an "averaged and summarized" potential for these wave patterns to be reproduced in the near future.

4) The ever changing and long term neural structural changes in the brain and where these collective changes "color" memory recall. e.g. when recalling an old memory the similarity of the current sensorium of that memory is influenced by multiple factors.  For example, the sum-total changes to the neural structures hosting the recall (many memories share the same neural substrate and thus influence each others future recall to some degree) and the fact that all recall is influenced by the current realtime sensorium and thus naturally will be perceived through the current mode of thinking, not the original sensorium.

 

In other words the brain is unconcerned with authentic recall, but rather is focused on structuring memory to best fit its current model of the world.  This is not to suggest that the brains ability to store and recall accurate events is not impressive, it often is, but unlike our current day computers the brains architecture has no apparent preference for authentic data storage when refined data better fits its current/refined models.



#7
pinkrose

pinkrose

    New Member

  • Members
  • Pip
  • 4 posts

Memory recall is a complicated topic and one that tends to be often oversimplified in these discussions. 
 
It is important to differentiate between the various kinds of brain interfaces to the sensorium phenomena:
 
1) Realtime experience (i.e. the most direct and temporally present version of sensorium phenomena, i.e. the chemo-electrical wave patterns that are still oscillating in the brain)
2) The ability to immediately recall details of a very recent event (neutral tissue that is still primed to re-fire and thus recreate aspects of recent wave patterns)
3) The neutral tissue connections formed long after the wave patterns have changed, encoding an "averaged and summarized" potential for these wave patterns to be reproduced in the near future.
4) The ever changing and long term neural structural changes in the brain and where these collective changes "color" memory recall. e.g. when recalling an old memory the similarity of the current sensorium of that memory is influenced by multiple factors.  For example, the sum-total changes to the neural structures hosting the recall (many memories share the same neural substrate and thus influence each others future recall to some degree) and the fact that all recall is influenced by the current realtime sensorium and thus naturally will be perceived through the current mode of thinking, not the original sensorium.
 
In other words the brain is unconcerned with authentic recall, but rather is focused on structuring memory to best fit its current model of the world.  This is not to suggest that the brains ability to store and recall accurate events is not impressive, it often is, but unlike our current day computers the brains architecture has no apparent preference for authentic data storage when refined data better fits its current/refined models.


I believe this too, we are unconcerned with 100% authentic recall but rather “schemas” of our lives... I think about Musk’s comment in his Neuralink demo though about memory uploading and life being like a “black mirror episode” where we can playback our entire lives and wonder how this will come into play in the future. How will we be able to playback the past before we have one of these devices if there is no standard format for memories or consciousness? Thoughts?

#8
pinkrose

pinkrose

    New Member

  • Members
  • Pip
  • 4 posts
Does anyone have any updates on this? I find this topic fascinating.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users