Dreyfus model?

I’ve read the Dreyfus book about 20 years ago and from what little I remember, their model of developing skill and expertise is largely anecdotal. As far as I am aware, which admittedly doesn’t say very much, the model with its distinguishable levels has never been put to a rigorous scientific test.

Notwithstanding, the Dreyfus model meme has gained some currency in geek culture—more due to its replicative prowess than its uncertain validity, I’m afraid.

What would you want a “rigorous scientific test” to tell you, exactly?

Has test-driven development been put to a rigorous scientific test? Does it provide you value anyway?

Andy, I think these two cases are not comparable at all. TDD is put to the test if and when I try it and decide that it improves my productivity. If I find this not to be so, I can toss the practice again without qualms.

The five stages of skill acquisition in the Dreyfus model are understood as descriptive or normative. If someone claims that when acquiring a skill a subject passes through these specific, discernible stages as a matter of psychological fact, then I’d like to see proof that this is indeed the case. In other words, facing a theory I’d like to know whether it is true.

No matter what notion of truth one subscribes to, be it staunchly realist or flexibly pragmatist, there’s always more to it than “We all read that book and thought it was plausible”.

When you write “Use the Dreyfus Model on the road to expertise” you imbue the model with a normative force, beyond the plainly descriptive. In effect, you’re saying that the stages of the model are not only real, but in order to attain expertise one ought to, even has to, master these levels in the game of learning.

So, what do I want? I’m looking for some reasonable assurance that I’m not lead down a garden path. Inadvertently and with the best intentions, but still. By contrast, I have no misgivings whatsoever when someone says or writes “Look this is what I did and it worked out well for me. It may or may not for you. Make of it what you will.” This is providing experience, not giving advice.

Compare that to “Use the Dreyfus Model of Skill Acquisition to become more expert” and “Leverage the architecture of the brain to strengthen different thinking modes”. These have at their core factual claims about the workings of the mind and brain. And these claims serve as the starting points for how to make the best use of mind/brain. Having good reasons to assume the foundational claims are valid would greatly increase my confidence in the implications.

There’s work based on the model that has been tried and proven in the field, but outside of computer science or software engineering. Dr. Patricia Benner started the ball rolling in the field of nursing, and has published extensively on the subject. So it’s not so much a matter of “We all read that book and thought it was plausible” as it is “yeah, we’ve seen that too.”

If you’d like to read further about this subject, and on the successful application of this model over the last 20+ years, Dr. Benner is a good place to start.

Also, you state “in order to attain expertise one ought to, even has to, master these levels”. That’s not the case, and I hope I haven’t given you that impression. These are not skills to master, they are observations of your capabilities along a spectrum. The levels are not rigidly fixed; you don’t wake up on Tuesday as an Expert. It’s a long journey, and these are helpful roadsigns along the way, indicating where you are on the path, and where you’re headed.

Another book which could be checked is from Gary Klein, “Sources of Power”, The MIT Press.

It talks about decision making: firefighters, nurses, ... , and even software developers. And differences between experts and novices.

There is examples about software development group from AT&T in connection with Dreyfus model (chapter 10). See also chapters 7 and 11.

Having just read the book, the focus on Dreyfus seems like its weakest point. His work (1), like so many popular items in education, comes down to applied common sense. It describes the apparent stages in a process without providing much insight in how those stages evolve. (I also stopped to read as much of Dreyfus’ published material I could get my hands on.)

I’d like to suggest a few things to fill in the gaps. The first is chapters 3 and 4 of Roger Schank’s “Virtual Learning” where he explains the roles failure plays in learning. (This will strengthen the explanation of a novice learner’s desire of rules and context - to minimize failure - and also lays out strategies for moving into self-directed learning and mastery.) The next is “On Intelligence” by Jeff Hawkins— a stunning examination of learning all the way down to the neural level. (Pay special attention to its explanation of trial and error, pattern matching, and how the hippocampus moderates learning.)

Etienne Wenger (in Communities of Practice), like Dreyfus, points to having many varied experiences as a road to mastery. But Wenger, like Schank, also refers to the need to construct and share explanations as key to learning effectively from those experiences.

There’s an interesting bit of educational research showing the key difference between someone who is proficient and someone is an expert is they both have about the same amount of knowledge but the expert has developed many “shortcut” connections between the pieces. These enable the seemingly magical “intuitive” leaps. (Unfortunately, I can’t recall the citation. I’ll have to ask a couple of colleagues.)

On a personal note, the best learning skill I’ve developed is also my best debugging skill—being able to recognize when what I’m seeing isn’t matching what I’ve expected and being willing to question/research the underlying belief. Technically, it’s a “metacognitive” skill, and there’s quite a bit of literature on metacognition and learning.

OK, hopefully I haven’t been too much of a nuisance and have supplied something useful. I’ve been teaching, learning, and designing instructional experiences for over 20 years and would be happy to help however I can. (I’m also a long-time reader of the pragprog books and an active software developer building simulations, decision support systems, instructional tools, etc.)

...Richard

(1) Note that Dreyfus doesn’t call his original work a “model”, but a “treatment of the phenomenology of skill acquisition” (see “A Phenomenology of Skill Acquisition as the basis for a Merleau-Pontian Non-representationalist Cognitive Science” available on Dreyfus’ web page.)

Richard,

Thanks for you posting—a couple of comments:

“the focus on Dreyfus seems like its weakest point”. I don’t necessarily disagree, but I’d like to point out that I’ve received a great many emails from folks who found this to be a great revelation. I agree with your assessment that his observations “comes down to applied common sense” but remember that much of this line of thought is brand new to many (most?) people.

I did read “On Intelligence” and was very, very impressed with it. I think he really emphasizes that everything is pattern matching—and prediction. I think that plays to the heart of your comment about “what I’m seeing isn’t matching what I’ve expected”.

Now here’s the question: I’m trying very, very hard to keep this book to a manageable size, under about 250 pages or so. I’m finding that quite difficult :-). This is such a vast area to explore, and every book I read has pointers to 20 other really interesting works and so on. So I have to be pretty ruthless in trying to stick to the tips of these icebergs, and also to keep the book very practical.

WIth that in mind, I don’t want to shortchange the reader either, so if you think I’ve really fallen short anywhere here do let me know.

/\ndy