Discussion on Overlap, Loop, and Cyborg Methods in Cognitive Models

2021-03-26 Complexity

Discussion Background

This article organizes an in-depth academic discussion between two scholars about cognitive model construction, focusing on "overlapping structures," "loop mechanisms," and "cyborg technology" in simulating cognitive processes, involving cutting-edge theories such as extended cognition, modal logic, and gamification.

Round One: Core Theoretical Exposition by Scholar A

Q1: Cognitive Science Foundation of the "Overlap" Concept

Scholar A explains:

We can start from several viewpoints in cognitive science, such as the concept of "extended cognition" manifested in embodied cognition.

Extended cognition does not consider any "unit computation" as the "ontology" of cognition. That is to say:

  • Computation cannot generate cognition internally

  • According to extended cognition, "boundaries," i.e., the "coupling" between computation and context, are key to the emergence of cognition

  • Boundaries rather than the interior of computational units determine the source of cognition

Core inference: The complexity of boundaries is an important indicator of cognitive levels

Q2: Guattari's "Polyphony" Concept and Subjectivity

Scholar A supplements:

Scholar Guattari has a "polyphony" concept:

  • All cognitive manifestations (understood as personality from human perspective) are "schizophrenic"

  • Continuous schizophrenia, in today's internet terminology, means continuous "persona"

  • This is the prerequisite for humans to possess (or be able to "use") subjectivity

Logical conclusion: Any attempt to construct models on cognition-intelligence-wisdom topics must probably face "overlapping structures."

Q3: Relationship Between Modal Logic and "Overlapping Structures"

Scholar A proposes:

A school of modal logic—possible world semantics—constructs metaphorical models that I believe are excellent logical approaches to "overlapping structures."

Design requirement: The interpreter's design concept must have ways to "manifest" the incommensurable (ontological boundaries).

Q4: Dual Value of Loop Mechanisms

Scholar A details the loop concept:

If boundary-based extended cognition can allow us to accept the "overlapping structures" provided by modal logic, then the so-called loop steps can take the stage.

Two values of Loop:

First: Loop extended from the halting problem of formal logic

  • Any formal logic-based design that wants to simulate the "reference-convergence" nature of cognition must involve recursion

  • But recursion cannot "internally" halt, necessarily requiring a step to achieve self-reference

Second: Self-reference under extended cognition approach

  • Self-reference is not traditionally "subject-ontology" based, but "subject-boundary" based

  • The complexity of boundaries requires that simulation magnitude (frequency) must yield to simulation "regression"

  • This regression probably cannot avoid loops in formal logic

Q5: Subjectivity as Foundational Existence of the World

Scholar A's core hypothesis:

If we think about this from the perspective of physics' presuppositional logic (worldview), then we can make a hypothesis: intentionality, which is the instrumentalized expression of subjectivity, might belong to world-constitutive elements like gravity.

Verification mode: Need to refuse to extract subjectivity-intentionality existence from the world's foundation, but use relevant instrumental methods (such as cyborg based on loop concepts) to attempt "manifestation."

Q6: Essential Function of Cyborg Technology

Scholar A clarifies:

Cyborg itself is not about connecting "human subjectivity" into models, but connecting model loops into subjectivity.

Key understanding:

  • Humans do not "possess" subjectivity from the completeness of formal logic

  • On the contrary, humans are just an already connected-activated loop

  • "Manifesting" subjectivity as a "phenomenon" of foundational existence in our world

Q7: Gamification as Grafting Solution

Scholar A proposes solution:

When overlapping models reach a complexity level, how should we use cyborg methods to activate the "intentionality" existing in this world?

Answer: Gamification

Gamification is what I consider the most feasible mode for grafting subjectivity into models through overlapping methods.

Core concept:

  • Gamification is a middle-layer design concept that uses behavior (computation) to replace presetting (algorithms)

  • Simulation of behavior, rather than "abstraction" of behavior, is the premise for ANN effectiveness

  • Basic learning utility is all constructed on gamification concepts

Technical implementation:

  • Cyborg's "recognition" of bodily electrical signals is actually just pattern recognition of behavior

  • Once we can "reversely" recognize models as manifestable "behavior" through cyborg technology

  • From a subjectivity perspective, this model is grafted into "autonomy"

Round Two: Scholar B's Understanding Confirmation and Questions

A1: Confirmation of Core Concept Understanding

Scholar B states understanding:

Given my limited involvement in mental science, I want to first state my understanding of some important concepts:

  1. Overlapping structure: Indicates an object where computation departing from specific subjectivity couples with environments constructed by other subjectivities

  2. Loop: Refers to the non-halting state of Turing machines, metaphorically representing a process where formal logic continuously deduces but cannot internally converge

  3. "Connection-activation" method:

A2: Concepts Requiring Further Explanation

Scholar B raises five key questions:

  1. Can you explain the specific relationship between overlapping structures and boundaries/polyphony? If this is an analogy or metaphor, I cannot understand the former through the latter.

  2. What is the metaphorical model constructed by possible world semantics in modal logic? The introductions to modal logic I've found seem unrelated to our discussion.

  3. Must formal logic-based designs involve "recursion" to simulate cognitive "reference-convergence" characteristics? Does reference here mean self-reference?

  4. Boundary complexity requires simulation frequency to yield to simulation regression - can you describe in simpler language what simulation regression means?

  5. Gamification is what you consider the most feasible mode for grafting subjectivity into models through overlapping methods - you mention overlap again, verbalizing it. I hope you can re-explain the original source of the word "overlap."

Round Three: Scholar A's Deepened Exposition

A1: Explanation of Terminology Issues

Scholar A acknowledges:

Sorry for my terminology problems, this is indeed a bad habit of mine. Because I don't engage in academic professions and have certain "resistance" to precise concepts, I indeed easily create understanding difficulties for others in concept application.

Semantic view: All semantics are necessarily obtained through context, so I might still try to use context rather than concepts to explain my semantics.

A2: Visualization of the Overlap Concept

Scholar A provides specific analysis:

My understanding of loop is indeed related to the Turing machine halting problem, but I hope this understanding has a visual "metaphor".

Core insight:

  • The Turing machine halting problem is like a linear statement that never stops (has no boundaries)

  • But the semantics of this statement itself has boundaries: the Turing machine "is" non-halting

  • Once this is such a statement, it is actually formal logic (complete)

Essential understanding: The halting of Turing machines is completely a bounded "infinite loop", which is loop.

A3: The "External" Space Where Semantics Exists

Scholar A's key argument:

If we strictly follow the proof principles of formal logic, then we can say semantics exist neither between relations nor within ontology (concepts).

Two choices:

  1. Deny semantics - reductionism takes this path

  2. Semantics exist in the "external"

Definition of "external":

  • Cannot be a homomorphic relationship, but must be realizable-formalizable in non-"comparative-homomorphic" expressions "external" to homomorphism

  • "External" must be expressed in homomorphic behavior (simulation) but cannot be completely covered by homomorphic behavior

A4: Analysis of ANN Working Principles' Overlapping Mechanisms

Scholar A's in-depth analysis:

Think about how ANN works:

  • Neural networks have many "homomorphic phenomena" in between

  • But the effectiveness (intelligence) of neural networks is not clear mathematical statement "functional homomorphism"

  • But phenomena driven by ML (which is gamification) from these homomorphisms in a network

Essence of machine learning:

  • Through repeated loops, various homomorphisms are "reconstructed" (re-homomorphized) in data

  • This "reconstruction" in formal logic description is not "reorganization" of previous structures

  • ML does not "understand-judge-choose" better structures

  • It just continuously engages in "infinite analogy" with previous structures

A5: Core Principle of "No Overlap, No Boundaries"

Scholar A's key statement:

Let us emphasize a most crucial statement: No overlap, no boundaries.

Logical derivation:

  • Analogical accessibility cannot be established on homomorphic linear boundaries

  • Ontological analysis of any linear statement will lead to incompleteness conclusions (Gödel incompleteness theorem)

  • That is, conclusions without clear boundaries

  • The "objectivity" of linear description is a thoroughly subjective problem

Necessity of overlap:

  • A single loop can only be linear that cannot manifest itself

  • A single loop is not a loop; a single loop cannot be "identified"

  • This world has no independently existing loops

  • Only when infinite loops generate numerous overlapping regressions, and numerous regressive loop behaviors cannot "reduce" each other

  • Yet are homomorphized in regression, semantics (which can be further analogized as awareness-consciousness) are established at boundaries

Round Four: Scholar B's Deepened Understanding and Further Exploration

A1: Deep Understanding of Overlapping Structures

Scholar B confirms understanding:

Your proposed "overlap" is an inspiring concept: "The formal network of possible worlds metaphorizes a universe where numerous forms overlap and reflect each other."

Core value: This worldview requires the intersection and overlap of all other forms to function.

Specific mechanism:

  • For a certain subjectivity (a certain "I"), it manifests "itself" through every form it uses to reflect upon itself

  • Displaying a rich, multi-faceted "I" in numerous such mirrors

  • "Boundaries" are demarcations of the external world after subjects recognize their own existence

  • The complexity of "boundaries" is homomorphic to the overlap amount of possible world forms

A2: Comparison with Existing Model Designs

Scholar B analyzes:

This overlapping structure corresponds to one of my model designs at a simple understanding level, using multiple independently running pattern worlds for model deduction, but:

Original design:

  • This design was only for optimization purposes (reducing information needed for deduction in single pattern worlds)

  • Considered that due to subjectivity being independent of specific model selection requirements, it could in principle operate in single pattern worlds

New understanding:

  • What you propose is not an optimization-computational approach

  • But a holistic model design approach

  • How this holistic design approach specifically unfolds is an important question to be explored

A3: Further Understanding of the "External" Concept of Semantics

Scholar B elaborates:

You mention semantics are in the "external," and this "external" must be simulated by homomorphic behavior but cannot be completely covered by homomorphic behavior.

Understanding of three key terms:

  • Homomorphism: More descriptive

  • Simulation: More operational

  • Metaphor: Can be understood as metaphor and homomorphism being interchangeable, both extracting analogically accessible facts from different possible worlds for connection

  • Simulation: Is reverse engineering of this, acknowledging homomorphism existence and inferring real-world properties

A4: Understanding the Relationship Between Loop and Consciousness

Scholar B summarizes:

You mentioned the significance of the loop concept for consciousness formation, combined with some of your blog posts I've read:

Consciousness formation process:

  • Humans have completed the process from one generating two to two generating three

  • If intentionality (or this is a kind of pending subjectivity, extracted subjectivity) is a universally existing world attribute

  • Then the "I" that humans feel is a result of completing "subject-autonomization"

  • Awakening "self" from vast intentionality

  • Thereafter more truly recognizing "I" and "boundaries" in the world of overlapping shadows

A5: Specific Implementation of Gamification Technology

Scholar B raises key questions:

You mention ML is a gamification process, and "gamification is a middle-layer design concept using behavior (computation) to replace presetting (algorithms)."

Specific questions:

  1. How do behaviors gradually transform into specific clear algorithmic structures? How does this process specifically proceed?

  2. In designing cognitive models, how does "self" awaken, form, and expand cognition in gamification?

  3. How to enhance understanding methods of the real world in overlapping structures?

Practical challenges:

  • The design difficulty of pattern universal language lies in its abstractness and purposeless, "meaningless" design

  • This is not a defect, because the problem we face is removing the previously uncontrolled binding relationship between subjectivity or intentionality and formal models

Discussion Summary

This in-depth dialogue reveals several core issues in cognitive model construction:

  1. Necessity of overlapping structures: Single computational units cannot generate true cognition; intelligence can only manifest through complex coupling of boundaries

  2. Dual role of loop mechanisms: Both formal logical recursive expression and foundational condition for subjectivity activation

  3. "External" space of semantics: True meaning does not exist in internal logical relationships but in "external" space that cannot be completely homomorphized

  4. Grafting function of gamification: Achieving organic combination of subjectivity and models through behavioral simulation rather than preset algorithms

  5. Foundational status of subjectivity: Subjectivity is not a product of cognition but a basic attribute of the world; humans are just one example of activating this attribute

This discussion provides a completely new theoretical framework for constructing artificial intelligence models with genuine cognitive capabilities.