Yiuno

Schema Theory

11 min read

Schema Theory

The idea that humans organise knowledge into mental frameworks called schemas — structured patterns that help us interpret experience, guide perception, and predict what happens next.

What is it?

When you walk into a restaurant, you already know what to expect: there will be a host, a menu, a table, food, a bill. Nobody taught you this as a formal lesson — you built this understanding through repeated experience. That internal framework of “how restaurants work” is a schema. Schema theory is the study of how these mental frameworks form, change, and sometimes fail.1

The concept was introduced by the British psychologist Frederic Bartlett in 1932. Bartlett showed that when people recall stories, they don’t reproduce them faithfully — they reconstruct them, filling in gaps and distorting details to fit their existing schemas.2 The Swiss psychologist Jean Piaget later expanded the idea into a theory of cognitive development, proposing two mechanisms by which schemas evolve: assimilation (fitting new information into an existing schema) and accommodation (modifying the schema when new information doesn’t fit).3

What makes schema theory relevant to knowledge engineering is the parallel it reveals. A schema is essentially an informal ontology — a personal, implicit structure for organising related concepts and their relationships. When you think of “bird,” your schema includes wings, feathers, flight, nests, and songs. That schema functions like a miniature knowledge graph in your head, with nodes (concepts) and edges (relationships) that let you reason about new situations.1

The difference between a schema and a formal ontology is rigour. Schemas are fuzzy, personal, and often unconscious. Ontologies are explicit, shared, and machine-readable. But both serve the same fundamental purpose: structuring knowledge so that new information can be interpreted in context.4

In plain terms

A schema is like a mental filing system that your brain builds automatically. When you encounter something new, your brain doesn’t start from scratch — it reaches for the closest existing folder, checks if the new thing fits, and either files it there or creates a new folder. Schema theory studies how these folders are built, used, and updated.

At a glance

How does it work?

Schemas as mental structures

A schema is not a single fact — it is a cluster of related knowledge organised around a central concept. Your “dog” schema includes what dogs look like, how they behave, what sounds they make, where you find them, and how to interact with them. This cluster activates as a unit: encountering one element (a bark) triggers the entire pattern.2

Schemas operate at every level of complexity. You have schemas for concrete objects (chairs), for social situations (job interviews), for abstract processes (how arguments work), and for narratives (the hero’s journey). Each one is a pattern that helps you process information efficiently by providing a ready-made framework for interpretation.1

Think of it like...

A schema is like a template in a word processor. When you open a “business letter” template, you don’t have to decide where the date goes, how to format the address, or where to sign — the template provides the structure. Your brain’s schemas do the same thing for experience: they provide the structure so you can focus on the new content.

Assimilation

Assimilation is the process of incorporating new information into an existing schema without changing the schema itself. When a child who knows about dogs sees a new breed for the first time, they recognise it as a dog — different in colour and size, but fitting the existing schema of “four legs, fur, barks, tail.”3

Assimilation is cognitively efficient. It lets you process familiar situations rapidly because you’re not building understanding from scratch every time. But it has a cost: it can force new information into categories where it doesn’t truly belong.

Accommodation

Accommodation is the opposite process: when new information cannot be assimilated, the existing schema must change to incorporate it. This is harder and slower than assimilation, but it’s how schemas grow in sophistication.3

Piaget argued that cognitive development is driven by the tension between assimilation and accommodation. When a child’s “bird” schema includes “all birds fly” and they encounter a penguin, the schema must accommodate — the definition of “bird” expands beyond flight to include other features like feathers and eggs.3

Think of it like...

Assimilation is adding a new book to an existing shelf. Accommodation is realising you need a completely new shelf — or that your entire shelving system needs reorganising. The first is easy. The second is disruptive but necessary for growth.

Schema failure

Schemas can fail in predictable ways. Because they are built from experience, they reflect the biases and limitations of that experience. Stereotypes are schemas applied to people — they provide rapid (but often inaccurate) categorisation. Misconceptions are schemas that were built from incomplete or incorrect information and resist correction because assimilation keeps reinforcing them.2

In education, schema failure explains why students cling to wrong ideas even after correction: the incorrect schema is deeply embedded through repeated use, and accommodation requires not just adding new information but dismantling the old structure.5

Key distinction

Assimilation is fast, automatic, and reinforces existing patterns. Accommodation is slow, effortful, and restructures existing patterns. Effective learning requires both — but most everyday processing is assimilation, which is why misconceptions are so persistent.

The parallel to formal knowledge systems

The connection between schemas and knowledge engineering is more than metaphorical. Schemas function as the brain’s informal ontologies — structured representations of domain knowledge that enable reasoning and prediction. The table below maps the correspondence:1

Cognitive conceptKnowledge engineering equivalent
SchemaOntology
Mental modelKnowledge graph
LearningKnowledge acquisition
AssimilationAdding instances to existing classes
AccommodationRefactoring the ontology
Schema failureOntology gaps or contradictions

This parallel matters for AI. Large language models don’t build schemas — they compress statistical patterns from training data. The result mimics schematic organisation (an LLM can discuss “restaurants” coherently) without the underlying structure. The LLM has no “restaurant schema” — it has a probability distribution over tokens that co-occur in restaurant-related contexts.4

Concept to explore

See ontology for how formal ontologies make the implicit structure of schemas explicit and machine-readable.

Why do we use it?

Key reasons

1. Understanding how people learn. Schema theory explains why prior knowledge matters so much — new information is always interpreted through existing frameworks, which means teaching must account for what learners already believe, not just what they don’t yet know.3

2. Designing better knowledge systems. If human knowledge is schematic, then knowledge systems should support schema-like structures — hierarchies, relationships, progressive refinement. This is exactly what ontologies and knowledge graphs provide.1

3. Explaining AI limitations. Understanding schemas clarifies what LLMs lack. They produce outputs that look schematic but are generated from statistical compression, not structured understanding. This distinction matters when deciding how much to trust AI outputs.4

When do we use it?

  • When designing a knowledge system and need to understand how humans naturally organise information
  • When building educational content and need to scaffold new knowledge onto what learners already know
  • When analysing why a misconception persists despite repeated correction
  • When evaluating whether an AI system truly understands a domain or is merely mimicking understanding
  • When deciding how to structure an ontology — schema theory suggests that categories based on experience are more useful than arbitrary logical divisions

Rule of thumb

If you’re building a system that humans need to learn from or contribute knowledge to, schema theory tells you how their minds will process it — design with that processing in mind, not against it.

How can I think about it?

The wardrobe analogy

Your brain is like a wardrobe that organises itself. Every new piece of clothing (experience) either fits into an existing section (assimilation — this goes with the t-shirts) or forces you to rearrange (accommodation — you’ve never owned a wetsuit, so you need a new section).

Over time, the wardrobe becomes highly personalised. Someone who lives in the tropics has a completely different internal organisation from someone in Scandinavia — even if they own some of the same items. This is why two people can encounter the same fact and interpret it differently: their wardrobes are organised differently.

  • Wardrobe sections = schemas
  • Sorting a new item into an existing section = assimilation
  • Reorganising the wardrobe = accommodation
  • A section that doesn’t match reality (winter coats filed under “swimwear”) = schema failure
  • The wardrobe’s overall layout = your personal ontology

The city map analogy

Imagine moving to a new city. At first, you know one route: home to work. That single path is your initial schema. Over time, you discover shortcuts, landmarks, and alternative routes. Your mental map becomes richer and more interconnected — you can navigate flexibly, predict traffic, and give directions.

But your map is always incomplete and sometimes wrong. You might “know” that two neighbourhoods are far apart because you always drove between them, when actually they’re adjacent on foot. That’s schema failure: your experience built a model that doesn’t match reality.

  • Your mental map = schema
  • Adding new routes to the same map = assimilation
  • Realising the map was wrong and redrawing it = accommodation
  • A formal city map with labelled streets = ontology
  • GPS navigation = machine reasoning over a knowledge graph

Concepts to explore next

ConceptWhat it coversStatus
ontologyFormal, explicit specifications of concepts and their relationshipsstub
constructivismThe learning theory built on schema theory — knowledge is actively constructed, not passively receivedstub
taxonomiesHierarchical classification systems that mirror the categorical structure of schemascomplete
knowledge-granularityHow knowledge is decomposed into atoms, concepts, and topicsstub

Some cards don't exist yet

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

Check your understanding

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 --> C[Constructivism]
    style ST fill:#4a9ede,color:#fff

Related concepts:

  • ontology — schemas are the brain’s informal ontologies; formal ontologies make the same structure explicit
  • constructivism — the learning theory that builds directly on Piaget’s schema mechanisms
  • taxonomies — hierarchical classification systems that mirror the categorical structure of schemas
  • knowledge-granularity — how knowledge is decomposed into units, paralleling how schemas chunk experience
  • agent-memory — how AI agents store and retrieve context, an engineered analogue to schematic memory

Sources

Further reading

Resources

Footnotes

  1. Polymathik. (2024). Your Brain Already Has a Knowledge Graph. Medium. ↩ ↩2 ↩3 ↩4 ↩5

  2. Bartlett, F. C. (1932). Remembering: A Study in Experimental and Social Psychology. Cambridge University Press. ↩ ↩2 ↩3

  3. Piaget, J. (1952). The Origins of Intelligence in Children. International Universities Press. ↩ ↩2 ↩3 ↩4 ↩5

  4. How to Think AI. (2025). A Schema Is a Mental Model Made Explicit. How to Think AI. ↩ ↩2 ↩3

  5. How to Think AI. (2025). Everyone Operates on Schemas. How to Think AI. ↩

Graph View

Backlinks