I admire when other authors showcase the best evidence against their position. It’s disappointing to finish a book and find the author ignored a good, well-known rebuttal.
In that spirit, I want to consider some of the best arguments I’ve heard against doing the real thing [1].
To recap my position: Most learning occurs by doing the thing you want to get good at. Skills are narrower than people think, and transfer is tricky. Practicing the real activity in real projects, real jobs, or for real results is more effective than many substitute or purely preparatory efforts.
The applications of this idea are broad-ranging. Some examples of doing the real thing include:
- Taking on projects at work outside your current abilities to advance your career.
- Pairing study with real-world use through co-op and apprenticeship programs in school.
- Focus on training what you want to be good at, rather than doing an unrelated activity for the purpose of strengthening mental “muscles.” (e.g. Brain training doesn’t work. Learning programming doesn’t make you a more “logical” thinker. Bilingualism doesn’t make you generally smarter.)
- Speaking a language instead of tapping through exercises in DuoLingo.
- Practicing mindfulness in daily life rather than only in seated meditation.
- Doing homework problems instead of just watching lectures.
Here, I’ll be focusing on arguments drawn from cognitive science, instead of practical concerns like convenience or cost. Not because practical considerations don’t matter, but because they’re more apparent. The idea that doing the real thing is harder is obvious; subtle contradictions from a research paper nobody has read are easier to sweep under the rug.
So let’s look under the rug and see what the best arguments I’ve heard against this view are:
1. Discovery Learning Doesn’t Work Very Well
Pure discovery learning is the idea of not giving instructions at all. Simply present the pupil with the problem situation and let them figure it out for themselves.
Unfortunately, the research seems to be against this.1 [2] Discovery learning is a lot less efficient than telling people what they ought to do and then getting them to do it.
My main takeaway from this research is that instruction matters. If there’s a good method for solving a particular problem, telling students the method is much more effective than leaving them to reinvent it themselves.
This research fits my own experiences. My portrait drawing challenge [3], for instance, began as pure discovery learning. I would keep drawing portraits until I got better. But my biggest performance gains didn’t come from extra practice; I was taught a better method [4] than what I discovered on my own.
A related issue comes up in language learning. While Vat and I made our best effort to adhere to the “No English Rule [5]” while traveling, this didn’t apply when asking linguistic questions of our tutor. Some language learning purists insist that the student never translate or break immersion, but this always seemed like trying to swim with one hand tied behind your back. If you can easily ask for an explanation and get it, why try to stumble through it?
There are no points for purity, only pragmatism. What matters is that you practice the right skills often and early. There’s no bonus for figuring them out yourself.
2. Cognitive Load Makes Problem Solving Inefficient
John Sweller’s cognitive load theory [6] argues that problem solving is often inefficient.2 [7] His studies showed that students learned to solve algebra problems faster when they were shown lots of examples of solved problems, rather than trying to solve them on their own.3 [8]
When we try to solve a specific problem, we need to keep details of our goal and how to reach it in our working memory. This additional cognitive load may interfere with schema acquisition (i.e. learning patterns that may be useful later).
Play may trump problem solving. When working on a problem without a specific goal, the student can try lots of things to figure out what works. In contrast, only one answer is needed to solve a problem with a single goal. A playful, exploratory mindset may map out the patterns of interactions better than a narrowly, solution-oriented perspective.
As an example of this, Sweller asked students to solve some math problems. One group was asked to solve the problems for a particular variable, and the other group was asked to solve for as many variables as they could. The latter group did better later, which Sweller explained in terms of cognitive load.4 [9]
These studies reconfirm that being told how to do something is generally more efficient than figuring it out for yourself. Second, they argue that exploratory-style practice may work better than overly rigid problem solving for difficult tasks. I have no quarrel with either statement.
Yet, Sweller’s suggestion to replace all problem solving with worked examples is based on the assumption that we understand what skills are needed to solve the problem.5 [10] This might be fine for algebra with its clearly defined steps, but it seems less applicable to more ambiguous problems. With cooking, for example, we all understand that watching a lot of videos does not make one into a chef.
3. Deliberate Practice Requires More than Just Doing Your Job
Deliberate practice is psychologist Anders Ericsson’s theory of expertise [11]. He argues that world-class performers get good through intensive, focused practice sessions.6 [12] These sessions challenge one to build or improve a skill rather than merely “using” it. The quantity of this practice, Ericsson argues, is what separates the best from the rest.
This argument is one I know well. I am a big fan of Ericsson’s work, and he even gave me feedback on my book Ultralearning [13] before it was published. His theory featured strongly in Top Performer [14], the career course I developed with Cal Newport, and I still reference deliberate practice in my work to this day.
Yet, it was working with students in Top Performer, that I found that the more typical cases often differ from Ericsson’s studies of violinists or chess grandmasters. For someone trying to advance in a big organization, the barrier often isn’t getting a better at the same skill you’ve been doing for twenty years. Instead, progressing in this environment requires developing completely new skills. The programmer becomes a manager. The start-up becomes a big company. New tools and technologies are required for rapid change.
Thus, in some ways, deliberate practice and doing the real thing deal with very different problems. The former suggests you’ve been doing the real thing long enough that you’ve plateaued in your performance and need to make special efforts to get better. The latter argues that you need to find real situations to acquire new, relevant skills to continue your progress.
For what it’s worth, I’ve found that taking on ambitious, real projects that require broader skill upgrades has improved my writing more than drilling specific skills I already have.
4. Theories are Useful, But Invisible
Thus far, most of the arguments against doing the real thing could be described as critiques of doing only the real thing. Getting instruction, seeing worked examples, playing instead of problem solving, and even deliberate practice are all activities that naturally dialog with real practice.
Consider, you’re a new programmer and want to get good fast. In my mind, getting a job or working on an open source project is a real activity. It might not be enough, though. You may get stuck because you don’t understand the syntax of a particular function. In this case, reading Stack Overflow code examples is perfectly acceptable. So is watching a video that explains how the function works. These detours may not be direct practice, but they complement direct practice in a way that you’d need a strong ideological commitment to ignore.
While this approach works for specific skills, it may not work if your issue is a failure to understand the broader theory. Suppose the problem you’re having with your code isn’t with a particular function but an overall design pattern. If you’ve never heard of the design pattern, you’re unlikely to rediscover it by chance. However, rarely will your difficulties coding suggest you ought to learn more about the design pattern.
The solution here seems to be to expose yourself to the theory broadly, regardless of your current project. Read books in your field, take classes, and learn theory to glean information that might someday be useful. Since the real situation is unlikely to lead quickly to discovery, or create the cues that a particular theory is needed, this kind of book learning needs to operate in parallel with direct practice.
This hybrid of doing the real thing and explanation feels true to me. It’s a big part of why I like courses and books. I don’t think they can replace direct practice, but neither can direct practice be a complete substitute for studying theory.
Other Arguments
This list isn’t exhaustive. Cost and convenience are significant factors, and there’s a trade-off between them and doing the real thing when choosing our approach to learning. Additionally, ill-defined goals make it harder to improve our practice—how do you know what the real thing is if you aren’t even sure what you’re trying to be good at? Dabbling may be a good way of testing one’s own interest in a subject, even if the activities involved don’t contribute much useful practice.
There’s also the difficulty of needing credentials, even if the subjects you study are dissimilar from the work. The “paying your dues” kind of work may be an unavoidable feature of many pursuits. It is fundamentally dissimilar to the end goal, but is needed to signal your commitment.
Despite these difficulties, the deeper I delve into this topic, the more I feel like it has important implications, not just for how we learn things but for everything we pursue.
Footnotes
- Kirschner, Paul, John Sweller, and Richard E. Clark. “Why unguided learning does not work: An analysis of the failure of discovery learning, problem-based learning, experiential learning and inquiry-based learning.” Educational Psychologist 41, no. 2 (2006): 75-86.
- Sweller, John. “Cognitive load during problem solving: Effects on learning.” Cognitive science 12, no. 2 (1988): 257-285. [15]
- Sweller, John, and Graham A. Cooper. “The use of worked examples as a substitute for problem solving in learning algebra.” Cognition and instruction 2, no. 1 (1985): 59-89.
- Sweller, John, Robert F. Mawer, and Mark R. Ward. “Development of expertise in mathematical problem solving.” Journal of Experimental Psychology: General 112, no. 4 (1983): 6 [16]
- Sweller, John, and Graham A. Cooper. “The use of worked examples as a substitute for problem solving in learning algebra.” Cognition and instruction 2, no. 1 (1985): 59-89.
- Ericsson, K. Anders, Ralf T. Krampe, and Clemens Tesch-Römer. “The role of deliberate practice in the acquisition of expert performance.” Psychological review 100, no. 3 (1993): 363. [17]