Categories
open content

Lying about Personalized Learning

Champions of personalized instruction tend to fall back on the assumption that one-on-one tutoring is the most effective instructional approach but is not scalable (implicit in Bloom’s two sigma problem), and since “we all know” that group instruction is poor, we’ve no choice but to personalize using an automated computer system as our best and most effective path forward.

Now, if you’ve ever taught, you know that many students love to talk. It seems that they live to ask questions, argue, and endlessly discuss. Now, I ask you: How can removing all possibility of engaging in their favorite approach to learning (by making the computer the only entity with whom they can interact) be said to be personalization for them?

(And let’s not forget my pet hypothesis regarding the increasing importance of social interaction as one moves further up Bloom’s taxonomy. When students are working near the top of the taxonomy, the absence of social interaction will greatly decrease the efficiency of all learners.)

Systems that want to make claims to “personalize” must include multiple options for students who prefer interacting with other humans. A “full palette” of personalization options that involve variations of interactions with a single entity (the computer) is basically a monochrome palette. “Personalize your new car with any color you want! You can get it in blue, light blue, midnight blue, sky blue, Carolina blue, azurite, ultramarine, cerulean blue, cobalt blue, even Prussian blue!” (Reminiscent of the great SNL skit, “He could be green, or lime green, or mint green…”)

If we’re going to talk about “personalization,” options rich with human interaction must be part of the palette. Otherwise, we should call it what it is – a power grab by administrators forcing learners into a learning environment they do not prefer for the sake of increased efficiency.

10 replies on “Lying about Personalized Learning”

In my courses, I am effectively forcing students to communicate in writing. I will answer as many questions as they can throw at me, but they need to be

1) well written
2) detailed and coherent
3) on topic

Turns out that few students can or will produce a lot of those. They are all for writing lots of vague questions, but once you ask them to be precise or to refer to specific arguments or facts, they complain that it is hard.

And once I do get the questions, very often, the answer is just a URI. They get very annoyed to get back as an answer: “I have answered this question in the notes, just see the second paragraph there”.

In the end, you see that no so many students like to chat a lot in writing. Very, very, very few. I can “out chat” then any day. After all, I blog to them, they never blog back.

I like those observations, and the hint of cynicism. I definitely think the “personal” in “personalized learning” is shifting toward the social, as being social is what we do as persons (e.g. the necessity of network(s) in “personal learning environments”).

Does the “personalized” of one-on-one and the “personal” of the social have to be mutually exclusive? I presume to say, not at all; we must be balanced.

Insightful observations Dave. As I was reading, I kept thinking of the phrase “it’s nothing personal” and the common reply “isn’t everything personal?”

What do we really mean by “personalized” learning? While the more common usage of the term implies learning that is tailored, adapted and customized to individual, i.e. personal, preferences, another equally viable meaning of personalized learning is learning that is more connected with other persons, i.e. interactive, full of the discussion, questions, and give-and-take you suggest.

So-called computer adaptive, “personalized” learning is really the opposite–learning that has been stripped of all human interaction. I’d call that *depersonalized* learning.

OK, I think we can all agree that having a student interact with nothing but a computer program – no matter how personalized – is a poor substitute for a proper program of learning.

This is not to say that such interactions cannot be employed very productively for spot duty – for example, students can learn to program a computer using an interactive module, create Flash animations through a series of computer demonstrations, or practice learning a language in a self-teaching manual.

Learning, though, when viewed more widely, typically involves some sort of interaction with others. This is not because ‘humans are inherently social’ or any such thesis about human nature but because what they are learning is composed to a large degree of social constructs – vocabularies, ways of living, ways of practice, and the rest.

So we want to include a dimension of social interaction in our online learning. We want students to engage with communities composed of practitioners, learners, instructors and mentors. And we want to organize these interactions in a way that best suits individual learners. ‘Personalized learning’ in this context means not merely personalized content but personalized interactions.

But we need to cash out what that means. As Daniel Lemire points out, while students can be a font of communications activities, not all of it will be useful or manageable for other people in the network. His experience is my experience – given carte blanche we can expect the full range from serious, detailed enquiries to long strings of questions semantically equivalent to a child asking “why… why… why?” The purpose of such questions can also range variously from genuine attempts to learn to facile attempts to annoy and irritate.

Thus, because we want to respect the need of others in the course – other students, instructors, mentors, and the like – to manage their own time and their own interactions, it follows that personalized interaction is not simply interaction tailored to the needs of an individual learner. It must be the result of a negotiation of process rather than a one-way catering to an individual learner’s needs and wants.

It is in this negotiation that the creation of the social environment of learning is created. It is in this negotiation that social learning consists. There is no ‘zone of proximal development’ of other such fictitious ‘common’ ground in which learner and mentor share some space – there is rather a series of trials and errors on the part of each in an attempt to negotiate an interaction, a mechanism for communication, a transaction, an engagement.

‘Personalized learning’ is therefore the creation of a mechanism in which this negotiation for engagement can take place. It is the creation of engagement opportunities – as Nancy White said this week, of ‘invitations’, of communication ports and protocols, of learning mechanisms on the part of both student and mentor. Much of this negotiation process is automated (as is, for example, our communications networks of telephones or email or RSS feeds) but the actual communication results in the end only from one human sending messages to another.

This does not man each instructor engages in a complete process of negotiation with each student. This mechanism – the supposed paragon of ‘personal instruction’ is neither expected nor desired. Not expected, because such a mechanism would require an immense resource of instructors, which society cannot sustain. Not desired, because the range of interactive possibilities would be limited to those that only two people could provide, and therefore insufficiently diverse to foster complexity of thought and understanding.

Engagement with a mentor or instructor, in such an environment, is typically the result of a larger set of interactions, a series of negotiations that occurs with members of a community as a whole, of negotiations with other learners (often resulting in a ‘student subculture’ within the community), with some more advanced learners, with practitioners, and with mentors and leaders of the community.

Any given negotiation build on the many negotiations that preceeded it – just as any given conversation in a language (English, say) builds on each participants’ previous learning and practice in that language. That does not mean that no further negotiation is necessary – typically, understandings of language vary widely – but it does reduce to a significant degree the negotiation required in a particular case.

So how do we understand ‘personalized learning’ in this context? It is the establishment of a mechanism (which may or may not be a technical mechanism – it could even be nothing more than a bunch of people standing in a field) whereby each individual can participate in the creation of engagement with others, where such engagements are directly negotiated by the participants to meet their own individual needs or interests.

People are better equipped and able to personalize than a computer. Computers don’t understand meaning, just programmed patterns. Artificial Intelligence isn’t up to the challenge that a normal teacher faces. However, it can handle things like helping a user identify procedural mistakes for standardized problems. Yet that still boils down to what Stephen Downes said. “Much of this negotiation process is automated (as is, for example, our communications networks of telephones or email or RSS feeds) but the actual communication results in the end only from one human sending messages to another.”

Then there is the question Jon Mott brought up, and his answer, “What do we really mean by “personalized” learning? While the more common usage of the term implies learning that is tailored, adapted and customized to individual, i.e. personal, preferences, another equally viable meaning of personalized learning is learning that is more connected with other persons, i.e. interactive, full of the discussion, questions, and give-and-take you suggest.”

Considering what has been said here, both definitions of “personalized” are the same. The interaction and adaptation are the same. The program, the environment, is merely a tool for people to interact in ways not normally possible. This is, in theory, why the personalized learning environment is/was thought to be the silver bullet to making one-on-one interactions scalable. Unfortunately the part that is commonly forgotten is just what Scott mentioned, discussion and engagement. Computers and artificial intelligence can only do so much in this, while humans are able to do much more. That’s why the social media is being pushed as a solution. It is an environment, tools, for humans to use to interact, rather than to replace that interaction.

Stephen–really enjoyed your characterization of learning [particularly the meaningful social aspects of learning] as negotiated. I hesitate to say, however, that the mere potential for such a negotiation [the creation of a mechanism that enables it] truly constitutes personalization.
Personalization connotes adjustment, direction or flat-out creation for or with an individual in mind. [Interestingly, the most common denotations I found were about monograms and engravings…there’s a blog post just waiting to happen there, I can feel it.] The mechanism you describe leads to personalization because each individual in the environment negotiates with others to “meet their own individual needs or interests.”
It seems to me that part of the reason that the “personalization” we see in one-on-one tutoring is so effective is that the personalization goes beyond preferences. It has some of what Jon was talking about; person-alization. Sometimes [perhaps even often] truly effective instruction meets needs learners don’t realize they have, in ways they don’t naturally prefer [and therefore wouldn’t likely negotiate]. Yet such learning is arguably quite effectively “personalized.”
I agree, David, that there are probably additional dimensions of personalization to be explored.

Good comments everyone! Would like to suggest that equating personalized learning with 1-1 learning or individualized learning or learning alone on a computer is incorrect. You can personalize by simply adding a person’s name to a screen, e.g., like Amazon does when you login. Personalization feeds the human emotional need to be recognized and engaged. Emotions in fact can trigger and fuel deep, active learning. Also, certainly lots of research to suggest the importance of the “meaningful social aspects of learning.”

Still surfing in Brazil David????

Kia ora t?tou Hello Everyone

I have to admit that I’ve been a lurker on this one. Simply because I was unclear what you meant by personalize, personalizing, personalization. So much of the usefulness of discussion relies on the understanding of words that are used.

T?n? koe Maggie. Your summary comment doesn’t cut the mustard as far as glossing the terms here. Which puts on the back foot as far as saying something that’s likely to be meaningful to you or others.

So saying, I am about to go into a soliloquy. I just thought I’d mention this in case some of you people didn’t realise that I’d a proclivity for that. BTW, I spell the terms with S not Z.

Yes you are right, Maggie. You can personalise a page on a computer by sticking the person’s name on it. Be sure you spell the name right though, and choose a font that they feel happy about. Oh, and, background colour. You’d better ask about that before you find that the person whose page it’s supposed to be personalised for likes the colour scheme. And that also applies to any font colour you might choose.

And pictures. Some students like to see their pets in pictures on their page – or their boyfriend or girlfriend or baby sister. I suspect that providing that sort of thing could well go down as personalisation too.

Some students like animation – some hate smiley faces. One of my past e-students wanted music so that it played when the page was accessed.

So personalisation of a computer page is all the niceties of how the student would like to see the page when it is displayed. Hardly a pedagogical wow, but neverthless significant – especially if you get the background colour all wrong or their name misspelt – oops.

But, hey. What do you do when you’ve got all that fixed and you discover that the student actually prefers to work from a book or booklet? Are you going to get the books printed with the right font on the right colour of paper? Are they going to be wired so they play the right music when they’re opened? Hmm. Could get difficult, not to mention expensive.

No, personalisation is more than just how the computer screen looks when it comes to learning. It is a whole raft of things, including pedagogy and the right blend of resources. Someone who loves videos might get a bit tired if ALL their lessons / learning resources / activities were delivered in video format.

Learning by doing and learning by observing are other balances that some may prefer. Many Science students just love the practical activities. So that’s a personalisation that creeps into their choice of subject. So the blended approach can be tailored to suit the individual.

Assessment is another galaxy in itself when it comes to personalisation. Does the student prefer exams, or would they feel happier with standard tests?

What about sychronous / asynchronous environments? What about chatroom discussions. What about Skyping and using webcams for 1 to 1 or 1 to many interactions. All of these can be learner preferences that can be considered personalisation.

Whether the teacher heeds the students wants, wishes or needs is part of how well personalisation is accomplished. Much of good personalisation involves cutting through all the learner foibles and finding out exactly what’s best for the learner. Getting the learner on track with that, with a blend of negotiating and what’s really necessary is, frankly, what teaching is all about.

Ka kite
from Middle-earth

Stephen Downes comment,

“I think we can all agree that having a student interact with nothing but a computer program – no matter how personalized – is a poor substitute for a proper program of learning.”

is absolutely correct.

However, how many students today really have access to a ‘proper program of learning’?. While it is right that we always demand improvement in learning technologies and transfer the responsibility of education from trained teachers to computer programs only under extreme conditions and in a limited fashion – perhaps we can reconsider the question in light of students are failed by institutions?

Christensen et al speak eloquently of a disruption of learning that will take place as online learning technologies ‘disrupt’ existing educational establishments. Their thesis is not that these technologies make a better substitute; rather they propose that new technologies will gain ground when they are the only available solution to a student, when they offer something that is better than nothing.

Christenson recently funded our startup to play in this space – Guaranteach. (http://images.businessweek.com/ss/08/10/1021_education_tech/13.htm). I think our take on customization through learning styles and real tutors feels like a worthwhile experiment in this space.

Going back to David’s original reference to Bloom’s 2 Sigma Problem, the original finding was that one-on-one tutoring often produces significant performance gains relative to conventional instruction–the average (50th percentile) student moves to the head of the class (98th percentile). One explanation for this is simple: the effectiveness and efficiency of instruction increases as the student is provided more opportunities to respond/perform with feedback. Who provides the feedback, as long as it is correct, is largely irrelevant. The human tutor can provide feedback, the student’s peers can provide feedback, or a computer can provide feedback. Bloom’s finding had nothing to do with peer interaction other than the fact that as the number of peers in the class increases, the number of opportunities for each student to respond decreases (with some exceptions such as group unison choral response).

Certain forms of “personalization” have been experimentally shown to increase performance–others are useless, or in some cases, counterproductive. (See Rich Mayer’s work on personalization and voice for positive examples and his investigations into computer avatars for a negative example.)

Group instruction is not inherently worse than individual instruction. Personalization is not inherently better than “generic” instruction. Learning can (and does) take place without social interaction. There are a *lot* of evidence-based strategies on what improves teaching and learning.

David wrote, “If we’re going to talk about “personalization,” options rich with human interaction must be part of the palette. Otherwise, we should call it what it is – a power grab by administrators forcing learners into a learning environment they do not prefer for the sake of increased efficiency.”

“…must be part of the palette”? “Power grab by adminstrators…”? “…forcing learners into a learning enviornment they do not prefer…”?

David: You’re ranting again—where’s the science? = )

Comments are closed.