Home > data science, math education > Online learning promotes passivity

Online learning promotes passivity

April 23, 2012

Up til I took Andrew Ng’s online machine learning class last semester, I had two worries about the concept of online learning. First, I worried that the inability to ask questions would be a major problem. Second, I worried about the possibility of building up material. I could imagine learning a given thing online but the ability to sustain and build material over an entire semester seemed kind of unrealistic.

On the second point, I think I’m convinced. Andrew definitely taught us a real semester’s worth of stuff, and he built up a body of knowledge very well. I now communicate with my colleagues at work using the language he taught us, which is very cool.

On the first point about asking questions, however, I am even more convinced there’s a crucial problem.

I want to differentiate between two different kinds of questions to make my point. First, there’s the “I’m confused” type of question, where someone literally doesn’t get the point of something or doesn’t understand the notation or a step in an explanation.

One can imagine tackling this kind of question in various ways. For example, one can strive to be a really good teacher, which Andrew certainly is, or to explain things at a high level but shove the details into black boxes, which Andrew did quite a bit (somewhat to my disappointment, especially when linear algebra was involved). If neither of those two things is sufficient, and the class is really important and/or really common, one can imagine teaching a computer to anticipate confusion and to ask questions along the way to make sure the students are following, and to go back and explain things in a different way if not.

In other words, the first kind of “clarifying questions” can probably be dealt with by the online learning community over time.

But there’s a second kind, namely the kind of question where someone is not confused but rather asks a question for one of the following reasons:

  1. they want to know how a certain idea relates to something else they know about,
  2. they want to generalize something the teacher said,
  3. they want to argue against an approach or for another approach,
  4. they see a mistake, or
  5. they see an easier way to do something.

Almost by definition, the above kinds of questions aren’t anticipated by the teacher, but the fact that they are asked almost always improves the class, certainly for the student in question but also for the other students and the teacher.

For example, one semester I taught three sections of 18.03 (exhausting! and I was pregnant!), which is a calculus class at M.I.T., and I remember thinking that in every single class one of the students made a remark or asked a question that I learned something from. It got to the point that, the third time through the same material, I’d be waiting for someone to explain how I should be teaching it. I loved that the students there are so smart but also so engaged in learning.

And that’s what I’m worried about- the engagement. When you embark on an online class, the best you can hope for is that you learn something and that you don’t get hopelessly confused. And that’s cool, that you can learn something, for free, online. But what you can’t do is what I’m worried about, and that’s to get instant feedback and discussion about some idea you had in the categories above.

I’m definitely one of those people who asks questions of the second type, and although I may sometimes annoy my fellow students, I really feel like the active engagement I pursue by coming up with all sorts of crazy comments and ideas and questions is what made me capable of doing original and creative things. For me, the most important part of my education was that training whereby I got to ask questions in class and got smart teachers who liked me to do so and would talk to me about my ideas.

How can that possibly happen with online learning? I’m afraid it can’t, and I’m afraid we will be training people to receive information rather than to engage in creation.

I imagine that in 200 years, almost everyone will be taught online, hooked into the machine and pumped up with knowledge. It will be only the elites who will have access to real live people to teach them in person, where they will be taught not only the material but also how to argue against a point of view and to propose an alternate approach.

  1. April 23, 2012 at 2:16 pm

    The Art of
    Problem Solving
    and others have embedded meaningful real-time interaction into on-line classes and supplemented with asynchronous discussion boards that encourage active learning. It won’t motivate a majority to engage actively but neither will face-to-face classes at top universities. After students work hard to get to MIT and be part of a vibrant learning community, you’d think freshman class attendance would exceed 95%. It’s not even close. Given the choice, a large percentage are skipping the lectures and learning on their own – despite the hefty price tag. Passivity is not just a problem for many MIT students but virtually every learning venue has problems getting students involved. Freshman class size is probably the main culprit at MIT and other universities. Many students will only ask probing questions and make insightful comments in one-on-one face-to-face interactions.


    • April 23, 2012 at 10:32 pm

      I’ve taught at AoPS a few times, and I can testify that the students there ask lots of questions of the second type, and I very much enjoyed fielding them live. It works well because the classes are live and not too big.

      This actually resolves MathBloke’s later issue as well, because students pose their questions privately to the teacher. I then had the option of forwarding it on to the entire class and answering it for everyone, or answering privately/asking my TA to answer privately (if it would distract the rest of the class too much). It was the best of both worlds, and as a bonus, students felt more confident asking questions because their classmates wouldn’t see them unless they were *good* questions!

      Of course, I still felt like I had less ability to engage my students because I wasn’t in front of them. There’s still something about being in the same room as another person that causes us to engage more actively.


  2. MathBloke
    April 23, 2012 at 2:25 pm

    I don’t understand why it would not be possible for questions of the second type to be asked in an online forum, possibly moderated by “smart teachers who … would talk to [you] about [your] ideas”?

    I’ve perpetually been annoyed by fellow students who ask questions of the second type in class. It’s always seemed to me profoundly rude to interrupt the thoughts of twenty other people to ask the teacher a question when it’s not clear that you’ve yourself given the answer more than a few minutes’ thought. (In other words: if you’re so smart, why haven’t you worked it out yourself, bigshot?)

    Thus, online learning would have been a dream come true for me when I was a student, particularly when paired with end of year/end of term exams for evaluation. I was an undergrad at Oxbridge, where in the first year there were at least a few lecturers who would discourage ‘adventurous’ questions during lectures. However, in graduate school in the US (particularly in courses attended by ambitious undergrads or first years bucking for attention from the faculty) the Teacher’s Pet Game was played with great enthusiasm.


    • MathBloke
      April 23, 2012 at 4:08 pm

      … upon rereading my previous comment (intended to point out that people with less ‘active’ learning styles might do fine in the online system you describe) I see that it may have come off as rather snarky. (Apologies if so.)

      My main point, I think, is that questions of the second type are probably most profitable when pursued as (in most cases) minor ‘research’ activities. In my own, admittedly limited, experience as both a student and a teacher, I’ve found that little such activity comes from these questions in class, and that the questions, if frequent, can be more disruptive than helpful. By contrast, I think that questions of the second type could be excellent fodder for an online forum associated to the class, especially if moderated or directed by a teacher with some research experience, and even more so if the student can present some of the calculations or work they have done related to their question. This would have the following immediate benefits: the student would be encouraged to flesh out their ideas more fully before bringing them to the attention of the class, thus yielding a more useful question, and there would be a semi-permanent record of the question that everyone in the class could refer to in the future.

      Perhaps whether or not live interaction is best probably comes down to individual learning styles and personality. Certainly I’m not predisposed to have a blog, in contrast to yourself. People are just different. However, I wonder if using the terms ‘passive’ and ‘active’ to describe people who are more or less reticent in class does their capabilities as students or scholars very much justice; ‘passive’ seems rather negative as a label. Is a productive thinker necessarily one who will actively ask questions during class? (This is a genuine teaching question, and if so, maybe I’ll shed my inhibitions and ask more questions in general from now on!)

      That said, I’d like to say that I do enjoy your blog very much, which I think also reflects upon and showcases your wonderful personality. Thank you for posting your thoughts on here, and for all the work you put in to keep it running!


  3. April 23, 2012 at 3:00 pm

    I don’t see how this is specifically and inherently a problem of on-line learning, rather than a problem of the scale of many of the new on-line classes.

    In Caltech’s first on-line learning class, the classes are broadcast live. The live students ask questions in class (they line up at a microphone to do it! I guess it must be a large class, as Caltech classes go. Other large classes elsewhere have TAs running around with microphones to enable students to ask questions in class.)

    The on-line students in Caltech’s class also queue up electronically in a chat window to share their questions and comments. There is some process of alternation that intersperse the questions from the live students and the on-line students.

    Obviously, it is impossible for every online student in a massive on-line class to ask a question during the course of a semester, but the same has long been true of true of large face-to-face classes.

    Universities have attempted to mitigate this problem in face-to-face classes by offering discussion sections led by TAs. I understand that some of the online classes are attempt to do the same sort of thing with electronic TAs. (Quality control is an issue for both venues, of course.)


  4. karen
    April 23, 2012 at 4:05 pm

    I have to agree with Greg. Passive students will be passive no matter what, and actively engaged students are able to do that just as well in an online class. Every online class I’ve taken has required weekly postings of thoughts and questions as well as commenting on other students’ posts. Lots of students asked the second type of question, because we were all required to post something unique, which would force us to think about the topic and come up with something to say.
    An advantage of online learning is more flexibility in the timeline, so you can actually think about your question for three days and flesh it out before getting feedback on it. That is impossible in the live classroom setting.


  5. April 24, 2012 at 8:46 am

    I am currently taking a free online course in Machine Learning class offered by Caltech, and taught by Prof. Yaser S. Abu-Mostafa. The course is sponsored by Coursera. Here’s the link.


    The lecture is 1 hour long and its is followed by 20 minutes of questions.
    One TA takes questions from the online audience and another TA takes questions from the in-class audience. Nothing is glossed over or left to the student to work on on their own.

    Yaser S. Abu-Mostafa is an extremely clear and excellent teacher. He does not skip any steps in presenting the material.
    E.g. – He’s dedicated 4 lectures to proving the mathematical basis for learning classification with perceptrons showing how to constrain the number of hypothesis in the Hoeffding inequality.

    His new book is also excellent:


    Were up to lecture 6 out of 18 but if you can still register, I highly recommend this course!


  6. John Lenihan
    April 27, 2012 at 8:28 pm

    The article points are well taken, but since the whole business is literally a bawling infant, a little patience could be a virtue. Some courses use the lecture at a blackboard style, which works well in the hands of a good teacher. Some use a disembodied voice driving a stylus with the professor lecturing in the background, really somewhat robotic and disconcerting.

    But much better technology and ingenious solutions can be expected to appear fast and solve the problem, as well as provide new experiences totally unimagined as of now.


    • Jim
      April 28, 2012 at 11:08 pm

      The whole business is NOT ‘literally’ a bawling infant. Please learn the proper definition and use of the word literally. Thanks.


  7. May 12, 2012 at 12:43 pm

    Two words: Stack Exchange.


  8. June 5, 2012 at 8:35 pm

    I imagine that in 200 years, almost everyone will be taught online, hooked into the machine and pumped up with knowledge. It will be only the elites who will have access to real live people to teach them in person, where they will be taught not only the material but also how to argue against a point of view and to propose an alternate approach.

    Nice point


  1. April 23, 2012 at 10:45 am
  2. April 23, 2012 at 11:54 pm
Comments are closed.
%d bloggers like this: