A Distributed Architecture Integrating Educational Philosophy and AI-Driven Learning Design: The PDP

Abstract

Purpose

Education in the age of AI faces an epistemic paradox: It can increasingly observe learning through data and analytics, yet it still struggles to understand learning as a process of meaning. While modern systems exhibit a degree of technical reflexivity, their feedback remains largely procedural and detached from interpretation.

Design/Approach/Methods

This article proposes the Parallel Distributed Processing (PDP)–Innovation, Creativity, and Entrepreneurship Education (ICEE) Learning System as a bridge from educational philosophy to design practice and computational modeling—a distributed architecture that enables education to learn about its own learning. Integrating the Human-in-the-Loop (HITL) principle with the philosophy of ICEE, the model redefines education as a reflexive and self-evolving system capable of transforming technical feedback into epistemic understanding.

Findings

The study develops this integration across five levels: (a) diagnosing the limits of current educational reflexivity, (b) articulating human–machine co-learning as systemic reflection, (c) operationalizing reflexivity through the Task Chain and Crystal System, (d) formalizing these dynamics within a PDP architecture, and (e) situating them within Panarchy theory to explain how education evolves as a complex adaptive system.

Originality/Value

By translating educational philosophy into computational and design logic, the PDP–ICEE Learning System offers a new epistemic form of education—one capable of learning from learning, integrating reflection, design, and evolution within a single self-observing framework.

Keywords

Human-in-the-loop learning learning architecture Panarchy and complex adaptive systems (CAS)parallel distributed processing (PDP)reflexive education system task chain

The Assessment Trap: When Education Stops Learning

Education has always aspired to make growth visible—to understand how learning unfolds and transforms the learner. From the dialogs of Socrates to the examinations of imperial China, this pursuit of visibility promised fairness, accountability, and improvement. Yet visibility has also become an assessment trap: The more we have tried to observe learning, the more we have replaced learning with what can be observed. In other words, the instruments of measurement replace the phenomena they were meant to observe.

From the Pursuit of Visibility to the Assessment Trap

Originally, education was an act of human encounter¹—a reflective dialogue between the learner and the world. Over time, as societies expanded, the need to evaluate and manage learning transformed this encounter into a system of observation. The invention of examinations, standards, and metrics made learning measurable and comparable.

Yet, as in quantum physics, the very act of observation changes the state of what is observed (Heisenberg, 1927). When learning is fixed into measurable form, its indeterminate qualities—curiosity, uncertainty, exploration—collapse into performance. What once lived as inquiry becomes a display of certainty. Education turns into a mirror that no longer reflects growth, but only what it can quantify.

This epistemic collapse parallels what physicists describe as the uncertainty principle: Observation alters the observed (Bohr, 1934). In education, each attempt to make learning more measurable restricts what can be learned (Biesta, 2009). The desire for precision produces blindness to emergence. What cannot be standardized becomes invisible, even though it may be the essence of learning itself.

When Evaluation Becomes the System

In theory, assessment was designed to serve learning; in practice, it has become the structure through which learning is organized. Curricula, pedagogies, and institutional goals all revolve around evaluative visibility—tests, rankings, and credentials. What began as an auxiliary function has evolved into the operating logic of the system itself. Education, once a generator of inquiry, now functions as a machine of verification.

This shift has produced a profound loss of feedback. Results flow outward, but meaning does not return. That is, assessment outcomes are produced and circulated, but they are not reintegrated into the system as interpreted feedback capable of reshaping how learning is understood, designed, or valued. Although modern education systems employ numerous feedback mechanisms—from standardized testing analytics to adaptive technologies—these mechanisms remain largely procedural. Without feedback loops that translate data into understanding, education struggles to learn from its own learning. It becomes, in cybernetic terms, an open circuit—signals exit, but none reenter to recalibrate the system.

Data Without Meaning

The early thinkers of systems theory—Wiener (1948), Ashby (1956), and Bateson (1972)—showed that systems persist by learning from their own errors. Education partially achieves this through technical systems of measurement and analytics, yet these mechanisms often aim at checking or confirming rather than understanding. Every act of measurement fixes what fluid was and every standard narrows what was possible.

Paradoxically, the advance of technologies has intensified this condition. Digital technology and AI have amplified this condition. AI can register and predict educational patterns of behavior with extraordinary precision—how students respond, how teachers intervene, how curricula adapt—but even its designers admit that AI cannot yet explain itself (Doshi-Velez & Kim, 2017; Gilpin et al., 2018). For education, this raises a deeper question: If our most advanced systems can observe learning yet cannot explain how learning happens, the problem is not technological but epistemological. It mirrors the same blindness within education itself—a system rich in kinds of “data” yet poor in understanding.

As with attempts to interpret neural networks—what researchers now call explainable AI—this illustrates the difficulty of making any complex system transparent to itself. The closer we look, the less we know why the system behaves as it does. That's saying that education risks mistaking information for insight. The challenge, therefore, is to transform procedural feedback into reflective understanding—to turn observation into interpretation, and measurement into reflection.

From Measurement to Meaning

To move beyond this impasse, education must extend its existing feedback loops beyond procedural optimization and restore the connection between seeing and understanding. It must design mechanisms that enable it to learn from its own evaluations. The key lies in reembedding human interpretation within the architecture of measurement (von Foerster, 1974).

The Human-in-the-Loop (HITL) principle offers this bridge. Originating in cybernetics, HITL refers to feedback structures in which human judgment is integrated into automated processes (Sheridan, 1992; Sheridan & Parasuraman, 2005).² Recontextualized in education, it becomes a principle of reflexivity: Humans provide meaning where machines perceive patterns. By linking algorithmic perception with human interpretation, the system transforms from one that merely evaluates learning to one that learns through evaluation.

This transition—from measurement to meaning—marks the reawakening of education's epistemic function. It prepares the ground for the following section, where the HITL principle converges with Innovation, Creativity, and Entrepreneurship Education (ICEE) philosophy to construct an ecology of colearning—a system capable of interpreting itself.

HITL and the ICEE Response

HITL: Restoring Reflexivity to the System

The preceding section concluded with a paradox: Education can observe learning but cannot understand it. What remains limited is the capacity for education to learn from its own learning—its reflexive processes remain fragmented and often technological rather than systemic.

The HITL principle, originally developed in cybernetics and later expanded in AI control systems, offers a pathway to restore reflexivity. In classical control theory, HITL denotes the inclusion of human judgment within automated feedback loops to ensure adaptive stability; recontextualized in education, it transforms a closed circuit of evaluation into an open process of colearning.

In practical terms, the HITL principle redefines the relationship between human and artificial agents: AI perceives patterns that exceed human scale, while humans provide the contextual meaning that exceeds algorithmic reach. The interaction between these two intelligences forms a new ecology of understanding, one in which feedback becomes formative rather than terminal. Each evaluative act—once a final judgment—now becomes a data point for reflection, inviting the system to learn from its own evaluative behavior. This reconfiguration is not a technical adjustment but a systemic reform.

The purpose of HITL in education is not to enhance efficiency but to recover interpretability: the ability of the system to see its own operations as processes of meaning. When human judgment and machine perception are continuously integrated, education ceases to perform learning and begins, instead, to learn from learning. HITL thus marks the turning point from observation to participation, from external assessment to internal evolution.

ICEE as the Systemic Response

To transform feedback into meaning, we need an epistemic foundation that integrates reflection, value, and agency. This foundation is provided by ICEE—a philosophy and system of practice that redefines learning as a self-evolving process within a complex adaptive ecology (Zhao & Zhong, 2024).

Through its three core principles—Personalizable Learning, Problem Finding and Solving, and Human Interdependence—ICEE transforms education from a hierarchy of evaluation into an ecology of colearning.

Personalizable Learning—the respect for learners’ interests, pace, and agency, recognizing that learning unfolds along diverse trajectories shaped by individual purpose.

Problem Finding and Solving—the cultivation of learners’ capacity to discover, frame, and address meaningful (related to real-life, related to real requirements) problems, transforming learning into inquiry rather than compliance.

Human Interdependence—the realization that learning attains meaning only through collaboration and value creation for others, forming the ethical and social ecology of education.

Together, these principles transform education into a reflexive ecosystem—one that can sense, interpret, and reorganize itself through the participation of all its actors, human and nonhuman alike. As articulated in Zhao and Zhong (2024), ICEE thus provides both a philosophical anchor and a systemic mechanism through which education can achieve what we call learning from learning.

Personalizable Learning: Sensitivity to Difference

In traditional assessment systems, standardization ensures comparability at the cost of individuality. ICEE reverses this equation by treating diversity as the very condition of learning. Personalization opens the system to variation, which means that personalizable learning enables the system to perceive variation as information. Each learner's trajectory becomes a feedback signal through which education calibrates itself. The greater the diversity of responses, the richer the system's capacity to learn from its own learning (Zhao, 2012, 2016).

In this sense, personalization is not an individual privilege but a systemic sensitivity—a way for education to remain adaptive rather than reactive.

Problem Finding and Solving: Reflection Through Uncertainty

Problem finding and solving convert uncertainty into understanding. Conventional assessment isolates correctness: ICEE transforms error into feedback—by framing learning as inquiry, it redefines what counts as success. The process of identifying, framing, and addressing problems generates the very feedback that fuels systemic adaptation (Bateson, 1972).

Human Interdependence: Ethics as Feedback Ecology

The third principle, Human Interdependence, ensures that reflection is not confined to cognition but extends to value and relation—learning gains meaning only when it contributes to others. ICEE posits that understanding is inherently social: Feedback must circulate not only between individuals and machines but also among communities, cultures, and generations. This is the ethical dimension of reflexivity—what Luhmann (1995) might call the “self-reference of meaning within social systems.” Human interdependence transforms assessment from an act of judgment into an act of participation, embedding ethics directly into the feedback loop.

ICEE and the Reconstruction of Educational Feedback

The three core principles of ICEE constitute a systemic response to the loss of feedback in education: Where assessment systems isolate learners, ICEE reconnects them; where evaluation extracts data, ICEE generates meaning.

The framework thus bridges technical reflexivity (HITL) and educational reflexivity (ICEE): The former supplies the mechanism of feedback and the latter supplies the logic of meaning.

Layered Feedback Within the HITL System

The ICEE framework operationalizes the HITL principle through layered feedback among learners, teachers (human facilitators), and AI systems. Learning no longer follows a linear trajectory from input to output but circulates through continuous perception and reflection.

The Micro-loop: Learner and AI

At the micro-level, the learner interacts directly with AI tools—adaptive prompts, multimodal portfolios, and reflective analytics. Each interaction produces data that the AI interprets as a representation of learning. The learner, in turn, interprets the AI's interpretation, refining both understanding and behavior. What emerges is a conversation of cognition—the smallest unit of systemic learning.

The Meso-Loop: Teacher–Learner–AI Ensemble

At the meso-level, educators (human facilitators) orchestrate the ensemble of human and machine participants. They curate contexts, interpret data dashboards, and transform algorithmic insights into pedagogical decisions. The teacher's role shifts from content delivery to epistemic mediation—helping both learners and AI models “learn how to learn” within shared ethical and cultural frames.

The Macro-Loop: Education as a Reflective System

Aggregated across multilayers, these micro- and meso-loops form the macro-level learning of the education system itself. Patterns extracted from distributed learning activities reveal where systems succeed or stagnate, enabling design feedback into curriculum (experience design), policy, and technology. Education thus becomes a panarchic structure—a nested ecology of adaptation, where change at one level triggers renewal or stabilization at others.

Ethics as Reflexive Literacy

Embedding humans in the loop also embeds ethics in learning. Transparency, agency, and value alignment cannot be external regulations; they must be literacies practiced through interaction. Learners must understand how their data signifies them, how AI perceives them, and how their choices recalibrate the system. Within ICEE, ethics is redefined as reflexive literacy—the ability to recognize and reshape one's participation in coevolution.

Toward the PDP–ICEE Learning Architecture

ICEE's triadic logic—personalization, creativity, entrepreneurship—naturally extends into a computational model of Parallel Distributed Processing (PDP; McClelland & Rumelhart, 1986). In such a model, each learner functions as an adaptive node and the collective network evolves through distributed feedback.

Education as a Parallel Distributed System

Parallelism: Learning occurs simultaneously across many human and AI nodes.

Distribution: Cognition and control are decentralized; intelligence emerges from their interaction.

Processing: Feedback continuously transforms raw data into meaning.

The PDP–ICEE integration translates educational philosophical principles into computational logic. ICEE provides the semantic grammar—what learning means and why it matters. PDP provides the structural syntax—how learning processes self-organize and communicate.

Systemic Understanding as Computable Reflexivity³

Rather than viewing learning as a linear transfer of knowledge, the PDP–ICEE framework conceives it as a recursive interaction between doing and understanding, between micro-level experience and macro-level adaptation.

Within this architecture, HITL serves as the dynamic interface between human interpretation and machine perception. AI detects patterns and regularities within distributed learning data, while humans—teachers, learners, and facilitators—assign meaning and ethical direction to these patterns. Through this continuous exchange, reflexivity becomes computable: Each act of evaluation generates feedback not only for individual improvement but also for system-wide learning.

The PDP–ICEE Learning Architecture thus connects two complementary dimensions: the technical reflexivity introduced by HITL and the epistemic reflexivity articulated through ICEE. It models education as a self-observing, self-organizing network—an ecosystem capable of transforming assessment from measurement into interpretation, and interpretation into systemic renewal.

YEEAI⁴ as an Epistemic Infrastructure

The YEEAI system translates the combined logic of HITL and ICEE into a functional infrastructure for educational reflexivity.

In practice, each interaction among learners, facilitators, and AI agents creates a dual record of learning: a pedagogical experience and an epistemic signal. These signals enable the system to recognize how learning unfolds and to adjust its own structures accordingly. The loop between data and meaning thus becomes computable without becoming deterministic.

Philosophically, YEEAI serves as an epistemic infrastructure, where AI seeks optimization while YEEAI seeks interpretation. Its architecture preserves the human role within automation: AI detects patterns across distributed learning contexts, while humans provide judgment and ethical direction. This human–machine symbiosis ensures that computation serves meaning, not the reverse.

In conceptual terms, YEEAI extends the PDP–ICEE architecture from the theoretical to the operational level. It renders the three principles of ICEE traceable and adaptive within a computational environment. Each feedback instance becomes a node of reflexivity, linking individual learning events to system-wide evolution. In this way, YEEAI embodies what the previous sections have called learning from learning: a continuous dialog between human and artificial intelligences that sustains education's capacity for renewal.

The following section translates this conceptual infrastructure into design practice, detailing how YEEAI's Task Chain and Crystal System realize the principles outlined above as operational mechanisms within a parallel distributed learning architecture.

System Design: The YEEAI Architecture and Its Core Mechanisms

While the preceding sections conceptualize education as a system that seeks to understand learning, the following sections describe how this systemic purpose is operationalized through the PDP–ICEE architecture. At the level of individual learners, each act of learning—when tracked, interpreted, and connected through feedback—constitutes a micro-operation of the system's self-observation. In this way, the PDP–ICEE learning system links the micro and the macro: The learner's feedback loops become the system's pathways of understanding, and personalizable learning becomes the operational basis for education's systemic self-evolution.

System Overview: From Educational Principles to Computational Architecture

The YEEAI system translates the three core principles (Zhao, 2012) of ICEE—personalizable learning, problem finding and solving, and human interdependence—into a computationally interpretable and educationally meaningful architecture. The YEEAI design does not merely implement a technical architecture; it instantiates the reflexive principles outlined in the previous sections. It extends existing forms of technical reflexivity into a systemic capacity for educational self-understanding. Grounded in Panarchy theory (Gunderson & Holling, 2002; Holling, 1973; Zhao & Zhong, 2024), the architecture views schools as semiopen subecosystems whose transformability depends on the interplay between curriculum, pedagogy, assessment, and environment (Zhong & Zhao, 2025). Within this framework, YEEAI serves as the technical infrastructure that makes such cross-scale interaction traceable and adaptive.

Structurally, the system comprises five interconnected layers—each corresponding to a level of semantic, structural, and computational operation (Figure 1):

Conceptual Layer—Defines the semantic universe of ICEE, encoding key concepts such as personalization, agency, and reflection into a knowledge graph of educational principles.

Ontology Layer—Maps relationships among concepts to generate formal linkages (“guides → Task Type,” “defines → Capability Set”) and prepare for entity-level representation.

Entity Layer—Implements these concepts as observable learning structures—Task Chains and Crystals—that organize actions and record cognitive change.

Algorithmic Layer—Hosts adaptive models (PDP distributed processing, Bayesian ICEE loop, kernel-based mapping) that analyze learning patterns and generate feedback. (Implementation details are proprietary.)

Application Layer—Provides user interfaces (Task Cards, Growth Maps, Feedback Dashboards) that enable learners, teachers, or other learning facilitators, and AI agents to collaborate within HITL cycles.

Figure 1.

The triple-alignment architecture of the PDP–ICEE learning system.

Together, these layers form a double translation mechanism: from educational theory to computational logic, and back from data to educational insight. YEEAI thus bridges what Zhao and Zhong (2024) call “anchor elements within adaptive cycles” of education and what the forthcoming “Double-Helix Logic of Curriculum” paper (Zhong & Zhao, 2026) describes as the integration of universality and personalization within dynamic learning ecosystems. Its goal is to make education both ecologically coherent and mathematically tractable—capable of being understood through human values while analyzed through computational models of learning behavior.

The Task Chain: Organizing Personalizable Learning as Distributed Action

Definition and Core Structure

The Task Chain serves as the structural and experiential backbone of YEEAI, transforming educational objectives into personalizable, distributed sequences of learner actions. The Task Chain, therefore, serves as a material expression of reflexive feedback—transforming procedural reflection into epistemic reflection through design.

Building on ICEE's first two principles—personalizable learning and problem finding and solving (Zhao & Zhong, 2024; Zhong & Zhao, 2025)—the Task Chain provides the operational framework through which individuality, curiosity, and competence development become visible and analyzable.

Each Task Chain begins with a Supra Task Card (STC), which acts not only as a pedagogical prompt but also as a personalizable experiential profile within the YEEAI system. The STC integrates the learner's background information, self-perceived interests, hobbies, and prior experiences—forming the first layer of personalization: the experiential layer. By connecting learners’ identities and motivations with ICEE's problem-based design, the STC provides an authentic entry point into the ICEE learning cycle and embodies the principle that personalization starts from lived experience, not algorithmic segmentation. This element is unique to YEEAI: Personalization is constructed from the learner outward, allowing technology to support, rather than prescribe, learning trajectories.

Within each STC, YEEAI structures the learning process through a four-level hierarchy (Table 1), these four elements function as structural roles within the learning architecture, rather than as content-specific task types.

Table 1.

Structure of the Task Chain in the YEEAI Learning System (STC–MTN–STL–STN).

Level	Description	Personalization Focus	Function
Supra Task Card	The initial entry point that integrates students’ background information, prior experiences, interests, and motivations. It provides the experiential foundation for generating learning tasks.	Experiential personalization: connects learners’ identities and past experiences with new learning goals.	Initiates personalized task generation and defines the learner's experiential profile within the YEEAI system.
Main Task Node（MTN）	Represents ICEE's core competence framework. Each node integrates specific competence dimensions and subdimensions (including indicators) with representational anchors and latent mappings (kernel-based feature space). Although structurally predefined, each node is dynamically adjustable—its competence dimensions may vary across learner stages or shift among subdimensions within the same node.	Conceptual personalization: competence-guided and meaning-centered learning.	Anchors conceptual understanding; connects ICEE competence framework to AI's representational layer for feature recognition and latent-space mapping.
Sub Task Line	The self-generated path created by the learner in response to their own interpretation of each MTN. It is constructed through personally preferred strategies, interests, and habitual approaches, reflecting how learners make sense of competence goals through action.	Path personalization: strategy-, preference-, and interpretation-based learning pathways.	Facilitates feedback integration between AI and learner. Captures learner-generated trajectories; records how individual understanding evolves into action, providing temporal data for adaptive feedback.
Sub Task Node	The smallest behavioral unit that records concrete learning actions (reflection, feedback, collaboration, exploration, creation, or AI-assisted generation/recommendation).	Behavioral personalization : real-time expression and adaptive experimentation.	Records fine-grained actions; supports crystal generation, tool recommendation, and real-time AI feedback.

Note. ICEE = Innovation, Creativity, and Entrepreneurship Education; STC = Supra Task Card; STL = Sub Task Line; STN = Sub Task Node.

References to representation, latent space, kernel mapping, and crystal generation in this section function as conceptual bridges linking educational design to the computational mechanisms that will be elaborated in Section 4.

They are introduced here to illustrate how educational structures—It is important to note that these references belong to the conceptual level of system design, not to the computational implementation level.

This expanded structure articulates three interacting levels of personalization—experience, path, and behavior—while embedding ICEE competence as the conceptual scaffold that connects them. It shows how learners’ experiences evolve into strategies and then into actions, creating a longitudinal record of personalizable development. The kernel-based representation within Main Task Nodes (MTNs) ensures that this process remains computationally interpretable without reducing its educational meaning. Further implementation details are provided in the section “The PDP–ICEE Learning Architecture.”

Through this multilevel structure, YEEAI maps the learner's experiential, procedural, and behavioral trajectories into a unified temporal network. Each Sub Task Node (STN) serves as an atomic trace of cognition-in-action, while the Task Chain as a whole becomes a living model of learner autonomy and system adaptivity.

Educational Logic and System Operation

From Structure to Living Process

The Task Chain provides the structural grammar of YEEAI. Yet, structure alone does not constitute learning. Educationally, the Task Chain becomes meaningful only when it is activated through the dynamic interactions among the learner, the learning facilitator, and the AI system.

This section explains how the YEEAI architecture operationalizes that process—how STCs evolve into experiential learning cycles, how MTNs adapt to competence development, how Sub Task Lines (STLs) are self-generated through interpretation, and how STNs anchor action, reflection, and feedback. In this living process, the Task Chain functions as a recursive, self-organizing loop of experience, understanding, and action.

Definition of the Learning Facilitator

In this article, the term learning facilitator refers to any human or nonhuman participant who supports the learner's development by curating experiences, mediating resources, and sustaining reflection. It includes teachers in schools, mentors in communities, peers in collaborative networks, or AI agents. YEEAI does not replace schools or teachers; rather, it augments their capacity to sustain personalization and reflection across contexts. Schools remain the most organized and scalable learning environments, but YEEAI is designed to operate across and beyond them, embedding facilitation into the broader learning ecology.

System Operation: From Supra task to Learning Loop

A learning cycle begins with the STC , which presents a contextual challenge derived from both the ICEE competence framework and the learner's personal profile. The STC integrates data such as prior learning experiences, interests, and motivational cues to form a personalizable experiential entry point. It frames learning not as a fixed assignment but as an open inquiry space—a bounded environment in which learner define their own approaches to the challenge.

Within this inquiry space, the MTN provides a competence-oriented scaffold that links conceptual meaning to learner action. While its indicators and subdimensions are predefined within the ICEE Competence Framework, they remain dynamically adjustable across learner stages or focus dimensions.

For example, Creativity in the ICEE framework refers to the learner's capability to transform ideas into purposeful actions under contextual constraints—to recognize, negotiate, and reconstruct limitations such as time, resources, and social expectations in order to realize an optimal and meaningful outcome. It represents a process of adaptive realization, in which creativity emerges not from freedom from constraints but from mastery within them. As learners develop, their creativity evolves from recognizing constraints to reconstructing them, aligning with ICEE's principles of personalization, problem finding and solving, and human interdependence. This adaptive flexibility ensures that every MTN can evolve with the learner's developmental trajectory.

The STL emerges as the learner's self-generated path, constructed from their interpretation of the MTN. Each learner reads the competence challenge differently—through personal strategies, interests, and habitual approaches—and reconstructs it into a concrete plan of action. Thus, STLs become expressions of agency in interpretation. The YEEAI system observes these self-generated trajectories, offering contextual prompts and recommending tools or collaborators that expand the learner's exploratory range, rather than enforcing convergence.

At the most granular level, STNs record concrete actions—creation, exploration, feedback, reflection, or collaboration—each becoming a behavioral trace within the YEEAI system. These traces are continuously logged, timestamped, and semantically tagged, forming the raw material for subsequent crystal generation (see the section “The Crystal System: Tracing Cognitive Growth Through Action”).

Together, these four levels transform abstract educational intentions into observable and analyzable patterns of action.

Role Interaction: Learner, Facilitator, and AI

Within YEEAI, learning unfolds as a triadic interaction loop where human facilitation and AI adaptation coevolve.

The Learner Acts as the Initiator of Meaning

They interpret challenges, generate STLs, and transform competence goals into actions—embodying ICEE's principle of creative agency, that is, the learner's capacity to create meaning and transform conditions through action rather than merely respond to them.⁵

The Learning Facilitator Functions as the Sense-Maker and Ecological Curator

In school contexts, this includes teachers guiding inquiry and providing formative feedback; in broader learning ecosystems, it may involve mentors, peers, or domain experts helping learners sustain curiosity and reflection. Facilitators do not prescribe learning but maintain its adaptive balance—what Panarchy theory (Gunderson & Holling, 2002; Holling, 1973) describes as steering the system within “safe boundaries of transformation.”

The AI System Acts as a Cognitive Partner that Amplifies Both Learner and Facilitator Capacities

It recognizes behavioral patterns across STNs, visualizes trajectories, recommends resources, and generates reflective feedback. In doing so, AI becomes not an evaluator but a mirror and catalyst—helping learners see their own thinking and facilitators perceive the learner's evolving autonomy.

This triadic loop ensures that personalization remains relational. Autonomy arises not from the withdrawal of guidance but from the learner's codesign of meaning with both human and artificial facilitators.

Educational Meaning

The educational logic of the Task Chain redefines the classic tension between structure and freedom. Instead of confining learning within a linear sequence, YEEAI structures openness itself: a system designed to sustain exploration while maintaining conceptual coherence. In ICEE's view, autonomy is not independence but coevolution—the capacity to grow through interaction and reflection. Each learner's Task Chain thus becomes a living ecosystem of agency, diversity, and growth, where human and artificial facilitators collaboratively nurture personalizable yet socially connected learning (Zhao, 2016).

While the Task Chain organizes how learning actions unfold, it does not yet explain how these actions accumulate into recognizable cognitive structures. The next section introduces the Crystal System, which traces how these distributed actions evolve into cognitive and metacognitive patterns—turning behavior into understanding, and understanding into reflective growth.

The Task Chain shows how learning unfolds in motion—each action leaving traces of thought and intention. It reveals an adaptive system of meanings, always in the making.

Yet every system hides a deeper rhythm: action folding into understanding, and understanding unfolding again as reflection. The next section turns to this rhythm—the emergence of cognitive “crystals,” through which the learning process remembers itself: the Crystal System, tracing how distributed actions evolve into cognitive and metacognitive patterns.

Application and Examples: The Task Chain in Real Learning Ecosystems

The YEEAI architecture was first introduced in pilot implementations that explored how ICEE's principles could be enacted in authentic learning environments. These experiences demonstrated that the Task Chain was not merely a digital structure but a living framework capable of adapting to different educational ecologies and learner stages.

Conceptual Validation: Hongfan (HF) Subsystem

The HF campus of Chongqing No. 8 Secondary School served as the first experimental ecosystem for implementing ICEE principles at the middle-school level.⁶ Long before the YEEAI platform was developed, HF's teachers and students began applying what would later be recognized as the task-chain mindset—organizing learning around authentic problems, iterative reflection, and competence-driven action. These early experiments demonstrated that the structure of an STC (defining the experiential entry point), an MTN (framing the competence focus), and learner-generated STLs (representing individual interpretations and paths of action) could operate effectively even without a digital system. Through projects conducted in both YEE GLocal and YEE Action, students learned to connect ideas to action, reflect upon constraints, and document their progress as evolving evidence of growth.

In retrospect, HF provided a conceptual rehearsal of the Task Chain logic. It validated the educational assumptions that later shaped YEEAI: that meaningful personalization must begin with learners’ lived experiences; that competence growth emerges through self-directed pathways rather than prescribed steps; and that reflection is most powerful when embedded within the flow of action. At this stage, the Task Chain existed as a pedagogical design rather than a completed technological system—a living prototype of the systemic principles that YEEAI would later formalize.

Adaptive Configurations: From the Eight Steps to the Six Phrases

The adaptability of the Task Chain became evident when ICEE extended from middle to high school programs.

In the GLocal projects for younger students, learning followed an Eight-Step model of entrepreneurial thinking—a concrete, action-oriented scaffold emphasizing collaboration and procedural creativity. At the high-school level, My Education, My Way (MEMW) introduced a Six-Phrase framework, a conceptual refinement that aligns directly with ICEE competence dimensions and supports more abstract, self-directed inquiry. This shift from steps to phrases marks a cognitive elevation—from learning through structured doing to learning through reflective design.

While the procedural logic of the Task Chain (STC → MTN → STL → STN) remains consistent, its framing evolves: Group-centered tasks become personally driven inquiries, external themes give way to passion-based questions, and AI transitions from supportive tool to colearner.

These variations confirm the Task Chain's capacity to scale across developmental stages without losing its conceptual coherence.

Systemic Realization: MEMW

The MEMW program represents the first full implementation of the Task Chain and YEEAI system. Each learning journey begins with an AI-assisted STC, contextualized by the learner's background and interests. Learners then navigate a fixed set of MTNs corresponding to the Six Phrases, each containing key elements mapped to competence indicators. From these nodes, students construct STLs—personalizable routes through inquiry and production. Every STN records an output, interaction, or reflection, generating the behavioral data that feeds YEEAI's learning analytics. Facilitators use these visualizations to guide reflection, while AI provides adaptive prompts that help learners expand their exploration rather than converge prematurely. Over time, the cycle produces visible learning trajectories: Students iteratively refine problems, test solutions, and trace their own growth across competence domains.

Across HF and MEMW, the Task Chain revealed itself as a meta-structure that binds agency, competence, and reflection. Its adaptability allows ICEE programs to maintain philosophical continuity while adjusting pedagogical texture. In practice, it enables a gradual evolution—from collaborative, theme-based learning to autonomous, passion-driven inquiry—without fragmenting the learner's developmental narrative. Facilitators, whether human or AI, curate the learning ecology rather than dictate its path, ensuring that personalization remains relational and purposeful.

Through these implementations, learning ceases to be a sequence of tasks and becomes a living architecture of meaning. With every STN, a trace is created; with every cycle, traces begin to condense—each an echo of learning's unfolding memory.

The next section turns to this condensation—the Crystal System—where distributed actions take shape as cognitive and metacognitive patterns through which learning remembers itself.

Systemic Significance

The Task Chain represents more than an operational mechanism within YEEAI; it embodies a systemic grammar through which learning, technology, and pedagogy become mutually intelligible. In traditional schooling, the design of learning is often separated from its observation—teachers plan, students act, and assessment comes after. The Task Chain collapses this distance: Design, action, and feedback coexist within the same temporal loop. Each cycle transforms educational intention into traceable behavior and, in turn, converts behavior into feedback that reshapes intention. In this recursive movement, the system itself becomes a participant in learning—an ecology that learns about learning.

As a structural bridge between educational ideals and computational processes, the Task Chain demonstrates how ICEE's principles—personalization, problem finding and solving, and human interdependence—can be rendered operational without losing their human core. It gives form to personalization not as algorithmic prediction but as relational design; it reframes problem finding as a generative dialogue among learners, facilitators, and AI; and it situates human interdependence within a dynamic ecology where meaning circulates across scales (Zhong & Zhao, 2025). Through this architecture, YEEAI does not digitize education but redefines its logic: Learning becomes both observable and self-organizing, both individual and systemic.

The structural rhythm of the Task Chain—design, action, reflection—sets the stage for a deeper question: How do these actions, once recorded and interlinked, begin to take cognitive shape?

The next section turns to this process of condensation—the Crystal System—where the traces of learning solidify into recognizable patterns of understanding and reflective growth.

The Crystal System: Tracing Cognitive Growth Through Action

The Philosophical Grounding: From Action to Reflection

The Crystal System is not an extension of the Task Chain but its reverse abstraction mapping. Whereas the Task Chain unfolds learning from educational philosophy to concrete actions/activities—from abstract idea to lived experience—the Crystal System moves in the opposite direction: It abstracts learning from action to understanding, from experience to cognition.

The two systems mirror one another in both function and philosophy. The Task Chain constructs personalizable learning experiences through YEEAI's design intelligence, while the Crystal System condenses those experiences into visible structures of meaning.

Together, they form a complete cycle of personalization—design and reflection, doing and understanding, outward exploration and inward crystallization.

We use the word “crystal” as an organizing metaphor to bridge education and computation, meaning and measurement, which captures how learning takes form—gradually, uniquely, and through the interaction of many small elements.

In ecology, “crystal” evokes self-organization within complex systems; in physics, the alignment of structure through repetition and interaction; and in computational learning systems, the stabilization of distributed representations through iterative feedback. Across these fields, a crystal marks the moment when variation finds order—a process that can be both observed and modeled. This duality makes it ideal for YEEAI: It represents how individual learning experiences accumulate into discernible structures of understanding that are interpretable both educationally and computationally.

Thus, the Crystal System is not about measuring performance but about mapping formation—capturing how learning, through countless micro-actions, evolves into recognizable cognitive structures.

Typology of Crystals

In the YEEAI framework, three interrelated types of Crystals operate at different levels of abstraction:

Here, the term “Crystal” is used as a conceptual label rather than a metaphor, referring to stabilized structures of learning meaning at different levels of abstraction.

Event Crystals

Generated whenever an STN is completed, each Event Crystal functions as a temporal trace of experience.

It includes a time-stamped record of actions, resources, and short reflections, capturing when and in what context a learning behavior occurred.

These traces serve as the raw material for later pattern recognition—micro-level evidence through which both learners and the system can revisit what actually happened.

Pattern Crystals

Formed when multiple Event Crystals resonate across time or context, revealing recurring ways of thinking or acting.

The YEEAI system identifies such patterns algorithmically, while learners recognize them through reflection—turning computational clustering into educational insight.

A Pattern Crystal thus represents a cogenerated awareness: a structure recognized by the system and made meaningful by the learner.

Cognitive Crystals

Emerge when different Pattern Crystals begin to interact, linking strategies, domains, or competencies into a dynamic conceptual network.

A Cognitive Crystal is not a fixed representation of understanding but a living schema—a structure that continues to evolve as new experiences reshape prior meaning.

It reflects how learning reorganizes itself over time, translating distributed actions into coherent, yet adaptable, understanding.

Together, these three levels reveal how the Crystal System transforms experience into structure:

Event Crystals record what and when learning occurs.

Pattern Crystals interpret how it tends to occur.

Cognitive Crystals illuminate why it matters and how it evolves.

Developmental Logic

Each crystal is formed through the ongoing dialogue between the learner and the YEEAI system: The Crystal System records what happens, while the learner interprets those traces and builds new meaning from them.

Through this interaction, learning becomes both lived and remembered.

Accumulation—Recording the Micro-Moments of Learning

Every STN produces an Event Crystal, a small record of experience that includes actions taken, resources used, and time-stamped context. At this level, learning is captured as a concrete activity—what the learner actually does.

These micro-moments do not yet show clear meaning, but they provide the raw material for feedback and reflection, much like individual experiences waiting to be connected.

Resonance—Finding Patterns Across Experiences

As Event Crystals accumulate, similarities and contrasts begin to appear. Certain behaviors, strategies, or emotional responses repeat across contexts. YEEAI identifies these recurrences and visualizes them as Pattern Crystals, while learners themselves start to notice their own preferred ways of learning.

This stage represents the first awareness of learning habits/approaches—a recognition of how one learns, not just what one learns.

Abstraction—From Strategy to Concept

When patterns extend beyond multi-STLs or a single MTN, they crystallize into Cognitive Crystals. Learners connect insights across projects, reexamine assumptions, and begin to reconstruct knowledge in new contexts.

Here, understanding and thinking become generative: Learners start to design differently because they have seen how their thinking forms and reforms. At this point, YEEAI does not direct the process but supports it by making cognitive connections visible.

Reorganization—Continuous Adaptation and Growth

Learning does not stop once a cognitive structure forms. New experiences may confirm, challenge, or expand prior understandings. Cognitive Crystals can therefore merge, split, or transform, reflecting the dynamic nature of knowledge itself.

This reorganization keeps the system alive; it allows the learner's understanding to stay open, adaptive, and responsive to change.

Together, these four phases show that the Crystal System is not a fixed archive of performance but a living ecology of learning. It traces how experience becomes pattern, and how pattern becomes cognition, without freezing individuality into metrics. For facilitators, these layers provide a window into each learner's evolving competence; for learners, they offer a mirror through which growth becomes visible and actionable.

In this developmental loop, YEEAI enacts ICEE's philosophy of personalization as both innovation and creativity⁷—supporting learners while allowing them to see, understand, and ultimately shape their own learning paths.

At the computational level, crystals and task events are represented through learned embeddings, relational graphs, or probabilistic descriptors, allowing structural aggregation without reducing educational meaning to single metrics.

Design as Reflexive Practice

The development of YEEAI represents more than a technical implementation; it is an experiment in reflexive design. Each component of the system—its layered architecture, the Task Chain, and the Crystal System—was conceived as a way to materialize the principles of HITL and ICEE. In this sense, YEEAI functions simultaneously as a technological infrastructure and as a research instrument: It tests whether education can observe, interpret, and reorganize its own learning through design.

Reflexivity within YEEAI operates at multiple levels. Technically, the system embeds human interpretation within automated processes; educationally, it allows learners and facilitators to perceive their learning as part of a collective feedback loop. The Task Chain embodies the generative dimension of this process—transforming individual actions into shared structures of inquiry—while the Crystal System enacts its reflective dimension, tracing how meaning stabilizes through feedback. Together, they instantiate what could be called design reflexivity: The system not only enables learning but also learns from the way learning is designed and enacted.

From a research perspective, YEEAI demonstrates how educational design can become an epistemic practice. It blurs the boundary between observation and participation, between technology and pedagogy, positioning design itself as a mode of systemic learning. In doing so, YEEAI extends the idea of technical reflexivity into epistemic reflexivity—where education, through its own artifacts and architectures, begins to understand the conditions of its understanding. This reflexive stance prepares the ground for the next section, which formalizes these dynamics in the PDP–ICEE Learning Architecture, modeling how design, feedback, and reflection operate as one continuous system of learning from learning.

The PDP–ICEE Learning Architecture

The previous section introduced the dual mechanism of learning within YEEAI: the Task Chain, representing the generative unfolding of learning actions, and the Crystal System, representing the reflective consolidation of understanding. These two mechanisms reveal that learning is neither linear nor isolated—it is a continual negotiation between doing and understanding, between the forward motion of action and the backward mapping of meaning.

Building on this foundation, the section “The PDP–ICEE Learning Architecture” formalizes these dynamics into a computational–educational architecture grounded in the principles of PDP. It views education as a parallel distributed learning system, where individual learners function as adaptive nodes evolving through time, and collective intelligence emerges from the shared reduction of uncertainty.

The aim of this section is not to show that the ICEE learning process can be described in computational terms without losing its educational meaning.

The PDP–ICEE model thus connects the behavioral dynamics of the Task Chain with the cognitive abstraction of the Crystal System, establishing a unified architecture for understanding how personalizable, reflective, and coevolutionary learning unfolds in time.

Theoretical Grounding of the PDP–ICEE Learning Architecture

While the section “System Design: The YEEAI Architecture and Its Core Mechanisms” outlined the architectural realization of YEEAI, the present section theorizes its logic within the broader field of reflexive systems. The theoretical foundation of the PDP–ICEE Learning Architecture lies at the intersection of connectionist cognition, ecological systems thinking, and Panarchy-based educational theory. Together, these perspectives construct a coherent view of learning as a distributed, adaptive, and recursively self-organizing process—a process that renews itself through cycles of feedback, reflection, and reconfiguration.

From Connectionism to Distributed Cognition

PDP, first articulated by McClelland and Rumelhart (1986), reframed cognition not as the manipulation of symbols but as the emergent pattern of activity across interconnected units. In a PDP network, meaning arises from the relations among nodes rather than the content of any single node. Learning occurs through gradual adjustments of these connections—a tuning process shaped by experience and feedback. In educational terms, this describes how understanding grows not from isolated knowledge but from dynamic relations among learners, tasks, and contexts.

In this article, PDP is adopted as an architectural and organizational principle rather than a specific neural implementation. Computational references serve to formalize system dynamics, not to prescribe algorithmic realization.

Here, “learning” does not refer to machine learning or algorithmic optimization. Rather, it denotes the human reorganization of meaning through interaction and reflection. The “adjustment of connection weights” represents shifts in the strength of relationships—among ideas, people, and experiences—through which meaning continuously evolves.

In this sense, the PDP–ICEE framework treats education as a humanly distributed intelligence system: one that learns by reflecting on its own interactions and reshaping the patterns that connect thought, emotion, and action.

Ecological Feedback and Nested Adaptation

While PDP explains how distributed adjustments occur, ecological and Panarchy theories explain why they stabilize, collapse, and regenerate. In education, this means that every level—tasks, classrooms, schools, and even cultural systems—follows its own adaptive cycle of growth, tension, and renewal. Learning at one level can either reinforce or disturb another, as individual changes accumulate into collective transformations.

This ecological framing extends PDP beyond the neural metaphor by situating learning within cross-scale feedback loops:

Micro-level feedback, which captures immediate adjustments within a learning activity—how a student modifies a response or rethinks an approach.

Meso-level feedback, which captures reflective reinterpretations that reshape teaching goals, group dynamics, or assessment criteria.

Macro-level feedback, which captures longer-term transformations that redefine the system's structure—new pedagogical norms, shared values, or institutional practices.

Through this lens, the ICEE model operates as a multilevel learning network. Each reflection, interaction, or innovation becomes a local update that ripples outward, gradually reorganizing the system around emerging patterns of shared understanding.

Micro-level feedback captures moment-to-moment adjustments within a task;

Meso-level feedback captures reflective reinterpretations that reshape goals or strategies;

Macro-level feedback captures transformations that redefine the system's structure—new ways of seeing, valuing, or organizing learning.

Through this lens, the ICEE model operates as a multilevel PDP network: Each reflection, interaction, or innovation is a local update that ripples through the system, rebalancing it around newly shared meanings.

Time, Uncertainty, and Bayesian Updating

Learning in this architecture unfolds over time not as repetition, but as Bayesian adaptation—the continual recalibration of beliefs and strategies based on new evidence. Each learning act updates the learner's internal probability model of “what works” and “why,” reducing uncertainty through comparison between predicted and actual outcomes. This process—known as time-difference learning—anchors the PDP–ICEE system in temporal logic. It allows learning to be expressed as a dynamic trajectory:

Δ W_{t} = η (E_{t} - {\hat{E}}_{t})

where ΔW_t denotes the change in connection weight (learning adjustment), E_t is the experience signal, Ê_t the expected outcome, and η the adaptive rate of learning.

In educational terms, this equation represents the gap between expectation and experience—the very space where reflection and growth occur.

Thus, learning is not merely the accumulation of knowledge but the system's ongoing dialog with its own uncertainty.

Within the PDP–ICEE framework, such adaptation takes place simultaneously at the individual and collective levels, linking cognition, interaction, and reflection within one evolving process.

Computational Model of Learning Dynamics

The PDP–ICEE architecture translates the philosophical logic of ICEE into a distributed computational structure that connects behavior, cognition, and reflection. It models education not as a linear cause–effect sequence but as a dynamic learning ecosystem where forward and backward flows of information continuously reshape one another across scales of activity.

The Task Chain: Forward Propagation of Experience

The Task Chain represents the behavioral dimension of learning—the outward movement of engagement, exploration, and action. Each task functions as a node within a wider network of learning events, generating data about how a learner interacts with challenges, peers, and tools (including AI). These activation signals propagate forward through the educational system, influencing how subsequent experiences are encountered and interpreted.

The strength of these signals corresponds to the depth of learner engagement: the more active the participation, the stronger the influence on future nodes.

Over time, these accumulative activations create recognizable learning trajectories—pathways that are both personal and systemic. What emerges is not a single sequence of tasks, but a field of interdependent adjustments, where local actions gradually shape collective patterns of practice and understanding.

The Crystal System: Backward Abstraction of Meaning

Complementing the Task Chain's forward motion, the Crystal System performs the reverse process—backward abstraction. It captures how dispersed experiences consolidate into understanding by mapping patterns across learning events.

Computationally, this resembles back-propagation in neural networks; educationally, it parallels reflection: the mind's effort to reconcile outcomes with intentions, to recognize patterns in diversity.

Each “crystal” represents a stabilized form of insight—a configuration of meaning that recurs and endures. As these crystals accumulate, they form a multilayered model of competence and worldview that guides future anticipation and choice.

Through this process, the learning system gradually stabilizes its own dynamics, allowing coherence to emerge from variability. What appears as reflection at the individual level simultaneously serves as structural reorganization at the collective level—a small-scale adjustment resonating within a larger adaptive pattern.

Integrating the Two Flows: The Bayesian ICEE Loop

At the center of this architecture lies the Bayesian ICEE Loop, which dynamically links the behavioral (Task Chain) and cognitive (Crystal System) layers through continuous probabilistic updating.

Each action produces evidence; each reflection updates belief. The system evolves as it revises its expectations in response to experience:

P (H_{{t + 1}} E_{t}) \propto P (E_{t} H_{t}) \times P (H_{t})

H_t represents the learner's current hypothesis or belief state and E_t is the evidence generated through interaction or reflection. The loop tightens when experience confirms expectation and loosens when new evidence requires reinterpretation.

This mechanism operates simultaneously at multiple scales:

At the individual level, it appears as self-correction and metacognitive awareness.

At the classroom level, as collective calibration of shared understanding.

At the school or cultural level, as systemic learning—the gradual redesign of practices, goals, and values in response to feedback.

In this way, the Bayesian ICEE Loop functions as the spinal rhythm of the learning ecosystem. It connects micro-level adjustments to macro-level transformations through continuous cycles of uncertainty reduction and meaning reconstruction. Learning thus becomes not merely personal progress but a coevolving dialog between learners and the systems they inhabit, where each feedback moment carries the potential to reorganize the whole.

As these recursive cycles accumulate, the system begins to exhibit recognizable patterns linking behavior and understanding—an emergent mapping between what is done and what is known.

Mathematical Representation of the PDP–ICEE Learning Architecture

The PDP–ICEE architecture can be represented through three interrelated mathematical expressions, each corresponding to a different dimension of learning dynamics: temporal adaptation, reflective updating, and structural mapping.

Together, these equations describe how the educational system evolves through feedback—how action and reflection form a coupled process that generates meaning over time.

Time-Difference Learning: Adjusting through Experience

At the most immediate level, learning can be viewed as a process of adapting to the difference between expectation and experience. This expresses how each learner, as a dynamic node in the system, modifies their internal state in response to feedback signals.

Δ W_{t} = η (E_{t} - {\hat{E}}_{t})

ΔW_t denotes the change in connection weight (learning adjustment),

E_t is the experience signal, Ê_t the expected outcome, and η the adaptive rate of learning.

In educational terms, this equation captures the reflective moment between anticipation and reality—the cognitive space where understanding deepens and growth begins.

This principle corresponds to ICEE's focus on learning through reflection: Each discrepancy between what is expected and what is experienced becomes the trigger for new meaning formation.

Bayesian ICEE Loop: Updating Belief Through Reflection

At a broader scale, learning involves the continual revision of internal beliefs and shared meanings across interactions. This reflective cycle can be modeled through Bayesian updating, representing how understanding evolves in light of new experience.

P (H_{{t + 1}} E_{t}) \propto P (E_{t} H_{t}) \times P (H_{t})

H_t represents the learner's current hypothesis or interpretive frame, E_t denotes the new evidence generated through experience, and P(E_t) acts as the normalization term ensuring internal coherence of belief.

The Bayesian formulation is used here to formalize epistemic updating rather than statistical estimation, emphasizing belief revision as a systemic learning mechanism.

This formulation shows that every learning act updates the learner's implicit “model of meaning”—tightening when experience confirms existing understanding, and widening when new situations require reinterpretation.

Educationally, this captures the reflective essence of ICEE learning: a continual adjustment of beliefs about self, others, and the world in response to evidence emerging from real interaction.

This Bayesian expression bridges the individual and collective dimensions of reflection. For example, within a classroom, each learner's update contributes to the group's shared understanding, while the group's dialogue in turn reshapes the learner's prior beliefs—producing a recursive alignment of perspectives over time.

The Bayesian ICEE Loop thus formalizes the relationship between learning as behavior (Task Chain) and learning as understanding (Crystal System)—each informs and reshapes the other through continuous feedback.

Kernel Mapping: Linking Behavior and Cognition in a Shared Field

At the structural level, the relationship between the Task Chain (behavioral layer) and the Crystal System (cognitive layer) can be formalized through a kernel mapping function. This describes how patterns of behavior and patterns of understanding are embedded in a shared representational space—how what we do becomes what we know.

Φ (C) = K (T, C) = \int k (T_{i}, C_{j}), d ∖ m u

T_i represents an element within the Task Chain, C_j a unit within the Crystal System, k the kernel function measuring their relational similarity, and μ the measure defining the integration space.

Here, the integration does not denote numerical integration over time or iteration count, but a measure-theoretic aggregation over a representational space defined by relational uncertainty. The variability is carried by the structure of the space itself rather than by an explicit scalar variable.

Educationally, this mapping shows how learning actions and reflective insights cohere into recognizable patterns of competence—the cognitive “crystals” that anchor individual and collective learning.

In simpler terms, this mapping illustrates the translation between experience and meaning. Every behavioral act (a Task) contributes to a distributed network of understanding (a Crystal), and the strength of their relation—encoded in the kernel—reflects how deeply experience has been transformed into understanding.

Taken together, these formulations describe the PDP–ICEE system as a feedback-sensitive ecology of learning—one that continuously reorganizes itself by reconciling differences between action and understanding, and between individual and collective experience.

Educational Implications: Reframing ICEE in Practice

The PDP–ICEE Learning Architecture translates ICEE philosophy into a living system—one that learns about itself through feedback (Personalizable Learning), adapts and evolves through reflection and (re)iteration (Problem Finding and Solving), and reorganizes through collective meaning-making (Human Interdependence).

Here, “system learning” refers to the system's capacity to reorganize its structures based on interpreted feedback, rather than anthropomorphic cognition.

Its significance lies in reframing education as an evolving system anchored in lived earning experience, where the ICEE principles operate as the generative core of educational renewal.

Personalizable Learning: From Customization to Self-Calibration

In the PDP–ICEE system, personalizable learning is a process of self-awareness rather than external differentiation; it is education's micro-mechanism of continuous renewal. Each act of learning is a small system of adaptation: a negotiation between what is expected and what is experienced. Through this temporal dialog, the learner gradually constructs internal models of competence. The educator's role shifts from managing diversity to orchestrating feedback environments—contexts where learners can sense, interpret, and respond to their own patterns of change.

Problem Finding and Solving: Learning as Reflective Reiteration

Within the PDP–ICEE framework, problem finding and solving constitute the recursive rhythm through which the educational system learns about its own learning. This process is represented by the system's capacity to update its internal state in response to new evidence.

When expectation and reality diverge, the discrepancy acts as a feedback signal that invites reflection and reiteration. Through successive cycles of questioning, testing, and reframing, learners transform uncertainty into insight.

Teachers (or broad facilitators) facilitate this dynamic not by minimizing confusion but by curating productive disequilibrium—designing experiences that allow surprise to become a source of understanding.

Human Interdependence: From Collaboration to Resonance

Human Interdependence, the third principle of ICEE, extends the idea of learning from the individual to the collective scale. Human interdependence redefines community as a cognitive structure: the place where reflection becomes collective and meaning becomes relational.

In the PDP–ICEE system, knowledge emerges through distributed resonance—patterns of connection among learners, ideas, and contexts.

Understanding is not the property of any single mind but the coherence that arises when multiple perspectives interact and adjust to one another. This distributed intelligence mirrors the kernel-mapping logic within the system: behavioral events (the Task Chain) and cognitive abstractions (the Crystal System) align through shared meaning.

Educationally, this calls for designs that balance autonomy and togetherness—spaces where difference becomes generative, and individuality contributes to the intelligence of the whole.

Toward a Systemic View of Education

Reframing ICEE through the PDP–ICEE architecture reveals that the essence of educational transformation lies in understanding how the learner learns.

Personalization expresses the system's ability to calibrate itself; problem finding and solving represent its reflexive capacity to reorganize through feedback; Human Interdependence captures its relational coherence across scales.

Together, these dynamics form an educational ecology that is capable of self-renewal. For educators, this implies a shift from control to curation—maintaining the conditions for openness and reflection. For institutions, it means cultivating adaptive balance—allowing cycles of innovation and stability to coexist. For learners, it means recognizing that autonomy grows through connection, and that to learn is to participate in the collective intelligence of the system.

In the age of AI, the challenge is not to automate learning but to design educational systems that can evolve with awareness. The PDP–ICEE Learning Architecture offers a path toward that goal: an education that mirrors life itself—anchored in experience, guided by interaction and coevolution with others (including AI), and sustained by human interdependence.

The Systemic Horizon⁸ of Educational Transformation

The rise of AI, together with the design of systems like YEEAI, enables education to see and reorganize its own process of learning, renewing a reflexive capacity that has long been constrained by traditional assessment. This marks a fundamental shift—from education as a passive object of reform to a living, self-reflective, and self-reorganizing system.

Within this view, the PDP–ICEE Learning System offers a new architecture for observation. It reveals that learning is a dynamic cycle between generation and reflection: The Task Chain drives the unfolding of action, while the Crystal System consolidates understanding. Together, they form the dual respiration of the educational system. When human judgment, values, and intentions interact continuously with AI's perception, pattern recognition, and precision, education acquires the capacity for self-observation and self-evolution.

This evolution unfolds through the adaptive logic of complex systems. The PDP–ICEE Learning Architecture exemplifies this adaptive logic within education, integrating HITL feedback with ICEE's ecological principles. Each layer of education—learner, classroom, institution, and society—moves through its own rhythm of growth, conservation, release, and reorganization. Across these layers flow feedback loops that link local innovation to systemic renewal.

As Panarchy Theory suggests, a resilient system sustains itself not by resisting change but by orchestrating transformations across scales: Local breakthroughs trigger new organizational forms, while macro-level stability provides direction and coherence for further exploration.

Within this adaptive ecology, curriculum plays the role of the slow variable—the element that changes gradually yet anchors the identity of the whole. It is not a static syllabus but a generative blueprint that sustains meaning while enabling flexibility. Amid fast-moving feedback and personalization, curriculum preserves the continuity of purpose; it functions as the system's memory, maintaining coherence so that innovation does not dissolve into fragmentation. By evolving slowly, it provides a stable horizon against which faster pedagogical and technological processes can move and renew themselves.

AI participates in this ecology as an intrinsic component of the system. Through pattern recognition, data reflection, and adaptive modeling, AI helps the education system perceive its own dynamics; human judgment, in turn, supplies ethical orientation and interpretive meaning.

The abstraction level adopted in this article is intentional, allowing multiple technological realizations without constraining educational meaning. Rather than prescribing a specific algorithmic pathway, the PDP–ICEE Learning System defines a structural and epistemic framework through which education can remain adaptive, interpretable, and human-centered in the age of AI.

Every design, trajectory, and reflection contributes to the system's self-knowledge. Over time, these fragments coalesce into a recursive intelligence through which education learns to perceive itself and to evolve by reflecting on its own evolution. Through YEEAI as its epistemic infrastructure, the system translates experience into understanding and sustains continuous self-learning. Education becomes a living process—regenerative like an ecosystem and adaptive like a neural network.

Ultimately, the PDP–ICEE Learning System reveals a new epistemic form of education: education as cognition. Its purpose is to sustain a state of ongoing learning and self-correction. The future of education is not to predict the future but to learn with it.

Footnotes

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Notes

References

Ashby

W. R.

(1956). An introduction to cybernetics. Chapman & Hall.

Bateson

(1972). Steps to an ecology of mind. Ballantine.

Biesta

(2009). Good education in an age of measurement: On the need for educational responsibility. Educational Assessment, Evaluation and Accountability, 21(1), 33–46. https://doi.org/10.1007/s11092-008-9064-9

Bohr

(1934). Atomic theory and the description of nature. Cambridge University Press.

Doshi-Velez

Kim

(2017). Towards a rigorous science of interpretable machine learning. arXiv preprint arXiv:1702.08608.

Gilpin

L. H.

Bau

Yuan

B. Z.

Bajwa

Specter

Kagal

(2018). Explaining explanations: An overview of interpretability of machine learning. Proceedings of the 2018 IEEE 5th International Conference on Data Science and Advanced Analytics, 80–89. IEEE. https://doi.org/10.1109/DSAA.2018.00018

Gunderson

L. H.

Holling

C. S.

(2002). Panarchy: Understanding transformations in human and natural systems. Island Press.

Heisenberg

(1927). Über den anschaulichen inhalt der quantentheoretischen kinematik und mechanik. Zeitschrift für Physik, 43(3–4), 172–198. https://doi.org/10.1007/BF01397280

Holling

C. S.

(1973). Resilience and stability of ecological systems. Annual Review of Ecology, Evolution, and Systematics, 4, 1–23. https://doi.org/10.1146/annurev.es.04.110173.000245

10.

Luhmann

(1995). Social systems (J. Bednarz Jr. & D. Baecker, Trans.). Stanford University Press. (Original work published 1984).

11.

McClelland

J. L.

Rumelhart

D. E.

(1986). Parallel distributed processing: Explorations in the microstructure of cognition (Vol. 1: Foundations). MIT Press.

12.

Sheridan

T. B.

(1992). Telerobotics, automation, and human supervisory control. MIT Press.

13.

Sheridan

T. B.

Parasuraman

(2005). Human–automation interaction. Reviews of Human Factors and Ergonomics, 1(1), 89–129. https://doi.org/10.1518/155723405783703082

14.

von Foerster

(1974). Cybernetics of cybernetics. University of Illinois.

15.

Wiener

(1948). Cybernetics: Or control and communication in the animal and the machine. MIT Press.

16.

Zhao

(2012). World class learners: Educating creative and entrepreneurial students. Corwin Press.

17.

Zhao

(2016). Counting what counts: Reframing education outcomes. Corwin Press.

18.

Zhao

Zhong

(2024). Paradigm shifts in education: An ecological analysis. ECNU Review of Education, 8(1), 21–40. https://doi.org/10.1177/20965311241296162

19.

Zhong

Zhao

(2025). Education paradigm shifts in the age of AI: A spatiotemporal analysis of learning. ECNU Review of Education, 8(2), 319–342. https://doi.org/10.1177/20965311251315204

20.

Zhong

Zhao

(2026). The double-helix logic of curriculum: Reframing universality and personalization in the age of AI. ECNU Review of Education, 9(1). https://doi.org/10.1177/20965311261421970

A Distributed Architecture Integrating Educational Philosophy and AI-Driven Learning Design: The PDP–ICEE Learning System

Abstract

Purpose

Design/Approach/Methods

Findings

Originality/Value

Keywords

The Assessment Trap: When Education Stops Learning

From the Pursuit of Visibility to the Assessment Trap

When Evaluation Becomes the System

Data Without Meaning

From Measurement to Meaning

HITL and the ICEE Response

HITL: Restoring Reflexivity to the System

ICEE as the Systemic Response

Personalizable Learning: Sensitivity to Difference

Problem Finding and Solving: Reflection Through Uncertainty

Human Interdependence: Ethics as Feedback Ecology

ICEE and the Reconstruction of Educational Feedback

Layered Feedback Within the HITL System

The Micro-loop: Learner and AI

The Meso-Loop: Teacher–Learner–AI Ensemble

The Macro-Loop: Education as a Reflective System

Ethics as Reflexive Literacy

Toward the PDP–ICEE Learning Architecture

Education as a Parallel Distributed System

Systemic Understanding as Computable Reflexivity 3

YEEAI 4 as an Epistemic Infrastructure

System Design: The YEEAI Architecture and Its Core Mechanisms

System Overview: From Educational Principles to Computational Architecture

The Task Chain: Organizing Personalizable Learning as Distributed Action

Definition and Core Structure

Educational Logic and System Operation

From Structure to Living Process

Definition of the Learning Facilitator

System Operation: From Supra task to Learning Loop

Role Interaction: Learner, Facilitator, and AI

The Learner Acts as the Initiator of Meaning

The Learning Facilitator Functions as the Sense-Maker and Ecological Curator

The AI System Acts as a Cognitive Partner that Amplifies Both Learner and Facilitator Capacities

Educational Meaning

Application and Examples: The Task Chain in Real Learning Ecosystems

Conceptual Validation: Hongfan (HF) Subsystem

Adaptive Configurations: From the Eight Steps to the Six Phrases

Systemic Realization: MEMW

Systemic Significance

The Crystal System: Tracing Cognitive Growth Through Action

The Philosophical Grounding: From Action to Reflection

Typology of Crystals

Event Crystals

Pattern Crystals

Cognitive Crystals

Developmental Logic

Accumulation—Recording the Micro-Moments of Learning

Resonance—Finding Patterns Across Experiences

Abstraction—From Strategy to Concept

Reorganization—Continuous Adaptation and Growth

Design as Reflexive Practice

The PDP–ICEE Learning Architecture

Theoretical Grounding of the PDP–ICEE Learning Architecture

From Connectionism to Distributed Cognition

Ecological Feedback and Nested Adaptation

Time, Uncertainty, and Bayesian Updating

Computational Model of Learning Dynamics

The Task Chain: Forward Propagation of Experience

The Crystal System: Backward Abstraction of Meaning

Integrating the Two Flows: The Bayesian ICEE Loop

Mathematical Representation of the PDP–ICEE Learning Architecture

Time-Difference Learning: Adjusting through Experience

Bayesian ICEE Loop: Updating Belief Through Reflection

Kernel Mapping: Linking Behavior and Cognition in a Shared Field

Educational Implications: Reframing ICEE in Practice

Personalizable Learning: From Customization to Self-Calibration

Problem Finding and Solving: Learning as Reflective Reiteration

Human Interdependence: From Collaboration to Resonance

Toward a Systemic View of Education

The Systemic Horizon 8 of Educational Transformation

Footnotes

Funding

Declaration of Conflicting Interests

Systemic Understanding as Computable Reflexivity³

YEEAI⁴ as an Epistemic Infrastructure

The Systemic Horizon⁸ of Educational Transformation