[ExI] Subjective/Objective

[Ben Zaiboc]

On 15/07/2026 21:40, Jason Resch wrote:
> On Wed, Jul 15, 2026 at 8:41 AM Ben Zaiboc via extropy-chat <extropy-chat at lists.extropy.org> wrote:
>
>
>     I wasn't trying to say that some messages inherently convey meaning, just the opposite. The ". .. ..." message is one that's familiar to us, precisely because of our SETI efforts and our attempts to create unambiguous messages that aliens might be able to understand. We are familiar with a context, so can make a guess at the meaning (which would be wrong, in this case).
>
>
> It is true that short of well-defined protocols of the type programs follow, there is always some degree of guess work and room for misinterpretation when it comes to an interpreter attempting to interpret a message. Even when two people speak the same language, each word for each person, is surrounded by a different set of associations and connotations of various strengths. It is a bit of a miracle that language works as well as it does to enable two minds to synchronize on more or less, similar lines of thought.


You don't have to tell me.
I'm constantly moaning about the dreadful language skills so many people display these days. Even mainstream news outlets, like the BBC keep printing howlers that make me cringe. Yes, the BBC. They used to be a bastion of the english language, but now they're ... well, let's just say they're not.
A few weeks long ago, I saw an advert on a billboard in my local high street, and I literally didn't know what it was saying. I'm not exaggerating. All the words were normal english ones, no jargon or slang, but the combination of words just didn't make any sense.

>  
>
>
>
>
>     Someone with a mathematical background would probably assume that the message says something about prime numbers, or maybe Pascal's triangle, or some other mathematical structure. And they'd be wrong.
>
>
> The longer the message, however, the more probable it becomes that a string of digits that perfectly matches the leading digits of Pi is actually meant to mean the digits of Pi, rather than some other thing that by chance, happens to follow the same pattern of Pi's digits.
>  

Certainly. You have to already know what the digits of Pi are, of course, and be able to compare the two to evaluate how close they are.
Without those two extra things, nobody would be able to tell that the digits are related to Pi.

>
>
>
>
>     What I had in mind (the actual meaning of the message) wasn't any of those. My point is that nobody can know what I intended those dots to mean, and no system can derive the meaning, without knowing my intention or having some decoding system that assumes that intention.
>
>
>
>     I'm sure you're right about compressible strings and discoverable patterns.
>
>
>
>     Where my idea possibly breaks down is, as you say, when there are large amounts of information. The example of LLMs is probably the best one.
>
>
> Yes, I agree with you that short messages are always ambiguous. But when it comes to something large, complex, self-referential, like the total English text of wikipedia, then even an alien from another universe, having no context, or understanding of the characters, could ultimately come to understand the meaning of this text. They might start by recognizing symbols for numbers, which appear in so many lists, and then come to understand the meaning of our symbols like +, -, and =, after seeing numerous examples of numbers added and the results. Then they would see patterns in language, like "a" and "the" always proceeding some types of words, but not others (the nouns), and then by seeing the pattern "an X is a Y" (e.g. "a dog is a mammal", "a mammal is an animal", etc.) it could generate a hierarchical classification, and come to realize the meaning of "thing" and "object" being the most general type of noun. Then they might notice the pattern "X is made of Y", and they could discover that atoms are more basic than molecules, and fundamental particles more basic than atoms, than then with its understanding of numbers, math, and knowing the basic equations that desctibe the fundamrntal particles that make up everything described in Wikipedia, these aliens from another universe, could have a pretty good idea of the stuff in our world, and how it behaves. No Rosetta stone is needed, the message (when large enough, and redundant enough) is self-describing (like the digits of Pi are when there enough of them).
>  
>

I would still argue that this is not 'self-describing'. The Rosetta stone in these cases is the existing knowledge of the world. A text that was about how magic works in Harry Potter would likely remain untranslatable.

And there's the question of still-untranslated scripts like Linear A, Olmec, Etruscan and the Indus Valley script. Some of those we may never understand.

>
>
>
>     When there is a very large amount of information, probably spread across many messages, it's probably possible to find patterns that correspond to patterns from another source, like the real world, and guess that there's some real link, and keep going on that basis until the messages can be deciphered to make sense with respect to the other source. It's very unlikely that, with enough information, this process will give false results.
>
>
> Yes.
>  
>
>
>
>
>     That's not the same thing, though, as a message that contains its own meaning, without any context. That's many messages, correlated with a guessed context (like the elliptical orbits example), and probably many attempts to fit the data to the context.
>
>
> Well there is no limit on message size. The entire multi-PB text-corpus used to train ChatGPT could be considered a single message, as one single (very long) string of 1s and 0s.
>  
>
>
>
>
>     So while what you say is correct, I don't think it invalidates my claim, which after all is actually quite modest. If those dots are fed into a nerve axon, in isolation, there's no way that whatever the axon leads to will be able to interpret my intended meaning. It will use its own pre-set mechanisms and context to interpret the message, and it will come up with the wrong answer. The meaning of the message is what is heard, not what is said (to quote one of the Assumptions of NLP).
>
>
> But consider the mostly blank state of a newborn's brain. All its brain receives are timing signals of nerve firings. And from these timing signals alone (a message without context), the brain builds an entire model of the external world, and imbues those mere timing signals with meaning. The list of all nerve impulses and their times, can be considered as one long message, that's gradually processed. And yet, somehow, from these "meaningless neural spikes" the whole meaningful world emerges. This, I believe, is accomplibhsed by the neural networ's capacity to learn to recognize, interpret, and predict patterns.


There's one factor that hasn't been mentioned yet: Feedback.

I think that feedback is essential to making sense of ambiguous signals, and is what enables us to build an internal world. We don't get signals only, we respond to them with actions, like blinking, moving our head, touching things, cramming them in our mouths, etc., and those actions cause further inputs, so over time we start to assign meanings to the signals, meanings that relate to our actions. I expect that a baby that was paralyzed would never be able to make sense of the sensory inputs it gets because it couldn't respond to them with its own actions (of course it would be dead before it had a chance, so verifying this would be difficult).

But all this is way beyond what we were originally talking about: The messages passed from one neuron to another. Each of the short, spike-train messages that a neuron sends along its axon are abstract signals, with no intrinsic meaning. The receiving neurons impose a limited meaning on them, because they provide the context, but without that, the signals could mean anything or nothing.
When the signals build up into a complex web of information over time (with feedback from our responses), then some kind of higher-level meaning starts to emerge, certainly.

This is not what Brent is talking about, though. He talks about '/fundamental/ colour qualities', being transmitted without any context or processing needed, but still somehow having meaning.

>  
>
>
>
>
>     What interests me is Brent's claim that there exists some other kind of messaging (what he's calling 'Non-causal messages', I think), and what that could possibly be (especially as it would, as far as I know, break the laws of physics). I suspect it's a figment of his imagination, but it's up to him to demonstrate that it actually exists.
>
>
> Brent, like I think all of us, struggle to understand how it is the brain accomplishes its great feat of producing qualia. It seems impossible. Carrying on with the assumption that it is impossible to create something like qualia, from a physical thing that doesn't have it, Brent reasons that the qualia must be intrinsic to the physical thing from the start.
>
> As I see it, however, this is the same mistake of those who thought dead matter is missing some "elan vital" and only special matter imbued with this elan vital property could build life. Their thinking was "life can't come from matter that lacks a vital force." much as Brent believes qualia (like the experience of red) cannot come from matter that lacks intrinsic qualities like redness. But today we understand the concept of emergence. Living things can be made of dead matter. New properties emerge when a bunch of smaller things are put together to interoperate in new ways. As difficult as it is for us to believe that a red experience can come from things that have no innate or intrinsic redness property, it seems this must be the case.
>
> The bridge to seeing this is to consider simple processes which have basic knowledge states. Think of a simple sensation like the knowledge that you feel the smallest touch on the back of your hand. What does this feel like beyond the mere knowledge of the location of where on your hand you feel this touch? That is ultimately what all conscious states are: knowledge states. For any process that is in a state of possessing knowledge, there is "something it is like" to be that process. For a process that can discriminate red from green blue and yellow, where red is the complement of green+blue (cyan), there is something it is like to be that process when it is in a state of having discriminated red from those other colors which hold those relations to the other colors. Knowing this doesn't reveal the "what it is like" as that knowledge is private to those processes which happen to be in those states having made those discriminations. It is frustrating that private subjectivity can't be described publicly/objectively, but that is the way of things. It logically cannot be any other way.

I agree with all that.

-- 
Ben