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

How My Views on Learning Have Changed Over Time

I’ve been writing this blog for almost seventeen years. From nearly the beginning, learning has been a central theme. Initially, as a college student, “how to study” was pretty much the only topic I could credibly offer advice on, but my interest in the subject of learning has endured.

However, the disadvantage of such an early start is that my naive opinions are encased in the amber of my archives. While some of my early ideas were outright bad, more often, they contained a mix of useful ideas and unhelpful suggestions.

With this post, I’d like to clarify how my thinking has changed over time. Not to undo my past writing or to claim my current views are final, but to help explain why I changed my mind about some of the things I believed in the past.

Early Views: Holistic Learning and Learn More, Study Less

Listen to this article

My first popular writing came from an observation that successful students seem to deeply understand subjects by linking them together. In contrast, less successful students attempt to memorize things by rote. I called the successful strategy “holistic learning,” and my earliest work centered on it.

None of this writing was grounded in research. That isn’t to say all opinions need a citation to be valuable, simply that I wasn’t basing any of my thoughts on a careful review of the scientific literature. Instead, I derived most of my advice from personal experience and reading other popular studying advice.

Having read more deeply now, I often see parallels to my thinking in formal research—even though I relied on none of this research when these topics were central to my advice. Mental models [1] have inspired considerable research. The associative character of memory that fascinated me resembles spreading activation [2] models of declarative memory. My idea about mental constructs is similar to schema [3] theory.

The philosophy I espoused was essentially a version of constructivism [4]—the idea that students construct meaning and ground understanding of abstract ideas in prior knowledge. In this view, it’s the active effort students expend in creating an understanding that leads to learning, an effort often stifled by drills and memorization.

What I Got Wrong in My Early Thinking

Some of the central pieces of advice I gave during this period included:

None of these are exactly false, but I wouldn’t make any of them central today. They all suffer from some problems:

Overall, the worst advice I gave was discouraging repetitive practice in favor of associative memory. My view on this is the opposite now. I believe associative strategies like mnemonics should be supplementary to retrieval practice, such as flashcards, rather than the reverse. Practice questions should be the cornerstone of studying, not a crutch to be avoided.

While it’s been interesting to see some of my initial intuitions about learning reflected back by Gestalt psychology or Constructivist thinkers, I don’t think I can take too much credit. Instead, I think I based my intuitions about learning on the same observations that inspired both the cultural Zeitgeist that made my early writing popular and also more serious research.

What I missed was that the process good students use is essentially what everyone does when learning subjects they find easy. When you’re learning hard things, and understanding doesn’t come easily, practice and memorization aren’t things to be avoided but essential building blocks toward deeper understanding.

Maturing Thoughts: Learning Projects and Ultralearning

After college, I embarked on a series of learning challenges: MIT’s computer science curriculum [5], multiple languages [6], art [7] and more [8]. These culminated in my 2019 book, Ultralearning [9].

This period spans nearly a decade, and thus my thinking during the MIT Challenge was quite different from when I had finished the research for Ultralearning. But, some consistent themes emerge.

One is the importance of practice. Unlike the conceptual understanding central to my undergraduate education, the cornerstone of my further learning efforts was practice, practice, practice.

During the MIT Challenge, I found practice problems to be the most effective tool for prepping for difficult exams. In my language learning odyssey, I spent nearly all my time practicing through conversations, flashcards and grammar books. With portrait drawing, I based my model of learning almost entirely on repetitive practice, to such an extent that I neglected learning more effective methods until almost halfway through the project.

This emphasis on practice is reflected in the research I did for Ultralearning. I wrote chapters on Directness, drawing upon the extensive research showing that people frequently fail to transfer skills to new areas; Drill, inspired by the work on deliberate practice showing effortful, targeted efforts at improvement are essential; and Retrieval, built on the robust literature showing that memory strengthens more from recall than review.

Mistakes Made in My Maturing Thoughts

Here I think my track record is much better than my early thinking. If I had to go through and edit Ultralearning again, there’s not much I would rewrite.

However, I think my focus on self-directed learning blinded me somewhat to the distinct challenges it poses.

First, I emphasized practice and de-emphasized examples and explanations. Part of this was because students have much less control over the latter. Some classes have tons of practice problems with worked out solutions. Some have almost none. Students generally don’t have much choice over which they have to take.

Second, following my interest in deliberate practice, I tended to view harder learning as more efficient. A cornerstone of the argument in Ultralearning was that greater efficiency came from more strenuous efforts. This view has some support: Bjork’s work on desirable difficulties [10], retrieval practice and others all lend some suggestion that students slack off to their own detriment.

But while practice is good, examples are too [11]! Watching other people perform a skill, especially with explanations for their decisions, is central to learning well. Similarly, while effortful practice is often necessary, not all effort is worthwhile. I now believe a lot of student struggles are wasteful [12]—failures of instructors to provide thorough examples and explanations rather than a sign that deeper learning is taking place.

Recent Adjustments: Direct Instruction and Cognitive Science

Since Ultralearning, I’ve delved far deeper into the science of learning. I collaborated with Jakub Jilek, a cognitive science doctoral student, on three research reviews on long-term memory [13], working memory [14] and self-control [15]. Afterward, I did a solo research project into motivation [16].

More recently, I’ve done a wide-ranging research project that has exposed me to the main currents in educational and cognitive psychology. Through this effort, I now feel well-versed in the history of different theories, current controversies and the main arguments and research used to support different opinions.

Reading this literature has made me more supportive of Direct Instruction [17] for learning. DI is a teaching philosophy that involves breaking down complex skills into simple concepts and actions, teaching with ample examples and practice. Critics accuse it of being mindless, much like I criticized rote learning in my early days. I now see this as its strength: when learning can occur without requiring exceptional cleverness, far more students will benefit, not just the brightest.

I now believe that practice difficulty comes in different flavors. Some difficulty is due to retrieval—you don’t know which knowledge to bring to a problem. In this case, the research seems to support the idea that moderate difficulty is better. We want problems easy enough that we’re typically successful, but not so easy that we need to rely on hints.

Other kinds of difficulty, however, are probably wasteful. Seeing good examples and explanations appears to involve different learning processes than learning by doing. The latter can not only be frustrating and slow, but it can lead to poorer generalization as well. For the majority of skills, it definitely seems like a lack of good examples is more of a bottleneck to learning than is an overreliance on easy practice.

Similarly, I now view issues of practice realism [18] in a different light. Realistic practice will be more efficient for skills of low-to-moderate cognitive load. But this same realism can make things harder to grasp for high-cognitive load skills, as in many traditional classroom subjects like math, programming or grammar in foreign languages. The right approach is a ramp: for complex skills, start with simplified problems with plenty of feedback and instructions and move onto more realistic applications in more ambiguous settings once the foundation is secure.

With So Many Changes, Why Listen to Me?

I’m not an expert. Part of this is an admission that I still have gaps in my knowledge I’m striving to fill (and likely always will). But part of it is that expertise is a social label. Belonging to that social category would require a career change I’m not particularly interested in making right now.

Given my lack of expertise, and my frank admission that my views have changed, it’s worthwhile to ask whether I should be listened to at all. What are the chances that I’m going to turn around ten years from now and deny everything I’ve said today?

After all, if you wanted to understand the psychology of learning, you won’t do much better than John Anderson’s textbook [19]; for memory, there’s Alan Baddeley’s [20]; from a neuroscientific perspective, there’s Stanislas Dehaene [21]; for teaching, there’s Daniel Willingham’s books [22] and essays [23]; and for expertise, there’s Anders Ericsson’s [24]. I’d trust all of them more than me.

However, if I might offer a weak defense of my work, I would argue:

But ultimately, the justification for my work is you! For some reason, people have stuck around reading this blog, despite it being a continual work in progress. Few people get to make a career about learning things, and for that, I owe everyone who listens to me a debt of gratitude.