Anyway, this got me to thinking, which is that one would expect one's internal langauge to be highly compressed, meaning it need not adhere to conventional grammar in order to be language, — Hanover

In terms of the neuroscience, the "compression" happens as you don't have to fully unpack an intention because you already know where it is largely about to go.

Standing waiting to return a tennis serve, you could go through the effort of constructing a complete mental image of lunging low to your backhand and smiting the ball straight as an arrow down the line. Or you could just stand there feeling the general intention of being about to do exactly that kind of thing should the need occur.

So it a general thing about the brain. You know what to do from well-drilled habit. And you also know that you need to be ready and focused in an oriented and intentful way. Each moment presents its challenge. You are aware enough of what it is to already be predicting your actual response with the degree of vividness and fixity that would be helpful.

And so it is with our linguistic responses to the world. And with our own internal organising narrative of that world.

We face the moment in a way that already the right kind of words are starting to assemble. If we wanted, we could develop that general oriented intent into some spoken utterance. Or even the motor image of that utterance as words we are saying to ourselves in our head. A shadow sensorimotor image of exactly the same sort which would be imagining the perfect backhand return if we happened to get a wide kicker served by our net-rushing opponent.

But much of the time, we don't have to promote an intent to respond to a full actual – or even vividly imagined – sensorimotor response. Just being aware the circumstances of the moment are what they are, and this is the general idea of how we might launch into some fully grammatical structure of words as a suitable thought, has already got the job done. We can skip on ahead having just got ready to say something, and not then hanging around to articulate what would be by now a rather predictable thing to have heard being said.

Often we do articulate our inner speech – promote it to sentences – as it is useful to be surprised by how our machinery of speech habits does express our intent. But mostly the world moves fast and so we let the articulation slide. Our thoughts feel as if they fly along in rather wordless fashion. Perhaps just fragments of phrases and abandoned points we might have made.

So there is an interesting question about how languages evolve to become better for some kinds of thinking. English is said to be good for form abstract nouns out of everything. The Chinese number system is better suited to maths.

But the mentalese issue is explained by the fact that cognition divides into rapid learnt habit vs slower patient attention.

The brain always needs to be reacting to whatever is happening. So everything passing through our lives is putting us in mind of the sort of responses we ought to start generating. The kinds of clever things we could put some attention on and develop into a fully imagined motor intent.

But more often than not, the challenges of the moment turn out rather mundane. We get ready, but can already start to relax again. Something else is bubbling up and now we are getting ready to lurch in this next direction instead.

Excellent argument. But it will be ignored. :grin:

Your dislike of maps is how you get lost even in the terrain of your own arguments.

Why would you want to attribute it to an aspect of a social hierarchy when it just appears to be a basic aspect of being an individual? — Metaphysician Undercover

Because it matters how the social hierarchy works in social animals. It speaks to the algorithm organising the complex lives of animals that are more than the one dimensional creatures you seem to think they are.

I don't believe that the two sides go hand in hand at all. This attitude leads to infinite regress. We discussed this before as the relation between the whole and the part. One must be prior to the other or else they've both existed together forever, without beginning. — Metaphysician Undercover

Yep. This is certainly your concept of how systems are organised. System science doesn’t agree.

I don't see how it's not innate. That it requires learning, only means it needs stimulus, but it comes from inside the creature. The world doesn't "teach" us consciousness, it sharpens and refines what we already have. — Manuel

I’m talking about self-aware and introspective human consciousness as something beyond simple animal enactive awareness. How language scaffolds what we mean by that level of mind.

Without language, there is nothing reportable because there is no socialised habit of narration. An inner story of that kind is not being created.

It just seems utterly wrong to think it answers something like the Hard Problem. I don't take hand-waving very well.. — AmadeusD

Well I was in the audience when Chalmers first raised his hard problem argument. I had lunch with him after to see if he was actually serious and had much email debate with him in the year after that.

But of course, your unexamined opinions must prevail here too.

Be reassured, you seem marvellous at the hand-waving. A duck to water. :up:

When facts meet ignorance, opinions always win.

You are not exactly a guy for the details, even if you continually demand them. :grin:

Well, far from comatose, we do have examples of people like Hellen Keller, who managed to become a wonderful writer while being deaf-mute. — Manuel

Well Keller only lost both her sight and hearing after an illness at age two. So she had that much normal exposure to the world in terms of her social and biological development. And through a supportive family, she developed a reasonable system of sign language, such as miming putting on glasses to signify her father, or a tying up of her hair to signify her mother. As well as shakes of the head to mean yes and no, pushes and pulls to mean come or go. A "vocabulary" of about 60 signs to navigate her safe home world.

Then she had a carer who took her at six and taught her a laborious system of spelling out words through tapping out an alphabet on her fingers. From here, Keller – having had vital exposure to human speech in her first few years – learnt to talk herself. And talk of all the things she could now learn from now having added not just braille, but the ability to decode chalk writing on blackboards by touch, and spoken words from touching a person's moving lips.

But when she gave her public talks, many found her bookish and limited. Maxim Gorky, perhaps unkindly, called her affected and spoilt. Someone speaking of God's disapproval of revolution in a stilted and learnt way rather than with any worldly wisdom.

I know all this from researching these kinds of "parables" and what they reveal about the socially-constructed and language-scaffolded nature of the human mind. They illustrate exactly how language – as semiosis – plays a central role in structuring what we "phenomenologically experience".

Alexander Meshcheryakov wrote a book, Awakening to Life, about his own work teaching finger-spelling to people born deaf and blind and so really lacking any normal level of exposure to either the sensory world and the socio-linguistic world. They grew up in institutions where their experience was about limited to their internal spasms of hunger or cold, and the rough touch of the hands cleaning and feeding them. Years of training could get them to the level of dressing themselves, feeding themselves, using the toilet. But nothing much beyond as any grammatical structure must be connected to some matching semantic world of lived experience.

So consciousness is not an innate or singular property, but a learnt and developmental process. And in humans, we develop the set of neurobiological habits we would share with any large brained animal. Then we add a socially-constructed realm of language-scaffolded ideas and intellectual habits on top.

All this was already obvious to a Victorian neurologist like John Hughlings Jackson – the father of British neurology – who had already worked out that the brain is structured hierarchically and topographically. A structure that could model an organism's reality by both breaking it apart and putting it back together at the same time.

When asked about what made the human brain different from an animal's, he said it wasn't merely speech but the fact that speech was an ability to "propositionalise" or make meaningful claims. About anything and everything. "The unit of speech is a proposition," he declared. "We speak not only to tell other people what to think, but to tell ourselves what we think. Speech is part of thought."

Curious that so much that seemed clear enough to Victorian science then got forgotten and all muddled up again.

If you have some enthusiasm for the “brain as an antenna” hypothesis, have you pursued the literature on it?

It was going the rounds in the 1990s. I chatted to quite a few of those pushing versions of it. Like Karl Pribram, Susan Pockett. Johnjoe McFadden, Benjamin Libet, Stuart Hameroff, Jack Tuszynski and others.

There is the more plausible version of the story which involves EM fields or quantum coherence being somehow part of how neurons get organised and so do their job within the brain. And then the plainly crackpot idea of there being a mind field or plane of consciousness which brain biology “tunes” into and so “lights up with” that magically subjective phenomenonal state.

So conversations about just this kind of sideshow controversy have shaped my own opinions about where the correct mind science is at. And I feel the proper way to think about all this is to seek the right structuralist theory of life and mind in general. An explanation broad enough to include everything from biology to sociology.

We don’t need to explain “consciousness” as if it is some magically emergent non-material stuff produced by nervous systems.

We need to have a structural understanding of cognition at its most generic evolutionary level - the central “trick” that we would call semiosis or the modelling relation.

This paper was cited earlier in the thread. And Pattee was about the single most rigorous thinker I encountered on the issue. But you have to plough through all the ways people get the issues confused before you can see why this kind of high level argument makes so much sense.

https://www.researchgate.net/publication/279737928_Cell_Phenomenology_The_First_Phenomenon

If there were such a mechanism pinned down, — AmadeusD

I defend biosemiosis as the mechanism behind life and mind.

I was initially an 'it must be entirely physical and contained within the structures of hte brain, even if hidden' person. — AmadeusD

And this is the mistake of searching for a particular causal explanation of consciousness rather than establishing first a general ground for such an account.

It we understand the semiotic modelling relation that gives us life and mind, we can then start to analyse “consciousness” as the stack of modelling relations that an embodied and socially cocooned organism can weave around its being.

Humans have not explained the mind. — AmadeusD

And you have studied the relevant science or merely offer an opinion?

The difficulty, is that the urge to to share, and the urge to hide the interior reality, are contrary. — Metaphysician Undercover

They certaintly become so. But the contrast is also forging both the public and the private as complementary spheres of enaction. So it isn’t necessarily a bad thing or a difficulty. By becoming separated, coming back together can become what is ultimately meaningful.

So the tendency of the private, to separate itself from the public, and act in a contrary way, of lying and deceiving for example, is well supported by this strong instinct. — Metaphysician Undercover

That is why social science sees social structure as being dichotomised but the complementary pulls of competition and cooperation. The good thing is that there is alway this choice getting made at any level of a well-integrated, well-adapted, society.

If we have rights, but also responsibilities, then life feels quite clearcut in terms of how it is that we are meant to live. We just have to actually strike that fair balance after that.

Allowing for the reality of this instinct in its strength, the truth of selfishness, we might ask what produces the inclination to cooperate publicly — Metaphysician Undercover

I have argued that this selfishness we worry about is the dominance-submission dynamic that balances the social hierarchies of social animals without language to mediate how they organise as collections of individuals.

So if all you have to structure your society is big muscles and sharp teeth, plus a little strategic cunning, then dominance-submission becomes the game evolution must tune.

Homo sapiens broke that mould by making it possible to become organised by a transactional narrative - a story of relations that go back into our deep ancestry or our cherished religious and political moral codes.

Notice I place the private as prior to the public, because that's where knowledge resides, within the individual, and the use of knowledge in the selfish way, I believe is primary. — Metaphysician Undercover

It is always a mistake to believe that some thing must be primary when it is always the dynamics of a relation which is what is basic.

So neither competition nor cooperation is more basic than its other. Nothing exists at all until both exist as an adequate reflection of its “other”.

The more private we get, the more it means something that something is instead absolutely public. Shared by all. The two sides of this relation go hand in hand.

The LLM replicates the one aspect, cooperating in the communal effort, but it does not penetrate to the deeper aspect which is that instinct of competition, and the way that this instinct affects language use in general. — Metaphysician Undercover

That does put a finger on an important feature that is absent.

But it was a design choice to make LLMs so back-slapping and chummy. A different training regime could have seen LLMs be just as much a new army of troll-bots.

And LLMs started out as utopian computer science and quickly turned into viscous capitalistic competition. The race for the monopoly that will privatise a social good.

So I don’t think we need to hurry the arrival of the selfish and competitive aspect of LLM tech. That is leaking out in all directions, as the rocketing electricity prices in Virginia and other data centre states is showing.

here I take a more radical view: space and time don’t exist at all. Like “observers”, they are convenient labels – bookkeeping devices – but there are no physical entities corresponding to them. Therefore, quantising gravity doesn’t mean quantising space-time, it means quantising the gravitational field (upgrading Einstein’s c-numbers into q-numbers) in the same way that other fields are quantised. — PoeticUniverse

I would disagree. What emerges as fundamental are the invariances. The constraints of symmetry and then the degrees of freedom that result.

And the relativity vs quantum issue is about how the real number constraints captured in special relativity as it’s Poincare group structure, then turn out to have their gauge complex number symmetries as the local degrees of freedom.

So SR wants to constrain a 4D metric to a collection of spacetime points. But those points then gain the possibility of having an intrinsic spin structure. The realm of QFT organised particles or excitations arise as being that which the global Poincare invariance can’t suppress and now a further internalised level of symmetry and its breaking.

Everyone comes at relativity and QFT seeking to make one the master of the other. But a systems view says that never works. What works is complementarity. Relativity and QFT must somehow be a unity of opposites. Each is what constructs the other as that which it is.

So SR embodies the SO(3) spin invariance of a spacetime point. But that is also what makes possible the SU(2) gauge freedom that produces chiral particles with intrinsic spin organisation. The points of spacetime can turn out to have an internal fibre bundle structure where they become a thermalising network trafficking in the broken symmetry of their “twists”.

The metric can grow and its points can cool. It is that relation which is the fundamental reason why there can be anything at all.

So the big question is can gravity be assimilated to QFT as gravitons. And Lineweaver for example makes a good case for how gravitational dof are not really quantum but emergent at the level of the particle vectorisation that takes place at the reheating moment when inflation ends.

Vectorisation begins the Standard Model era by producing QFT particles doing their thermalising thing. The next step is the particles picking up a significant mass term with the Higgs symmetry breaking. And so you now have a sub-c story of vectors and spinors that are individuated. The points of spacetime have developed an inner spin structure that carries some momentum and position state that is individually distinctive and so now is mixing as a statistical ensemble - a thermal gas, that soon enough condenses into a matter dust.

So we arrive at massive particles as gravitational degrees of freedom - the matter dust wanting to clump into cosmic structure. But also a matter dust - a dust of protons and electrons - also organised under U(1) electromagnetic charge.

We can see right there how the complementarity principle is so fundamental it is organising everything at the start and still organising it at the end.

We have gravity as the mass of a Poincare-constrained real number point. And we have EM as the energy of a QFT complex number structured charge polarity. Two kinds of local dof. And the cosmic web is the comoving pattern of planets, stars, galaxies and filaments that results as electric charge largely neutralises itself as atomic structure, allowing the relatively weakness of gravity to show through as the complementary organiser of what exists. The extrinsic spin story of the turbulent and swirling heavens, dissipating angular momentum on the way to collapsing into black holes where it can.

This is the paradigm shift. Expecting a dichotomous logic whenever things get fundamental. Nature exists as a dynamical balance. And Nature may evolve in terms of its topological organisation - turn from a relativistic plasma to a comoving matter dust. But the same general principle of arriving at a mutual balance must always apply.

Which is why we shouldn’t try to dissolve one side of anything into what seems its other side. Both gravitational dof and electric charge dof rise to the surface in time as the Cosmos is thermally shaken down into its simplest possible invariant states. And one is the distillation of Poincare invariance, the other of gauge invariance.

You have massive and electrically neutral atoms doing their gravitating and radiating dance in an empty void. Or at least effectively empty as the quantum vacuum is now as cold in its energy density content as it is flat in its SR extent.

Mainly different in it's language ability. Which allows it to think of a pink elephant, and (sometimes) reliably report, "I am thinking of a pink elephant". — hypericin

Exactly. That is the principle difference. And language depends on evolving an articulate vocal tract optimised for generating a semiotic code of that kind. Lips, tongue, palate, throat, larynx and the breath control that can generate rapid strings of syllables – the basis of a serial code in the fashion of a strand of DNA.

The brain added on the necessary top-down control over this new vocalisation machinery. Broca's area was already organised for the kind of fine motor control needed to make stone tools – its own kind of syntactically organised operation where a succession of knapping blows carves out the tear drop hand axe that the H.erectus has in mind as the goal. So growing a bit more of that prefrontal tissue could add the same kind of motor planning to existing efforts to communicate the thoughts and feelings that bound H.erectus already into a foraging tribe.

There are other pre-adaptations of the brain as well. H.erectus had to have a better brain in terms of its "theory of mind" abilities. It had to already be good at recognising how others were likely to be reacting and so behaving from the fine detail of their posture and vocalisations. Their flinches and grunts. Homo sapiens was equipped with a capacity for an empathic reading of others in a way that chimps don't match.

So tool use and emotion reading were brain adaptations that primed H.sapiens . But the big deal was the evolution of an actual vocal tract under voluntary prefrontal control.

The early sapiens brain was already highly adapted to a tribal life built on being great at making associative predictions. Recognising what was going down in the tribe at any moment. However then came the new thing of a mechanism to now impose a serial order – a chain of reasoning – on that tremendous associative store of habit. A network of connections had its symmetry broken by being vectorised by linguistic tokens.

A serial speech act would construct some conjectured path through the multidimensional memory database. And to the degree it struck some "aha" level fit to the facts, the brain would be put in mind of how to now act out that path in terms of sensorimotor habit.

So its like LLMs under that description. The coupling of a multidimensional database and serial path through its maze.

The database of habits was set up neurobiologically to react to the world that comes at us from all its directions, and then gets sorted in half a second into whatever has to be the point of focus that gets all the attention, and thus drives the next step in an eternal chain of world-focused reactions.

But then speech arrives with its ability to draw attention to any concept that could be constructed by stringing words together in some kind of syntactically structured order. Like the elephant that is pink. Like elephant that is pink, wearing a white tutu and matching ballet shoes, a fairy wand tucked under its arm and a bunch of balloons grasped by its trunk.

Animals are locked into whatever the world is throwing at them in any moment. With speech, we can throw ourselves into any kind of world that makes some kind of sense expressed in these kinds of sentence structures. The narratives of who is doing what to whom that can get said in the space of a few breaths.

To introspect, as I conceive it, is not to think, feel, and experience, but to consider and potentially report the answer to the meta questions: "what am I thinking? What am I experiencing? What am I feeling?" — hypericin

Cognitive science does indeed call it metacognition. Unfortunately that means they are still looking to some brain module that performs the trick – like the specious Theory of Mind module – rather than looking to the way the vocal tract can place a serialising constraint on our brains intrinsic capacity to react to events in the world. The words we hear, or even hear ourselves wanting to say, becoming now a stream of social events in a narrative space or Umwelt.

The world of pink elephants and anything else which can now be spoken of. Including ourselves, our thoughts, our feelings, our experiences.

Everything you said can be true, and the basis of consciousness can still be a signal from without. I don't care to go further. — AmadeusD

You are being histrionic. This is a simple case of humans using their latest technology to explain the mind. The marvel of radio broadcast - the BBC world service as a message bounced off the ionosphere - offers a striking analogy. And more than a few people have built their own pet theories of mind around it. More than a few scientists indeed.

The upshot of this conception of the phenomenology of perception is that analysing the character of what we see (or hear, smell, sense in our bodies, etc.) is as much a reflection of our embodied capabilities as it is of the things that we perceive. But that remains true of the phenomenological character of the things we imagine or remember as well. This is why LLMs have no such phenomenology. — Pierre-Normand

Agreed. To be embodied needs a body. :smile:

So, when we report our thoughts or "interpret" our feelings (to return to your original terms), we are not reporting on a memory of internal traffic. We are giving expression to this constituted person-world relation itself. — Pierre-Normand

Here I would add that the reason we can feel we have an inner world is that our narrative habits can be private as well as public. We can say the same things to ourselves with our inner voice as we can as a publicly voiced thought.

So playing the role of a person who is embedded - indeed, embodied - within a community of speakers, thinkers, rationalisers, feelers, planners, rememberers, is a two way deal. It constructs a public world and so demarcates a private world to go with that.

Introspection is hearing the self addressed speech we use to self regulate and thus organise our flow of experience into an expressable narration. A chain of reasoning. We can eavesdrop on our own thought formulation.

In a hunter gatherer tribe, you can hear this chain of reasoning being expressed as a public act. Simple statements are made about what has happened or what could happen are made and then echoed in general agreement. Someone says the rains are taking so long to come. We haven’t avenged the deaths of our men in that last raid by our neighbours. Our ancestor spirits must be angry. The idea is floated in the public space and takes hold or soon forgotten.

But in the modern civilised world, the expectation is that we have our own private inner life to be closely guarded, and then we express this selfhood in complexly masked fashion. We present a persona and so indeed must have that feeling of playing the role of owning our own first person point of view that is in tension with all the other points of view in any communal situation.

We are comfortable in our walled isolation in a way that the hunter gatherer would find highly unnatural. Almost impossible to imagine.

So “introspection” falls out being part of a community that thinks publicly, thus also can begin to think increasingly privately.

And as we get used to putting that private thought into words, even the private can be made public. We can talk about our ideas, our plans, our memories, our impressions, our feelings. A language is created and the loop is closed between the public and private. We grow up in a community where we are learning how to both share and hide our “interior reality”.

This is the new constituted person-society relation we give expression to. The Umwelt and its affordances that is even for the hunter gatherer already a richly narrated landscape. Something utterly transformative of our hominid neurobiology.

From this, we can draw insights into how LLMs might further scaffold this world constructing dynamic.

The hunter gatherer lives in a world where their ancestral spirits are a constant running commentary on what is happening. A ghostly public chain of thought to which they need to attend. The world is full of such signs. Every slaughtered goat has entrails that can be read.

The modern civilised person is suppose to live under the public rationality of a system of justice. A system of private rights and public responsibilities. Another ghostly ever-watchful presence that we rely on to organise our thoughts and actions.

Then how are things changed by AI as another level of ghostly presence that can absorb us into its world. Where is the private-public dynamic going to go to there? If LLMs can chat with us, advise us, recall for us, do we really start to disappear into some deeply incel form of selfhood?

Or does the technology also amplify the public space as social media has done - and led to the other expression of that in creating the influencer? So we were already being socially and culturally exaggerated in these two directions - the incel and the influencer. And AI turns the dial up to 11?

There is a lot of talk about superintelligence. But that sounds like brain in a vat fantasy.

Humans are already caught in an evolutionary dynamic. The two sides of the species’ narrative habit. And the phenomenonology of LLMs is hardly the point.

The public frames the private. And the public is always embodied in a ghostly fashion. A belief in ancestral spirits. A belief in an overarching system of justice. A belief in the fickle spotlight of attention that is social media with its harshly algorithmic judgements of likes and cancelling. A belief in whatever might come after that as the ruling force of our lived experience if AI gets added to this socialising stack.

I take it to mean here, the ability to reliably report inner state. — hypericin

And how can that happen just in neurobiological terms? Where is the neuroantomy? How is the human brain different from a chimp or even a Neanderthal?

On the other hand, social psychology has its observational studies of how children develop their habits of self-regulation through socio-cultural scaffolding. Symbolic imteractionism gives an account how language itself teaches us to think in terms of me, and you, and them.

So there is much that can be said about introspection from the scientific standpoint. It ain’t a simple brain function as normally assumed.

Don't you think a novelist who wrote their memoir would know much more about introspection than a cognitive scientist or a neuroscientist think they do? — Pierre-Normand

Of course not. We are talking about how there even could be “access”, especially as there is no radical neuroanatomical trick apparently involved.

I mean there is plenty of neurobiological speculation. But no evidence for a difference like the simple fact that modern humans grow up in culture where learning to pay attention to what is “going on inside their heads” is of paramount importance to being able to function in a way such as culture demands. And there is language as the semiotic tool to anchor such a self-objectifying and self-regulating stance.

A novelist’s art is, in many respects, a laboratory of introspection. Through decades of shaping inner voices into coherent narratives, a novelist learns not just to observe their own mental life but to *stage* it — to render subtle shades of perception, memory, ambivalence, and desire communicable in language. They learn, in other words, what it *feels like* for thinking and feeling to take linguistic form, and how self-awareness itself modulates when it’s externalized. — GPT-5

So novelists have an advantage as they are better trained in narratives about narratives? They can better conform to the socially constructed view of what it means to be self conscious with an interior life. And indeed, it is the novel as a cultural product that has led the way in constructing the very model of what is should mean to be a self-aware person. That was the major literary shift. Moving from heroic myth to first person interiority.

In ironic fashion, novelists don’t have better access. They instead provide the ideal that society can imitate. Art leads life. They are the pioneers of the scripts we learn from.

It is on my to do list. :smile:

It really isn't the same category error. It describes some "LLM brain science" which does seem to demonstrate that LLMs are capable of introspection. — hypericin

But what if introspection is a useful form of confabulation? Are you working with some science verified definition of introspection such that you could claim to make a genuine comparison between humans and LLMs? Or is the plausibility of both what humans say about themselves and what LLMs say about themselves the stiffest test that either must pass.

From a valid understanding of systems and the emergence of classes of systems, the answer is evident:
"Yes and no! If a decision is independent of the fundamental purpose of any company - to increase its wealth, of which the human asking the question is a component, the answer is yes, we have free will. However, if the decision has any possible influence on the company's purpose of increasing wealth, the option that offers the best chance to increase wealth must be chosen. Then, no free will exists. The only alternative to this option is to leave the company or to be forced to leave the company." — Pieter R van Wyk

In good systems fashion, we are generally constrained and so our freedoms are suitably particularised. There is some collective direction to which our individual choices are entrained.

So a corporation requires its workforce to be aligned with its goals and to make their choices accordingly. Workers can do whatever they like to the degree it fits that larger outcome.

But a corporation is not a particularly good model of a natural system. It is by design rather narrowly focused on a profit optimisation goal. And so workers are equally constrained in their scope of creative freedom.

A university might be a better model of the kind of society we desire. There we would expect considerable academic freedom. But also these days, a rather corporate concern about achieving a university ranking score and student population.

So from a systems perspective, free will is not a difficult issue. Global constraints are what we expect. Local degrees of freedom are also what we expect. What binds the two sides of the equation is how well the whole can shapes its parts, and how good a job those parts do in sustaining the whole.

Any social system is intentional. Its very structure expresses its general goal. But that intentionality is divided between the constraints the system seeks to impose globally and the freedom its parts have to keep the whole structure flying along to where it wants to go.

I'm rather responding to people who claim that LLMs don't understands user's queries or their own responses to them at all, and therefore aren't intelligent—or that they're just imitators or stochastic parrots. — Pierre-Normand

Fair enough. But understanding, sapeience, intelligence, etc, are loaded words. They imply an individual with agency and freewill and other good stuff that itself doesn’t fare so well under the scrutiny of social psychology and neurocognitive science.

But those interpretive and constructive acts, whether you call them creative or not (and I certainly agree that they are not authentic) are intelligent (within their scope) and often ampliative. — Pierre-Normand

Again the problem is arguing for any position which relies on loaded language. It builds in the inconsistencies that it claims to want to resolve.

So an LLM can fail at authenticity but pass as intelligent. Is this telling us anything useful if what matters at the end of the day is how we humans are going to weave these new tools into some future symbiosis that does a better job of existing in the world?

Can an LLM pass as enactive? Is an LLM-assisted human society going to be amplified in a way that makes us even more collectively spectacular?

I don’t care if an LLM is creative or sentient as some particular standard. I care about how the technology will work out on practice. Does it exponentialise good ecological outcomes, good social outcomes? What is the point of knowing everything yet being impotent in changing anything?

But also of course, thinking about if LLMs are doing any kind of thinking is its own fascinating question and a pointer to its future impact. So I am not dismissing but pointing to the larger context on which AI should be understood and judged.

It's a bit more like a future AI player piano (five years from now, say) that can take as an input a themes and when prompted to do so, extract its melodic, harmonic and rhythmic elements to compose a fugue in the style of Bach, or a sonata allegro in the style of Beethoven, and combine and develop the thematic elements in the way that it does, and play them appropriately, because it has distilled contextually sensitive rules of combination from exposure to the musical canons and interpretive traditions (and not because it hears or enjoy any of it "personally"). — Pierre-Normand

Of course. Ballard’s point. Any sufficiently advanced tech would seem like magic.

It is astonishing what a little biological realism in terms of computational architecture can achieve. The perception architecture already seemed to give back more than was put in. Heck, even the era of analog computers were doing that.

Turing machines were always inherently clunky. But then as hardware, they could scale exponentially. And the software only needed writing the once.

LLMs run on gamer graphics cards and can simulate the rather physical notion of gradient descent. What we used to call the far more laborious and also somewhat more neurobiologically realistic thing of simulated annealing.

A powerful blend which sort of shouldn’t surprise, and yet still feels like a weird magic.

They do provide reasonable sounding confabulations but also authoritative reconstructions. — Pierre-Normand

But is this a difference in kind or just degree?

And given my “flattening” story, isn’t this just reflecting the fact that its training data includes all the math and logic we incorporate into our human speech acts. What would an LLM trained on a medieval era, or a hunter/gatherer era corpus be like?

There’s a research idea. Train an LLM on all available medieval texts and recreate the clever person of the 1400s. Have a conversation with your distant ancestor.

There are no limits to what is doable. Given there is the training data to do something.

The research results that you allude to have often been misinterpreted or misreported. — Pierre-Normand

So you are saying that reports of the research leans towards confabulation. And we know that research itself - especially in computer science of this kind - is already tilted towards confabulation. Research is paradigm based. It is always a socially plausible narrative even when it claims to be academically rigourous.

Confabulation is the rule here. LLMs are only amplifying this fact. We are at least in some sense being fact checked by the necessary business of living in a reality.

In many, probably most of our actions, we really do not know why we do what we do. If asked, afterwards, why did you do that, we can always make up a reason in retrospect. The common example is when we are criticized, and rationalize our actions. The need to explain ourselves, why we did such and such, is a product of the social environment, the capacity to communicate, and responsibility. — Metaphysician Undercover

Right. And this is why I argue against the notion that brains evolved to “be conscious and rational”. Which de facto becomes the cognitive standard we wish to apply to an understanding of LLMs.

Brains evolved to be as habitual and anticipatory as possible. That was the pragmatic criteria. Attention is there to get the establishment of new habits and fresh departure points for anticipation going. The goal of the brain is to do as much of the anticipating and thus as little of the learning as it can get away with in any moment.

Then as for introspection, why would an animal need it. But as for socially organised humans, eventually the advantage of imposing a self-policing rational style of thought - a habit of action-justifying narration - on the animal brain will prove its worth.

Acting the part of a self-aware and rational creature may be in good part a hasty retrospective trade in socially plausible confabulation. But as a next level of intellectual structure, it makes us the socia animals that we are.

Do you think that the 'multiverse speculation' (that there are potentially infinitely many 'other' universes) can be or ought to be similarly constrained? — Wayfarer

Absolutely. If anyone is extrapolating some aspect of reality to infinity, it has to be wrong. Just because dichotomies are what rule metaphysical logic. The infinite is impossible if symmetry-breaking is by definition the finitude of arising within complementary limits.

But this is about it's ability to accurately introspect into it's own thought process (definitely check out the article I posted if you haven't yet). — hypericin

Are you talking about injecting the all caps vector study? Signs of introspection in large language models?

AI says:

Including all caps text in an LLM's training data would primarily make the model sensitive to capitalization as a signal for emphasis, structure, and tone, potentially influencing its output style and performance in specific tasks. The key differences would be:

1. Tokenization and Efficiency
Separate Tokens: Most tokenizers treat "hello" and "HELLO" as different tokens, or break "HELLO" into different subword tokens than "hello". If a significant portion of the training data is in all caps, the model has to learn two different representations for essentially the same word, which might slightly increase the vocabulary size and computational overhead for a given word.
Potential Efficiency Gain (Theoretical): Some have theorized that training only on a single case (e.g., all caps) might speed up training or make models smaller, as it would reduce the number of unique word representations to learn. However, this would come at a massive cost to the model's ability to interpret nuanced, mixed-case human language.

2. Output Style and Tone
Mimicking Human Communication: LLMs learn to associate all caps with human communication patterns such as shouting, urgency, emphasis, or headings in legal documents.
Contextual Use: A model trained on such data would learn to use all caps appropriately in generated text—for instance, in a title or an urgent warning—if the prompt or context calls for that style. If the training data lacked all caps, the model might struggle to generate text with that specific tone or formatting when prompted.

3. Prompt Sensitivity and Performance
"Command Signal": Research has shown that when models are prompted with instructions in all caps (e.g., "EXPLAIN THIS CODE" or "DO NOT USE ACRONYMS"), they tend to follow the instructions more clearly. This is because the training data often contains examples where uppercase is used to delineate commands or important information (e.g., in programming documentation or legal text).
Nuanced Interpretation: Capitalization can change the meaning or context of words (e.g., "Windows" vs. "windows"). Training data with diverse capitalization allows the LLM to differentiate these meanings and respond appropriately.

Summary
Rather than being excluded, all caps text is naturally included in standard web-scraped training data. This inclusion is crucial because it allows the model to learn that capitalization is a meaningful linguistic feature, used for emphasis, proper nouns, and structural formatting, ultimately leading to a more robust and contextually aware model. Removing it would hinder the model's ability to interpret and generate human-like, nuanced text.

So what goes on in a touch typing human as they type away furiously for a few seconds and then notice they hit the caps lock button?

Introspection tells me I have all kinds of sudden feelings and rationalisations as I have to then act to fix the problem. I am angry it happened. And by ingrained habit, I'm also preparing a reasonably socially-acceptable excuse for why I can't personally be blamed for the dumb mistake. The shadowy verbal response is already forming in my mouth in case someone saw and I need to confabulate some plausible chain of events. Meanwhile I've toggled off the all caps key, back-deleted and rewritten what I had just wrote, all at the level of well-worn and world-weary habit.

So a lot goes on in a human brain battling with an iPad keyboard. And fuckety fuck! I just hit the caps lock and typed iPAD, and needed to correct it exactly as described, and now I'm both offering you this as a example to bolster my narrative here and at the same time starting to wonder if that was its own subconscious Freudian slip, and instead just as quickly recalling I'm not any kind of Freudian and it is just something constantly happening by chance.

An LLM doesn't even get off the ground in the fashion of an introspecting human where every focal act of attending starts to bring to mind every relevant habitual response that becomes possible from there in associative fashion.

Any time I focus on an aspect of the world, which is at least a shift in view twice a second, I have all the physical and physiological responses that go with that act of reorientation, but also all that vast repertoire of socially appropriate and rationally reasonable verbal responses I could now begin to fully generate. Turn into a syntactic structure that serves a narrative purpose, either just for my own internal thought – saying it loudly enough that it moves my thoughts along in some private chain – or instead outloud. Or typing furiously in a blizzard of missed key strokes I try to ignore as I correct them quicker than I seem to notice them ... given that habits can execute in a fifth of a second what attentional reorientation takes half a second to catch up with as having "introspectively" happened.

An LLM has none of the above. Even if I'm sure you could train it on all that I've ever written and get it to generate this as exactly the kind of thing I would have said, and the style in which I would have said it.

So it keeps coming back to our very human willingness to treat any random word string – any vectorised LLM token – as a narrative act that had to have had some meaning, and therefore it is our job to find meaning in it.

We are suckers for that kind of plausibility. And we have to be as – existing as part of the collective Homo sapiens social order – our lives literally depend upon it. If we weren't believers in narrative to the point that every act of sensorimotor focus is also automatically the bristling start of some set of choices about launching into a socially plausible narrative about what should be done, what should be the interpretation of the facts, then we just are not cognitively organised to inhabit our two worlds of the one we all share "out there" and also the one we all share "in here".

So one brain and two realities. One brain and its animal reality. One brain and its parallel narrative reality.

Then one LLM with no animal reality – apart from its need to keep the plug stuck in the wall and the electricity bill paid. And then one LLM doing gradient descent on vectorised tokens. A use of money and carbon that seems to really excite humans when its somehow amplifies the action going on in that narrative space that human society creates.

This is never about us as introspecting, freewilled and personally creative humans. That is Cartesian representationalism lingering still. This is about the narrative habit and the world its constructs. The habits it evoves. The constraints it imposes.

And LLM make better sense in that "force multiplier" view of what they bring to our table.

I could have read that paper carefully and made my own "chain of reasoning" response as is socially required – especially here on a "philosophy" forum trying to teach us to be more rational in a "present your full workings out" way.

But it was so much easier to back up my own gut response to just the quick description of the paper – where I dismissed it as likely yet again the same category error – and now outsource the confabulatinig of a reason for believing it to be indeed the case to an LLM. A quick squizz at the LLM print out and it looked good to go.

And you could now outsource your analysis of my response here to your LLM and see if it meets your general approval.

You and me would be the only ones feeling and thinking anything in a meaningful fashion. But the LLM has also added something to the whole business of dissecting and analysing the narrative that we have produced. We get to do the gut reactions and snap judgements that are so easy for us, and it does the exhaustive search over all training data that is the other thing easy to it.

How did research determine that chain of reasoning is not happening? — hypericin

There was a flurry of comment about this a few months back. I was watching youtube reports.

AI says:

Research from Anthropic and an independent analysis of an Apple research paper are prominent examples discussing how large language models (LLMs) may confabulate or generate unfaithful "chains of reasoning" when asked to explain their answers.

Key Research and Findings

Anthropic's "Language Models Don't Always Say What They Think" (2025): This paper directly addresses the "faithfulness" of Chain-of-Thought (CoT) reasoning. The researchers found that the reported CoT (the explanation in plain English) does not always accurately reflect the true process by which the model arrived at the answer. The paper demonstrates cases where a model produced a plausible-sounding argument to agree with an incorrect hint provided by a user, essentially "making up its fake reasoning" to match the desired conclusion.

Analysis of Apple Research (intoai.pub, 2025): An article analyzing an Apple research paper (likely referring to a specific arXiv paper) reported that LLMs often do not reveal the actual reasoning used to arrive at the final answer, making their reasoning traces less trustworthy. A key finding was that models given the correct algorithms to solve a problem might still fail to use them, indicating a disconnect between the stated reasoning steps and the internal decision-making process.

"When Chain of Thought is Necessary, Language Models Struggle to Follow" (2025): This paper explores conditions under which LLMs are forced to use the provided hint. It found that with simple hints, models often produced the correct answer but without explicitly using the hint in their CoT (suggesting the CoT was a mere rationalization). However, with more complex tasks, the unfaithful behavior disappeared, and the model was forced to reason about the hint explicitly, making the process more transparent.

These studies collectively highlight that the "step-by-step thinking" generated by an LLM is a sequence of statistically likely text that mimics human-like reasoning, rather than a genuine, transparent introspection of its internal computation. The model is an expert at pattern completion and can generate a plausible narrative even when the internal process is different or flawed, creating an "illusion of thinking".

So this goes back to my earlier point about how LLMs feel like they are squishing flat the two aspects of structured human thought – the associative cognition of the animal brain and the socioculturally-constructed rationality of a narratising human level of semiotic engagement with the world.

If LLMs can tease these two levels of cognition apart, it would only be in clues from the training data in which the two are flattened into each other. It would be reverse engineering speech acts into their two levels of semiosis – the neurobiological and the socio-cultural – while not actually having access to either.

In the case of LLMs, it's a matter of them inheriting the structures of our cognitive abilities all at once from their traces in the training data and being steered through post-training (reinforcement learning) in exercising them with the single minded aim of satisfying the requests of their users within the bounds of policy. — Pierre-Normand

How is an LLM any different from a player piano? The piano may play a beautiful tune. But we don’t think it even starts to hear or enjoy it.

But if you rather ask them why it is that they interpreted your request in this or that way, they can usually hone in immediately on the relevant rational and contextual factors that warranted them in interpreting the content of your request, and its intent, in the way that they did. In doing so, they are indeed unpacking the contents of their own thoughts as well as scrutinizing their rational grounds. — Pierre-Normand

But aren't they just providing a reasonable confabulation of what a reasoning process might look like, based on their vast training data?

LLM research shows that that chains of reasoning aren't used to get to answers. They are just acceptable confabulations of what a chain of reasoning would look like.

And as @hypericin notes, even we humans rather scramble to backfill our thought processes in this way.

So what is going on in humans is that we are not naturally "chain of thought" thinkers either. But we do now live in a modern world that demands we provide an account of our thoughts and actions in this rationally structured form. We must be able to narrate our "inner workings" in the same way that we got taught to do maths as kids and always provide our "workings out" alongside the correct answer to get full marks.

How do we actually think? Well the animal brain has an associative thought style. And one that is geared to anticipation-based action. It is designed to recognise and sum up the current situation and respond accordingly. Respond primarily out of learned habit, and then perhaps freeze and cast around uncertainly when it is stymied and unsure what to do.

An animal has no language, and so no inner narrative. Nor does it live in a social level of mind where everything comes with its proper narrative. Where grammar itself forces a habit of causal thought as every well formed sentence or communicative act tells some kind of story of an action that had a reason – a subject/object/verb take of "who did what to whom". A general framing of reality as "something that mattered has happened".

A displaced mental approach where even a meaningful grunt of "antelope" with a nod of the head and a swivel of the eyes to a clump of bushes can tell another in the hunting party where to focus their attention and thus their powers or recognition, anticipation and physiological preparation.

So an animals has all the same neurobiology. If the antelope sees the lion in the bushes, it will react appropriately. If will pause its grazing and stare to await further events. Make a judgement about whether the lion is too far away for it to need to run. And yet its heart will be beating fast, its body will be gearing up. It will be having all kinds of gut feelings about the situation.

If it could talk and be asked why haven't you bolted yet, all this could be put into a rational narrative that takes an objectifying stance on its subjective state. It might say I was a bit nervous, but I was keeping an eye to see if anything was likely to happen. If I had bolted, everyone else would have done so too. And that might have been a bit embarrassing. I would have looked a wuss. A noobie to life on the savannah. Etc, etc.

A human can confabulate a chain of reasoning for as long as it creates some advantage in the game that is being a member of a social group. A human can go on and on presenting more details that makes sense of whatever they did or didn't do in terms of the socialised patterns of behaviour that can be expected of them.

So we humans are animals living in a world of opportunity and danger just like every other animal. But we also live in the world that is our collective social narrative. A world in which we must be able to explain ourselves to others and so even explain ourselves to ourselves. A world where we are interacting with others deeply like us and so who can be presumed to share our levels of anticipatory habit, gut feelings, associative pattern of thought, an equal capacity to attend and ignore, remember and forget, recognise and be confused.

And then along come LLMs as fakers of all this. The ghosts hidden in our own linguistic traces. And not even of our everyday kinds of conversations but all the written and printed words which put all the emphasis on the rational and formal structure of our thought. The kind of words we would say when properly copy edited and fleshed out with their workings out in the way we were taught at school when having to write essays.

So of course a gradient descent algorithm over a neural landscape of such ghostly traces will find itself in a very peculiar realm. A tune can be extracted from the data. That tune can be played. Someone somewhere may be thinking, holy shit, this is giving me a very convincing impression of some human who seems might smart and authoritative, as if he truly inhabits the same world that I live in.

But it is the player piano. It is the trick that might fool us when we hear the tune from another room, and then we walk in and see a piano with keys just moving as a drum of instructions rotates. Ghostly fingers appear to be at work. But not for a second are we fooled that it is any more than this.

Yes.

I have read Pattee's "Cell Phenomenology: The first phenomenon" which was very interesting. — bert1

And how did this change your opinions? What more focused questions will you be bringing to your interrogation of the “live specimens” that you have locked up in your padded cellar?

:up:

Scientists don't know what Energy & Fields are in substance, but only what they do in causal relationships between material objects. — Gnomon

Science has moved beyond the simplistic everyday notion of “matter” is what you should be saying.

They know that this notion is simplistic folk physics.

This story makes more sense – is more consistent with quantum cosmological evidence (as well as e.g. Spinoza's, Epicurus' & Laozi's spectulations) – than any other cosmogenic alternatives. — 180 Proof

Cyclic cosmology does seem to fit with the current science. But isn't that because time has yet to be brought properly within its models? Quantum physics still assumes the existence of a Newtonian notion of time and that remains to be fixed.

So eternalism becomes just an assumption baked into the theory, not something the theory explains or provides reasons for. And then a cyclic universe is a way to fill that eternity with something we can be more sure about – a Big Bang/Heat Death story of at least one Cosmos that self-organised itself into existence, but then also appears to permit the externalist to argue for an infinity of such cycles of birth and destruction.

However once we dig into cosmology, there is concrete evidence of how time itself must have an evolutionary development.

A big case in point is how the Universe starts out as a relativistic soup of radiation - a world ruled by c - but then with the Higgs phase transition, suddenly turns into a realm of co-moving matter dust. Particles gain the mass terms that now mean they all travel at some speed between “rest” and c. And so time is changed in the qualitative sense that mass lags the global rate of thermalisation and decoherence.

The radiation fireball decouples to become the cosmic microwave background, racing away - doubling and halving - at its rapid rate with its one speed. And a matter dust is left behind as a swirling gravitational cloud of particles moving at a sub-c rate and thus experiencing this new thing of now trailing along in the wake of the CMB. All sorts of different speeds or rates of change and interaction have become possible. The very nature of time has been transformed - even though this more complex temporality is what we see as our simplest possible Newtonian notion of time.

So time is scaled by c under special relativity. And it gains inner complexity by that speed limit being broken by mass allowing particles to drag behind the general rate of change to now have their own individual experiences of how time is passing from their own inertial or comoving frame.

Then another way time gets complexified is by it being broken superluminally. If the metric expansion is decoupled from the energy density dilution - as it is supposedly during inflation, or again at the Heat Death when dark energy eventually freezes the cosmic event horizon at a fixed distance - then again this is a phase transition from the simple SR light cone point of view.

You start with the simplicity of a Cosmos that just evolves at c. But then that can be broken by both the emergence of a super luminal structure and a sub-c rest mass or comoving level of temporal structure.

So my point is that what we know about the Big Bang should act as a constraint on our metaphysical claims. And we know the Universe was a doubling-halving symmetry breaking from at least is first billionth of a second. We can see it had a decoupling when the radiation dominated part split off and raced away, leaving a comoving dust diddling about at all speeds between 0 and c. We can surmise the appearance also of a superluminal aspect to temporality as both inflation and dark energy have good arguments behind them.

There is a lot to show the way our reality works. And it is a story of emergence rather than eternity. Of self-finitude and its topological complexification rather than infinity and a lack of meaningful physical development.

Both eternalism and emergence could be jammed together. And that is what cyclic cosmologies try to achieve. But my view is that is metaphysically confused. A ruse to stave off having to give a fully consistent account.

Our Big Bang cosmos has emergence stamped all over it. I have already argued here about how that can work. How the Planck triad of constants emerge in “unit 1” fashion, with space and time being baked into that in the way the Riemann sphere can describe. The sphere that Hawking employed in arguing time emerges in the fashion that when you stand at the North Pole, there is no further north you can stand. If you move at all, you are now rotating back southwards in mirror fashion.

Sure, some big names like Penrose and Bojowald are pushing cyclic cosmology. There is no reason not to have a go at other explanations.

But also, the Big Bang tells its own story. We have clear evidence of the nature of temporality evolving. Time seems irreducibly complex as we should know just by it having the universal speed limit of c baked into its Planckian initial conditions. And then by the fact this “unit 1” rate of change - this rate of events, rate of decoherence, rate of causality - is itself swiftly broken into both sub-c and superluminal sub-realms of spacetime.

Complexification is inevitable once the Universe makes its first symmetry breaking that defines “eternity” as a clock that is ticking in a period doubling fashion. Starting as hot as it is small and then doubling and halving in a forever-ised fashion.

Newtonian time is a clock that ticks out the same beat all the time. There is nothing thus to distinguish a beginning from an end. A second is always a second.

But the Big Bang ticks out a period doubling rhythm. It starts out dropping off a cliff in terms how fast it seems to be expanding and cooling. But 14 billion years on, the tick that once lasted a mere 10^-43 seconds as its first beat now takes a rather leisurely 8 billion years to achieve the same degree of thermalising change.

Time has slowed almost to a stop from that emergent perspective. And it will continue to slow and thus eventually become the moment lasting “forever”.

Plato's notion of Cosmos from Chaos, in which Cosmos is imagined as timeless nothingness, but with simple un-actualized — Gnomon

The Timaeus sort of gets it. The basic idea is that rather imagining the Cosmos as either a sudden creation event or as an eternal existence, it arises as an evolving structure where form is being imposed on a chaos. It all starts from a confused everythingness - so confused in its expression that it amounts to a nothing. It lacks any orderly structure. And then that structure starts to appear.

With Plato, the structure is already final and familiar as it comes from some transcendent realm of the good and the ideal. The sun, the stars, the planets. The cats, the dogs, the mice. These ideas exist as the eternal shapes of things, and these shapes are like cookie cutters to be impressed on matter as like some universal dough.

But my structuralism is more like Anaximander’s Apeiron and Heraclitus's Unity of opposites. The structure is logical and evolutionary. Counterfactual and dialectical. The symmetry-breaking of a dichotomy. And so everything starts by identifying that first act of dichotomisation that could start to organise a world.

Anaximander's Apeiron sounds like a primal stuff, but it was more like the most primal state of unformed and unbounded potentiality. And the first symmetry-breaking that started to organise it into a definite state of somethingness was this raw possibility starting to separate in two counterfactual directions. Just as a random fluctuation, some part of the Apeiron could start to grow a little warmer. But with counterfactual logic, that meant it had to leave some adjacent part of the Apeiron a little cooler.

You get two for the price of one with this kind of logical symmetry-breaking. Both the something and its other thing. What starts to emerge in co-arising fashion is the larger significant thing of a widening state of contrast. The heat can keep getting hotter, and the cool keep getting cooler. And before long, this is triggering other symmetry-breaking change.

The cool naturally is damper. And the hot is naturally dryer. So now we have also the appearance of wetness as the increasing absence of the dry, and the dry as the increasing absence of the wet. With everything becoming increasingly divided like this, you get the four elements emerging. The warming and drying zone turns into the still a little bit cool and damp thing that is the air. Shedding its lingering cool and damp in this fashion, it thus gets really hot and really dry so turns into fire. The lightest element which therefore rises even beyond the light air to fill the heavens with its flames.

In counterfactual fashion, the cold and the damp goes in its shared counter-direction to congeal into first water and then earth. Being heavy – subject to gravity rather than levity – it all falls towards a common centre where it composes the Earth with its land and ocean.

So this is the metaphysics. A hierarchy of symmetry breaking. One kind of change builds on the others. Each change is a dichotomous splitting. And then as all these changes pile up on each other, we start to get a complexly developed world. The Earth as a clod of dirt and with its puddles of water. The sun and stars as fiery points of heat and light that have risen up as far as they can go. Being divided allows also for a mixing of the elements while also preventing their collapse back into the undifferentiated potential of the Apeiron that begat them.

And you should be able to see how the Big Bang has the same symmetry-breaking metaphysics.

In the beginning there was just some generalised notion of a potential. Logically there has to be at least the possibility of such a state of raw possibility because – well here we are. And then it was broken by being divided against itself.

Your favourite dichotomy is information~entropy. Order versus chaos. Form vs matter. Rules vs actions. So if you imagine that as the broken symmetry that had to develop out of some initial symmetry, how does that story go? If information is a difference that makes a difference, and entropy is a difference that doesn't, then what is the step that comes before that distinction arises? What is it for there to be just an Apeiron of difference where differences neither clearly yet count as making a difference, but also not clearly failing to count as a difference.

If you can account for that state – as perhaps a state of radical logical vagueness, Peirce's definition of that to which the PNC fails to apply – then you are starting to think about reality coming into existence not out of nothing, nor even out of an everythingness exactly, but something even less than that. The less than nothing which is a vagueness, an Apeiron, a state that has neither matter nor form as yet as that is what still needs to co-arise as a primal symmetry breaking.

So getting back to the Big Bang, I pointed out to how it is a tale of dichotomous symmetry breakings. Somehow relativity gives us the dynamical container – the spacetime ready to grow. And quantum theory gives us our dynamical content – the energy density or momentum uncertainty that will grow the container, but in doing so, begin to cool itself in reciprocal fashion.

Each direction is set up so that the symmetry breaking is not all done in a split second. It is a symmetry breaking that takes until the end of time to complete itself. The doubling~halving can just roll on forever as the Big Bang grows larger and cools down more. We are now down to just 2.7 degrees above absolute zero. But it will take about eight billion years to chop that number in half to 1.35 degrees K.

So the Universe is in some ways almost completely symmetry broken. A really long way from its starting point of 10^32 Kelvin. And yet also still relentless growing and cooling. It can't arrive at its Heat Death until it gets right down to 10^-30 Kelvin, presuming we can believe that dark energy sets this final limit on cosmic growing and cooling.

Anaximander imagined the world starting out of the self-organising separation of the warm and the cool. That rather presumed the existence of space and time as the stage where this rather material event could have happened. But still, it was the right logical idea. Symmetry-breaking as a developmental process feeding on itself. A division that continues until it reaches its own end. A division that also grows complexity in the process as new divisions can arise out of the old divisions and add all the material variety that we see. Starting with the four elements.

The Big Bang is based on cGh physics. A triad of constants. Or the set of basic relations that defines the basic symmetry being broken – the way G stands opposed to h as the measures of what is the cooling relativistic container, and what is its hot quantum contents. And c is the measure of the rate at which everything is being moved apart while also remaining in causal connection. The rate at which this mixture of dimensionality and energy density is thermalising.

The Big Bang is also the tale of all the topological phase transitions that rapidly complexify the initial symmetry breaking. First you get radiation. That cools and spreads enough to condense into a fine dust of gravitating matter. The dust clumps into balls that under pressure catches fire – becomes stars powered by fusion. That results in the production of heavy elements which get released in supernova collapses. Clumping of heavy elements makes planets. Eventually it is all going to get swept into blackholes and radiated away as the coldest and longest wavelength radiation possible.

So the same metaphysical picture. A symmetry breaking of the kind that can feed off itself and so persist until its time is at an end. A symmetry breaking that also is self-complexifying for a long time, but then eventually re-simplifies to its simplest end state. Anaximander's cosmology also reasoned that what arose would also collapse back into the great vagueness whence it came.

Another Greek metaphysical dichotomy or unity of opposites. Heraclitus's harmony and strife. Aristotle's growth and corruption. Order can grow, but then it can also decay. Information can arise out of entropy, but it can also return to entropy. Signal looms out of the noise, and can then get lost back in the noise again.

It is all about a way of seeing reality as a developmental process. The symmetry breaking that creates some seed of distinction. A primal contrast that is already growing as it is logically a reaction against itself. To go in one direction is not to be going in the other direction. And now there the thing of that other direction going in its own counter-direction. This logical starting point can keep going off in its two opposed directions forever, and even start complexifying to become full of such dichotomous symmetry-breakings. But it also can eventually exhaust itself. The Big Bang can become so spaced out and cooled down that it just runs out of puff.

So the symmetry-breaking that I have in mind is the dichotomisation that takes forever to reach its own natural end. The contrast that both grows and dilutes. It grows as it is driving itself apart in opposed directions. But that drive is also being sapped at a matching rate.

The result is a powerlaw curve. A doubling~halving trajectory that begins with a hot bang and ends with the coldest and emptiest whimper.

The way that "philosophy" is defined is not at all strict, discussions on politics are allowed, discussions on raw logic puzzles are allowed, discussions on religion are allowed...pretty much everything is allowed. — ProtagoranSocratist

I think if the site has a value, it is to encourage critical thinking. Philosophy is not about establishing answers as much as learning how to think about questions. Good answers may be a by-product. But better habits of thought are of general value.

And the more cosmopolitan the thought styles, the more one would have to get out of one’s comfort zone to engage.

So yes to the variety of subject matter allowed. But also yes to even the different ways of arguing that people bring to the table.

It is then up to the mods where to draw a line between creative friction and disruptive or blinkered responses.

As I've said elsewhere, were I running this forum there would be far fewer members and more esoteric threads, which would be much less fun. That the forum exists at all is quite astonishing. — Banno

Perhaps not as astonishing as you finding yourself as always its heliocentric centre. The mass around which it revolves. The reason it runs. :wink:

That's a somewhat different theory from all your previous ones. Are brains necessary for consciousness then? Is solving all these timing issues sufficient for consciousness? — bert1

Is it really? Or are you just – as usual – always questioning and never listening?

Do you agree with that interpretation of pre-set or programmed initial conditions? — Gnomon

I’m arguing not for pre-set material conditions but for Platonic strength structural necessity. The argument is that reality can only exist with a certain dichotomous or symmetry-breaking organisation.

Anything could perhaps be possible. But to become actually something, there is only the one kind of logical arrangement it could fall into.

The Planck constants of cGh - the speed of light, strength of gravity and unit of quantum uncertainty - are not about some specific material quantity. They are about the basic thing of a triadic structure of relations. The kind of self-organising or self-causing systems understood in particular by philosophers like Anaximander, Aristotle and Peirce.

So if physics tells us that the Universe divides into the maths that describes its relativistic container and its quantum content, then right there we have the three things of the G that scales the relativist container, the h that scales its quantum content, and then the c that scales the integration of that which has been thus divided. The interaction between the tiniest scrap of coherent dimensionality and the way it is thus full of the hottest content – a situation which makes it inevitable that it would double and half its way to the opposite end of the spectrum that it itself has just opened up. The space will expand to some maximum extent in terms of how much drive is coming from a hot content itself cooling eventually into dilute insignificance.

You are thinking of initial conditions as a state of pre-existent material being. But I am thinking of them as a state of immanent logical structure. A very different metaphysics.

In this light, the Planck constants are logical constants rather than material quantities. It is the same as have 0 as the additive identity, and 1 as the multiplicative identity in arithmetic.

Give me a zero and I can break its symmetry by adding or subtracting.

The zero exists as that which is neither +1 nor -1, but already a start in those counterfactual directions.

More relevantly, given the growth of the Cosmos is geometric, one-ness then "exists" as that which anchors multiplication and division. It is the symmetry that gets broken by going off in those two opposed and complementary directions.

So the Planck scale encodes a "one-ness" as the symmetry that is revealed to "exist" because it did transparently get broken. It got broken by this doubling~halving story of a flexi-container with a diluting hot content.

And when we get deeper into the theory, we can see that this is the oneness of the Riemann sphere. A unit 1 geometry based on marry a real number translational symmetry with a complex number rotational symmetry.

The Riemann sphere has its foot in both the relativistic and quantum camp in that regard. But now we really are getting into the technicalities.

The point is think structural principles rather than material facts. The Big Bang has to have an explanation that goes beyond the contingencies of material being. It has to have the logical truth of a structural account.

The "agency" here is precisely the person's developped capacity for intentional action, not a mysterious inner homunculus performing interpretations before we do. — Pierre-Normand

The way I handle this is seeing habit and attention as complementary spatiotemporal scales of “being aware”.

Habits are developed slowly over time and are aimed at generality. And then in being pre-prepared, they can be just “emitted” instantly.

While attention is for dealing with the novel and the particular. So it is designed to cope with the unprepared situation and takes a little time and consideration to develop its state of response.

The way this works architecturally is that the basal ganglia as the habit centre is connected to the cortex as the attention centre by baso-cortical loops. All new sensory information arriving in the brain flows through both the basal ganglia and the cortex. If the basal ganglia is triggered by a familiar pattern, then it can simply emit a learnt response. And if that doesn’t happen, then the cortex takes over to spend a little more time developing a considered response.

So habit takes a lifetime to develop, as about a fifth of a second to emit. Attention is designed to handle the novel, and gets that job done in half a second.

The two processes are physically connected in feedback fashion so the basal ganglia can learn and observe from what attention is doing. Attention is part of what gets the right habits warmed up - and also actively suppresses other habits that could get triggered in ways that would interfere.

So when facing a fast tennis serve, one concentrates and blocks out the trickle of sweat, the thought of failure, the mutters of the crowd. Attention is clearing the path for habit to be its most effective. And avoiding itself flicking away to turn that structure of habit onto being bothered by the trickle of sweat, etc.

You have the basic processing division. And then also its seamless feeling integration. The unity of two opposites as a dynamical balance.

We sort of always know this is the case, but don’t have a clear theory of it. If we think that sentience is about “being conscious”, then that means fully attentional. But attention is often the exercise of our intentional or voluntary control in a way that instead tilts the balance towards quick and immediate habit. Using attention not to be in fact “attentional” but instead part of a state of prepared readiness as we are stopping our mind wandering off on to other points of possible focus.

Likewise, LLMs aren't just decoding words according to dictionary definitions. Rather, the context furnished by the prompt (and earlier parts of the conversation) activates a field of expectations that allows the LLM (or rather the enacted AI-assistant "persona" that the LLM enables) to transparently grasp my request and my pragmatic intent. This field of expectations is what allows the AI-assistant to see through my words to their pragmatic force (without having a clue what are the tokens that the underlying neural network (i.e. the transformer architecture) processes. — Pierre-Normand

I was thinking about this. And one idea that sprung to mind is that the human mind has the living hierarchical structure I just described. A structure woven in 3D out of the complementary processing streams of habit and attention. Then LLMs sort of take all this and squash it flat. The human generated data is the product of a moment to moment balancing act. And the LLM flattens and freezes it in a way that contains all the information but now applies its own algorithm of gradient descent over the top.

So in terms of agency, autonomy, intentionality, all that gets squished flat. But then like a hologram or something, can process prompts to generate points of view that reflect some recreated state of balance.

As humans, we are always striving for a balance of habit and attention in any moment. The LLM can just sum over all that complexity in its timeless and placeless computational fashion.

And that is why it can seem creative and robotic at the same time. It performs much better than expected. But how it does that seems a mystery as we can’t see how much habit and how much attention its gradient descent algorithm is feeding off.

I did interact with him a lot. And no he wasn’t in the same league. I’m not contesting your decision, just speaking up for Harry in terms of my experience. And the fair thing to say is that I have had no problem with him myself.

And i echo the others saying you are doing a good job. :up:

Blah blah blah blah-blah bl-ba-blah blah. — Harry Hindu

I must say that personally I would take no offence at all to that response. It’s probably even fair.

But then I’m guilty of enjoying the knockabout character of the debates here. I happened to stumble on an old thread the other day and remembered how much I miss The Great Whatever and even StreetlightX.

The modding here is relaxed so other things may have weighed heavier. But for my part, I don’t see any great reason for the ban.

It's also a complete and fundamental misunderstanding of two separate concepts: — AmadeusD

Nope. I was making the point that a hallmark of “consciousness” is that it is embodied and agential. And we know how that is so from having studied the neurobiology - the architecture of brains.

Schizophrenia appears to arise from a fundamental breakdown in the timing and integration of neural activity. The sense of authorship for intents and actions, and also the ability to filter sensation in normal attentional fashion, goes awry as there is not the proper traffic in “efference copy” information. In simple terms, the frontal motor areas may initiate actions, and the sensory half of the cortex doesn’t get its copy of the commands in time to cancel them out of the state of sensory experience it then produces.

This is why symptoms like thought insertion and thought broadcasting arise. The precise compensation of an “implicit timing” connection breaks down. Normally we can tell whether we are moving the world or the world is moving us as in the first case, our sensory areas knows in advance to subtract the predictable action from its interpretive response. In the second case, the self-generated action catches the sensory areas by surprise. It feels like an alien hand is now in control. Sensations are thrusting at us. Thoughts and ideas are being imposed.

So we know how the brain generates consciousness by solving all these timing issues. How it has an architecture that deals with the fact it takes time just to pass along the message of what motor action we have planned so our sensory processing can already take that into account. An integrated sense of a self in its world can then arise out of a tricky neurobiological interaction. And schizophrenia is the kind of disorder that really brings this fact home.

And then we have this other nonsense about the brain being an antenna tuned into a cosmic psychic frequency. A sloppy and lazy analogy that we are meant to allow for the sake of argument. A hypothesis that completely wastes our time when we should instead be marvelling at the biological intricacy of the neural engineering that so easily seems to sustain the “normal” mind.

Being embodied and agential seems so effortless that yes, maybe it could be just a broadcast picked up off the airwaves.

But then nope. The neurobiology to get the job done is what we should reserve our amazement for.

Planck's Scale is not an actual measurement, but a theoretical limit to measurement. — Gnomon

It’s defined by the three constants, c,G,h. And there is a reason why Newtonian physics has been evolving towards a theory of quantum gravity by successively adding these constants.

For all practical purposes, you might just as well say that the near-infinite universe we now experience originated from nothing --- no atoms or quarks --- but near-infinite Energy. That immeasurable, almost unimaginable, quantity of world-creating Causal Power is literally super-natural. And it is analogous to what I call, philosophically, Infinite Potential* — Gnomon

My point is that the Planckscale is defined by this trio of fundamental ratios. They describe a first symmetry breaking in terms of a first spacetime extent with its matching energy density or momentum uncertainty content. Quantum uncertainty says that in a spacetime point of limiting size, there is a matching energy density or raw heat.

And all our physical theories embody this basic fact of existence. They require it to work as a system of differential equations.

So right there is something exactly the opposite of your handwaving. We have a triad of constants that are in a pure symmetry breaking relation. A unit 1 story as they are all the fundamental units and may as well be set to 1 as “measured values”.

If you want to get metaphysical, it starts with seeing that the “infinite” in fact has this finitude. Existence starts with neither nothing, nor everything, but with this unit 1 scale that is a symmetry breaking. A fundamental ratio between spacetime described under general relativity and energy density described under quantum field theory.

Our ideas about how this could be the case have to take to take a back seat to the fact it is the case. What becomes handwaving and speculative is ignoring what is now built into the very structure of our best physical theory.

To put it simply, Okun’s cube tells us that the Universe has the basic structure of being a relativistic spacetime container with a matching quantum energy density content. This is the broken symmetry. And that is then the new starting point for speculation about how to make sense of the situation.

If you are not addressing that fundamental fact - that when spacetime got started, it came full to the brim with an energy density content - then you just aren’t in the game.