- Scott H Young - https://www.scotthyoung.com/blog -

14 Books On Learning by Doing, Apprenticeship and Transfer

Recently, I embarked on a research project [1] exploring the topics of learning by doing, apprenticeship and transfer. I’ve just finished a big reading binge of a few dozen books loosely related to this topic.

Since I’m fully aware that at most 1% of the material I read will make it into any future writing, I thought I’d highlight some of the more interesting books I’ve read while they’re still fresh in my mind.

Here are some of the books that made me think the most…

1. Democracy and Education [2] by John Dewey

Listen to this article

John Dewey was one of America’s most prominent public intellectuals a century ago. Today, few people would know him by name (when I mentioned him to a friend, she thought he was Melvil Dewey [3] who came up with the Dewey Decimal system [4]). Yet, as the father of the progressive education movement, his ideas are still being felt.

I won’t even try to summarize Dewey, but an idea that stood out was Dewey’s argument that the division between “higher” education and mere vocational training is nothing more than the philosophical baggage we’ve inherited from a society where aristocratic elite consumed useless arts and the lower classes did all the work.

2. Antifragile [5] by Nassim Nicolas Taleb

When this book was published, I remember reading about a hundred pages of it and then putting it down. I was off-put by Taleb’s pugnacious literary style, in which everyone he disagrees with isn’t just wrong, but also a con artist. Perhaps I’m older and more cynical now, because I found the book a lot more enjoyable the second time around.

Taleb essentially argues that natural systems gain from disorder, whereas artificial inventions often exhibit fragility. He attacks the epistemic overconfidence of expertise, particularly in areas of economics and politics, where elegant mathematical models often have a dubious pairing to reality.

3. Cognition in Practice [6] by Jean Lave

Do people use the math they learn in the classroom in real life? Studies of transfer point to the abysmal performance of many skills we were supposed to have learned in school. Jean Lave’s ethnographic research, studying how people use math in their everyday lives, adds an interesting twist to this story.

Following people in the grocery store, she found the math they used in real situations was almost perfectly accurate. However, when the exact same calculations were transposed into a test-like format outside the store, people failed miserably. Lave argues that the picture of school as imparting general purpose cognitive tools to be applied elsewhere is misguided, and that our performance in everyday life is much better than experiments often give credit.

4. Apprenticeship in Early Modern Europe [7] Edited by Maarten Prak and Patrick Wallis

When one thinks of learning by doing, apprenticeship naturally comes to mind. This volume was one of a few books I read on the topic, in this case looking at historical patterns of apprenticeship across Europe.

While I’m still a fan of learning by doing, the book highlighted some of the ways that apprenticeships fall short of an ideal. The tension between a master (who mostly gains from cheap labor) and an apprentice (who wants to one-day become his master’s competitor) creates interesting problems. Still, the authors credit the institution as a major cause in the Great Divergence [8], by which Europe emerged economically dominant, prior to the Industrial Revolution.

5. The Lean Start-Up [9] by Eric Ries

Somehow this book escaped my notice until now. It’s excellent. Ries argues that start-ups, due to the highly uncertain nature of their offering, need a fundamentally different approach to management than bigger firms. But, he argues, it is a management approach (as opposed to the seat-of-your-pants style that often pervades entrepreneurial ventures).

The problem, Ries argues, is that most start-up employees are talented builders. Given them the plan and they’ll make it. Unfortunately, too often, what is made isn’t what the customer actually wants—or the engine of growth the business creates is too slow to survive. Instead, entrepreneurs need to steer towards validated learning, even if it sacrifices big-business notions of efficiency.

What I found particularly interesting about the book is that it’s essentially a conceptual reorganization of what the “real [10]” work of starting a company is. Many think it’s building a product and hoping for the best. Ries argues the real thing is figuring out what customers want.

6. How Innovation Works [11] by Matthew Ridley

Ridley argues that innovation is more evolution than insight. The vision of the lone genius experiencing a “Eureka!” insight is not only wrong, but it holds back our future innovation through ineffective policies. But this is more than idle speculation, Ridley offers a vast array of stories to back up his thesis.

Ridley argues that, understanding how innovation works, we ought to rid ourselves of the patent system which fails to recognize simultaneous discovery and embroils inventors in costly legal battles. Also we need to recognize the need for learning by doing. Misguided regulations aimed at safety can occasionally backfire, by keeping us stuck with nascent designs that are actually less safe than what might be possible.

7. Laboratory Life [12] by Bruno Latour and Steve Woolgar

An anthropologist spends a few years observing life in a leading biomedical laboratory. Observing how science is actually done, as opposed to how it is often idealized, he finds many common theories of science simply can’t be true. He argues that science, like all knowledge, is socially constructed.

Weeks after reading, I’m still not quite sure whether the thesis is obviously true or total nonsense. On the one hand, science is done by people. Issues of deciding which problems to work on, what evidence really ‘means’, whose work gets celebrated and whose gets ignored obviously matter.

On the other hand science seems to make contact with reality in a way that astrology or divination does not. Trying to understand science while remaining agnostic to its purported subject seems a bit like trying to understand the murmurings of art patrons while sitting in a gallery blindfolded.

Still, the fact that the ideas of the book have stayed in my mind for weeks after means it was definitely worth reading.

8. The Unschooled Mind [13] by Howard Gardner

Gardener, best known for his theory of multiple intelligences [14], argues that students don’t arrive at the classroom as blank slates. Indeed, they come with strong prior assumptions about the world that are often wrong! From economics to politics, physics to poetry, students arrive with strong preconceptions that years of schooling often fail to modify.

I’m sympathetic to Gardener, as I believe there are many unintuitive things worth knowing in schools. However, I’m also suspicious of the a priori belief that certain scholastic ways of thinking are necessarily best. Gardener uses the example of students liking poetry that rhymes, being needed to be educated that good poems don’t need to rhyme. But one could also argue that this is simply elite tastes being imposed on the masses.

Much is made of studies showing students have a naively Aristotelian model of physics. This is seen as being simply fallacious, but I suspect it’s a natural byproduct of extensive experience with real world objects. “An object in motion tends to slow down” is physically false by Newton, but it is usually true, given that friction is ubiquitous. I definitely think physics is worth learning, but it also seems wrong to discount the impressive tacit knowledge that human beings have in dealing with the physical world.

However, I largely agree with Gardener that if we want to educate students to appreciate new physics or poetry, we can’t just get them to memorize the right answer—they need to experience real situations where the educated approach is more profitable. Otherwise education risks becoming segregated from the rest of life.

9. Mind over Machine [15] by Hubert Dreyfus and Stuart Dreyfus

In a sense, this book is completely outdated. The villain is good old-fashioned artificial intelligence [16] techniques, expert systems and the assumption that all human thinking can be reduced to a simple set of rules. This was, at the time of the writing, the principle application of the computer metaphor to the mind.

Minds aren’t like machines in this way. We don’t reason through rules, consciously or subconsciously, simply because there would be an infinite number of rules to consider. These frame problems [17] plagued artificial intelligence research. Yet deep learning [18] seems to bypass many of these objections by learning in much the way the Dreyfus brothers argue should be done in humans—through pattern matching large libraries of “experiences” not simply observing hard-coded rules.

Still, I enjoyed the book because it represents a line of thinking that was largely derided by prestigious experts at the time, but which now pretty much everyone agrees was correct.

10. Shop Class as Soulcraft [19] by Matthew Crawford

Why do we denigrate the manual trades as being lower skilled than office work? Why do we encourage people to study hard only to spend their lives doing meaningless bureaucratic tasks from within a cubicle? Maybe those people would enjoy fixing things more?

Crawford’s broad-ranging attack on our cultural bias towards abstract education over hands-on work is an interesting journey. The more I probe into this topic, the more I sympathize with the view that the prestige value associated with many fields of study rests on a questionable pedestal.

Yet, I depart from Crawford in many respects. He blames capitalism for giving us too many conveniences, robbing us of the joys of repairing one’s own car. I tend to be more sympathetic to the economists on this point—a world of self-reliance is a world of unending toil. While non-college work deserve more of our esteem, I’m happy that most of my devices break down so little that I don’t know how to fix them. Progress is good, even if it sometimes has side-effects.

11. Transfer on Trial [20] Edited by Douglas Detterman

Beginning with “A Case for the Prosecution” Detterman takes the strong view that education’s lauded goal of teaching wide-ranging, all-purpose skills has utterly failed. After summarizing the evidence, he explains:

“When I began teaching, I thought it was important to make things as hard as possible for students so they would discover the principles themselves. I thought the discovery of principles was a fundamental skill that students needed to learn and transfer to new situations. Now I view education, even graduate education, as the learning of information. … In general, I subscribe to the principle that you should teach people exactly what you want them to learn in a situation as close as possible to the one in which the learning will be applied.” [emphasis mine]

While not all of the contributors to this volume are nearly so pessimistic, I’ve found reading the constellation of studies on transfer helpful for narrowing down my understanding of what learning actually is. What remains clear to me is that the issue of transfer isn’t a peripheral worry, but the central concern if we want to learn and grow.

12. Transfer of Cognitive Skill [21] by John Anderson and Mark Singley

Singley and Anderson attempt to provide an answer to the question of “what is actually learned?” when we do things. They formulate their answer in terms of “production systems” which are little IF -> THEN rules applied to the situation. These rules can be extremely specific (e.g. IF you see the variable x next to a constant THEN divide both sides of the equation by the constant) or they can be rather abstract (e.g. IF you see a complicated problem THEN break it into sub-problems).

For practical purposes, the production system approach is the most useful one I’ve read when trying to explain what gets transferred between situations. It accommodates both the fact that actual performance of skill depends on getting many low-level details right, along with the fact that more abstract ideas are useful and can occasionally cross disciplinary boundaries. The system also helps explain why sometimes we fail to transfer ideas that should be obviously useful.

Still, production systems suffer from the same flaws as good old-fashioned AI critiqued by the Dreyfus brothers above. The empirical tests of their system, while suggestive, are also far from proclaiming production systems is how we really think.

13. The Tacit Dimension [22] by Michael Polanyi

Tacit knowledge, the idea that there are skills we know how to do but can’t explain how we do them, was first introduced by Michael Polanyi in his magnum opus Personal Knowledge. This shorter book revisits the concept, which Polanyi made an essential part of his justification of science.

14. The Craftsman [23] by Richard Sennett

Manipulating objects in the physical world is essential to thinking. Philosopher Richard Sennett covers considerable ground arguing that we need our hands to use our minds. The writing is engaging, covering examples from the perils of computer automated design, to Linux developers and how Julia Childs teaches you how to cook a chicken.

====

Some other books I also enjoyed included Phil Agre’s blend of continental philosophy and computer programming in Computation and Human Experience [24], Michael Polanyi’s vigorous defense of science in Personal Knowledge [25], and David Goggins feats of human endurance in Can’t Hurt Me [26].