 | What is the connection between connectionism and connectivism? |
|
Hi classmates, I was thinking of writing a more in depth blog on this topic but my current research paper on learner control and instructional topic sequencing is taking most of my time. So I'll just post my quick questions here and try to rely on the cck09 network for weightier/stronger connections. This question has been touched in the CCK08 Moodle forum but it was spotty. See a search of the term connectionism in the CCK08 moodle forum here: http://ltc.umanitoba.ca/moodle/mod/forum/search.php?search=connectionism&id=20. Ken in particular had asked the question: What does connectionism and connectivism have in common?. Stephen said: ...Not all of connectionism applies to connectivism; otherwise they would be one and the same theory. I am interested in finding out what principles of connectionism do apply to connectivism. Had Stephen elaborated on this in some blog somewhere? I think this is important in discussing the role of memory in connectivism. George mentioned that the role of memory in the theory is adaptive patterns, representative of current state, existing in networks. Connectionism is involved in trying to explain memory in the form of neural networks. But in my very shallow understanding of research in connectionism, their studies are limited to one shot, short (hours), one topic, perceptual forms of treatment/instruction/stimuli. Most of the models i've seen of neural nets are in the form of input units->hidden units->output units. This are feedforward neural nets that are activates in one direction. I doubt it if these models will be a good analogy to the growth of knowledge in connectivism where the emergent knowledge of one learner becomes input for another and so on. There might be other models like recurrent neural networks where there is some cycle. So my questions are: 1. How can we model connectivist learning? Are symbolic cognitive architectures like ACT-R and SOAR useful? 2. How does connectivism make sense of artificial neural network theories when the former deals with highly complex, discontinuous, long-term learning activities like CCK08 and currently CCK09, while the latter deals with short term, perceptual studies. 3. I assume neural network nodes are non-differentiated and the inputs are simple, while the outputs are emergent and more complex behavior. If these ideas are correct and they are mapped in connectivism, what would be the simple inputs? What would be the nodes? What would be the emergent outputs? Let's say in CCK09, wherein I assume the desired goal is to learn about the theoretical concept connectivism (the emergent output?) from blogs, forums, conferences, wikis, twits etc. (the inputs); are the nodes whole articles, messages, sentences, paragraphs, words, letters, or the very bits of our 16x16 (?) grid pixels? I can't see whole blogs for instance as nodes, since I might reject some word associations being emphasized by the author, while accepting some other word associations from the same author. But if we go down to words, how much complicated would accounting for concept formation would be. As well as how much more expensive a simulated model would be in terms of computer power. And there's the issue of network boundaries. If it is possible to follow the activation (firing) of a network from its root node how far and when do we go? When can we achieve quiescence or equilibrium; or some stability? Roel |
 | Re: What is the connection between connectionism and connectivism? |
| I wonder if human 'consciousness' is merely an emergent output of simple neural inputs, or an epiphenomenal output, or a qualia or something like that? |
 | Re: What is the connection between connectionism and connectivism? |
|
Just found this in a twine of mine: Israeli scientists discover "center of consciousness" in brain http://bit.ly/ypLkK |
| Re: What is the connection between connectionism and connectivism? |
|
Have you read this exchange? http://ltc.umanitoba.ca/wiki/Kerr_Presentation George acknowledges connectionism to be the internal (neural) workings and connectivism to be the external networking (of connectionist entities including, perhaps, humans, machines etc?) ...connectivism...."The aspect that I’m more focused on is the external networks, the networks that I create when I mediate through technology relationships and ongoing understanding". ...connectionism...." arose...partly with artificial intelligence, partly with neuroscience, and connectionism is something that I’ve validated, something that I feel has a strong role in being able to move forward as a model for how learning occurs" Perhaps connectivism is an inferred, inductive theory of learning based on connectionist theory using the methodologies of neuroscience at the neural level and 'network theory' at the social level. Churchland's suggestion that someday two distinct human brains might 'connect' in a similar manner to the way in which one human brain's two hemisphere connect is interesting (pg 88). |
 | Re: What is the connection between connectionism and connectivism? |
|
Roel: My vague knowledge of ACT-R, which comes from taking a course with John Anderson a long time ago, would indicate that ACT-R is much different than connectivism, at least the way it has been discussed so far. If you look at Anderson's publications on the ACT-R website, you'll notice that many of them are in journals with the word "cognitive" in them. That should tell you something. In this model, knowledge is propositional and it is held in long term memory until it is called upon to solve a particular problem. It doesn't really emerge from the chaos of the network; rather it is inputed in a top-down fashion. I'm uncomfortable with the neural net analogy with connectivism primarily because individual nodes are not neutral; some are more important than others and that is decided by the consciousness of the individual learner. One of the things I got out of the last class was that I felt that some nodes were more interesting than others. When the class began, I wasn't feeling well, so I had to carefully choose which nodes of hte discussion I was going to read. I choose the one with the most replies, feeling that tthe one with the most replies was going to be the most interesting. That was a rational decision based on inputed knowledge I held in long-term memory. It may be incomplete knowledge, but it was the best I had. I think that approach is more consistent with the ACT-R model for problem solving. I think the roots of connectivism go back to congition and the way an individual makes decisions to choose one node over another. One can see this in the reseach on hypertext and the way the reader makes decisions to work through the links. Again, the individual is making sense of the nodes based on prior knowledge. I think that model is more consistent with the ACT-R model although I admit this is not my field and I'm really sure about it. Thanks for the link; I hope to catch up with this and make some new connections. |
| Re: What is the connection between connectionism and connectivism? |
|
@Ken, thanks for the link. My initial understanding of connectivism was also in the network level. It is the reason why my entry point was social network analysis. I am even trying to analyze the phenomena of the social network of the CCK08 Moodle forums. But in a private communication with George Siemens he said: "The more difficult challenge, I believe, comes in understanding how connections contribute to concept formation. For example, how does my understanding of a concept change due to being connected to you (by connected I mean having read and replied to a post you’ve made)?" A question he reiterated here http://www.elearnspace.org/blog/2009/09/18/untangling-the-web/: "Network analysis reveals the flow of information in an organization. As important as the structure itself is the why and how of connection forming. Why do two people share information with each other? What impact does a connection (social or conceptual) have on a learner’s level of understanding a subject?" This is a question that I am deeply interested in as a teacher. I thought perhaps the phenomena CCK09 can speak of a learning communities' emergent theoretical conception of "connectivism" (do we need to imagine a social mind for such purpose?), but imho it would be inapplicable to students that must be treated as individuals. Even if I can aggregate the understanding of my class after a "connectivist" task, what about the individual? Initially I thought of combining social network analysis and semantic network analysis to be able to answer the question. The social network will provide the structure and the semantic network will provide the "what resonated in the structure?" But semantic nets are just a teeny weeny part of the conceptual network of an individual. And it seems improbable that I would be able to reconstruct the dynamic process of a multi-agent interaction between learners in a learning network. So I turned to simulation models. @Joel, thanks for the information on ACT-R. I was trying to learn SOAR, and you're post affirmed my impression of cognitive architecture. I initially thought SOAR has an answer to George's question. The understanding of a concept changes in the same way "Who wants to be a millionaire?" works. Whenever the brain reaches an impasse you can "call a friend" as one of the proposed operator. "Calling a friend" could literally mean that, or googling, or any use of connections you can think of. So a learner does not have to go into a inefficient search of procedural and declarative memory. But that answer appear to me too good to be true. I turned to another simulation model, the emergent neural network simulation simulation system. When I looked at the examples, there's the input unit->hidden unit->output unit model. And you train the model with binary data  . If I feed this model matrices of the coword analysis of let's say CCK08 is it going to give me the emergent phenomenon "connectivism theory of learning"?Emergent seems so low level, and the problem we're dealing here "learning" seems so high level. SOAR appears to deal with learning in higher levels than Emergent (but I really don't know since I'm a total newbie in these models). I can't shake of the idea that to say color "red", the brain would have to fire connections between the neurons that are working on the electrochemical signals from the retina, with the neurons connected to the ear that remembers the activation of the sound "red', and then so on and so forth connected with red until quiesence. Does the brain contain only what to me are settings of the sense organs (e.g. excitation due to a certain pattern of light frequencies). And that all these primitive settings/memory emerges into let's say a child pointing to a new object as red? (is this behaviorism in the level of neurons?) How do you model that with the word "connectivism"? If the question could be answered in CCK09, I'm hoping I could find or build a simulated student that will display connectivists learning behavior. For this simulation I would need a connectivist task like the water-jug task or towers of hanoi of cognitive architectures. What would be a good simple connectivist learning task? Btw. The question "how does a learner choose which node to connect to?" and related "how does a learner weight connections?". (Assuming the learner has a choice) is also fascinating. |
| Matthias Melcher's work on the relationship of the conceptual and the social layers of connectivism. |
|
Lest I forget, Matthias sent me links to his awesome work on the relationship between the conceptual and social layer of connectivism. I am really interested in furthering his ideas on resonance. I think new students to this course will learn a lot from his swf presentation (4 mb so please be patient): http://www.rzuser.uni-heidelberg.de/~x28/en/195/ Matthias blogs on the topic: Conceptual and Other Network Levels cited here http://connect.downes.ca/cgi-bin/page.cgi?post=44125 CCK08 Week 3 Leftover: The Three Layers http://x28newblog.blog.uni-heidelberg.de/2008/10/02/cck08-week-3-leftover-the-three-layers/ CCK08 Conceptual and Social layer http://x28newblog.blog.uni-heidelberg.de/2008/11/23/cck08-conceptual-and-social-layer/ CCK08 Final presentation http://x28newblog.blog.uni-heidelberg.de/2008/11/24/cck08-final-presentation/ Blog vs. forum preferences http://x28newblog.blog.uni-heidelberg.de/2009/05/01/blog-vs-forum-preferences/ He also said that Jenny Mackness elaborated in depth on the concept of resonance, but I can only find the following from Jenny's blog: http://jennymackness.wordpress.com/2009/05/05/resonance-in-blogging/ http://jennymackness.wordpress.com/2009/08/07/connected-resonance-what-does-it-mean/ |
| Re: What is the connection between connectionism and connectivism? |
|
I've recently realized that Stephen Downes (2007) did elaborate on the relation between connectionism and connectivism. He also discussed concept formation in connectivism in this presentation: http://ltc.umanitoba.ca/wiki/Downes_Presentation . Here's the juicy section where I've highlighted the parts I'm interested in and hope that I did not take it out of context. ***start of quote*** (the formatting appears to be being filtered) Here’s the theory: concepts are not words and that’s why it’s not going to be a rule-based system; they are patterns in a network and that like the human brain or a network like society as a whole. In these networks, there’s no specific place where the concept is located. The concept is distributed as a set of connections across the same network and other concepts are embedded in the same network; they form parts of each other and they affect each other. <snip...>What I’m going to do is defer that question until later on in the presentation. Right now I simply want to establish that concepts are a pattern, OK? The short answer – I guess I’m not deferring it, am I? – the short answer is that the principles that govern the operation of rules are sentential principles, they are logical principles, they are principles of symbol systems. And these principles are different than the principles that govern operations of patterns, and to express that sort of difference in a slogan type form, symbol systems are based on the concept of equality, identity, this is that. Patterns are based on the concept of similarity, this is like that. And that’s the short answer. Right, and as was indicated in the comments – not the comments, but in the discussion in the Moodle threads – constructivism, to my understanding of it, is based in physical symbol systems. You don’t get constructivism without symbol systems. And since this is not symbol system theory, it is not constructivism. My take on this, and we may have to bat this around a bit, George, connectionism is a computational theory that has been around for quite a while. It’s the principle of how patterns in networks are recognized, and then we can go on and start talking about that. Connectivism adopts all of that; all of connectionism is absorbed by connectivism. And then connectivism becomes the study of how we use this knowledge, this knowledge of connectionism in a pedagogical sense. I’ve represented it in the past, although I don’t think you liked this: connectivism is the pedagogy of connectionism. That would be my take on it right now, but at heart I’m a connectionist, my roots are in connectionism. That’s not a bad characterization. I’m not sure that that’s the distinction that would have been intended, but that’s kind of how it happened. A lot of discussions about networks that occur in society fall under the heading of discussions of social networks, so you get people like Duncan Watts and others and you’re looking at stuff like scale-free networks and all of that. And you’re quite right, connectionist literature focuses on what they call neural networks or simulations of neural networks and it’s focused on things like the brain and perception and recognition by computers and so on. Part of my position is that the two phenomena are one and the same, that what we’re seeing at the micro level in the brain is the same kind of thing that we’re seeing in society, that we’re seeing in different ways in different places in society. The same principles that govern crickets interacting with each other govern bloggers citing and quoting each other, govern the development of river systems and trees – those principles are also the principles that govern things like human brains and computer networks set up in certain ways. Now, the interesting thing, and the reason I’ve taken the tack that I’ve taken in this particular paper or presentation is that we have a network here and a network there, a network there, we have a brain, we have a society, and how do concepts go from one of these networks to the other? This is where we get to connectivism. The whole point of teaching is to get some sort of knowledge, if you will - whatever you want to call it - from the big network, that’s society, to the little network, that’s the student’s brain. But at the same time we’re also working on the problem of how to get stuff that’s in the little network, say my brain, into the big network, society. And so we’ve got this process that goes back and forth here, and what’s really important – it turns out to be crucial – is that this is not strictly a causal process. You don’t just simply cause a pattern to occur in a straightforward X causes Y sort of thing. Me having a thought in my brain and then saying it publicly like here, doesn’t automatically cause that thought to be part of the wider social network. So we need to understand what that process is, how concepts go from network to network, and the first or beginning steps of that is to understand what a concept in a network is to begin with, and that’s why I’ve taken the approach I’ve started with here. So we get to the point concepts are not words. You’ve got to get away from that idea, talking about concepts as though they are words; concepts are patterns. So the properties that concepts are going to have are not going to be logical or syntactical properties; they are going to be network properties. ****end of quote**** |
| Re: What is the connection between connectionism and connectivism? |
|
Hi Roel. I am glad you found this excerpt. Clearly connectivism is born of connectionism, and connectionism has strong ties to behaviourism. The method of connectivism is operant conditioning, I think, and have written about that in CCK08 and in my blogging. We saw an example of OC at work in CCK08 with the moodle forum force subscription (MFFS) application. More on this later, off to work..... |
 | Re: What is the connection between connectionism and connectivism? |
|
Hi Ken, Thorndike's connectionism certainly has behaviorism connections (!). I'll wait until you write more about the connectionist reference before I respond in any depth. For now, I would ask you to consider the neural/cognitive-conceptual/external-social dimensions of connectivism. When you emphasize OC, I assume you are addressing that aspect of connectivism. How would you apply OC to neural/conceptual levels? |
| Re: What is the connection between connectionism and connectivism? |
|
Hello George. Thanks for running CCK09 again, I love having this forum in which to debate and dialogue. I hope you don't mind me 'kicking the tires' of your connectivism vehicle. I believe I am recognizing patterns in what is stated about connectivism and what is stated about behaviourism. I also am interested in connectivism's lineage. I think the relevant concepts are connected something like this:
But I may have it wrong. What do you think? Does connectivism employ any behaviourist methodologies? |
 | Re: What is the connection between connectionism and connectivism? |
|
Where are 'concepts' in this diagram? But certainly as connectionism goes, if I were to try and develop a pedagogy based on connectionism alone it would include lots of rote learning and practice drills. Which is indeed how artificial neural networks are usually trained, and a method that is also being used a lot in our house at the moment as my daughter learns piano... But I don't think that rote learning and drills are intended to be cornerstones of a connectivist pedagogy, and I think I'd be surprised if George or Stephen advocated such. So at what points do connectionism and connectivism diverge and disagree? |
| Re: What is the connection between connectionism and connectivism? |
|
@Ken, I see OC in the training of artificial neural networks (I could be wrong though). But whenever I step back and look at the network as it becomes more complex from two neurons connected to society in general, I can't help but feel that OC is just a small part of the explanation, there's just more to it than that. I just had this geeky vision. Out of zeros and ones comes Second Life. The source code is the blueprint of the patterns of Second Life's zeros and ones. We have the source code for Second Life but not for Life. Learning is part of Life. Is it possible that we are looking for the source code/s of Learning and Life? As Java is to Assembly, is Connectivism to Connectionism? Is it possible that connectionism and connectivism are different levels of languages for dissasembling Learning? Would it be possible for connectivism to subsume connectionism as the human machine level language? btw. some of those zeros and ones are predetermined e.g. instructions sets for ICs. Is this true for Life as well? |
| Re: What is the connection between connectionism and connectivism? |
|
Hello Roel. I think your vision may have some legs. As Siemens stated elsewhere: "I would state, however, that connectivism attends to the primacy of the connection in both knowledge and learning. Everything flows from the connection. Everything grows from the connection." It would appear he is seeking the source as well, in some way, and believes it to be found in the connection. I still wonder about the conditioning of the connection. Is stimulus-response the means for growth/flow/life? |
| Re: What is the connection between connectionism and connectivism? |
|
Thanks Ken, glad to know I'm not seeing things. Your last two sentences made me think back of what you said about consciousness. Since I can only see the stimulus-response in the neural network (and I'm open to other explanations), I find it a bit freaky to think that all the s-r activity (not just an output) of the brain does emerge into consciousness. And then I think if connectionism is already working on the neural network where the node is a neuron(?), would it not be more productive for connectivism to talk about basic patterns found by connectionism as the building blocks of a conceptual network rather than conceiving of other forms of nodes (e.g. words, people, etc.). The patterns are identifiable connections that could be built into more complex behavior and knowledge. Similar to how higher programming languages hide the machine language with libraries, apis, or functions. |