Constructivism

The learning theory that knowledge is actively built by the learner through experience and reflection, not passively received from a teacher or a text.

---

What is it?

Constructivism starts from a deceptively simple observation: you cannot transfer knowledge directly from one mind to another. A teacher can present information, but the learner must construct understanding by connecting new information to what they already know. Learning is not copying — it is building.¹

This idea has two major branches. Cognitive constructivism, developed by Jean Piaget, focuses on the individual learner. Piaget showed that learners build mental schemas through two mechanisms: assimilation (fitting new information into existing schemas) and accommodation (restructuring schemas when new information doesn’t fit). Learning happens when there is a productive mismatch between what you expect and what you encounter.²

Social constructivism, developed by Lev Vygotsky, adds a critical layer: learning doesn’t happen in isolation. It happens through dialogue, collaboration, and social interaction. Vygotsky introduced the zone of proximal development (ZPD) — the gap between what a learner can do alone and what they can do with guidance. Effective teaching targets this zone: challenging enough to push growth, but scaffolded enough to prevent failure.³

Why does this matter for knowledge engineering? Because constructivism tells us something fundamental about how knowledge systems should be designed. If knowledge must be constructed (not copied), then dumping a complete ontology or knowledge graph in front of someone won’t produce understanding. The system must support progressive construction — starting from what the learner knows, introducing complexity gradually, and providing the connections that let meaning emerge.⁴

In plain terms

Constructivism says that learning is like building a house, not like receiving a delivery. The teacher can bring materials and show you how to use them, but you have to put the walls up yourself. If the foundation isn’t there, adding a roof won’t help — you have to build from the ground up, one layer at a time.

---

At a glance

The constructivist learning cycle (click to expand)

graph TD
    P[Prior Knowledge] -->|encounter| E[New Experience]
    E -->|fits schema| A[Assimilation]
    E -->|challenges schema| AC[Accommodation]
    A --> I[Integrated Understanding]
    AC --> I
    I -->|social interaction| R[Refined Understanding]
    R -->|becomes| P

Key: Learning begins with prior knowledge. New experiences are processed through existing schemas via assimilation or accommodation. Social interaction refines the result. The refined understanding becomes the new prior knowledge for the next learning cycle.

---

How does it work?

Cognitive constructivism (Piaget)

Piaget’s core insight is that learners are not blank slates. Every new experience is filtered through existing mental structures — schemas — and those structures determine what the learner can and cannot understand at any given moment.²

This means learning is inherently sequential. A learner cannot understand multiplication before understanding addition, not because the information is unavailable, but because the schema needed to make sense of multiplication hasn’t been built yet. Each layer of understanding provides the foundation for the next.

Think of it like...

Learning a language works the same way. You can’t understand conditionals (“If I had known, I would have gone”) until you’ve built schemas for tense, pronouns, and clause structure. The grammar of understanding has prerequisites, just like the grammar of language.

Example (click to expand)

Consider teaching someone about knowledge graphs. A constructivist approach would start from something they already know — for example, a family tree. “You know how a family tree shows who is related to whom? A knowledge graph does the same thing, but for any kind of relationship — not just family.” The learner assimilates the new concept into their existing “family tree” schema, then gradually accommodates the differences (knowledge graphs can have multiple relationship types, cycles, and attributes).

Social constructivism (Vygotsky)

Vygotsky argued that Piaget’s model was incomplete. Learning isn’t just an internal process — it’s fundamentally social. We construct meaning through conversation, argument, collaboration, and shared problem-solving. The language we use to discuss ideas shapes the ideas themselves.³

His most influential concept is the zone of proximal development (ZPD): the space between what a learner can do independently and what they can do with guidance from a more capable peer or mentor. Learning happens most effectively in this zone — below it, the learner is bored; above it, the learner is overwhelmed.³

Scaffolding is the instructional practice derived from the ZPD. A teacher provides temporary support structures — simplified examples, guiding questions, worked-through problems — that the learner can use while building their own understanding. As competence grows, the scaffolding is removed.⁵

Think of it like...

A swimming instructor doesn’t throw you into deep water and explain hydrodynamics. They start in the shallow end, hold you up while you practise arm movements, then gradually reduce support until you’re swimming independently. The instructor hasn’t “given” you the ability to swim — you constructed it through guided practice.

Why transmission doesn’t work

The opposite of constructivism is instructionism (sometimes called the “transmission model”) — the idea that knowledge can be packaged and delivered, like a parcel. Lecture, absorb, repeat. Constructivism’s central claim is that this model is fundamentally flawed: hearing information is not the same as understanding it.¹

Evidence from educational research consistently shows that passive reception (reading, listening to lectures) produces shallow retention compared to active construction (problem-solving, discussion, teaching others). This is because active engagement forces the learner to connect new material to existing schemas, which is the mechanism by which durable understanding is built.⁵

Key distinction

Transmission assumes the teacher holds knowledge and the student receives it. Construction assumes the teacher provides materials and conditions, and the student builds knowledge. The first treats learners as containers. The second treats them as builders.

Implications for knowledge systems

Constructivism has direct implications for how knowledge systems — including knowledge graphs, documentation, and AI-augmented learning tools — should be designed:⁴

1. Progressive disclosure. Don’t present everything at once. Reveal complexity as the user’s understanding grows. Start with core concepts, then expose relationships, then edge cases.

2. Scaffolded complexity. Provide worked examples, analogies, and simplified models before introducing formal definitions. Let the learner build intuition before precision.

3. Active engagement. Design systems that require the user to do something with knowledge — answer questions, make connections, apply concepts — rather than passively consume it.

4. Connection-first architecture. Make the relationships between concepts visible and navigable. Understanding comes from seeing how ideas connect, not from memorising isolated facts.⁶

Concept to explore

See knowledge-granularity for how knowledge can be decomposed into atoms, concepts, and topics to support progressive construction.

The AI parallel

An LLM does not learn in the constructivist sense. It does not build schemas through experience. It compresses statistical patterns from a training corpus — a process closer to transmission (absorb everything at once) than construction (build understanding incrementally).⁴

But the systems we build around LLMs can embody constructivist principles. Retrieval-augmented generation (RAG) provides contextual scaffolding. Knowledge graphs provide the connective structure. Context cascading progressively reveals relevant information. These engineering patterns compensate for what the model itself lacks — structured, connected, progressively disclosed knowledge.⁶

Concept to explore

See human-in-the-loop for how human judgement provides the constructivist element — active meaning-making — in AI-augmented workflows.

---

Why do we use it?

Key reasons

1. Designing systems that produce understanding, not just exposure. If knowledge must be constructed, then a system that dumps information without structure, scaffolding, or progressive complexity will fail — regardless of how good the content is.¹

2. Explaining why knowledge management often fails. Most knowledge management systems treat knowledge as content to be stored and retrieved. Constructivism explains why this approach produces bloated repositories that nobody learns from: storage is not construction.⁴

3. Building better AI-augmented learning. Constructivist principles — scaffolding, progressive disclosure, active engagement — provide a blueprint for designing RAG systems, knowledge bases, and AI tutors that support genuine understanding rather than superficial retrieval.⁶

---

When do we use it?

• When designing a knowledge base or documentation system and deciding how to structure and sequence content
• When building an AI tutor or learning assistant that needs to adapt to what the learner already knows
• When evaluating why a knowledge management initiative is failing despite having good content
• When deciding how to onboard new team members — constructivism says you need scaffolded complexity, not a document dump
• When designing concept cards or learning materials that need to build understanding progressively

Rule of thumb

If your system presents information but users don’t develop understanding, the problem is almost certainly a constructivist one: the system is transmitting when it should be scaffolding.

---

How can I think about it?

The LEGO analogy

Giving someone a complete LEGO model isn’t the same as giving them the bricks and instructions to build it. The person who built it understands how it holds together, can modify it, and can build something new from the same pieces. The person who received the finished model can describe what it looks like, but can’t explain its structure or adapt it.

Constructivism says that understanding is always built, never delivered. The instruction manual (scaffolding) helps, but the learner must place each brick themselves. As they become more skilled, they need fewer instructions — until eventually they can build without any manual at all.

• LEGO bricks = knowledge atoms (facts, concepts)
• Instruction manual = scaffolding (analogies, examples, guided questions)
• Building the model = constructing understanding
• Receiving a finished model = passive transmission
• Building without instructions = expert-level schema mastery

The cooking analogy

Reading a recipe doesn’t make you a cook. Following a recipe once gives you a procedure. Cooking the dish ten times, adjusting for taste, substituting ingredients, and recovering from mistakes — that makes you a cook. Each iteration builds a richer schema for how flavours, textures, and techniques interact.

A cookbook that only lists ingredients and steps is transmissionist. A cookbook that explains why you sear meat before braising (Maillard reaction creates flavour compounds) gives you the scaffolding to understand what you’re doing — and to improvise when you don’t have the exact ingredients.

• Ingredients = raw information
• Recipe = structured knowledge
• Cooking the dish = constructing understanding through experience
• Explaining the “why” behind techniques = scaffolding
• Improvising a meal without a recipe = deep schematic understanding

---

Concepts to explore next

Concept	What it covers	Status
schema-theory	The mental frameworks that constructivism builds upon	complete
dikw-hierarchy	Data, information, knowledge, wisdom — the layers of understanding	stub
knowledge-granularity	How knowledge is decomposed into atoms, concepts, and topics	stub
human-in-the-loop	How human judgement provides active meaning-making in AI workflows	complete

Some cards don't exist yet

A broken link is a placeholder for future learning, not an error.

---

Check your understanding

Test yourself (click to expand)

1. Explain the difference between the transmission model and the constructivist model of learning. Why does constructivism claim transmission is insufficient?
2. Distinguish between Piaget’s cognitive constructivism and Vygotsky’s social constructivism. What does each emphasise that the other does not?
3. Name Vygotsky’s zone of proximal development and explain why it matters for designing learning experiences.
4. Interpret this scenario: a company builds a comprehensive internal wiki with thousands of well-written articles, but employees consistently say they “can’t find what they need” and prefer to ask colleagues directly. Using constructivism, explain what is happening.
5. Connect constructivism to knowledge graph design. How would a constructivist approach to building a knowledge graph differ from a purely logical or taxonomic approach?

---

Where this concept fits

Position in the knowledge graph

graph TD
    KE[Knowledge Engineering] --> ST[Schema Theory]
    KE --> KG[Knowledge Graphs]
    KE --> MRF[Machine-Readable Formats]
    ST --> CON[Constructivism]
    style CON fill:#4a9ede,color:#fff

Related concepts:

• dikw-hierarchy — constructivism explains why progressing from data to wisdom requires active construction at each level
• claims-and-propositions — constructivism treats knowledge as constructed claims, not objective truths waiting to be discovered
• knowledge-granularity — progressive construction requires knowledge decomposed into appropriately sized units
• human-in-the-loop — the constructivist element in AI systems, where human judgement provides the active meaning-making that models cannot

---

Sources

---

Further reading

Resources

• Constructivism and Knowledge Building in Education (Preprints.org) — Academic overview of constructivist theory and its application to knowledge systems
• Building Your Graph Is Building Your Mind (How to Think AI) — The constructivist case for personal knowledge graphs as tools for understanding
• A Schema Is a Mental Model Made Explicit (How to Think AI) — How making schemas explicit supports constructivist learning
• Decomposition for Dummies (David Didau) — Practical strategies for breaking complex ideas into learnable components
• The Theoretical Foundations of Metadata in Knowledge Management (Fehlau) — How constructivist principles inform metadata and knowledge structure design

Footnotes

1. Preprints.org. (2025). Constructivism and Knowledge Building in Education. Preprints.org. ↩ ↩² ↩³

2. Piaget, J. (1952). The Origins of Intelligence in Children. International Universities Press. ↩ ↩²

3. Vygotsky, L. S. (1978). Mind in Society: The Development of Higher Psychological Processes. Harvard University Press. ↩ ↩² ↩³

4. How to Think AI. (2025). Building Your Graph Is Building Your Mind. How to Think AI. ↩ ↩² ↩³ ↩⁴

5. Didau, D. (2025). Decomposition for Dummies: How to Break Down Complex Ideas. Substack. ↩ ↩²

6. Fehlau, M. (2025). The Theoretical Foundations of Metadata in Knowledge Management. fehlau.de. ↩ ↩² ↩³