Unedited AI-generated transcript of a talk given online April 27, 2021.
After remember to start to record. There we go. Now, I've started the video recording as well. Now, this is we have two hours and I've got lots of slides and lots of talk. And we can do that and that's no problem. But I want to emphasize that this isn't supposed to be a whole lot of information that I present to you and that you remember.
This isn't that at all. I am creating a record of this so you can go back and look it up anytime you want. You don't need to learn it. I want you to you know, kind of sit back. I don't be idle don't just passively sit and listen. Maybe keep a note to the side.
Of your page or something like that. Not to record what I said but to record your thoughts and I want you to think as I go through this presentation think about what I'm saying from your perspective from the perspective of your experiences and think about how you would use the ideas that you see me presenting in something that you were doing perhaps a blog post that you're writing perhaps a class that you're.
Creating. Maybe think of a conversation that you might have between you and your colleagues. Perhaps even focused on one single slide not the whole talk. So don't just take this as me presenting a whole bunch of information. I know it will look like me presenting a whole bunch of information.
But I don't want you to think of it that way. So far as interaction is concerned. I have access to the chat and I've set it up so I can watch the chat while I'm giving the talk. And so, that means that if you type something into the chat.
I'll see it. I might not respond to it, but I might You never know. Also, if you would like to speak that's fine with me, but you have to figure out how to get the attention of the organizers. I'm not sure there is an opportunity for too much speaking, but if there is.
That's fine as well now your words will not show up on my recording. But if you are recording on your side as well, then your words may well be captured by the WebEx recording.
So that's. The ground rules if you will to start. So. The topic and I hope you appreciate my flower picture is simply connectivism. And. I haven't spoken about connectivism for quite a while. I wrote a paper a couple of years ago called recent work in connectivism, but you know, I haven't been developing the theory or writing on the theory.
Although I could say and I would say that all of my work really is involved in different aspects of connectivism. But my approach to the theory isn't one where I have to develop a theory and then spend the rest of my life promoting it and defending it. I don't see theory or science in general as you know, being like politics or popularity contests the theory stands or falls on its own merits whether or not I promote it.
Indeed whether or not I talk about it. And I want you to think of this theory as. My perspective. On these topics. I'm I'm pretty sure that I'm on the right track otherwise. I wouldn't be talking about it but everything I have to say here today should be subject to empirical test.
Should be subject to credibility you should not simply take my word for it. I'm only one person. I only have one hifetime to work with and even though I've been able to collect from and read a large number of very talented and insightful people. I still only have one perspective on that and our understanding of how learning works and how knowledge works will transcend any.
Thing that any one person could could say could write could do. Oh I only have three I actually have four parts.
First of all, I'll talk about what connectivism is. Secondly. I'll talk about connectivism as a learning theory looking at how learning actually does occur. Third. I'll talk about interpreting connectivism. And then fourth and pretend it's there. I'll talk about connectivism as a theory of pedagogy or a theory about how to learn now.
I've roughly set this up as one half hour for each section. I'm gonna try to keep to that schedule. I'm very bad at keeping the schedules but I'll try. So let's begin then with what is connectivism. Well connectivism has to do with learning. And we ask to begin what is learning and there have been all kinds of choices or options or theories presented over the year.
Gonyea, for example says learning is a change and human disposition or capability. Which is a theory that reflects a behaviorist approach as characterized by say gilbert rail. From a more cognitive cognitive as to perspective mayor talks about learning being a change in a person's knowledge. Changes also. At the cornerstone of Bingham and Connor's argument that learning is a transformative process of taking in information.
And we also have the sense of acquisition or acquiring knowledge and skills from both Smith and Brown. I don't think learning is any of that and I think these theories are incorrect in some important ways they are what I call blacks box theories. And what I mean by a black box theory is that they don't tell us exactly what is happening.
When somebody says learning is a change in disposition. What that means is that they behave differently after learning than they did before learning but how does that happen what makes that happen we don't know if somebody says somebody acquires information. Again, that's something that's happening inside a black box but does it mean to say they acquire information did they put information in their head so I don't really think that's the case.
For me 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. I've talked about and use that definition many times and today I'll talk quite a bit about what I mean by that.
Okay. Yours is saying are you using a presentation slide in this show it's not shown to us, oh, I see what's happening.
Okay. I'm using a single video camera to share my video and my screen but I see it's only showing up is very small in your presentation, there should be a way to make. To to make it. Yeah, you can set the speaker in the stage, so the organizer has to do that.
I believe or I can change it. Ridge.
Okay good. All right. Yeah, you side by side layout right that's it use side by side layout if you're seeing it only very small and then if you do that, you can see me nice and big and you can see my slides nice and big. So if that continues to be a problem, let me know every every video conferencing system is slightly different in this regard.
But I do prefer doing it like this rather than just sharing my screen because I think that you being able to see me speaking as well as the slides as well as other things that I will show you as an important part of the presentation and it's better than just staring at a static slide, which is incredibly boring.
I know I've been there.
So. What does it mean then for a connectivist what does it mean for me to talk about learning well when I say that connectivism? When I say that learning is in the formation of connections in a network, I mean that quite literally this is not a metaphor there are a lot of theories of learning are based on metaphors, we'll talk a bit about that this is not a metaphor when a person learns or when something learns a connection actually is physically created between two nodes or two entities in a network.
And what do I mean by a connection? Again, this is an actual description of a physical event not a metaphor not a black box. I say a connection exists between two entities. When a change of state in one entity can cause or result in a change of state in the second entity.
Hello, there are different ways of thinking about that we can think of one neuron being connected to another neuron and the activation of the one neuron changing another neuron or we can think of one person saying something at another person hearing it and so the other person is changed by the first person there are different ways to think of that but the connection needs to be an actual connection not a metaphorical connection not.
A relation not a conceptual relation there needs to actually be a change of state. And again this is all about getting away from the black box right getting away from mysteries and well we don't really know and you know, you just have to conceptualize it for me. I want to know what learning actually is.
So. On that account, what is it to learn? Well learning is a thing that networks do it's a thing that all networks do and arguably a thing that only networks do. And it consists of the following either the addition or subtraction of nodes in the network or you know, the the entities that are connected to the rest of the network.
The addition or subtraction or strengthening or weakening of the connections between those notes. And those first two are known collectively as plasticity and sometimes you'll hear people talk about neural plasticity and what they mean is that in the brain. We sometimes well, we more often lose neurons and gain neurons but the connections form and break between neurons as well the the subject of neuroplasticity is talking about how the brain is learning.
And then we can also talk about changes in the properties of the nodes or the connections. I mentioned, for example the strength of a connection can vary can be a stronger or weaker connection. It might take more or less energy for a change of state in one node to result in a change of statement another node and inside neurons as well as we'll see there can be changes in activation functions.
I want to give you an example of the sort of thing that I mean, hopefully you can hear this as well, we'll see. I'm not hearing it that's not good.
Well what you should see here. Is a bunch of well, they're called metronomes and basically they just beat back and forth and they were started off randomly, so they were not all going at the same time. And as you can see in this video. They slowly. Become synchronized and now if you look at them, they're all going at the same time.
And the question is well what's happening here well each one of these metronomes is connected to the others and the way they're connected is through that piece of wood and that piece of wood is as you can see there that piece of wood is sitting on those two pop canes so that each time a metronome goes back and forth it pushes a little bit on the piece of wood which moves the word back and forth in this feedback.
Between each of these things and you can actually describe it mathematically results in the metronomes becoming synchronized each metronome indirectly reacts to what's happening in the other metronomes. This is an example of what is called self-organization and the idea is that independent things like metronomes in this case that are connected together can by virtue of that connection alone become organized or synchronized themselves without needing any other intervention, it doesn't need direction, you don't need to organize the synchronization there isn't at?
A head metronome. Nothing like that. And that's the sort of thing that I have in mind. I do wonder why that didn't give me audio but oh well.
There are different things that can learn because there are different kinds of networks. This is a really important diagram. On the left hand side. I've stolen a drying all the references for my images are in the notes of this PowerPoint and this PowerPoint will be available after. So on the left we see a neural network.
And on the right we see a social network. And connectivism is about learning both in neural networks and in social networks. And one of the big differences between George Siemens and myself is the way we interpret this picture. And the way George would say it is that this is all one big network that our knowledge consists of all the connections inside our mind and the way this is connected to everything else that's in the world and the way it's connected with itself.
So, our knowledge is partially in ourselves and partially in Twitter or in WordPress or Pinterest or our network of friends, etc. I keep these two separate. I think personal learning is one network and social learning is another network and they're two separate networks. But they interact with each other through the process of perception.
And perception is the way a neural network is able to interact with the social network. Unmanned communication conversation is the way the social network is able to interact with the neural network. So I think that would describe this interface and we'll talk about this a little bit later is a process of emergence and recognition.
Let's look at George's principles for a bit. George Siemens wrote in his paper on connectivism in 2004. And he came up with these eight principles. Learning and knowledge rests in a diversity of opinions and that of course is to state that learning and knowledge existing networks. Not just in one place.
And learning is a process of connecting these notes in the network and that's his second point. His third point learning may reside in non-human appliances, that's the idea. That personal learning and social learning are all one part of one large learning network. He also says the capacity to know more is more critical than what is currently known.
And that's an important point. I won't talk a whole lot about it, but we both agree that learning isn't just the acquisition of content learning is about developing a way of seeing and interacting with the world. For example, we reach his fifth point nurturing and maintaining connections is needed to facilitate continual learning.
And indeed the sixth point the ability to see connections between fields ideas and concepts as a core skill. Now, I ask what does that mean? Because I always ask what does that mean? And we'll talk about that later. Currency says George meaning up to accurate or up-to-date knowledge is the intent of all connectivist learning activities and decision-making is a learning process.
And I'll talk a bit more later on in this presentation about that. I have a video of George speaking. I don't know if we'll be able to make the audio work. It was working before. And now it's not.
And it's still not is it. Now, I'm not hearing it either this is my fault. I messed up the sound let's try opening the sound settings. Nope they're learning and the way in which they connect and well that'll work, okay. Somebody has an idea on a blog over here.
I read about it. I've the late night. Here we go. Sorry about that I started blogging in a late nineties or oh you can ask and as I was blogging I found I had a very different relationship with other people and with knowledge than what I had had when I was learning in a classroom or a university program.
The biggest difference I found was this the way in which people are aware of one another while they're learning and the way in which they connect and build and improve. And so somebody has an idea on a blog over here. I read about it. I write a post and corporate some.
Of those ideas but try to expand it and add my own perspective. I posted someone else comes by and take some of those bits and pieces and expands it and augments and so on. We could say at a very broad level that's exactly what science has done. This notion of combinatorial creativity right that we we we connect knowledge we'd be build we grow we advance.
But at that point it was more for me it was a very odd experience as mentioned as a Red River College and so I was involved in a training or in the in my interactions with faculty. I was one of the few folks in the system at that time that was very active around trying to find new ways to teach with technology.
But I found online all the sudden there was a rich group of blogs and colleagues from other universities other systems and not just in Canada or US, but globally there there were some fascinating things going on in the space. And so my learning in that setting wasn't the instructivist way.
It wasn't necessarily a purely self-regulated way. I Were. I was taking books out of the library and reading them a reading journals and saw it instead it was really a social connected process of learning and that's what eventually resulted in a paper I did in 2004 where I argue that in a network world learning is a network forming process knowledge is a network product, so when we are knowledgeable it essentially is reflection how we've connected concepts and ideas over a period of time that's getting a little bit more detail this happens at three distinct levels, so on the one hand literally at a biological.
Level learning as a network farming process neurons firing neurons connecting if we move that up one level and this is a big level to move up on actually but at a at the same level network learning consists of taking concepts and forming conceptual connections, so as we get a new idea we connected to what we already know and that rounds out our perspective and often deepens our understanding that the final level is one that we're all quite familiar with actually but that's network learning through external social spaces, we might use tools like Twitter or Facebook or any number of few.
Er to emerge network tools, we might be connected through a mobile devices staying and if we have a question why don't say we're a salesperson in a workforce we have a question we can text colleagues or those kinds of activities, but that's where the social systems and the technology systems are now part of human knowledge more and more part of human knowledge become a part of our overall capacity to know.
Okay, so that was George Siemens.
So. Connectivism can be distinguished from other learning theories and a few important ways. One way is to distinguish it from theories that are based on content. For example. Instructivism or transactional distance theory and distance education. Connectivism says that the brain is not a book or a library it's not an accumulation of facts and sentences and propositions that we bring in we organize and we store like a whole bunch of stuff the brain does not get full.
Of too much information there's nothing resembling the pages of a book or the books of the library there's nothing resembling the text and the sentences inside the brain if you cut open a brain or if you analyze a brain you will not see any of that all you will see is this network and this signals that go back and forth between the different entities in the network.
What? I mean by that is that therefore connectivism is non-cognitivist. You've probably seen a lot about knowledge and learning but is based on cognitivist theories of mind. Cognitivist theories of learning where they talk about sensory memory and working memory and long-term memory. They talk about encoding and constructing schemas and cognitive load.
All of this is from a metaphor of the computer is being like an information processing system. Very much like a computer. But the mind does not like that. There are ways that we could interpret some cognitive phenomena using the metaphors of working memory for example or cognitive load, but these are not descriptions of processes.
These are not descriptions of actual learning that occurs. As I say the brain is not a computer. If you think about it. You know, we we have theories that tell us that learning is about constructing knowledge or representing reality. A lot of constructivist theories tell us this. But what does that mean and again we run into this black box problem?
You know for the brain to be like this. It has to be some kind of representational system. Like a language or a logical system, or even in a graphical image system or some other symbol set. Some other system of signs where the signs represent objects. I guess out there in the world.
And there would have to be rules or mechanisms for creating entities and manipulating entities in that representational system. And there's nothing like that if we look at actual cognition, there's nothing like that that's happening. And it's interesting because you know people are using this computer metaphor to talk about human learning when even artificial intelligence doesn't use this model anymore.
This model characterizes what we used to call expert systems. Or symbol based systems or rural based systems of artificial intelligence. But in fact almost all artificial intelligence has moved away from this model and uses the model of neural networks. So we'll talk quite a bit more about that.
Connectionism to me is a non-representational theory. That makes it very different from other theories. What? I mean by that is there's no real concept of transferring knowledge making knowledge or building knowledge. Rather learning and knowing our descriptions of physical processes that happen in our brains. And when we learn when we know what we're doing is more like growing and developing ourselves the way we might build and that has to take to do this but the way we might build a muscle.
Okay is the way we build learning and you don't tell a muscle okay muscle. Now you will get bigger. You don't acquire new physical strength. That's not how it works. And I don't know why people when they're talking about learning would think that it's different. We're working with a physical system the human body composed of physical properties and in particular a neural net that grows and develops based on the experiences it has in the activities and results of those activities in undertakes.
We're running a little bit behind but not too bad. How does learning occur? Now we get into some of the fun stuff about connectivism. So, they're a little out of theories over the years about how learning occurs and in fact there's a whole domain of learning theories about processes.
There's called for example. There's Dewey's model of experiential learning. You notice they're all kind of like loops and they're all kind of doing the same sort of thing. You know, concrete experience observation theory and deductive inferences on and on called for example, now what these are doing. Or talking about the processes that create learning and in fact.
We could go on forever talking about the processes that create learning and talk about whether this process is better than that process. But what we're doing is we're describing the conditions around the person rather than the person themselves. And we're talking about the sorts of activities like gun gays nine events of instruction.
You know talking about the activities that are set up and organized by an instructor or a teacher rather than what learning is like from the perspective of the individual.
Here's what learning is like from the perspective of the individual now this is just one of many kinds of neural network, we'll talk about that but this kind of gives you an idea. Look at the way this network works we have what we might call an input layer, whoops.
We have an input layer of neurons these are connected to a second layer which identifies edges these are connected to a third layer that combines edges these are connected to a fourth layer that identifies features now this is the sort of processing. If we can call it processing that happens in the visual cortex that's located back here comes in through your eyes and sits back here and what it does is it takes all the input impacting your eyes and detects what?
I believe it was mar described as the two and a half dimensional sketch edge detection and all of that. Now these are our interpretations of what these neurons are doing these neurons aren't actually saying oh I'm looking for an edge right that's not what's happening these neurons are simply receiving input and then sending output that's all they're doing they're not intended to or created to detect edges that's the interpretation that we put on that's what we say that they are doing from our perspective.
Now George talks about networks quite a bit and he talks about networks having various characteristics and I don't think that any of this is wrong at all but I think we want to sharpen our discussion of it so he talks about networks having content. I'd rather talk about networks having signals than content people talk about the data or the information.
In the brain or in in our mind, these are very technical terms. A data represents a fact information strictly speaking if we follow say dreets key is the reduction of possible states of affairs in the world from the point of view of the receiver. Not really specific thing to say is in a network, so I'd rather just say signals.
In that works as George says there are interactions there are connections forming. There are also signals sent through these connections that's how we get interactivity one entity sends a signal to another entity that has the potential of changing the state of that entity. There are static knowledge structures and we'll talk a little bit about that when we talk about distributed representation but it's also dynamic it's also constantly receiving new signals as George has new information new data.
I would say new perceptions. There are he says self-updating nodes. Will be a little bit more precise and then emotive elements. You know, it's not just about visual sensation it's not even just about the five senses our neural network is connected to all aspects of our body so as David Hume what's famously said, you know, you know a dream might just be a pain in my gut yeah our emotions our sense of movement a vertical sickness nutrition all of these have an impact on our neural networks.
We still want to talk about this more precisely, though. When I've talked about learning theory in the past, I've talked about ways of creating. Connections between entities ways of creating these networks and I talk about four major. Types of connectivity something called heavy and rules. Which is basically the principle with fires together wire together if you have this neuron and this neuron and they both fire and they both stay silent and they both fired and they both stay silent at the same time, they will eventually grow a connection between each other fire silent growing the connections fire sign on a growing the connection.
That's the simplest form of network formation another type of. Network formation is contiguity. That works at our beside each other will organize so that they form I don't want to say a lattice or anything like that. It's a bit too complicated but for example the different cells in your eye are arranged beside each other and that informs how they're connected to the different layers of the visual cortex.
Another method for. Learning in networks is something called back propagation. Now, it's not clear that back propagation works in human neural networks. Although we do say, you know, people learn from feedback. Certainly back propagation is used in artificial neural networks. It was developed by Rebel Heart and McClellan in the 1980s and for a long time was the most promising form of neural network learning.
And my personal favorite is bolts and connectivity and that's the idea that a network tries to achieve the most thermal dynamically stable state. So the connections in the interactions are all the network trying to settle in to the most stable configuration. Sometimes we talk about that process as being similar to annealing or the way you hardened metal by heating it up and cooling it down.
I like to think of it is like when you throw a rock into a pond. Oh, you know a pond is composed of atoms of water and these atoms of water slosh and jostle but eventually they all settle into a nice flat state again. Now, these are rough generalizations and when you actually look at actual neural networks that people are creating you won't see these rules in particular.
When these are ways of thinking about types of rules. When we look at actual artificial neural networks, there's two major categories machine learning and deep learning machine learning involves some human intervention to identify or classify the data deep learning uses layers of neurons to identify and classify data on their own deep learning is much more like human learning than machine learning because we don't have little men inside our heads.
Classifying features for us. Some of the theories that I read make it sound like that, but we don't. There aren't.
Now. What makes and this is one of the key questions what makes neural networks work. Well, you know, I mean right now in the field of artificial intelligence, the answer is trial and error they're trying out a lot of different neural networks and some of the work. But really what it boils down to is the different way she can organize and set up these networks and then you set them up so that they can adjust to meet if you will the sort of output that's desired.
And that's where the concept of training will come in will come back to that. So here, for example is what might happen in one single neuron. We have some connections coming in from other neurons. Other entities in the network. Each connection has an input each connection has a weight.
And then our neuron in this case adds them all up. That's what that little sigmoid says adds them all up. Now we can easily imagine different kinds of functions there where it doesn't just add them up maybe it takes the strongest one, maybe it takes the strongest two, maybe takes the middle to and drops the strongest in the weakest, you know, in many ways we can discuss we can describe this and as an aside.
In artificial neural networks, we're only working with a few different types of neurons but in a human brain there are thousands of different types of neurons and I think that kind of diversity in human brains is really important but we're a long way from understanding how that plays into this how different types of neurons interact together.
Anyhow. This neuron will take in all of this information and do something with it and then it goes through what we might call an activation function and what that means is does the neuron send the signal on to the next neuron. Now these in actual neurons these activation functions are built from electrical potentials, and somebody who is far more versed than I am would talk about the chemistry of how the electrical potential.
Of the neuron builds until it reaches a certain point and then the neurons spikes where sends a signal and then the potential drops back down to zero until new signals increase it. There are very detailed descriptions on a molecular level of what's happening inside an iron of this type.
Also, now you can have. Different types of networks. I talked earlier about a heavy in network. Here this is the same sort of thing only here we're not going to call it a heavy network. We're going to call it a feed forward neural network or perceptron. This was one of the earliest types of neuron artificial neural networks.
And you can see the signal goes through the layers and in an individual neuron the signals come in they're added the activation function. Decides to fire or not and then it fires. These are simple. Networks they can only be used for what we call linear regression. In other words, straight line or curved line linear relationships between things.
So can look at tabular data image data text data, but it can't learn about complex relationships. A different model called a recurrent neural network will actually feedback into itself. There's a little loop there you see that and that way it preserves data that comes as a series, for example the words in a sentence or events that happened in your life.
And so here these neurons are able to get. You know, each word in turn as part of an overall. State internal state and then it's going to determine whether or not it fires off again in similar manner as the other neurons that we talked about. And it's used to solve problems related to time series data text data, like language translation audio data such as automated transcription.
I'm using such a system on my phone right now. It's recording the audio and as I record the audio it's actually transcribing the text of what I say and it's using a network similar to this to do that. There's another type of network. Called a convolution neural network that captures in this case the spatial features of an image.
And then bring some together. It looks at in other words the arrangements of pixels in what's presented. This is similar to the contiguity type of network that I talked about earlier where the neurons are related to each other according to how they are organized on the retinal cells how they're organized in the eye.
And there are more. There are even more networks. But I want to give you a feel for what they're like, I feel for what the different types of neural networks look like.
You see how it's recognizing the different numbers from the perceptual input?
So, What I hope you get from that. Is not not only. The idea here that you know, the the recognition and interpretation of in this case these numbers but you know pretty much anything is or can be done by a network but also the incredible complexity of the networks that are doing this processing these were, you know, networks of thousands in some cases millions of connections.
And that's still a small fraction of the types of connections that we can have in the human brain. We have a hundred billion neurons in the brain and many more connections. So we have incredible complexity incredible capacity. To. Well I was going to say process information, but that's not what we're doing is it to receive and transmit and reorganize and grow.
To produce interesting and relevant phenomena for me to see something and respond. You know, the change in disposition as some people would say but a change of disposition that is learning that can be explained by a change in the neural network. Now I don't think that will ever be able to say this specific change in the network produces this specific type of learning.
Because although we are all human and we all start off with a similar set of neurons our experiences. Change their different for each one of us from the moment of our birth. So your neural net and mind your own net although there may be similarities, they are nonetheless going to be different and the sort of things that would lead you to say something and would lead me to say something are very different and a good example of that is language.
I have a word flower that I used to represent or not even a represent but to talk about things that I see you have a different word in your language neither of those words is right, they're both based on the different background and the different experience that we've had in our lives.
And that leads us to training. Because in all cases. Neural networks are created by what is called training now in the world of education the word training has a bad connotation, let's leave that aside for the moment, you know, we're not worried about you know, whether something is vocational or academic or anything like that.
Training in this context is just the process. That a person goes through or even more accurately the process that a network goes through through repeated iterations of experience to acquire the configuration the set of connections and the set of weights of connections. Appropriate to whatever it's experiencing. But there are different kinds of training young epochs and iterations there's a whole language describing that but essentially it's this ongoing configuration on ongoing.
Feed its new century data process and again processes the wrong word adapt to that data.
So a long time ago, 2005. I was thinking about all of this and thinking about what George had done and thinking about the earlier works earlier working networks that I had done and thinking of this from the perspective not of artificial neural networks, but of networks generally. And I was asking myself.
What would make a good network? Well, because we want a network that's able to learn what we mean by that. Well, we want it to be able to respond appropriately. To the stimulus the perceptions that it encounters whatever that means and we want it to avoid. Some of the problems that might be caused in a network.
I remember in the 19. 90s watching a talk by Francisco Varela. Where he talked about the immune system and he talked about you know, finding that ideal weight of or the ideal structure of connections. And you don't want no connections. Things have to be connected somehow otherwise milk communication happens.
But by the same token you don't want everything to be connected to everything. That results in chaos is just nothing but loud noise. Kind of like our social media situation today where everybody's connected to everyone on Twitter and Facebook and nobody can make sense of anything. So I was thinking about well, what would constitute good design principles?
Now, these are hypotheses, you know, they're not rules. They're things they're principles that I think would be relevant when we're thinking about designing networks. So, here's some of them. First one is decentralized. This is the description of how you connect the different entities together and they're different ways entities can be connected.
But we saw in the the diagrams of artificial neural networks some decentralization. In the form of layers of neurons and that is a good way to decentralize but we can think of this as a more generic principle thinking of. Network in general as resembling a mesh more than a star.
And here's what I mean by that. This is. A network and so is this but in this kind of network we have this characteristic star formation as contrasted to this kind of network, which is a mesh. As contrasted to this kind of network, which has no connections whatsoever. So this obviously has too few connections there are no connections this also arguably has too few connections it puts too much emphasis and too much reliance on a few neurons a few entities which creates the possibility for failure if this neural neuron failed right here, that would be a catastrophe.
So what we're thinking about networks maybe in terms of organizing a company or a society putting all of your emphasis on a single neuron is to create a single point of failure and there's a lot of discussion about how a mesh kind of structure is a better structure for society and a better structure for networks in general.
Here's another. Another principle distribute to distribute your entities. Even in the brain are neurons aren't all located in the same part of the brain, in fact our brain actually has different different lobes. You know, there's a lot about to start listing parts of the brain because I've forgotten them but you know, there's a hippocampus and the cerebellum etc.
What we're thinking about networks in society. We don't want to put our entire network inside a single building. That would be a bad idea. Because if the building fails the entire network fails on the internet. We've seen the strength of distributed networks such as peer-to-peer networks. Like email. Or Nutella or content syndication networks like RSS.
And those often are more reliable than centralized networks like Facebook. Or Twitter. In fact, just yesterday. Microsoft teams went down. They went down for the entire network because it's not a distributed network. If we really wanted a good, you know team application, it would not depend on the single central source in a single central application.
Distribution also applies to representation and this is a really important concept. We sometimes think as though in our head there's a specific place where we have an idea. You know, you you point to your head. And you say in my idea of a cat is here my idea of a dog is here, but that's not actually how it works nor should it work that way.
And the reason for this is we want our ideas to be able to relate to each other. In. A natural and easy way. So, The way it works is. This is very simplified obviously but our concept of a cat if you will that is to say the neurons that are activated over time when we see a cat here they are.
And our concept of a dog, these are the neurons over time that are activated when we see a dog. Similarly, these are the neurons are activated when we see a fish. Now, what's important what's important is they're all using the same network of neurons. And that means the cat and dog can overlap.
The dog and fish can overlap the cat and fish can overlap. So what we know about cats influences what we know about dogs. What we know about fish influences what we know about cats and vice versa. That's a really important concept and I wish I had several hours to talk about just this concept.
But what's really important here is this is not the word cat in our brain. There isn't a formal logical structure or classification of cats dogs and fish. They just happen to overlap our perceptions overlap in the neural network the way our neurons are organized, that's it. And that's what we might call a sub symbolic representation.
I again, I don't like the word representation for various reasons, but it's sub symbolic we are not using symbols we are not using. Words. We're not using rules of grammar or anything like that, it's all the activations of neurons the connections of neurons. Another one of these principles was disintermediation.
And this is the idea of removing the barriers between the source and the receiver. This is more of something that works better in in social networks, it's sort of hard to describe in neural networks because when there's mediation in a neural network, we call that a disease and we think something's gone wrong with the brain.
But it happens a lot in social networks where we have editors and publishers etc that stand between somebody saying something to somebody listening to something. And. Where there should be. Mediation is only to mediate the flow that is to say. To make sure that we're not overwhelmed by too much information.
There are there are types of mental diseases where that kind of filter just doesn't work and the person is overwhelmed, you know, they have sensory overload all the time.
Disaggregate. By this what I meant said and this is similar to the whole idea of distributed representation the idea that we should stop thinking of. You know. Our knowledge our information or content as one single unified in divisible whole like, for example, take the concept of dog. It's hard for us to conceive of you know, a dog is not a single entity, you know in one sense we can we can think of a nose and ears and things like that, but on the other hand it's sort of hard not to right the dog really presents itself as an object in the world.
But you know linguists have had a lot of discussion about that, you know. Wilford van or man, for example, what does the word dog mean? Does it also include puppy? Does it also include a three-legged dog? Does it mean the overall state of the universe in the current form of a dog?
There are different ways we can interpret that concept in that perception and none of them is necessarily right and between you and me and another person we might all have different concepts. So we should not assume that the world comes presented to us a nice human shaped objects or human-sized objects it probably doesn't.
And these are interpretations that we are opposing onto the world.
Disintegrate. Entities in a network are not components of each other, they're independent. The structure of an entity sending is logically distinct from the entity receiving that's an important concept in systems theory. Which is different from connectivism all of the different parts of a network are all part of a whole and typically in systems theory that whole.
Is trying to move towards some objective or some purpose. Hey a lot of times people interpret say Darwin's theory of evolution that way that we are all evolving and that there are higher and lower forms of life. But evolution doesn't have a purpose. You know, we don't grow wings in order to fly.
Rather see other way around we're able to fly because through this process of random selection we've grown wings and that gave us an advantage. And that's why I say disintegrate, you know, the the entities we think of them as a single unit, but they're not. I sometimes represent this.
By talking about the distinction between learning using stories and learning using maps. Suppose. I'm discovering in new city. Well there's always somebody in the city who tells me how to get to Franklin's square say and they say oh you take this road to the end of the park then you take this street and then turn right here and then turn right again and then go up 17th and you're at Franklin Square.
And that's nice except that presents something very complicated very complex as though it were a single thing. But there are as you can see these alternative roots or I could choose to go to different places but even more to the point if my knowledge of the city is represented whoops as only this line then if this line breaks in any point if I miss a turn or something like that then I have no understanding at all.
But if I see the city as a set of distinct entities all kind of related to each other with multiple routes between them then my understanding of a city is much better.
I've also talked about the principle of democratizing networks. And when I first formulated it. I talked about the need for the entities to be autonomous and the need for them to be diverse and again I talked about the many different types of neural networks and we can talk about the many different types of people the mainly different points of view in a society which was also an asset.
I think that's something Malaysia in particular has learned to appreciate over the years with the many distinct cultures that it has. I know that I experienced that when I was in Kuala Lumpur. It was one of the really interesting things about the city. Over time. I've come to talk of this as the semantic condition.
This is the condition that for better or worse ensures that the network will stay coherent and the network has the capacity to respond to changing circumstances in the environment. Networks of characterized between groups and networks you might recall before I showed the diagram with the star and the mesh where the group is the star and the network is the mesh on this picture.
These four principles diversity autonomy openness and interactivity are principles that describe networks that are capable of adapting to changing circumstances. These principles by contrast represent a type of brittleness or fragility on the part of the network, it makes it more difficult for the network to adapt it makes it more difficult for it to learn and to grow.
So for example, unity. Now if everything is the same in a network. There's nothing for the members of the network to talk about. You can't be any change because everything's the same. Now of course, nothing is ever completely the same but a lot of times people push for more sameness in a society rather than less sameness.
In a human that would be a disaster. And in a society that makes it more difficult for the society to recognize different points of view that might allow it to recognize changes in say the environment different. Similarly with autonomy. If the individuals in the network, although still related to each other.
I'll still although still influencing each other if they are nonetheless each of them. Deciding we're not deciding but each of them sending signals according only to the input that they receive on their own. Then that network is more responsive than a network that requires coordination. Sometime now in a human it almost makes no sense to talk about coordination now some cognitive theories talk about an executive function.
That's like a little man talking in your head. And there is no little man talking in your head. In a society the more coordination you require the longer it takes to adapt to change. So the more autonomy you are able to have you know, this isn't him absolute but no more autonomy you're able to have the faster a society will be able to respond we've seen that a lot in military organization we I've been involved in various studies of military networks and a military is the classic example of coordination right everybody does what the general says but in fact when a military is engaged they try.
To make each of the units as autonomous as possible so that they can react independently to whatever situation they see in front of them because otherwise they'd be unable to react and they'd be vulnerable. Another principle is openness and there are different ways of thinking of openness openness of membership, so a network is able to add entities this makes growth a possibility for networks also open this from the perspective of if you will experience.
Or concepts. It's like having bridges to the rest of the world. And again, this gives it the sort of input that it needs in order to be able to react as compared to closed networks. Classic example of a closed network is the Apple ecosystem. Apple computers. Right. And a haploid computers famously don't work well with other things.
And you're as they say locked in to the Apple computer. If you want a new capacity, you can only one source to get it. You have to buy it from Apple. Then finally interactive versus distributive. That is to say Knowledge is created by the interaction between all of the different components in a network.
As compared to distributive where knowledge is created at the center and then sent out. And you see the distinction here. The best you can get from a distributive system is to have copies of whatever there was at the source in each of the different entities. You can never go beyond the capacity of one member of the network.
But when knowledge is created through an interactive process of all the individual entities working and interacting together then the knowledge of the whole can be greater than the knowledge of any individual. And that's a very important principle. We can see that in a human brain, it's absolutely essential because an individual neuron is not very smart.
An individual neuron cannot do anything except receive signals and send signals. But a human brain a hundred billion neurons all connected together is incomparably smarter. Same in a society. Right? Compared to a whole society and individual person is not very smart. Even the smartest person in the world. Is not as smart as society as a whole.
Now, we sometimes act as though they are. But arguably that's because we've organized our society very badly and really we're just comparing one smart person with another smart person. But a society that creates and forms knowledge as a whole. Is much smarter. Than any individual in fact the danger is when the individual substitutes his or her own perspective.
Over and above the society as a whole and and and you get kind of a dissonance. In the network again in a human there's no real example of that because there's no capacity for a single neuron to take charge couldn't happen.
Another principle is dynamize and you can see how this comes into play with the semantic principle. I just finished. And that network needs to be thought of as fluid as changing as growing, you know, it's only through the process of change that a network learns it's only through the process of change that the possibility of growth exists.
And then desegregate as a final principle. By that what I mean is although it's tempting to take your entities and compliment to different groups and say this is one group and this is another group and this is another group. It's probably a mistake to do that because even if we can see patterns and shapes and and you know clusters in a network, they network is nonetheless a single undivided whole and the individual elements we see in it are just our interpretation of what's in the network.
To give you a couple quick examples. This is the web of science now this diagram was created by looking at scientific papers that are published. And looking at the references in those papers. And then just sort of grouping them according to those references and we can see some clusters here's nursing for example, that's the red over here social work and dermatology and the study of skin.
Here's psychology. Down here we have biodiversity plant biology over to the right here we have biotechnology. Moving up into the middle here here up here at the top we associology Asian studies this group over here to the left is education anthropology, all of the different sciences are all on the same graph because they're all related to each other the people who work in these different domains all interact and communicate with each other now one person here doesn't talk to everybody.
Again that's Facebook or Twitter where everyone talks to everybody and all you get is noise right but they're connected just enough so that a concept in firm pharmaceutical research like say the discovery of MRNA as a cure for a virus could connect through the different parts of the network to have an impact say on social work.
And understanding that everything is connected in that way and structuring science so that everything is connected in that way. Probably in my view results in. A web of science that is more likely to grow and develop than if we just kept all of these different disciplines apart from each other.
Here's another example the art network from Barabasi and here's the museum of modern art and that can't even read the other text but again there are relations between all the different institutions and the different artists and the different schools of art and are similar pictures for music and for any domain that you care to think of you know in music we have big debates in this country, do you like rock music or do you like country music?
And you know, it's easy, you know, you want to categorize them into different things so they're so different from each other you have to pick one but they're really two parts of the same thing just different parts of a much more complex network and seeing the world that way and representing in structuring the world that way is more likely to be successful than keeping it all segregated into different parts.
Interpreting connectivism now. I'm running behind quite a bit but these sections are shorter. So I still have hope. So. What do we make of this? I mean. You know, when I talk about interpreting connectivism, it's kind of like interpreting say probability, right? Yeah, you have this theory in case probability all these mathematics but what does that say about the world, you know is probability really the frequency of events that happen like breaking box says or is it the number of total possible states of affairs the way Karna would say, Or is it how much he would bet on a certain outcome the way Ramsey would say.
Well, these are different interpretations different ways of taking this theory and applying it to the world. Now. Connectivism, in fact the whole theory of knowledge has networks. Can be found in the world in a number of different ways and it's interesting one of the advantages of a network way of looking at the world is what might be called parsimony.
Persimmony is the way I'm using it here means that the same theory applies in many different domains. And you know by the theory I might be talking oh I'm not talking necessarily about an underlying foundational theory, but we see the same thing in different places in mathematics. We have graph theory.
In computer science, we have connectionism. Neural networks and artificial intelligence. In biology, we have ecology and ecosystems. In sociology, we have social network analysis and actor network theory. In physiology, we have theories of perception and neuroscience and I've talked about a lot of these and if philosophy information theory and distributed representation.
All of these are different places in very different fields. That we can find the principles of networks instantiated. We see them out in the world as well. Networks in nature have a video. I'll show you in a little bit called a murmuration. We see it in social organization corporate networks political networks.
We see networks same infrastructure such as the electrical grid or the internet a worldwide information network. And we even see networks in social networks. Some of them better designed like the network of websites some of them badly designed to like Facebook and Twitter. Now. Connectivism was postulated in Georgia's original paper as a theory for a digital age.
And you know, I look at that now, he's look at these things unless I say well, what does that mean? Well to me. It comes down to the question of what is knowledge. Now here I mean, I've already said knowledge is the organization of connections in a network. But what does that mean on a day-to-day basis?
If I say Fred knows something. Yes, I'm saying Fred is organized in a certain way. But that's not very helpful to somebody. I'm talking to because that doesn't tell them what to expect. You know, so when I say Fred is organized in a certain way. Fred has a certain neural network.
He's grown his neural network in a certain way and therefore he knows something say mathematics, what does that mean? Does it mean that the person has memorized the multiplication table or the Pythagorithm theorem? Well, no, now let's not what I mean. Knowledge is more than memory. In fact, I don't think knowledge is memory at all.
Memory is like a parlor game where you're able to recite back what you were told. It's like the spelling bees that they have in the United States where you have kids memorizing how to spell obscure words with and without ever using them in a sentence other than in the sentence they use in the spelling bee.
They have no idea what they mean or why the word would be useful. Now on knowledge is something more than that. In my original discussion of connective knowledge, I said, there were three types of knowledge. Qualitative knowledge based on figure or property the color of something the size of something.
Quantitative knowledge, which is based on mass or quantity the number of things the weight of something. These are the types of knowledge you're familiar with and people talk a lot about qualitative and quantitative research, but I think there's this third kind of knowledge which is connective knowledge, which is based on organization and structure.
Okay again though, how was that helping me? How is that making it possible for me to understand what it means to say? Fred know something. Okay. What does it mean to be organized in a certain way? Well, We could represent it using the concept of emergence. And the idea here is that from the interaction of all the entities the way I talked about earlier something over and above that network emerges.
And here I want to show you the picture or the video of a murmuration. You might not be familiar with a murmuration, but here it is.
What you're seeing there are birds. In particular, they're starlings. Now. Stop looking at network like a brain or a computer network, but it's still a network. The Starlings are interacting with each other. And. We can see them for me shapes in the sky. And this behavior of starling is called a murmuration.
Now, we can see circles or kidney shapes we can see something like well we would call it a flock. Or it's moving in a certain direction. But there's no head starling. There's nobody in charge. And in fact the starling at one end of this murmuration over here on the left can't even see the starling on the right.
And in fact in a mermaidation, each bird is responding to a few of the birds around it. And you can actually see the waves that happen when one bird moves and then that causes the next to the move the next to the move and eventually you get a whole group of them.
And in this way suppose, for example, there was a predator that arrived at the edge of the flock all it takes one starling to the act and the entire flock of the reacts. Now, the reason why you say that this is the, Because what we can say of the flock does not apply to any individual.
And above the knowledge of any individual starling. And so when we look at the image that I shared. The same thing is true. If we look at that what we have actually is just a series of circles, isn't it? And yet when we look at this image we see.
A woman wearing a hat. This is a very common phenomenon. In fact, you're looking at me on a screen right now. And I'm not actually on your screen. What you're looking at is a whole series of dots. And. Because they're organized in the way they are you perceive me or an image of me.
And that's what we mean by emergence. So. When I say somebody knows something. What I'm saying is that they are organized in such a way as to be able to recognize. Phenomena in the world just in the same way you recognize a picture of a woman in that image or a mermaidation of starlings.
That's what a person who knows does. To know something is to be organized in a certain way sure to learn is to acquire patterns to learn is to have experienced something frequently enough so that we can form a characteristic response to that thing. So I say that knowledge is recognition.
If you think about it. It makes a lot more sense than traditional theories of knowledge. A traditional theory of knowledge might say you understand the parts of something or you understand the rules or you remember the specific configuration but suppose you are meeting your mother and a bus station.
And there's a crowd of people and you're looking and the crowd of people for your mother and then yeah, there's my mother now did you follow a set of rules no you did not. Did you look for something specific like her red shirt known you probably did not know what happened is out of all of those people one of them matched the pattern in your head and you just become suddenly huh there she is.
It's recognizing it's an instant awareness and instant matching of what's out there in the world with the configuration of your neural network as it exists and it's very individual. That is very context specific. I thought of showing you different images on that slide. I thought about showing you a dot matrix picture of Arnold Schwarzenegger.
Who's a well-known cultural figure at least here, but I decided no. I won't use that because you might not actually recognize him, maybe you do. I don't know. I don't know how popular are not. Schwarzenegger is in Malaysian. But that's the idea right if I showed the same picture to my grandparents before Arnold Schwarzenegger was even born.
Then they would not recognize him. I am being told he is quite popular in Malaysia, okay, well I could have used that example. But recognition depends on. You know what you have already experienced?
So from my perspective and I don't want to say that George shares this but from my perspective what we think of as literacy isn't about language, it isn't about rules and grammars, it's about patterns, it's about recognizing common patterns and phenomena. And so this to me gives me a very practical application of connectivism.
I don't have time sadly to talk about this kind of literacy in any detail. But if you search for my speaking in law cats presentation, you'll see this described in some detail. But basically I identify six major types of patterns that cost you the sort of social interactions that we would think of as literacy.
Now there might be more there might be fewer they might be better ways of describing these patterns, it doesn't matter. It's just a way of getting our mind around the idea. So here they are syntax semantics pragmatics cognition context and change.
And you'll recognize all of these ideas through most of the people that you've written about and read about but now think of them from the perspective of these are patterns in the environment these are things that you are wrecking recognizing because your neural network is shaped to certain way.
So syntax, for example. Syntax well syntax is not just rules and grammar, in fact. I would argue it's not rules and grammar at all no matter what chomsky says we can think of forms archetypes grammar might be repeated patterns in language, there's a series of grammar series called the code build series that represents grammar that way it's not a set of rules it's a description of common patterns that people use we see it an operation some procedures motor skills.
How to fix an engine how to remove a carburetor there are patterns in regularities in mathematics and substitutivity and things called egg corns. I encourage you to look that up. And similarities are types of patterns. We also have semantics theories of truth meaning purpose goals and but again, these are kinds of patterns sense and reference interpretations.
Remember I talked about interpretations earlier the different interpretations of probability. Forms of association. Again, the heavy and contiguity back propagation. And then other kinds of semantics in the world making decisions, voting finding consensus emergence itself is a pattern. Pragmatics covering use action impact. And people like Austin and John Searle have talked about speech acts and the way we do things with words.
The way we direct people or express things or make statements, but also harm people harass people bully people. We can talk about hate speech, for example under this category. And what hates bitch actually is? Interrogation and presupposition or patterns recognized by header and meaning in the idea of meaning is use is an important pattern recognized by Ludwig Vickenstein.
Cognition. Again, we tend to want to say that we need cognition in order to know and to learn but I turn this around and I say that cognition itself is a very involved form of pattern recognition and there are different ways we use patterns to do different kinds of inference.
Overall, I divide inference into four major categories description definition argument and explanation. There may be other forms, it doesn't matter again. Context. What do I mean by an explanation? What are the elements in common to explanations? Bossman frassin, for example says an explanation is an answer to a y question.
And you can only answer a y question. In terms of what else was possible. And that's a very common pattern a common way of reasoning about the world. Cline remember? I talked about him and the interpretation of words that refer to animals. He talks about meaning. And the different ranges of possibilities.
That are affected by context with respect to meaning Jokerina talks about vocabulary in the space between our concepts. George Lakeoff talks about frames and worldviews. These are all again. Patterns change classic way of talking about patterns the different kinds of change, you know, the e-ching describes change but so did Marshall McLuhan so did Hagel talking about the dialectic we talking about game theory progression logic branching tree scenarios scheduling timetables, all of these are ways of thinking about change all of these are patterns layered on patterns layered on patterns.
So we could take anything any concept that we have. And represent it along these lines according to these elements. I gave an example, for example, let's take performance. As in acting as in movies or theater or whatever. And what is the syntax of performance what are the forms rules operations patterns are similarities?
Well we analyze the concept of performance we have very basic things like you should know your lines or stenosis system the method acting, ritual performance and funerals and weddings comparing tales the art of storytelling and so on. All of these are ways of seeing all of these are ways of recognizing different things in the world.
We think can be evaluation of learning then what are we thinking of how do we evaluate learning? Do we simply have somebody repeat back something that they were told? Well, no. We think of community what is community but the place where these patterns are instantiated. We think of social learning what is social learning?
Well again, this is the ways we interact with each other creating patterns recognizing patterns.
And that takes us to connectivism as pedagogy. Well traditionally in pedagogy we get things like instructional theory. You know, the the elements that Bruner describes things like the learning predisposition the design of concepts how to present an instructional design the successful and proper progression of ideas beginning from foundations, for example, and then of course the administration of rewards and punishments well to me none of us has anything to do with connectivism.
I describe it in terms of what I call the ARF method. And ARF stands for aggregating remixing repurpose and feed forward. Whoops. I hate when I do that. Now. This is not unique to me and you know, I I think that you will see this pattern. Repeated all over the place everywhere.
Gathering things together remixing them. Adopting them to your own purpose translating to your online language and then sharing. I developed this from the perspective of well what does it neuron do? If you were a single entity in a network, what are you doing you're receiving the signal your processing the signal in some way and then there's an activation function and then you fire fourth a signal of your own.
And so, I tell people in a connectivist model of learning be the node be the entity in the network. As an instructional theory. We see that the core skill is to see these connections between these information sources and to make decisions. You know, even in rapidly changing environments based on our capacity to recognize things.
To be able to continually update and change our knowledge based on new phenomena that are presented. So it's not simply can you repeat something back but can you work in a rapidly changing and dynamic environment? And you know, we talk about assessment. We when we have people who are working in high skill high stakes environments and here I'm thinking of for example people like surgeons or doctors in a hospital airline pilots, maybe military commanders.
Etc. Lawyers, even. We don't simply give them a test and we don't simply ask them to state knowledge back. No what we do. And you can see this for yourself we put them in a real environment. We put interns in a hospital we have lawyers article with law firms, we put pilots into flight simulation systems or on an actual airplane.
And then somebody who is already an expert in the domain watches their performance. And their assessment is not you know, there might be a checklist but their assessment is not based on any particular set of rules or principles or even knowledge that a person has what the evaluator is doing is looking at the person overall.
I'm asking themselves do I recognize this person as a doctor as a pilot as a lawyer. Do I recognize them and what what I mean by that is do they function in their environment as though they recognize the things around them and are responding appropriately they're using the words and the right way they're asking the right questions they're treating the right things as data and information they're performing skills in the right way the whole combination of thing but you know, You can list all of these things and see if they're doing each one you have to look at it overall and say do I recognize this as performance in this environment?
And so that's what the learning activity is based on in connectivism putting people into that kind of environment not in a formal course, but in an overall sort of environment like that, that's what we tried to create George Siemens and myself when we created MOOCs now originally we did it just because we could scale it better.
But ultimately what we wanted to do is give an example of connectivism by putting people in an actual functioning learning network. Going into that network ourselves showing how we learned in that network and giving them the chance to practice learning for themselves in the network, we actually did not care what they learned.
There was no content we wanted them to learn and we even said to them. You determine what counts as success for you we don't have a body of content here what's important to us is that you are able to function in a network and learn things from the network and that's connectivism now people later came along they created X MOOCs which were to go back to the traditional model of courses as presentations of content but nonetheless we continued with what we then called the CMOC or connectivist move.
Now this can be applied in the classroom. I probably wouldn't but it can be yeah creating scenarios having active learning using network resources using the internet itself in the classroom, but really connectivism is intended to work in as authentic and environment as possible and one of the things that I've always said from the very beginning that the advantage of being able to learn online.
Will be to get people out of the network out of the classroom and into the community. Some people are using connectivism to justify micro learning and in a way it is because in a way micro learning takes and you know a center presenting a course as a seamless hole looks at all the different individual things.
But of course, it's not just the individual things. It's how these things come together form a pattern give us some emergent perspective that we can learn to recognize. And personalization connectivism gives us a story about that as well. Now what we're trying to teach people facts and give them information then personalization means more.
More options more choices more types of tests and you have to do more work to customizing environment but it should be clear by now based on what I've said that in the connectivist approach personalization typically means less. Fewer rules fewer constraints, greater autonomy for the individual within the network.
I've talked about in this beyond the scope of this obviously personal learning. Where the connectivist approach is personal learning where you do something for yourself, you're trying to achieve something. And then you produce content. That's the feed forward part. You get a response. That's feedback or back propagation. You try again produce more content and it goes around and around.
And that's very different from a content-based approach where you take a test and you get corrected and you take a test and you get corrected. Connectivism is about personal learning not personalized learning. And I've developed over time and approach based on personal learning environments and personal learning networks, this is not my picture.
This is someone else's. It's in the notes and the slide but it gives you an idea of you know, you can see connectivism playing a role in personal learning networks and other theories playing other roles the whole idea here is putting oneself in the center of a learning network and then using various capacities of.
Forences to create connections and interactions with the wider world. And we think about connectivism. We think about concepts like open education open educational resources and open networks and and connectivism very much is a theory about openness and it's ethereal about openness because. Networks work best in that environment what once a network is structured and restrained and controlled it's no longer able to respond just no longer able to grow and to learn.
And if you saw from the different literacies that I described connectivism also fosters critical thinking and deep learning we're not just looking at the surface features of something we're not just looking at what somebody said or what somebody did but we're thinking about the patterns the patterns of patterns that are created.
And this is the last slide. And some of you are saying at last. You know thinking of connectivism and we go back to the beginning of this talk and I said, you know, I'm not really talking about connectivism a lot and I'm not really promoting it and if we think about it the success of connectivism is not going to be determined.
By traditional measures of learning success, it's not going to be for example established by the OECD's PISA tests that evaluate for how much a 15 year old knows about mathematics and language and maybe geography connectivism is focused on a wider understanding of learning and understanding of learning that's based not just on facts and information, but rather a person's capacity.
To live work and thrive within a wider interconnected community, you know, I've said before, you know, you you look for the success of learning not in task scores and grades or even graduations you look for success in learning in social index indices, like lower crime rates better health greater happiness among the people these are.
To me the proper measurements the proper indicators of learning and the proper ways to evaluate a learning theory. So that's my presentation this just goes to some links to the videos that I showed during the session and again, you'll be able to find these links in these slides on my webpage and if we go all the way back to the beginning, I don't know if I can go all the way back to the oh I can let's go all the way back to.
Come on all the way back all the way but yeah all the way back all the way back doesn't go all the way back, come on, come on, come here and 28 yards is close to 47. Oh it's not gonna let me go to slide one okay, well I'll cheat then.
Here we go.
You can see right here www.downs.ca/presentation/547, these slides will be there. The audio recording will be there a video will be there and a transcript and unedited neural network artificially intelligently created transcript will be there which all later edit so we got actually get proper sentences so I hope you didn't try to memorize any of this.
I hope that it stimulated some thought and now you have a resource that I am feeding forward into the community. For you to share for you to use for your own purposes, whatever they may be so I thank you for your time for your attention. I know that most of you are still here, which I really appreciate and it's been an honor and a pleasure to be able to talk to you today.
Okay. Yeah, there's a bunch of stuff happening in that question and you know to provide a full response I would need to take it apart and look at the different parts of it but overall you're asking if you're teaching and you concept to somebody some new information you want them to acquire it, how do you do that, how do you integrate it if you will into what they already know because that sounded yeah, you're not even good so the first thing I would say is that.
We we don't want to think about the student acquiring this new piece of knowledge, okay, that's not something that can happen right you can't take a decent knowledge and give it to them and say here like an apple or an orange. It doesn't work that way so what we need first of all is to ask what is it that you are expecting from the students, okay that's a different question, isn't it you put the question in the form.
I want the student to acquire a piece of knowledge but I'm asking you now what do you expect of the student and you can't say I expect them to know such and such because there's no way for you to look in their brain and see if they know. Right so what would count as shall we say evidence that they know something well, it's gonna depend on what the piece of knowledge is obviously and it's gonna depend a lot on the person but usually.
We say some sort of performance is what counts as evidence that they know something right and usually the performance that we have in mind is a test. So we give them a piece of knowledge we say Paris is the capital of France how do we know that how do we know that they know that well if we ask them what is the capital of France they come back with the word Paris we said they know but do they know right well we know that they've learned to associate the word Paris at the capital of France they've got that pattern.
But doesn't have any concept of what it means to be the capital of France, you know, well what I would do? Is you know, and again, I'm still thinking of evaluating right what I would do is. I'd put them in an environment where knowing that Paris is the capital of France is somehow relevant.
There used to be a show called where in the world is Carmen San Diego and it did that sort of thing right or you know, we there's all kinds of different scenarios we might have a model united nations. Gathering in the capital of France or you know, any number of scenarios where the point here is that the concept that parishes the capital of France is somehow embedded in that scenario.
And then we watch people perform in that scenario. And see if they stumble over you know, see if they make a mistake when it comes to the capital of France or not. So it's like they're in a completely different context right where they're not just reciting back. Paris is the capital of France but they're actually using that knowledge or applying that knowledge or at least being informed by that knowledge in this scenario.
Now you ask how do you teach that well see now we have a problem? Because if I need a whole scenario for each piece of knowledge, I'll never be able to teach them because the evaluation would be just so immense right so I need to rethink what it is that I'm doing when I'm teaching them.
I'm not teaching them that Paris is the capital of France I'm not actually giving them a concept to acquire that's not what I care about what I care about is can they function in this model United Nations can they function in a game that involves knowing capitals can they function as a lawyer sometimes dealing with French people you see what I mean?
And so I'm focused on. That kind of capacity in that kind of environment the individual facts don't matter or more accurately if they do matter they will be learned. And if they don't matter they won't be learned. And it's it's a very different picture of what learning means. You know, mathematics trigonometry.
I don't know if you study trigonometry or not, but maybe you have lots of people have so trigonometry is to study of angles and relations between angles and and sides of figures. And I studied trigonometry four times believe it or not. I started it in grade. 13 high school.
I passed it, but I didn't remember it. And then I started it in college. I took one year of college at Algonquin College and it was hand studying it and passed it still didn't know it. Then I studied it in first year university when I passed it still didn't know it and then I was just making a computer game one day the Star Trek game now one of the things I wanted to do in my computer game is rotate a cube on the screen.
That's what I wanted to do rotate a cube well, how do you rotate a three-dimensional cube on a two-dimensional screen you use trigonometry? So in order to figure out how to rotate a cube. I went and dug out my old trigonometry text looked at all applied the formulas made my cube rotate and it was brilliant and I learned trigonometry.
And I could have done without the previous three if I had just in high school rotated a cube. I would have learned trigonometry. And you wouldn't have taught me trigonometry you would have taught me how to rotate a cube but I would have learned trigonometry. And that's what I need, right?
You know, we think when we're teaching we're giving people sets of facts to acquire but we're not we're teaching them. How to do complex things in complex environments. Because the parts that we try to teach them will never add up to the whole the whole is an emergence property of the parts and you will see all kinds of theories that are all kinds of arguments that say well direct instruction is better because people are more likely to remember this and remember that and remember that and I said well, that's true.
You could probably just yell at them and get them to yell back and they'll remember right you walked into the classroom and you yell at them Paris is the capital of France and you do that 20 times and have them yell back. Paris is the capital of France 20 times they will remember Paris is the capital of France but they won't know it.
And that's the key distinction.
Yeah, we've got several in the chat there.
Yeah. I'm. Yeah, that's that's actually an interesting question this theory. I mean part of the motivation for this theory is not only that it also works offline it describes and explains learning in small children infants in animals cats right in all kinds of things that we know learn but who do not have what might be called cognitive tools right a cat my cat learns?
My cat does not have language my cat, you know cannot do logic it cannot construct knowledge right it does not make mental representations of the world but it's here every day at noon to get some tuna because it's learned it is figured out time and tuna and me and and somehow associated those things so that it comes on a regular basis to get tuna.
Now. You might be thinking well how is that relevant to Kenneth work offline well the whole point here is that thinking of learning is thinking of what networks in general and brains in particular do that's the key, right? Connectivism was presented as a theory for the digital age, but it's so only because we have better tools to understand some of these concepts online, but these concepts still apply offline and in fact they broadly apply.
There are many similarities there are some differences. I mean human beings are different from neurons right so for example a human being is much more mobile and can go places and they near on camp so there are gonna be some differences obviously. But there are many similarities and and the similarities have been described by people like Duncan J.
Watts and Alfred Laskey who focus on social network theory. Um, One good example of the way social behavior works in the same way as the brain is in emergent phenomena what I described during my talk is recognition.
And you can see a society. Recognize. Say a change of a state of change of you know, just you know, how do I want to phrase this? Because you know, there's lots of colloquialisms, you know a change in a change in the wind is the colloquialism. I don't know if it translates well.
For example. I mean in in revolutions, this is very common. Take the the some of the Arab spring events right take the events into nesia where everything was normal one day and the next day all of society is decided okay, no now we're gonna change the government. And it's funny right because no person told everybody in society okay, we're gonna have a rebellion now right it's just that it was an emergent phenomenon all of Tunisian society suddenly, you know, as a whole recognized okay, something has to change.
And that's kind of an example another example.
If you're ever watching a sporting event. And it's a close game between two teams and there's a big crowd of people watching the game. And. The momentum shifts and then at some point. You can just tell by looking at it that one team is going to win in the other team is going to lose.
And the whole stadium comes to that realization all together all at once and in Canada, we have a tradition in hockey games. In the playoffs when the home team is going to win people start singing none then hey, hey goodbye. It's very nice right but not happens at different points in the game depending on whether the home team think the home team crowd thinks they're going to win because you don't want to sing it if you're going to lose right the only thing if you're going to win and.
At a certain point during the game the crowd starts to sing how did the crowd know how to sing then they all saw what was happening they all recognized as a whole as as a unit they all recognized one more example memes. I write about memes quite a bit in my presentation on speaking in all cats, but a meme.
Strictly speaking is a computer image usually with some text on it, that is funny and relevant to something socially there was one meme, it was a cat saying I can has cheeseburger and. It means nothing but it means everything it's one of these things. And the thing is. When a meme is shared everybody knows what it means well not everybody but but people get the sense of what it means even though there are no rules nobody told to them right it's just there's this bit of social knowledge out there that the creator the meme tapped into and they said are they realized or they just by accident they said I put this image but these words everyone will know what it means.
And and that's what a meme is so. Yeah, I mean and I can multiply these examples, you know over and over and over and arguably and I would argue that there is a mathematics roughly analogous to graph theory that describes this and you know, I don't know if it'll be a simple mathematics.
I do know that there are thousands maybe millions of people working in artificial. Intelligence who are working on this mathematics and maybe there'll be a nice simple ways explaining it, you know, maybe Wolfram's theories are right saying or maybe they'll be really complicated but there is an underlying mathematics that describes the social behavior of networks and the social behavior of people that there are similarities there.
Okay connection is um is a theory in computer science, it's specifically a theory and computer science and it's specifically a theory about how to produce. Intelligent behavior and we talk about what that is, but how we can produce intelligent behavior using artificial neural networks. Okay, so. It's you know, obviously influenced by you know, network theory and other domains like neurophysiology or even social network theory but it's specifically a theory in computer science.
Connectivism. Draws on connectionism, it's very heavily influenced by connectionism. And it says let's suppose connectionism works. What does that mean for education? Right, what do we take from connections connectionism, what do we take from that and how do we apply it to learning? Now connectivism is influenced by more than just connectionism it's also influenced by social network theory and graph theory in the rest.
But it's intended specifically as an educational theory, it's the application of connectionism and similar theories to education.
Yeah now we we need to be cleared here that we night we might not be able to distinguish between them simply on the basis of the practices that I'm that are employed right there, you know different theories can result in somebody taking the same action so we understand that but there is a way to clearly distinguish between connectivism and constructivism at least as far as I'm concerned.
I don't know if. George will completely agree with me here but he might he might. And that's that's the following. On my theory. Connectionism, they're sorry connectivism is a non-representational theory while constructivism is a representation authority. Now what do I mean by that?
In constructivism. People are engaged in an intentional process of constructing or co-constructing knowledge. Right? When we ask what that means exactly because I always ask that what it means or what it tends to mean we have to be careful because there are many different varieties of construction constructivism. But what it means is basically they are designing different ways of representing the world, they're building different representations of the world.
So all constructivist theories. Basically have a two-part structure. One part of the structure is the nature of the representation what processes govern it how it's put together how it's created and the other part of it is how this applies to or stands for or refers to the world. That's what it means to be a representational theory right the construction is a is it's a picture or a model or a tool that you've created in order to work with the world but it's not the same as the world.
So far so good. Okay now. In connectivism. There's no picture. There's no representation. There's no sense in which we can say this neural network stands for or represents something in the world. That does not mean there's nothing in the world. There probably is. And I don't know, but there probably is right.
But. The thing that's in our mind doesn't represent that. Our knowledge is. Completely and only the structure of what's in our mind. The structure of the neural networks the connections. And we can talk about them as though they stand for things. But. When we're talking about knowledge. We can't talk about it in the sense of it inherently stands for things.
We can't talk about it as we created this mental structure for a purpose to stand for the world. To see the distinction. It's a hard distinction because we might end up doing the same thing. And there's a lot of overlap. I agree. There's a lot of overlap between constructivism and connectionism.
And there's a lot of good in constructionism. I think there is I think a lot of the practices that it recommends are sound learning practices. But, The way it talks about what knowledge is and what learning is is wrong. It's based on. You know a constructivist epistemology. It's it's based on like the works of bass man, frassin and constructive empiricism.
It's based on work ultimately goes back to about your current. It's based on. You know. The in some way the bringing in or incorporation of. Principles. Or methodologies that come prior to an or independent of our experiences in the world goes back to current right? What are the prior conditions for the possibility of perception and cons as they are space and time and as soon as Kant said they are space in time he was outlining the fundamental principles for creating a representation.
He began with space and time. Or if you're if you're what's his face hide your beginner being in time or if you start you begin with being and nothingness, you know, and there are different ways of constructing Nelson Goodman ways of world-making, right? And all of these you know, they're constructions they're representations we build them and then we apply them to the world.
So, that's the difference. And it's a hard different hard difference to think about but but that's the difference.
Oh, you're on mute.
Exactly. Yeah. That's right.
Yeah. Yeah, that's that's that's a good understanding. I think I think that's a good starting point.
I did a presentation on this one so long presentation and if you look it up, it's called the representative student.
Yeah. The short answer is yes, although I say that risking making everybody who believes other theories angry with me, but yeah I do. I do think that I wouldn't say it's a level above I would say it's underlying or more foundational than these other theories, but I'm not a foundationalist so that's hard for me to say.
Sure. Bye everyone, thank you for taking part.