- Complex learning, step by step
- Complex learning (coffee on the side)
- Ten little steps, and how One grew
- Problem solving, scaffolding, and varied practice
- Step 2: sequencing tasks, or, what next?
- Clusters, chains, and part-task sequencing
- Step 3: performance objectives (the how of the what)
- Criteria for objectives–also, values and attitudes
- Step 4: supportive info (by design)
- Learning to learn (an elaboration)
- Step 5: cognitive strategies (when you don’t know what to do)
- Step 6: (thinking about) mental models
- Step 7: procedural info, or, how to handle routine
- Procedural in practice
- Step 8: cognitive rules, or, when there IS a right way
- Step 9: prerequisites, or, ya gotta start somewhere
- Step 10: part-task practice (getting better at getting faster)
- You? Auto? Practice.
- Media’s role in complex learning
- Self-directed learning: stepping out on your own
- Where do the Ten Steps lead?
In the Ten Steps to Complex Learning, I’m still working through Step 1: Design Learning Tasks. The previous post focused on support related directly to the learning tasks: the givens, the goal, and the solution(s). A central idea for van Merriënboer and Kirschner is the primacy of working with whole tasks; by varying these factors (e.g., having learners work with a partial solution), you can provide richer learning experiences.
Complex skills are, in vM&K’s terms, non-recurrent: you apply the skills differently to new problems. To do so effectively, you need to build cognitive strategies. You’re creating for yourself mental models for grappling with problems in this field. Step 1 gives three examples of such guidance: modeling, process worksheets, and performance constraints.
Modeling: learning by example
One common form of on-the-job learning is seelou training–“See Lou? Do what Lou does.” The intention is good; Lou is (theoretically) a skilled performer, and if you could do what Lou does, you’d be one too.
As Ten Steps points out, modeling as a form of learning guidance is more complex. You need a skilled professional to perform the task and also to explain or think out loud as she does.
Becoming this kind of model, like becoming a mentor, isn’t something that happens because the boss dubs someone an expert. Successful modeling requires not only a skilled and credible role model, but explicit attention to to the process of solving problems. I think it’s critical to prepare that skilled performer, both in terms of highlighting whole tasks and in terms of focusing on the process.
(You can, of course, use virtual models, such as a recorded performance or even, in theory, a simulation of an actual performance, though that’s easier to say than to do.)
One form might be what’s often called shadowing–following the skilled performer through the workday. Without the explicit focus on how the skilled person approaches problem-solving, though, shadowing isn’t likely to teach the novice performer any complex skills.
Process worksheets: cookbooks for chefs
Ten Steps describes a process worksheet as way of guiding learners through problem-solving. Here’s an oversimplified example to guide a paralegal in writing an internal memorandum of law. The left column lists the task steps; a right column deals with the processes involved.
vM&K favor using questions when the process worksheet is a training tool: “learners are provoked to think about the rules-of-thumb.”
You could provide this type of guidance at a much more elaborate and detailed level. vM&K refer to SAPs (systematic approaches to problem-solving), of which the process worksheet is an example, and note that some SAPs include branching so that the learner goes to different subparts for different types of problems.
And you’re right, that looks a lot like a performance-support system. One difference I see is that we’re talking about learning, whereas the performance support system is about… well, performance. So it tends not to deal explicitly with the cognitive processes behind the performance.
Rolling right along: performance constraints
A third way to guide learners is by constraining what they can do. Ten Steps calls this the training wheels approach. When a child’s learning to ride a bicycle, there’s a lot of skill to focus on: pedaling, steering, maintaining balance. Training wheels remove balance from the equation; they constrain the performance. The authors believe that performance constraints are particularly useful for early phases in the learning process.
There’s a certain amount of relativism here. It’s like the rhetorical question, “How long is a rope?” If the overarching skill is riding a bicycle, what constitutes a whole task? “Moving the bike along a path” might be one; the kid’s still moving along even when the bike has training wheels attached. Removing the wheels creates a whole task (moving) with less support (moving and balance on your own).
Which makes a good link to the final point in Step 1: scaffolding
Scaffolding: support that fades
“Scaffolding” is a technique that combines providing support to the learner (like modeling, process worksheets, examples) with gradually reducing or fading that support until the learner faces what vM&K call a conventional task (“here’s the problem; solve it”).
Research on expertise reversal indicated that highly effective instructional methods for novice learners can lose their effectiveness and even have negative effected when used with more experienced learners….
There is overwhelming evidence that conventional tasks [meaning, those with no support] force novice learners to use weak problem-solving methods that bear little relation to schema-construction processes…
Or: if you just have novices work at solving complex problems but don’t provide them support while they’re learning, they won’t know what they’re doing, and they won’t learn how to do better–at least not in any efficient way. I think this explains why so much alleged on-the-job training does so poorly with complex skills: novices focus on externals, on what looks important (or on what Lou thinks is important).
A corollary is that when someone’s already skilled, the support and guidance can actually hinder his performance. The skilled performer has built her own mental models, and they work (or else she wouldn’t be skilled). When you start providing other models to skilled performers, you’re struggling against what’s already there.
And you’re probably losing.
Scaffolding (support that gradually faces as you move through the training program) is a tool to replace external guidance with self guidance.
Variation: the road from concrete to abstract
vM&K talk about “variability of practice,” though I think “variation” is a better word (not that they asked). I’ve discussed this in earlier posts; mentioning it again stresses the emphasis that Ten Steps puts on inductive learning.
Students construct general cognitive schemas of how to approach problems in the domain, and of how the domain is organized, based on their concrete experiences offered by the tasks.
Variation in the tasks makes for richer learning experiences, which in turn makes for stronger schemas. Generalization and discrimination are obvious ways that people build their cognitive maps. So is mindful abstraction, deliberate efforts to generate alternative concepts or solutions.
New to me was the concept of implicit learning. Some tasks lack straightforward decision guides but have a lot of information. Giving learners a wide range of positive and negative examples can help them derive useful schemas. The situation that vM&K talk about involves showing air traffic controllers thousands of examples of control situations–some dangerous, some safe.
A final recommendation: learning tasks adjacent to each other should have learners practice different versions of the constituent skills. The higher “contextual interference” will actually help learners solve new problems better.
Let’s say you’re working on effective writing. Imagine three elements like using parallel structure, using personal pronouns, and avoiding passive verbs. Low interference would come from a cluster of problems, each of which involves parallelism in email, followed by another cluster involving personal pronouns in email. Eventually you get to a second batch: parallelism in reports, personal pronouns in reports, and so on.
Higher interference would come from practice tasks that interspersing both the elements and the situation: parallelism in an email, then passive voice in a report, then personal pronouns in the comments section of a form.
Since vM&K end the chapter with a warning, I’ll repeat its essence here:
- Novices don’t learn well from conventional tasks (“here’s the problem, what’s the answer?”).
- A randomized set of conventional tasks is even worse.
- High variability with ample support and guidance helps novices learn.
- High interference (how the variability is presented) also helps them learn.
So, for novices learning complex skills: without support, your learning won’t hold up.
CC-licensed dubbing photo by The Other Dan
CC-licensed “I know this already” photo by orionoir.
2 thoughts on “Problem solving, scaffolding, and varied practice”
These are the kinds of articles that are very helpful to learning experience designers. As a group, we need to better understand a multitude of approaches so we can use the appropriate strategy to meet varied situations. Thanks for a great post.