Connectivism

Personal comment: I disagree the theory's definition of learning, but agree with their view of knowledge in the digital era and the consequences of it.

Also, their idea of relating learning with a network/graph is quite useful (even though they may have done it for other reasons).

Lastly, connectivism is distinct but related to connectionism. The latter is a AI/DL framework inspired in biological neural networks.

Connectivism (2004) is "a learning theory for the digital age".

Origin and Definitions

As the digital era unfolds a few important changes become clear:

• Knowledge grows faster and faster;
• Computers and algorithms role changes, and can now store, retrieve, analyse information and follow instructions.

So the environment requires that we:

• Select and discriminate what to (and not to) learn;
• Develop ways to search the best knowledge-sources;
• Understand how knowledge is distributed in human networks, how it grows, changes...

Definitions of Learning

Connectivism defines learning as (bold is mine):

Learning is a process that occurs within nebulous environments of shifting core elements – not entirely under the control of the individual. Learning (defined as actionable knowledge) can reside outside of ourselves (within an organization or a database), is focused on connecting specialized information sets, and the connections that enable us to learn more are more important than our current state of knowing.

It also states:

Learning is a process of connecting specialized nodes or information sources.

So if learning is redefined as knowledge that we can use, then indeed it is being conceptualised like a recipe in a database.

This does not seem a definition most people subscribe to.

What seems clear though is that there may be less need to know (how & what) and more need of knowing-where the knowledge needed is (plus select and delegate).

If instead we define learning as a persistent change in the object's interaction with the world, as a result of a previous interaction then it isn't learning what has changed. What changed are the specific tasks. For example, learning how to search may become more and more important.

By this definition, one may argue that the aspects of learning that matter remain in place: making sense of a new problem.

Also, at a lower pace, we also had to know how to search before the rapid informational changes: find the right expert for a job, record information on a piece of paper and so forth.

But this insight into the dynamics of knowledge-rich era is still a useful idea:

When knowledge, however, is needed, but not known, the ability to plug into sources to meet the requirements becomes a vital skill. As knowledge continues to grow and evolve, access to what is needed is more important than what the learner currently possesses.

Systems That Learn

In their view, not only individuals can learn but also other objects, such as organisations and human (or social) networks.

And, as mentioned earlier, the digital era involves even faster knowledge growth, and what we know will become proportionally less:

When knowledge, however, is needed, but not known, the ability to plug into sources to meet the requirements becomes a vital skill. As knowledge continues to grow and evolve, access to what is needed is more important than what the learner currently possesses.

Either way, representing learning as "CRUD" operations on a graph / network is a useful idea.

They define connectivism as:

Connectivism is the integration of principles explored by chaos, network, and complexity and self-organization theories. (...) Connectivism is driven by the understanding that decisions are based on rapidly altering foundations. New information is continually being acquired. The ability to draw distinctions between important and unimportant information is vital. The ability to recognize when new information alters the landscape based on decisions made yesterday is also critical.

Notice that all the above fits well with a "behaviourist" view of learning but adding an interpretation of some aspects of learning via a graph (e.g. how the brain or an organisation learns via modifications of the graph) and considering that now information changes and grows at a faster rate. Also, considering a network of people as the "knowledge storage device" is quite insightful, as it is also meaningful to consider it from the perspective of a node.

I'd also consider that the process of "digitally storing and updating information" isn't by itself organisational learning. But how the organisation responds to and prepares for new problems and events does fit the behaviourist definition of learning quite well.

Networks

A central object of the theory are networks; as they state:

A network can simply be defined as connections between entities. Computer networks, power grids, and social networks all function on the simple principle that people, groups, systems, nodes, entities can be connected to create an integrated whole. Alterations within the network have ripple effects on the whole.

And also:

The starting point of connectivism is the individual. Personal knowledge is comprised of a network, which feeds into organizations and institutions, which in turn feed back into the network, and then continue to provide learning to individual. This cycle of knowledge development (personal to network to organization) allows learners to remain current in their field through the connections they have formed.

(...)

This amplification of learning, knowledge and understanding through the extension of a personal network is the epitome of connectivism.

Hence structure of interconnected nodes (a network or graph) is an important part of the naming connectivism. The aspects that the links and the nodes represent, how it grows or shrinks, what the link-pattern and architecture look like and so forth will change from problem to problem.

A set of concrete instances of the idea of a graph seems to inspire connectionism, which appears in the area of human and organisations learning.

Examples of Networks

1. The Internet, and later on the World Wide Web
   • Seems to have inspired this theory the most: nodes/documents becomes more useful, by being accessible and interlinked; suddenly, the individual is connected via a "fast"-link to a world wide network of documents; this idea also shines light on what an organisation is and how to improve it (more accessible, diverse knowledge for example)
   • These both were from the start problems about networking; heterogeneous devices or resources linked together,
   • These technologies opened up access to a network of resources and people.
   • The world changed towards quicker decision-making, but I argue pattern-matching, and thinking strategies have not changed much.
2. Neural Networks, both biological and artificial;
3. Certain aspects of organisms and organisations. The paper linked above states:

   The organization and the individual are both learning organisms. Increased attention to knowledge management highlights the need for a theory that attempts to explain the link between individual and organizational learning.

As said, networks are the main object of the theory, so let's briefly define them:

Networks: collections of linked nodes or items.

What is considered a node is defined by us. It could be an individual, a group, a government, countries that trade, libraries, and so forth.

There are types of nodes and edges (or links) with different properties. For example, a node may have restricted access to a resource. There are also topologies with specific wiring pattern.

Knowledge and Learning

The theory views knowledge and learning as properties of networks, and defines them as:

• Knowledge

  • Network view: it is stored distributed across the net (in nodes, edges) in a latent manner. Distributed knowledge means that no single node or edge is uniquely responsible for a concept, rather, many parts of the whole network are. There will be nodes that are more relevant in some situation than other, but the knowledge is still distributed across nodes and links.
  • Behavioural view: It is evidenced as an appropriate response to an input signal or stimuli.

• Learning

  • Network view: a change of network connections by exposure to experience and reasoning.
  • Behavioural view: a persisting change in knowledge. Learning is evidenced through a change towards more appropriate behavioural responses (to a signal or stimuli.)

In a network, one node or link changing means much more has changed due to the ripple effects given by the links.

Discussion on whether all networks really learn are left for the next post. This one focuses on learning-human-networks and how the theory can help understand and design them.

A human in the net

A promising area for connectivism is as a paradigm to think of human networks (or human-and-resources networks), how to organise them (or letting them self-organise), which properties make nodes and the net grow and so forth.

It appears of especial important in the digital, interconnected world; but also classrooms, organisations, and social networks in general.

Let's consider the case of a human or agent embedded in a wider network such as a classroom, a group, community, organisation.

This networks can be seen as an metaphorically organism, or just as a whole: when the connections and nodes change and so does the behaviour in response to input signals, which in certain cases will imply learning, and formation of knowledge.

Example

Think of the response of an organisation in the event of a fire or an emergency.

The input (fire alarm) ripples through the network as the nodes act and propagate activity: learning isn't only intra-personal, the whole network can be said to have learnt in this example.

In the case of a classroom, this framework may aid us with questions such as:

• Centralised human-learning-network guided by teacher or a decentralised network? Should the teacher be more like a routing node?
• Are all items considered part of it, such as resources, or agents such as humans? Is it useful to split or join different networks?
• An open network where agents can change resources, or closed networks where they only have read permissions (if even so)?

A human is a net (theory of mind)

We also want to focus on the individuals. Can connectivism help?

Their general claim is that any network can learn by modifying connections and nodes of the network. And stores distributed knowledge.

So can parts of the brain, which are networks of neurons, and learning is later on evidenced on the behaviour of individuals.