Failing to succeed

May 1, 2012 5:10 AM

The learning paradox is at the heart of “productive failure." While the model adopted by many teachers and employers when introducing others to new knowledge — providing lots of structure and guidance early on, until the students or workers show that they can do it on their own — makes intuitive sense, it may not be the best way to promote learning.

With one group of students, the teacher provided strong “scaffolding” .... Meanwhile, a second group was directed to solve the same problems by collaborating with one another, absent any prompts from their instructor. These students weren’t able to complete the problems correctly. But in the course of trying to do so, they generated a lot of ideas about the nature of the problems and about what potential solutions would look like. And when the two groups were tested on what they’d learned, the second group “significantly outperformed” the first. Via. Paywalled paper. Bonus Feynman learning technique.

posted by unSane (29 comments total)

I completely agree with this concept. I've noticed I learn best using an algorithm like this:

1) Attempt.
2) Examine results in light of information on how I should have done this.
3) Go to 1 until results satisfactory.

I use this on the kids too. They ask me how to write a function or build something or beat a video game or dig a hole or whatever I tell them to try first, then ask me how.

That said, it isn't so much which teaching technique works better as a matter of what you are trying to teach. For instance, let's say you are teaching someone how to build a house. You could throw them a couple of 2x4s and a random selection of tools and tell them to figure out how to fasten the boards together. They would indeed learn a great deal about the nature of the problem. But getting from there to an understanding of how a house is built would take quite a while.

There has to be some bottom level of "here is how you do X. no, don't ask questions just do it and you can think about the reason it works later because it's only a tiny part of what I'm teaching you".
posted by DU at 5:37 AM on May 1, 2012 [3 favorites]

Bonus learning paradox.
posted by Beardman at 5:38 AM on May 1, 2012

Thanks, unSane. My personal guideline has long been that when trying a new skill or task, I accept that I'm going to screw it up the first time, and that's OK, because it really cements "Hey, don't do THAT, do this instead."
posted by MonkeyToes at 5:40 AM on May 1, 2012 [2 favorites]

I really like the general idea of this approach (learn knowledge by assessing it) and use it in the form of flash card software like Anki, but teachers really need to be careful so that they don't frustrate students too much.

I've experienced a few college classes where you are given a group assignment to apply a model/framework/theory (which is great). As you and your group are working hard on doing the assignment you begin to realize that the reason why you are so confused all the time, why you keep having heated discussions about the most basic concepts, is that you never got the required knowledge to do the assignment in the first place. It pisses you off because everything becomes a slow and confusing trial and error experience and you have to use precious tutoring time to learn the basic concepts in the first place. This becomes particularly frustrating if you're studying a small or near obsolete topic (hello computer/info/management science!) with no hope of finding additional resources on the web.

Obviously a professor cannot tell you everything regarding a topic but you should at least have enough knowledge to figure out where and why you are stuck and how to remedy this.
posted by Foci for Analysis at 5:47 AM on May 1, 2012 [2 favorites]

This would fall under constructionist learning, I'd say.

I use Constructivist techniques with my first graders, but you have to be really careful. For example, telling first graders to measure something without teaching them how a ruler works will create all kinds of chaos. They often start measuring from the 1" mark instead of the '0' or end (depending on the type of ruler), or use centimeters, and if you then test them they'll get every question wrong.

If you show them how a ruler works first, though, and measure a few things together, and then challenge them to measure the whole room, you get them thinking and problem solving and also correct answers (mostly).

The other problem with letting kids figure things out in a group that I've experienced first hand is that one child will dominate the group and do all the work, and the others just go along passively. I'm curious about how the study controlled for that, or whether the scores jumped up for the dominate students but not the passive ones.

As with every good educational theory, this has its place in an effective classroom, and is being used by many, many teachers. It was even part of the teacher training I went to a couple of weeks ago.
posted by Huck500 at 6:00 AM on May 1, 2012 [12 favorites]

I accept that I'm going to screw it up the first time, and that's OK, because it really cements "Hey, don't do THAT, do this instead."

Almost without exception, this is the attitude of people who do the coolest things.
posted by Rykey at 6:15 AM on May 1, 2012

Without getting too nitpicky, there's a difference between learning a subject well enough to pass an exam and actually learning a subject.

This article really addresses the latter, and it resonates very strongly with me. The things that I've taught myself (and done wrong the most amount of times, often) are the things that I've learned the best. It's because I've been forced to actually understand what's under the hood, not just the end product.
posted by Benny Andajetz at 6:26 AM on May 1, 2012 [1 favorite]

Doesn't this kind of depend on what you want to do? I've been part of a bunch of projects where we've had to learn a radically new skill in a new context...and there are times when you want or have the time to reinvent the wheel, even if the whole reinvention process gives you more perspective and bigger skills. If you have to have something up and running by next week and you're responsible to a bunch of other people to get it right, flailing, failing and having a working project three months down the road doesn't help.

Part of what you learn about learning is how to make things work under the constraints you're facing. This is something that I've struggled with as an anarchist, since IME we tend to lean toward "reinvent the wheel over and over and over while maintaining that real-world constraints can be overcome if we will it hard enough". Which is partly why we have trouble building projects that last.

This is one reason why I don't like a lot of education theory - it assumes that the classroom is the best place for learning and that the learning which takes place there can be ur-learning which prepares you for just about anything. Plus people tend to take learning theory designed for this intensely artificial environment and assume that it has deep meaning about humans.
posted by Frowner at 6:32 AM on May 1, 2012 [5 favorites]

er...don't want to reinvent the wheel.

The unconscious is a funny thing, eh?
posted by Frowner at 6:33 AM on May 1, 2012

My wife is required to teach in this manner. The kids are amazing at solving complex problems, but completely incapable of figuring out how much change they should get when buying a cheeseburger with cash.
posted by blue_beetle at 6:35 AM on May 1, 2012 [3 favorites]

That's interesting. But I think in the real world you might run into some problems like kids just giving up or deciding to spend all their time gossiping about popular kids or whatever. I suppose a teacher could patrol and try to keep them on task.

This reminds me of middle school when teachers all tried to have us do "Concept Attainment" exercises. I remember one in particular where we had pictures of some kind of animals, and we had to sort them into groups, without being told what the groups were. We spent some time discussing the various attributes of animals and stuff, but at the same time we had no idea what the concept we were supposed to attain was. Of course it must have been somewhat effective, since I still remember the exercise well over a decade later - on the other hand I don't remember what the actual concept we were supposed to attain was. It's probably something i learned about over and over again (like whether or not an animal is a vertebrate or whatever). But, at the time it seemed like it would have been much more efficient to just tell us what the concept was.

Anyway, I can see this might be effective, but kids might really get annoyed by it.
posted by delmoi at 6:44 AM on May 1, 2012

I was really hoping the Feynman technique was bongo-related.

The kids are amazing at solving complex problems, but completely incapable of figuring out how much change they should get when buying a cheeseburger with cash.

yeah, i think a mix of styles is probably appropriate. I think everyone's had the experience of meeting an adult who was great at school but lacks "common sense" in other areas.

Sometimes kids respond well to "because that's the way it is" and don't want further explanation. When I'm making bread with my son, I don't say, "here, try and figure out what ratios of ingredients will work". He's three; stirring up the dry ingredients and dumping in water is thrilling enough.
posted by dubold at 6:52 AM on May 1, 2012

This is good news because extrapolating from the way I've lived my life as a whole, during my second life I'm going to significantly outperform like a motherfucker.
posted by Ritchie at 7:18 AM on May 1, 2012 [6 favorites]

I am reminded of a helpful technique: the process of phrasing a question to ask someone else can help you figure out you know the answer. I've done this a lot, often reverting to "okay, what problem are you trying to solve?"
posted by rmd1023 at 7:51 AM on May 1, 2012 [2 favorites]

Isn't this a lot ilke "build one to throw away"? That's how I do pretty much everything, and it drives my dear wife craaaazy. :7)
posted by wenestvedt at 7:52 AM on May 1, 2012

Hm. I'm trying to think of how to operationalize this in my classroom without, as Delmoi said, frustrating students. I teach a gen-ed class at 8:30 in the morning to a group of 50ish students who won't even snicker when I mention farting gorillas ... I think setting them a problem to fail on purpose would totally shut them down. But there has to be a way to incorporate this somehow.
posted by ChuraChura at 8:15 AM on May 1, 2012

I feel like this approach would only work really well if the students received a lecture about the problem solving methodology beforehand and guided practice problems before being left to figure things out on their own. E.g., the core principles of Geometry can be learnt intuitively so long as students are memorizing the theorems as they go along.
posted by lotusmish at 8:19 AM on May 1, 2012

I've often thought that the video game model was the right one. In video games you are often tossed into a situation, faced with repeated failure and its consequence (character death). Then you have to construct productive strategies for which you are rewarded. Someone has to figure out how to make educational video games that don't suck.
posted by dances_with_sneetches at 8:38 AM on May 1, 2012 [2 favorites]

I think this kind of environment would be good for kids, because I think a lot of the stress for young people these days is the "if you fail it will go on your permanent record forever" kind of stuff. Instilling a hefty dose of try and fail and try again, along with the just as important "don't laugh at the kid who is trying but failing" would be good. As long as it is tempered with a little bit of visualizing what can go wrong first. Figuring it out as you go is a nice skill, but sitting down and figuring out a project is important too.

Without getting too nitpicky, there's a difference between learning a subject well enough to pass an exam and actually learning a subject.

Not if the exams are designed correctly. I'm doing some work certification stuff, and I can guarantee you, passing the exam requires actual knowledge of the subject.
posted by gjc at 8:41 AM on May 1, 2012 [1 favorite]

I had an argument in class with a professor about the elaborate processes used in software engineering to make sure that what comes out at the end is actually good enough to be useful. Having done some reading on the subject on my own, I came at him with "Instead wasting time with all of this process, why don't you just build a quick one first and see where it doesn't work?"... "Well, that would just be wasting time!".

In industry, sometimes the monetary risks seem too high to make failing once an option. And yet, companies that ARE willing to fail and really learn for their next product, end up doing quite well (The first xbox being a very prominent one).
posted by Nutri-Matic Drinks Synthesizer at 8:56 AM on May 1, 2012

"See one, Do one, Teach one" addresses the same basic idea that, for most people, it takes active participation for something to stick.
posted by madajb at 8:57 AM on May 1, 2012

This learning about learning is fascinating stuff. Thanks, unSane!
posted by Kevin Street at 9:47 AM on May 1, 2012

As a language teacher, it seems increasingly clear to me that to design someone's learning path for them often robs them of the most important part of learning: making connections and fitting your motivations to the tasks at hand.
posted by mammary16 at 9:59 AM on May 1, 2012

I've been teaching philosophy using the Socratic Method for long enough now I unconsciously Socratize people all the time, most particularly my toddler. Which mostly works really well in getting him to fail in interesting ways that lead him closer to a solution, but it turns out the Socratic Method is not actually an appropriate way to teach a two-year-old which shoe goes on which foot. After three minutes of asking him increasingly leading questions, it dawned on me that I could actually just TELL him which shoe went on which foot, and explain how to tell the difference, and we might actually get to leave the house before nightfall.

I think this is the most important point, however, especially with younger students: First, choose problems to work on that “challenge but do not frustrate.” That's really hard to structure properly and requires a lot of work by the teacher. We've switched to a problem-based science curriculum in my K-12 district this year, and the kids are a lot more engaged in science, but it is really, really hard work and you have to structure it SO carefully so that the problem-based learning is actually building their knowledge (and aligning to curriculum standards) and isn't just "screwing around with stuff in a disconnected way." (Not that screwing around with stuff isn't also valuable.)

You also do have to create situations where people can fail safely -- I swear this is why it took me so long to learn to cook vs. bake. Baking has clear instructions and can be completed accurately by anyone who can more or less follow directions, the necessary skills are not too challenging (measure, mix, pour, bake) and the fail states are pretty clear from the baked good. Cooking has a lot more flexibility and things you only learn by doing and the fail state (for meat, anyway) is DEATH BY FOOD POISONING or FINGER CUT OFF IN ONION-CHOPPING ACCIDENT. I probably would have learned to cook a lot faster with a little scaffolding and someone to tell me, "That's probably safe to eat." If failing ends in mockery, or loss of digits, or an F on your permanent record, students are going to be very resistant to failing.
posted by Eyebrows McGee at 11:41 AM on May 1, 2012 [3 favorites]

That's funny re cooking/baking, because I have the exact opposite problem. I can't follow a recipe to save my life but the trying/failing mode of cooking has made me a pretty good cook, because I generally completely understand what's going on with the thing I'm cooking.

Baking, on the other hand, is inscrutable because most recipes don't tell you WHY there's 1/2 tsp of baking powder in the cake, and you don't want to screw it up by leaving it out, so you never actually find out what the point of it is at all. And it bores me silly, because it's a purely mechanical process.
posted by unSane at 11:55 AM on May 1, 2012 [2 favorites]

It's also called "Desirable Difficulties", a suite of cognitive principles concerning effortful processing. Google it.
posted by stroke_count at 1:44 PM on May 1, 2012

unSane - you should read something like McGee's "On Food and Cooking". Amazing stuff; explains a lot of the science behind the process of cooking. It has improved my cooking for sure.
posted by dubold at 7:28 AM on May 2, 2012

Cookwise by Shirley Corriher is what you want, field figuring out why particular ingredients and techniques matter in baking (and lots of other cooking processes.) Great book.

In chemistry ed there's a movement called Process-oriented guided inquiry learning, POGIL for short, that is along these lines. Evidence seems to show better learning outcomes, but the students completely fucking hate it, and woe betide the instructor who teaches the trailing section full of the kids who failed at that approach the preceding term. I still have nightmares...
posted by Sublimity at 4:11 PM on May 2, 2012

They often start measuring from the 1" mark instead of the '0' or end (depending on the type of ruler), or use centimeters, and if you then test them they'll get every question wrong.

The kids are not the ones making the mistake there.
posted by srboisvert at 2:58 AM on May 4, 2012 [1 favorite]