Last updated: 2026-10-04

U
Undergraduate level

Learning as a Network: Connectivism, Knowledge and Conceptual Navigation

Connections among people, artefacts, sources, and systems, and what a learner has to do with them

Most accounts of learning picture knowledge as something placed inside a mind. A learner receives representations, stores them, and retrieves them when needed. That picture has real explanatory power, but it struggles with the conditions most learners now work in. Knowledge is spread across textbooks, colleagues, databases, search engines, software tools, and increasingly across models that generate answers rather than store them. If what a learner needs is distributed in this way, then learning cannot consist only of putting representations inside one head. It also involves creating, strengthening, pruning, navigating, and judging the connections that link a learner to the wider network.

1. A Connectivist Foundation FoundationalKnowledge that endures for decades — core principles

George Siemens proposed connectivism in 2005 as a learning theory for the digital age. His starting point was that the established theories of learning, behaviourism, cognitivism, and constructivism, had been developed before networked digital technology reshaped how people find and use information. He argued that a theory of learning should account for informal learning, for the growing role of technology in mediating what people know, and for the way knowledge changes faster than any individual can track it[1]. Several principles follow: learning is the connecting of specialised sources of information; the capacity to know more is as significant as what is presently known; learning includes recognising relations across fields; and maintaining connections is itself a learning activity.

Stephen Downes developed a more relational formulation. On his account, knowledge is distributed across a network of connections, and learning consists in constructing and traversing those networks. Learning is the growth, modification, and strengthening of connections among entities, and a change in one entity can alter the connections around it[2]. This shift matters. The unit of analysis moves from the item of knowledge to the relation between items, and from the learner's store of content to the learner's place in a structure of relations.shift from content to topology

2. What Connectivism Is Not Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

Connectivism is often reduced to the claim that people now learn from the internet. That reduction hides the theory's content. Three positions are easily confused.

Technology-enhanced learning uses technology to assist an otherwise familiar educational process. A lecture is recorded, a quiz is automated, a document is shared online. The structure of the teaching has not changed.

Networked learning treats the relationships among learners, sources, communities, tools, and representations as part of what constitutes the learning environment. The network is an object of design, not only a delivery channel.

Connectivist learning understands learning through the creation, recognition, traversal, evaluation, pruning, and transformation of connections. Its claim concerns the form of knowledge and the form of learning, not the medium.

Giving students access to websites therefore does not create a productive learning network. A network needs diversity, openness, interaction, learner autonomy, critical evaluation, and connections reliable enough to be used. A set of links that all point to one source, or to one platform's ranking, is a narrow network whatever its size.designing for agency, not just access

3. Several Kinds of Connection FoundationalKnowledge that endures for decades — core principles

The word "connection" operates at several levels, and it helps to keep them distinct:

Neural connections
        │
Conceptual connections
        │
Connections among records and representations
        │
Social connections among learners and experts
        │
Technical connections among systems and repositories
        │
Trust relationships regulating reliance

These levels are not one mechanism. A synapse, a conceptual link between two ideas, a hyperlink, a friendship, an API call, and a decision to rely on a source are different kinds of thing. What they share is relational structure: each ties one entity to another in a way that can be strengthened, weakened, or broken. The theory concerns that shared form. It does not claim the levels are identical.

4. Forms of Knowing Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

If connections are the substance of learning, knowing takes several forms, which can be arranged in a rough sequence:

Knowing that
Knowing how
Knowing where
Knowing whom
Knowing how to verify
Knowing when to trust

A learner may know something because they can recall and explain it. They may know because they can reconstruct an answer from connected concepts, find an appropriate source, identify a relevant expert, operate a suitable tool, coordinate several sources, or judge whether a result is adequate for the purpose at hand. Most of the forms beyond the first two are invisible to a test of recall, yet they govern how competently a learner acts in a networked environment. A learner who can recall a fact but cannot locate its best source, or judge when it has been superseded, is poorly equipped for a world in which facts change.

5. Connectivism and Compression Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

A learner does not internalise complete copies of every resource in a network. The learner develops compressed structures instead: concepts, schemas, disciplinary frames, navigational landmarks, memories of sources, trust relationships, retrieval strategies, procedural routines, and expectations about where useful knowledge is likely to be found. The compression discards most of the detail and keeps what is needed to act. The page on the compression of knowledge sets out the general pattern: a representation keeps some properties and discards others, and the question is what has been kept.cf. the compression page: what is kept vs discarded

What a learner remembers is often a pointer rather than a proposition. The learner may recall that an issue was discussed in a particular article, that a certain person has relevant expertise, that two concepts are connected, that a source needs independent checking, or that a question belongs to a particular disciplinary conversation. Each memory is a route into the network rather than a copy of its contents.

This leads to a central claim of the connectivist approach: connectivist learning includes learning the topology of a knowledge environment, not only its propositional contents. A learner who has mapped where the strong arguments sit, how the sources relate, and where the disagreements lie has learned something the individual propositions do not contain.

6. The Literature Review as a Navigational Instrument Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

The site's project-navigation work treats inquiry as continuing orientation among three things: the implementation or evidence being produced, the existing literature, and the wider problem the work belongs to. The triangulation page argues that fixing one's position depends on checking against sources with independent origins, and that the literature review is repeated rather than performed once. Seen through connectivism, the literature review stops being a static preliminary chapter. It becomes a living instrument for locating oneself in the network and for finding the connections that matter next.

7. Connectivism, Critical Pedagogy, and Trust Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

Connectivism can become too celebratory about networks. Networks reproduce as well as enable. They can reward popularity as a substitute for reliability, reflect platform power and algorithmic visibility, encourage conformity, carry inherited inequalities, create dependence on opaque systems, and repeat one original error through many apparently independent channels. A dense network with a shared origin is not a body of independent evidence.

Critical pedagogy supplies the questions a connectivist account needs. Paulo Freire argued that education should develop learners' capacity to examine and transform the conditions of their own knowing[3], and that argument applies directly to networks. Who constructed this network? Who is visible within it? Who is treated as authoritative? Whose knowledge is absent? Who controls ranking and recommendation? Which connections does the institution reward? The pedagogy, andragogy, heutagogy, and rhizome page addresses the related question of who decides what is learned, and shows how authority over a curriculum can move between institution, instructor, learner, and community.

The calculus of trust adds a further discipline. The calculus-of-trust page treats trust as relational, depending on who trusts whom, in what context, with what history. For a learner the question becomes: on what basis should I rely on this node, for this purpose, in this context, with these consequences? Judgements of this kind are part of what a learner must learn to make, not a background assumption.trust is not binary. it scales with stakes.

8. Maintaining the Network and Recognising Its Shape Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

Maintaining connections is not enough. A learner has to recognise the structure of their own network, evaluate the reliability of its nodes, distinguish independent evidence from repeated transmission, and notice when an established connection no longer serves its purpose. Each of these requires a model of one's own position within the network: not only of the subject studied, but of which sources one leans on and where one's picture is thin.logs help you see the shape

Building that model is the point at which connectivism leads to metacognition. A learner who can see the shape of their network is in a position to ask how well their current judgements are calibrated, and what to do when they are not. The next article in this series examines the self-model a learner uses to regulate their own activity.

References

  1. G. Siemens, "Connectivism: A Learning Theory for the Digital Age," International Journal of Instructional Technology and Distance Learning, 2(1), 2005.
  2. S. Downes, "Learning Networks and Connective Knowledge," IT Forum, 2006. https://philarchive.org/archive/DOWLNAv1
  3. P. Freire, Pedagogy of the Oppressed, trans. M. B. Ramos, Herder and Herder, 1970.