Meta-Rationality // Post-Rationality

[r0ckyreed]

I would define meta-rationality as thinking about the limits and strength of rationality. Post-rationality, I would define, are features of consciousness outside of rationality. In other words, there are certain limits in rationality/logic and there may be levels of thinking beyond rationality.

But what is rationality really? Rationality is using reasoning and logic to arrive at certain conclusions. It uses chains of reasoning, consistent structured, critical thinking to arrive at certain conclusions based off of the validity of the premises/evidence.

I would say that rationality is a higher form of thinking than strict empiricism or observation. Empiricism/observation/experience is the foundation of all knowledge. Without observation, there could be no knowing and no rationality. Rationality is dependent on what we observe. But knowledge does not strictly stop at observation. Rationalism is the view of knowledge that suggests that we can derive knowledge from rationality. We can use mathematics and logic to figure out truths that we may not be able to get a direct experience at. For instance, the fact that humans knew that they had to invent space suits to go to the moon. They had no direct experience of this, but they arrived at the conclusion based off of reasoning. Rationality has its strengths, but there are limits.

Post-rationality suggests that rationality is one perspective out of many. One of the main objections is that rationality is based off of certain assumptions that were not itself rationally justified. Rationality discounts that logic, reasoning, and evidence are highly subjective and contentious. Context plays a lot in rationality, and rationality may not be flexible enough to understand deeper implicit and intuitive facets of understanding and reality. 

Understanding itself is a highly intuitive process. Rationality also faces the issue of how to discern and differentiate the differences between what is rational vs. irrational. Everyone thinks they are rational, so who is right? Whose method of rationality is actually rational and how do we decide what is more rational than another? Is there one way to be rational or many?

Also, how do we know that rationality is a valid method for deriving knowledge without self-justifying it with rationality? Rationality may not be able to handle self-reference, which entails contradictions and paradoxes. But the Universe is full of contradictions and paradoxes. Contradictions may be a feature of reality and not some fallacy as rationalists might suggest.

Of course, we are not going to abandon rationality. We are going to take the good parts of rationality and incorporate them in post-rationality. Post-rationality will encompass intuition. But you see the issue right now is that I am using rationality right now to discuss the limits of rationality! Even to argue with me about rationality is you also using rationality. We really cannot escape rationality, but the most rational form of rationality I would call post-rationality or meta-rationality, which incorporates rationality with what normie rationality misses. Rationality without encompassing intuition and without self-observing its own biases in terms of its limits is itself an irrational system! Of course, I am using rationality to make this post and you will respond with rational reasons for why rationality is irrational. But this is the standstill we are at. In some ways, we can go beyond rationality, but we could never explain those means without using rationality because rationality is the means to which our minds are able to make the implicit explicit.

[rachMiel]

Bohm and Krishnamurti inquired deeply into the nature and limitations of thought. Lots and lots of links out there. Here are the seminal 12 dialogues: 

 

[Davino]

Rationality free of bias or reasoning uncontaminated by like and dislike is a powerful tool for discerning relative truth from falsehood.

Let's have an example:

Imagine we are lovers in a honey moon. You'll give me everything

Imagine we are in a street fight to death. You'll give me nothing

This is under the influence of like and dislike

Under the influence of pure reason, one can see whether I deserve or not. 

If I deserve you will give me, even if I'm your worse enemy. If I don't deserve you won't give me even if you love me to the bones.

These virtues are to be cultivated in the Mind.

 

I've been having great insights into the nature of reason lately (many thanks to Shivapuri Baba). The truth is that after all, reason is such a marvelous gift, that has been coopted by our ego taking advantage of our blind spots and cognitive biases.

So it's not the tool but the maturity and skilfulness of its user. It's the consciousness shining through the subject which determines the output of Rationality.

[Ero]

@r0ckyreed That’s pretty spot on. You have been doing a lot of good contemplation, thanks for sharing! 

The only thing I would point out as fruitful ground for deeper examination is the following quote from your post:

”Empiricism/observation/experience is the foundation of all knowledge. Without observation, there could be no knowing and no rationality”

The nuances of what is an “observation”/ experience are pretty tricky. What are we to make of schizophrenia or psychosis? Altered states of consciousness are especially tricky to point down within this framework without digging into the nuance.

[Yimpa]

5 minutes ago, Ero said:

Altered states of consciousness are especially tricky to point down within this framework without digging into the nuance.

Without direct experience.

[Ero]

2 minutes ago, Yimpa said:

Without direct experience.

What I was pointing to is that direct experience alone is slippery slope. In the example of schizophrenia/delirium, I may experience being stung by bees and scorpions and it feels as real as reality. Does that make it “true”? “Yes, for you” you may answer, but then we get into the trap that has become apparent in the other thread. If I experienced Jesus Christ, does that make him any less of a social construct? 
Which is why I chose the phrase “without nuance” instead of “without direct experience”.

[r0ckyreed]

17 minutes ago, Ero said:

What I was pointing to is that direct experience alone is slippery slope. In the example of schizophrenia/delirium, I may experience being stung by bees and scorpions and it feels as real as reality.

Of course direct experience alone is unreliable. But observation is the foundation. If you do not observe something, you cannot rationalize about it. If you just observe but don’t rationalize, then you are missing out on another half of reality. People with hallucinations cannot trust their own senses in the same way. And maybe they can’t trust their rationality. But there is a good movie based on a true story called A Beautiful Mind where a schizophrenic learned how to distinguish his hallucinations from reality. You can only do that if you are observing something, which is an active reflective process. 

Edited October 7 by r0ckyreed

[Scholar]

Large Language Models are a perfect example of the limits of rationality, and so is evolution. 

The most significant limitation of rationality are actually it's metaphysical limitations. Rationality is a particular form of existence, namely concepts-symbols and their relatively simple relationships to each other.

Rationality is fundamentally linearlistic and linguistic and that comes with obvious limitations as far as Creation goes.

A rational mind can create a complicated machine. Machines, like computers, are a perfect example of how rationality functions and how it constructs functionality.

Namely, each component in a machine relates to another component, but usually how many relationships each component has to each other is very limited. It works more like A -> B -> C -> D, in a very complicated way. That is in essence what processing is.

 

If you compare this to a complex system like the human body, you will notice that each cell is in relationship to every other cell. Everything inherently affects everything else. A rational mind could not possibly arrive at a system of such complexity because its' linearlistic nature is too limited to take into account the amount of relationships you would need to hold in a given space (consciousness) to recognize it's functionality.

In fact, sufficiently complex systems are not only impossible to construct within a rational space, they are also virtually impossible to understand and grasp.

Even if scientists would understand the function of every single molecule in the human body they would not be able to predict the holistic motion of the body, simply because the functions as they relate to each other create such high complexity that new, metafunctions would emerge that would be impossible to predict via a linearlistic framework.

 

A rational mind has to basically process any complex problem by putting it into a linear language that it can grasp and process. In essence, this is an attempt to translate complexity into complicatedness. "Part A does X, therefore if Part B does Y the resulting interaction will be Z."

There will be no rational mind that will ever truly "understand" what a Large Language Model does when it is constructing an image or intuiting the shape of our language and collective understanding. And no rational mind could have ever constructed a machine that would have achieved what a Large Language Model does.

 

 

A neural network fundamentally achieves what a complicated program or machine cannot achieve in the same way, and that is the emergence of true complexity. A large language model is not understanding what an image is and how to create an image, and then recreating an image. No, what a large language model is, is a neural network that literally is the function of what it means to imagine an image. It doesn't need to understand itself or it's process, because it's very nature is the function itself.

It literally became function, whereas rationality attempts to imitate or grasp functionality through linearlistic understanding. This hard limitation cannot be overcome because it is actually metaphysical.

 

This is the same way evolution functions. We look at evolution and we think of how impossible it is that such complexity could exist. How could we possibly ever understand the human machine?

But this is flawed thinking. The human body is not a machine, the human body is an organism. It is not imitating function, it literally is function. And the forces which allowed the complexity of the human body to emerge were not limited by linearlistic processing. The universe doesn't have to think in "If A, then B, then C.", because the universe in it's very nature is complexity. It is every single interaction and their infinite relations to each other.

 

This is why, given the right conditions, it is utterly effortless for the universe to construct life from dead matter, to have it self construct itself in the oceans and eventually evolve into all the diversity we are seeing today. There was not a single rational thought put into any of it, and no rational mind could ever hope to imitate this kind of complexity.

 

For that, it would require a creative mind, a mind that that can become Complexity itself. In other words, a mind that can Imagine.

 

 

 

In the future, we will create a Non-Machine Machine, whose very hardware will be organic, meaning it's components will have truly complex relationships. Today's machines will seem like a simpletons invention in comparison. It will not be processing, it will be doing what the human brain does, in which billions of components relate to each other and create an emergent function.

This will be simulated using computer processing, but the very nature of computer processing is linearlistic and laughably limited for the reasons explained above.

Edited October 7 by Scholar

[Ero]

17 minutes ago, r0ckyreed said:

. People with hallucinations cannot trust their own senses in the same way. And maybe they can’t trust their rationality. 

And here we get to the meat of the issue. How do you know you can trust your senses “in the same way”? In what way? How are you sure everything you experience isn’t just a fever dream?

Bottom line is you don’t.

[Ero]

@Scholar LLMs are not rational in any sense of the word. They are not “reasoners” and do not possess ability to perform algorithms or abstraction. They are stochastic next-token predictors. So they say nothing about the limit of rationality. 

Rationality is limited for other reasons. Your arguments about complexity are closer to the heart of the issue. However I would contest that undecidability/ chaos/ non-determinism does not prevent us from making concrete statements about the system. In fact, your description of processing as relationships using maps is part of the language that modern math uses to “rise above” the chaos. It’s called category theory and is at the foundation of what will be the Tier 2 paradigm. For a concrete example, study Poincare and von Neumann. Despite the provable unsolvability of the three-body problem, Poincare was able to prove the existence of unique stable orbits using techniques that later grew into Topology - the field that gave rise to Category Theory!

If you are interested in this type of stuff, you should check out my current journal here:
 

 

[Scholar]

9 minutes ago, Ero said:

@Scholar LLMs are not rational in any sense of the word. They are not “reasoners” and do not possess ability to perform algorithms or abstraction. They are stochastic next-token predictors. So they say nothing about the limit of rationality. 

Rationality is limited for other reasons. Your arguments about complexity are closer to the heart of the issue. However I would contest that undecidability/ chaos/ non-determinism does not prevent us from making concrete statements about the system. In fact, your description of processing as relationships using maps is part of the language that modern math uses to “rise above” the chaos. It’s called category theory and is at the foundation of what will be the Tier 2 paradigm. For a concrete example, study Poincare and von Neumann. Despite the provable unsolvability of the three-body problem, Poincare was able to prove the existence of unique stable orbits using techniques that later grew into Topology - the field that gave rise to Category Theory!

If you are interested in this type of stuff, you should check out my current journal here:
 

 

You should re-read my post because you seem to have misunderstood. I never made the claim that LLMs are rational. They are fundamentally intuitive, that was the whole point my post.

Rationality is a function of individuated consciousness.

[Ero]

40 minutes ago, Scholar said:

Large Language Models are a perfect example of the limits of rationality, and so is evolution. 

Then how are they examples of the limits of rationality if there is nothing rational about them? 
 

If you are instead referring to them as the “result of human rationality” and taking their faults as examples for the latter’s limit I will say that they are not even the cutting edge (saying this is as a researcher in the mathematical foundations of AI)

Edited October 7 by Ero

[Scholar]

5 minutes ago, Ero said:

Then how are they examples of the limits of rationality if there is nothing rational about them? 
 

If you are instead referring to them as the “result of human rationality” and taking their faults as examples for the latter’s limit I will say that they are not even the cutting edge (saying this is a published researcher in the foundations of AI)

They are examples for things that contrast rationality, namely things that construct function through evolutionary selection within a field of complexity.

 

It demonstrates the limits of rationality, because linearlistic processing could never arrive at functions that require the level of complexity necessary to achieve something like visualization or replicating the relationships of a language.

 

LLMs are precisely not the result of human rationality, that is why they are so good at what they are doing. They are only a result of human rationality insofar as humans having constructed the conditions necessarily for evolution to allow the functions they sought to emerge through selective emergence.

 

The researchers of AI have no idea what they are doing, most of them are highly specialized baffoons.

Edited October 7 by Scholar

[Ero]

1 minute ago, Scholar said:

They are examples for things that contrast rationality, namely things that construct function through evolutionary selection within a field of complexity.

 

It demonstrates the limits of rationality, because linearlistic processing could never arrive at functions that require the level of complexity necessary to achieve something like visualization or replicating the relationships of a language.

 

LLMs are precisely not the result of human rationality, that is why they are so good at what they are doing. They are only a result of human rationality insofar as humans having constructed the conditions necessarily for evolution to allow the functions they sought to emerge through selective emergence.

 

The researchers of AI have no idea what they are doing, most of them are highly specialized baffoons.

I see what you mean. I guess your definition of rationality is more related to “linearity”, “closed-form”/“interpretable”, etc. However, I would contest that is not all there is to rationality.

Consider that there are highly non-linear, stochastic and emergent forms of rationality, ones that work on the macroscopic level instead of the microscopic. This comes from the process of “abstraction” that I have spoken previously about and it is a large part of my definition of “rationality” (at the root of math). The fact LLMs work in the first place is because we constructed them through this type of “abstract rationality”. And we will keep doing it, building better and better models. And with respect to your characterisation of most researchers, I, believe it or not, largely agree. Hence, why I work at the foundations rather than specialize in a particular field.

[Scholar]

Just now, Ero said:

I see what you mean. I guess your definition of rationality is more related to “linearity”, “closed-form”/“interpretable”, etc. However, I would contest that is not all there is to rationality.

Consider that there are highly non-linear, stochastic and emergent forms of rationality, ones that work on the macroscopic level instead of the microscopic. This comes from the process of “abstraction” that I have spoken previously about and it is a large part of my definition of “rationality” (at the root of math). The fact LLMs work in the first place is because we constructed them through this type of “abstract rationality”. And we will keep doing it, building better and better models. And with respect to your characterisation of most researchers, I, believe it or not, largely agree. Hence, why I work at the foundations rather than specialize in a particular field.

Can you give me an example of what you mean by macroscopic levels of rationality?

What is the process of abstraction and how do you define rationality?

 

Do you recognize rationality as a form of existence?

[r0ckyreed]

1 hour ago, Ero said:

And here we get to the meat of the issue. How do you know you can trust your senses “in the same way”? In what way? How are you sure everything you experience isn’t just a fever dream?

Bottom line is you don’t.

Correct. However, you can make observations and apply rationality, and your best bet is to see if what you come up with is something that is verifiable and has gone through critical thought. This is a topic for another post, but you can chew on the fact that some people see ghosts, others don’t, and some see demons. Who has the correct view of reality? Is a person who sees demons more true than a person who sees a ghost? What if they had a mystical experience. We could never verify if the observation had any basis in so-called reality. However, if your hallucinations become unreliable/ inconsistent, then that could cause problems. Some people hallucinate and it is completely wrong and some hallucinate and it could be right. For instance, if someone hallucinates that you killed someone but you actually didn’t, then that would be a false hallucination. But if someone hallucinates a ghost telling them you killed them when you actually didn’t, then it is a true hallucination. How do I know I am not hallucinating right now? The only way I could know is through observation. I know it’s circular. But we are getting off track into meta-observation.

Edited October 7 by r0ckyreed

[Ero]

@Scholar Sure. To be clear, I don't think we have any disagreements, since I believe you are pointing to some pretty important points I have myself brought up before. Your critique is essentially that of the contemporary notion of 'rationality', as has been defined largely by the works of Newton (determinism), Hilbert  (formalism), Russel (logicism) and many more. I fully agree with your assessment of its limitation for handling complex phenomena (which are essentially all of existence). 

 In what has largely been the death of the latter rationality programs through QM's uncertainty principle, Gödel's incompleteness theorems and Turing's undecidability, a new avenue has opened up for redefining what 'rationality' means, one that I place at the root of what I describe as the "Chaos, Entropy, Order" paradigm. It goes something along the lines of :

1) Chaos = Fundamentally unknowable nature of existence. All our models are incomplete (in the formal meaning) and will never be able to fully describe reality. There will never be a 'fundamental theory', because if there were one, it would be undecidable by means of inclusion. 

2) Entropy I will skip for now, because it is a very tricky and somewhat unrelated, but the part that is important is that 'rationality' is then defined as the process of 'minimizing surprise', i.e maximizing entropy through the Active Inference/ Free Energy principle, i.e it is the process of iterating on our generative model (which I contest in its present form is science, but will largely transform due to the precise point you are pointing for the failure to reconcile AI models in the current framework )

3) Order = Abstraction - the definition of "emergence of order" is the existence of layers of abstraction, such that the properties on the level above can be explained and studied without the need to reference the underlying details (i.e forgetting the unimportant parts). Here is where this becomes so powerful: 

All of the "complex systems" that are undecidable exhibit emergent behavior of some form - Fluids, Plasma, Quantum Systems, Spin Glasses, DNA, Cellular Automata, Turing Machines, NNs - all of them fall within the computational class. Here is the important part - even though we provably cannot know with certainty the position and exact state of the system after some time t, we can still infer global "topological features". This is what:

- Boltzman and Gibbs did with Thermodynamics (showing the emergence of global gas properties out of microscopic interactions), 

- Poincare did with the 3-body problem (showing stable orbits in an undecidable system)

- Kologomorov did with turbulence in fluids (microscopically undecidable, yet there are still global properties, such as viscosity, and forms, such as 'vortexes' and various types of flows)

- what von Neumann did with Operator Theory for QM, abstracting above the microscopic undecidable nature and proving the macroscopic invariants

 

All of these examples are clearly highly rigorous and "rational" - they work by abstracting away the undecidability into stating global phenomena. It is at the root of what we need to do to transition to the next stage of science.

[zurew]

On 2024. 10. 07. at 2:17 PM, r0ckyreed said:

Without observation, there could be no knowing and no rationality.

I reject this. Straightforward counter examples - analytic truths and apriori truths.

For example: "Bachelors are unmarried men" doesn't require any kind of observation or investigation in the world -  it is just true by definition.

 

On 2024. 10. 07. at 2:17 PM, r0ckyreed said:

Everyone thinks they are rational, so who is right? Whose method of rationality is actually rational and how do we decide what is more rational than another? Is there one way to be rational or many?

Also, how do we know that rationality is a valid method for deriving knowledge without self-justifying it with rationality?

Most of these questions are hard to answer because they are vague. You would need to clear up what is meant by certain terms like "right" or "valid" or what is meant by "rational" and "more rational" and then after that more fruitful conversation can be had.

 

On 2024. 10. 07. at 2:17 PM, r0ckyreed said:

Contradictions may be a feature of reality and not some fallacy as rationalists might suggest.

Yeah, but there you are making claims about metaphysics and most "rationalists" don't really make claims about metaphiscs but make objections in the context of epistemology. Accepting or being open to the idea that contradictions may be a feature of reality is very different from accepting arguments that are inherently contradictory as justificatory or as a valid form of argumentation.

But regardless, there are PhD logicians who hold the view that there are true contradictions. 

 

Btw , be careful and avoid category errors. Notions like "justify logic or justify justification or ground justification or logic" is meaningless or doesn't make sense under certain semantics.

[r0ckyreed]

15 hours ago, zurew said:

I reject this. Straightforward counter examples - analytic truths and apriori truths.

For example: "Bachelors are unmarried men" doesn't require any kind of observation or investigation in the world -  it is just true by definition.

A bachelor is a concept and is not something that actually exists outside your mind. You wouldn't even know what a bachelor is if you cannot observe anything. I am claiming that you cannot form concepts without your senses. If someone has always been blind, deaf, then how could they form concepts about sight and sound? Your rationality is slave to the senses. I will coin that quote there.

It is like rationalizing about what color is when you are blind. If you cannot observe something, you can't rationalize about it. And bachelor is also technically a social construction anyways, so it is true only under relative conditions.

15 hours ago, zurew said:

Most of these questions are hard to answer because they are vague. You would need to clear up what is meant by certain terms like "right" or "valid" or what is meant by "rational" and "more rational" and then after that more fruitful conversation can be had.

 

Seems pretty clear and straight forward to me. Who is more rational, a flat earther or a round earther, a trump supporter or a liberal? How do we know?

15 hours ago, zurew said:

Yeah, but there you are making claims about metaphysics and most "rationalists" don't really make claims about metaphiscs but make objections in the context of epistemology. Accepting or being open to the idea that contradictions may be a feature of reality is very different from accepting arguments that are inherently contradictory as justificatory or as a valid form of argumentation.

But regardless, there are PhD logicians who hold the view that there are true contradictions. 

 

Rationalists definitely have metaphysical biases. The world isn't just rational as a rationalist would have you think. The world holds many contradictions that may not be suited for a rationalist approach to epistemology. Remember, epistemology is always about metaphysics in someway shape or form because knowledge is always about the reality/existence of something. Sure, you can ask how do we know what is right or wrong in the context of morality, but it is easy to forget how much our metaphysical biases shape our worldview. There is a reason why worldview is called world-view.

Edited October 9 by r0ckyreed

[zurew]

1 hour ago, r0ckyreed said:

A bachelor is a concept and is not something that actually exists outside your mind. You wouldn't even know what a bachelor is if you cannot observe anything. I am claiming that you cannot form concepts without your senses. If someone has always been blind, deaf, then how could they form concepts about sight and sound? Your rationality is slave to the senses. I will coin that quote there.

It is like rationalizing about what color is when you are blind. If you cannot observe something, you can't rationalize about it. And bachelor is also technically a social construction anyways, so it is true only under relative conditions.

If by observation you just mean that at some point in your life you had to have some  access to qualia , sure I can go with that,  although I would be curious if you can make an argument that can actually defend that claim. But regarldess, thats wildly different from claming that in order to form a concept about something you need to go out and observe that thing / object. Like  - I have a concept of Mars, but I have never done any kind of empirical investigation or observation about Mars. 

Do you need to go outside and investigate the world to conclude bachelors are unmarried men, or the truth value of that statement is reliant on the  definition of bachelor?

Its not like you going outside and doing an investigation in the world will suddenly change the fact from 'bachelors are umarried men' , to 'bachelors are actually married men'. No amount of empirical investigation or observation will change the definition.

 

Other example where you don't need any kind of empirical investigation  - logical entailment and using the rules of inference:

for instance : P1) All men are mortal P2) Socrates is a man C) Socrates is mortal. The conclusion that Socrates is mortal can be derived using a deductive inference without any kind of empirical investigation about Socrates' mortality or immortality.

 

1 hour ago, r0ckyreed said:

Seems pretty clear and straight forward to me. Who is more rational, a flat earther or a round earther, a trump supporter or a liberal? How do we know?

Seems pretty vague to me. I can parse that question using my concept of rationality, but its not clear at all whether you have the same concept in mind when you use that phrase. 

 

1 hour ago, r0ckyreed said:

Rationalists definitely have metaphysical biases. The world isn't just rational as a rationalist would have you think. The world holds many contradictions that may not be suited for a rationalist approach to epistemology. Remember, epistemology is always about metaphysics in someway shape or form because knowledge is always about the reality/existence of something.

But wait, surely you can acknowledge that accepting  contradictions in the context of metaphysics is very different from accepting contradiction in the context of justification or in the context of rules of inference.

 

Edited October 9 by zurew