The Empty Brain

[studiot]

Several members have congratulated @Alkonoklazt for bringing this paper to our attention and I believe it is so interesting it deserves adiscussion thread all of its own

 

The Empty Brain 

Robert Epstein

 

https://aeon.co/essays/your-brain-does-not-process-information-and-it-is-not-a-computer

 

The paper is short but I am not sure I have permission to reproduce it here.

 

Perhaps a mod would advise please ?

 

I have a couple of reservations about the paper in that it makes absolute declarations or assertions although I do appreciate the 'market' in which the concepts are being promoted.

For instance it says there is no 'memory' and no algorithmic programming.

In this case I wonder how I can correctly produce long numbers such as social security numbers, telephone numbers and the like on demand.

Edited September 17, 2023 by studiot

[swansont]

12 minutes ago, studiot said:

Perhaps a mod would advise please ?

A summary would be appropriate. You can post excerpts, as necessary.

[CharonY]

I skimmed the essay (or at least a similar one) quite a while ago, and I think most of the time I had the term no shit sherlock in my mind. I agree with the general gist, the way the brain works is not the way a computer works. It is also true that the way we describe cognitive activities are superficial narratives (including specifically "information processing theory", as we do not really understand the underlying biology.

And that is in my mind the overall issue, we have mostly a black box, we can see what comes in and what comes out and we make a story about what might happen in between. Bits and pieces are known, but we do not really know how they fit together. I think with respect to memory, we do not store memory, we (re)create memory, in part when certain paths are activated in certain combinations. I think a claim that memory does not exist is overreach, but memory as expressed in information processing theory, which was and perhaps still is something that has been heavily promoted, in cognitive psych. It also has resulted in quite a vast arrays of self-improvement theories (usually with little evidence) and which also has been liked a lot by tech folks. 

[Genady]

Just as a mental experiment...

Let's imagine an elaborated rainbow crystal. Depending on direction of an incoming white light, different multicolored patterns are projected out of it. The query for social security number is associated with a specific direction of incoming light. When light comes onto the crystal from that direction, the social security number is projected out.

Edited September 18, 2023 by Genady

[CharonY]

7 minutes ago, Genady said:

Just as a mental experiment...

Let's imagine an elaborated rainbow crystal. Depending on direction of an incoming white light, different multicolored patterns are projected out of it. The query for social security number is associated with a specific direction of incoming light. When light comes onto the crystal from that direction, the social security number is projected out.

This seems to suggest that the number is somehow hardcoded in the crystal, and it is a matter of finding it. This likely not how it works in the brain, though. 

There are quite a few different models, created by disciplines ranging from cognitive psych to more neurological sciences. One is called memory engrams , and the idea is that connections and activities in the brain (classic examples are LTPs and LTDs) are changed upon learning and acquiring information, and that activation of those patterns results in recall (i.e. it is an active and creative process as these pattern are themselves plastic and the patterns do not recreate themselves fully). So memorizing a number does not necessarily store the number itself, but it may be part of a the context in which the number is stored. One example is passwords, for example. If you often type  complex password, you might note that typing it in sometimes requires little thinking, here the memory is linked to movements that have been strengthened or otherwise activated over time. Yet trying to recall them verbally can be a bit more challenging, especially if one uses many of those. Also, depending on type of memory, those can be quite distributed across the brain, so it requires many parts to recreate memories.

[Genady]

6 minutes ago, CharonY said:

This seems to suggest that the number is somehow hardcoded in the crystal

No, I hope it does not. I meant to make it obvious that the hardcoded number is nowhere in the crystal, but the number appears as a result of the white light going through the entire crystal. And there is no algorithm anywhere in the crystal for retrieval of the number. 

So, I tried to make a simple example of a thing where "memorizing a number does not necessarily store the number itself", etc.

If my example failed to demonstrate it, too bad for the example.

(Do people know what I mean by rainbow crystal? My wife has one hanging on top of the window. It makes beautiful moving and changing colorful patterns on the wall in the morning.)

[TheVat]

7 minutes ago, Genady said:

No, I hope it does not. I meant to make it obvious that the hardcoded number is nowhere in the crystal, but the number appears as a result of the white light going through the entire crystal. And there is no algorithm anywhere in the crystal for retrieval of the number. 

Is this similar to holographic memory theory?  

But even if not, I think it illustrates how a brain could recreate an experience without retrieving it from any kind of register.  It would make reflection a better term than processing.

2 hours ago, studiot said:

For instance it says there is no 'memory' and no algorithmic programming.

IIRC, Penrose has also speculated that human brains are not algorithmic.  I will try to find a link if the chat goes that way.

[iNow]

1 hour ago, CharonY said:

with respect to memory, we do not store memory, we (re)create memory

This is a part I believe is often missed by many.

The process of remembering things relies on many of the same neural processes as are used in writing or creating entirely new things. It’s basically narrative creation with heavier anchor points. We’re literally rewriting our past experience as if it were happening anew right now in the present in our current mood and mind, and we further erode fidelity to the original event(s) each time we do.

Paradoxically, memories tend to get less accurate the more we remember them. 

https://www.news-medical.net/news/20200814/Study-Adopting-a-third-person-to-recall-memories-changes-how-our-brain-processes-them.aspx

[Genady]

The statement in the article, which I did not fully understand is this:

Quote

a snapshot of the brain’s current state might also be meaningless unless we knew the entire life history of that brain’s owner

I don't think it is explained in the article, why it might be so. 

[TheVat]

10 hours ago, Genady said:

The statement in the article, which I did not fully understand is this:

I don't think it is explained in the article, why it might be so. 

I think it relates to the idea that two people can have the same thought but each brain will show a different set of pathways for that same thought.  Each brain develops connections and signal paths in an idiosyncratic way.  (that's the price of plasticity) So the converse of that is that you could have two identical connectomes which would be having entirely different thoughts and memories.  A connectome map does not consistently correspond to a particular pattern of thought.  Human neural architecture is not built of off-the-shelf standardized components.

So you would need my entire developmental and experiential history to know what my current connectome snapshot means.  And maybe not even then?

[Genady]

46 minutes ago, TheVat said:

I think it relates to the idea that two people can have the same thought but each brain will show a different set of pathways for that same thought.  Each brain develops connections and signal paths in an idiosyncratic way.  (that's the price of plasticity) So the converse of that is that you could have two identical connectomes which would be having entirely different thoughts and memories.  A connectome map does not consistently correspond to a particular pattern of thought.  Human neural architecture is not built of off-the-shelf standardized components.

So you would need my entire developmental and experiential history to know what my current connectome snapshot means.  And maybe not even then?

I understand that there are no two identical brains. What I don't understand is, how history of a given brain can affect its current thoughts and memories without affecting its current state.

[CharonY]

1 hour ago, Genady said:

I understand that there are no two identical brains. What I don't understand is, how history of a given brain can affect its current thoughts and memories without affecting its current state.

Well, it will affect its current state, but it depends a bit on perspective. Basically, memory formation is an active process that is associated with some sort of changes in structure and activity patterns. However, the precise activity is highly dependent on the state the brain is currently in at time of creation, as well as retrieval. I.e. if you stimulate the same pattern in different brains, the results for most cognitive elements will be different. 

[Genady]

31 minutes ago, CharonY said:

Well, it will affect its current state, but it depends a bit on perspective. Basically, memory formation is an active process that is associated with some sort of changes in structure and activity patterns. However, the precise activity is highly dependent on the state the brain is currently in at time of creation, as well as retrieval. I.e. if you stimulate the same pattern in different brains, the results for most cognitive elements will be different. 

The brains are different - the results should be expected to be different. As in a simplest mechanical system, the same force being applied to different masses causes different accelerations.

[CharonY]

Well, yes and because the brain constantly changes, you need to know the history of the brain if you want to reconstruct what things mean from a snapshot.

Or think of it that way, cognitive functions are a process and in order to understand what is going on, it is not enough to see what is happening right now. You have to understand the full context, as the brain keeps changing (unlike a crystal perhaps?) to interpret what is happening at any given time. 

This is not entirely true, as there are areas of the brain that are somewhat fixed and can be associated with certain activities, but I think the author refers to more complex cognitive activities.

[Genady]

4 minutes ago, CharonY said:

Well, yes and because the brain constantly changes, you need to know the history of the brain if you want to reconstruct what things mean from a snapshot.

 

I think my lack of understanding of this point is caused by my lack of understanding of what "to reconstruct what things mean" means.

My point (perhaps different from the one in question) is that to describe the next state of a system all one needs to consider is its current state and the current state of all external factors that affect it.

[CharonY]

2 hours ago, Genady said:

I think my lack of understanding of this point is caused by my lack of understanding of what "to reconstruct what things mean" means.

My point (perhaps different from the one in question) is that to describe the next state of a system all one needs to consider is its current state and the current state of all external factors that affect it.

To figure out the misunderstanding here, where do you see the difference between what you describe here and:

 

15 hours ago, Genady said:

a snapshot of the brain’s current state might also be meaningless unless we knew the entire life history of that brain’s owner

I suspect the issue might be in the details and/or phrasing.

[Genady]

1 minute ago, CharonY said:

To figure out the misunderstanding here, where do you see the difference between what you describe here and:

 

I suspect the issue might be in the details and/or phrasing.

I guess the snapshot of a state is not the same as the state.

Using a mechanical analogy again, is a snapshot like a 'position' of particle, while a state is its 'position and momentum'?

[TheVat]

5 hours ago, Genady said:

I understand that there are no two identical brains. What I don't understand is, how history of a given brain can affect its current thoughts and memories without affecting its current state.

I think a snapshot is static, but its meaning is dynamic.  So your later analogy might be somewhat valid.  We don't really get the dynamics of a mental state from just seeing the connectome map.

In your particle analogy, we don't know the momentum of the particles.  In the connectome map, we don't have the kind of dynamic picture that would allow us to say When Alice's C4905 transmedial fiber fires, plus (etc), then she is seeing a soft red glow.

We don't know rising or falling activation potentials or numerous other dynamic conditions back through time to when baby Alice was figuring out the soft red glow and configuring those transmedial fibers and synaptic sensitivities and so on.  Maybe it's like penetrating a lot of noise and we just can't.

But who knows?  Maybe someday.

[Ghideon]

2 hours ago, CharonY said:

Well, yes and because the brain constantly changes, you need to know the history of the brain if you want to reconstruct what things mean from a snapshot.

I agree. An one addition; I don't think the brain has a "pause button" or a "global clock" that allows for a well defined state even if history wasn't an issue. Here the computer analogy may work as an illustration?
A single computer typically allows a snapshot to be persisted and the state can be recreated on a different machine.
In a large network with many computers and many concurrent network connections it gets tricky. Assuming constant network traffic and concurrent changes to local computers it may not be simple to take a global snapshot that correctly represents the complete state including signals in transit between computers.

[Genady]

Certainly, any static picture, even one that includes not only the connectome map but also momentary distribution of all neurotransmitters etc., is not sufficient. Dynamic data are necessary.

Here is another analogy, weather. Static satellite pictures of current weather conditions do not allow to make predictions. Dynamic data are necessary. But I doubt that it necessitates knowledge of the entire history of Earth. Cf.,

Quote

unless we knew the entire life history of that brain’s owner

 

Edited September 18, 2023 by Genady

[CharonY]

2 hours ago, Genady said:

Certainly, any static picture, even one that includes not only the connectome map but also momentary distribution of all neurotransmitters etc., is not sufficient. Dynamic data are necessary.

Here is another analogy, weather. Static satellite pictures of current weather conditions do not allow to make predictions. Dynamic data are necessary. But I doubt that it necessitates knowledge of the entire history of Earth. Cf.,

 

Well it does, if you do not know how things interact with each other. For example, in case of weather, we have a generally idea e.g. temperature trends, factors influencing precipitation and so on. So you could just take the recent history (or even just the model) and try to predict things. In the brain our knowledge is much more limited so if we reconstruct what is happening at any time, we have to move back further and further to see how things are connected.

The other issue is that things might be handled very differently by each respective brain. For example an apple might be associated with food in one brain, but with a traumatic even in another. Without knowing that, the different activities in a brain in response to showing an apple might not be understandable.

 

[Genady]

11 minutes ago, CharonY said:

if you do not know how things interact with each other

i.e., as long as brain is a 'black box,' I agree. I know that now, it pretty much is.

I think, I know now what the quoted sentence referred to. Thanks a lot.

[studiot]

12 hours ago, Genady said:

Certainly, any static picture, even one that includes not only the connectome map but also momentary distribution of all neurotransmitters etc., is not sufficient. Dynamic data are necessary.

Here is another analogy, weather. Static satellite pictures of current weather conditions do not allow to make predictions. Dynamic data are necessary. But I doubt that it necessitates knowledge of the entire history of Earth. Cf.,

 

Yes in general I agree.

The Brain is the most complicated thing in our known universe so it is little wonder that we have barely begum to get a handle on it.

Even more so when you realise that the brian alone is only part of the whole human consciousness system. We have at least found that bit out.

 

As regards the weather here is a true story from my experience.

We have also found another way to predict the weather.

Weather prediction is quite important to many industries not least the gas supply industry, especially in the winter.

So some years ago British Gas, which normally uses the met office predictions, wanted a backup system if for any reason the normal weather forcasts became unavailable.
So they experimented with statistical methods on the basis "Null hyypothesis : Tomorow's weather will be the same as today's."  What are the chances ?

Conditional probability, Markov and Bayesian methods allowd them to conclude that the developed statistical calculator "Was at least as accurate as the met office in predicting the temperature in the next 24 hours and therefore the gas demand".