What Connectivism Is

Posted to the Connectivism Conference forum (which hits a login window - click 'login as guest' (middle of the left-hand column) - I'm sorry, and I have already complained to the conference organizer).

At its heart, connectivism is the thesis that knowledge is distributed across a network of connections, and therefore that learning consists of the ability to construct and traverse those networks.

It shares with some other theories a core proposition, that knowledge is not acquired, as though it were a thing. Hence people see a relation between connectivism and constructivism or active learning (to name a couple).

Where connectivism differs from those theories, I would argue, is that connectivism denies that knowledge is propositional. That is to say, these other theories are 'cognitivist', in the sense that they depict knowledge and learning as being grounded in language and logic.

Connectivism is, by contrast, 'connectionist'. Knowledge is, on this theory, literally the set of connections formed by actions and experience. It may consist in part of linguistic structures, but it is not essentially based in linguistic structures, and the properties and constraints of linguistic structures are not the properties and constraints of connectivism.

In connectivism, a phrase like 'constructing meaning' makes no sense. Connections form naturally, through a process of association, and are not 'constructed' through some sort of intentional action. And 'meaning' is a property of language and logic, connoting referential and representational properties of physical symbol systems. Such systems are epiphenomena of (some) networks, and not descriptive of or essential to these networks.

Hence, in connectivism, there is no real concept of transferring knowledge, making knowledge, or building knowledge. Rather, the activities we undertake when we conduct practices in order to learn are more like growing or developing ourselves and our society in certain (connected) ways.

This implies a pedagogy that (a) seeks to describe 'successful' networks (as identified by their properties, which I have characterized as diversity, autonomy, openness, and connectivity) and (b) seeks to describe the practices that lead to such networks, both in the individual and in society (which I have characterized as modeling and demonstration (on the part of a teacher) and practice and reflection (on the part of a learner)).

Response to comments by Tony Forster

A link to my paper 'An Introduction to Connective Knowledge' will help with some of the comments in this post (long, sorry).

Tony writes, "Knowledge is not learning or education, and I am not sure that Constructivism applies only to propositional learning nor that all the symbol systems that we think with have linguistic or propositional characteristics. "

I think it would be very difficult to draw out any coherent theory of constructivism that is not based on a system with linguistic or propositional characteristics. (or as I would prefer to say, a 'rule-based representational system').

Tony continues, "The Constructivist principle of constructing understandings is an important principle because it has direct implications for classroom practice. For me it goes much further than propositional or linguistic symbol systems."

What is it to 'construct an understanding' if it does not involve:
- a representational system, such as language, logic, images, or some other physical symbol set (ie., a semantics)
- rules or mechanisms for creating entities in that representational system (ie., a syntax)?

Again, I don't think you get a coherent constructivist theory without one of these. I am always open to be corrected on this, but I would like to see an example.

Tony continues, "I am disturbed by your statement that "in connectivism, there is no real concept of transferring knowledge, making knowledge, or building knowledge" I believe that if Connectivism is a learning theory and not just a connectedness theory, it should address transferring understand, making understanding and building understanding."

This gets to the core of the distinction between constructivism and connectivism (in my view, at least).

In a representational system, you have a thing, a physical symbol, that stands in a one-to-one relationship with something: a bit of knowledge, an 'understanding', something that is learned, etc.

In representational theories, we talk about the creation ('making' or 'building') and transferring of these bits of knowledge. This is understood as a process that parallels (or in unsophisticated theories, is) the creation and transferring of symbolic entities.

Connectivism is not a representational theory. It does not postulate the existence of physical symbols standing in a representational relationship to bits of knowledge or understandings. Indeed, it denies that there are bits of knowledge or understanding, much less that they can be created, represented or transferred.

This is the core of connectivism (and its cohort in computer science, connectionism). What you are talking about as 'an understanding' is (at a best approximation) distributed across a network of connections. To 'know that P' is (approximately) to 'have a certain set of neural connections'.

To 'know that P' is, therefore, to be in a certain physical state - but, moreover, one that is unique to you, and further, one that is indistinguishable from other physical states with which it is co-mingled.

Tony continues, "Connectivism should still address the hard struggle within of deep thinking, of creating understanding. This is more than the process of making connections."

No, it is not more than the process of making connections. That's why learning is at once so simple it seems it should be easily explained and so complex that it seems to defy explanation (cf. Hume on this). How can learning - something so basic that infants and animals can do it - defy explanation? As soon as you make learning an intentional process (that is, a process that involves the deliberate creation of a representation) you have made these simple cases difficult, if not impossible, to understand.

That's why this is misplaced: "For example, we could launch into connected learning in a way which which forgets the lessons of constructivism and the need for each learner to construct their own mental models in an individualistic way."

The point is:
- there are no mental models per se (that is, no systematically constructed rule-based representational systems)
- and what there is (ie., connectionist networks) is not built (like a model) it is grown (like a plant)

When something like this is said, even basic concepts as 'personalization' change completely.

In the 'model' approach, personalization typically means more: more options, more choices, more types of tests, etc. You need to customize the environment (the learning) the fit the student.

In the 'connections' approach, personalization typically means less: fewer rules, fewer constraints. You need to grant the learner autonomy within the environment.

So there's a certain sense, I think, in which the understandings of previous theories will not translate well into connectivism, for after all, even basic words and concepts acquire new meaning when viewed from the connectivist perspective.

Response (1) to Bill Kerr

Bill Kerr writes, "It seems that building and metacognition are talked about in George's version but dismissed or not talked about in Stephen's version."

Well, it's kind of like making friends.

George talks about deciding what people make useful friends, how to make connections with those friends, building a network of those friends.

I talk about being open to ideas, communicating your thoughts and ideas, respecting differences and letting people live their lives.

Then Bill comes along and says that George is talking about making friends but Stephen just ignores it.

Bill continues, "Either the new theory is intended to replace older theories... Or, the new theory is intended to complement older theories. By my reading, Stephen is saying the former and George is saying the latter but I'm not sure."

We want to be more precise.

Any theory postulates the existence of some entities and the non-existence of others. The most celebrated example is Newton's gravitation, which postulated the existence of 'mass' and the non-existence of 'impetus'.

I am using the language of 'mass'. George, in order to make his writing more accessible, (sometimes) uses the language of 'impetus'. (That's my take, anyways).

Response (2) to Bill Kerr

Bill Kerr writes, "Words / language are necessary to sustain long predictive chains of thought, eg. to sustain a chain or combination of pattern recognition. This is true in chess, for example, where the player uses chess notation to assist his or her memory."

This is not true in chess.

I once played a chess player who (surprisingly to me) turned out to be far my superior (it was a long time ago). I asked, "how do you remember all those combinations?"

He said, "I don't work in terms of specific positions or specific sequences. Rather, what I do is to always move to a stronger position, a position that can be seen by recognizing the patterns on the board, seen as a whole."

See, that's the difference between a cognitivist theory and a connectionist theory. The cognitivist thinks deeply by reasoning through a long sequence of steps. The non-cognitivist thinks deeply by 'seeing' more intricate and more subtle patterns. It is a matter of recognition rather than inference.

That's why this criticism, "Words / language are necessary to sustain long predictive chains of thought," begs the question. It is leveled against an alternative that is, by definition, non-linear, and hence, does not produce chains of thought.

Response (3) to Bill Kerr

Bill Kerr writes, "I don't see how what you are saying is helpful at the practical level, the ultimate test for all theories."

This is kind of like saying that the theory of gravity would not be true were there no engineers to use it to build bridges.

This is absurd, of course. I am trying to describe how people learn. If this is not 'practical', well, that's not my fault. I didn't make humans.

In fact, I think there are practical consequences, which I have attempted to detail at length elsewhere, and it would be most unfair to indict my own theoretical stance without taking that work into consideration.

I have described, for example, the principles that characterize successful networks in my recent paper presented to ITForum (I really like Robin Good's presentation of the paper - much nicer layout and graphics). These follow from the theory I describe and inform many of the considerations people like George Siemens have rendered into practical prescriptions.

And I have also expounded, in slogan form, a basic theory of practice: 'to teach is to model and demonstrate, to learn is to practice and reflect.'

No short-cuts, no secret formulas, so simple it could hardly be called a theory. Not very original either. That, too, is not my fault. That's how people teach and learn, in my view.

Which means that a lot of the rest of it (yes, including 'making meaning') is either (a) flim-flammery, or (more commonly) (b) directed toward something other than teaching and learning. Like, say power and control.

Bill continues, "Stephen, your position on intentional stance sounds similar to Churchland's position on eliminative materialism."

Quite right, and I have referred to him in some of my other work.

"Other materialist philosophers, such as Dennett, argue that we can discuss in terms of intentional stance provided it doesn't lead to question begging interpretations."

Well, yes, but this is tricky.

It's kind of like saying, "Well, for the sake of convenience, we can talk about fairies and pixie dust as though they are the cause of the magical events in our lives." Call it "the magical stance".

But now, when I am given a requirement to account for the causal powers of fairies, or when I need to show what pixie dust is made of (at the cost of my theory being incoherent) I am in a bit of a pickle (not a real pickle, of course).

The same thing for "folk psychology" - the everyday language of knowledge and beliefs Dennett alludes to. What happens when these concepts, as they are commonly understood, form the the foundations of my theory?

"Knowledge is justified true belief," says the web page. Except, it isn't. The Gettier problems make that pretty clear. So when pressed to answer a question like, 'what is knowledge' (as though it could be a thing) my reponse is something like "a belief we can't not have." Like 'knowing' where Waldo is in the picture after we've found him. It's like recognition. And what is 'a belief'? A certain set of connections in the brain. Except not that these statements entail that there is no particular thing that is 'a bit of knowledge' or 'a belief'.

Yeah, you can talk in terms of knowledge and beliefs. But it requires a lot of groundwork before it becomes coherent.

Bill continues, "Even though we don't understand 'constructing meaning' clearly we can still advise students in certain ways that will help them develop something that they didn't have before."

What, like muscles?

Except, they always had muscles.

Better muscles? Well, ok. But then what do I say? "Practice."

"I think it's more useful and practical to operate on that basis, for example, Papert's advice on 'learning to learn' which he called mathetics still stands up well."

But what if they're wrong? What if they are exactly the wrong advice? Or moreover, what if they have to do with the structures of power and control that have developed in our learning environments, rather than having anything to do with learning at all?

"Play is OK" has to do with power and control, for example. "Play fosters learning" is a different statement, much more controversial, and yet more descriptive, because play is (after all) practice.

"The emotional precedes the cognitive." Except that I am told by psychologists that "the fundamental principle underlying all of psychology is that the idea - the thought - precedes the emotion."

And so on. Each of these aphorisms sound credible, but when held up to the light, are not well-grounded. And hence, not practical.