Learn by Doing, Then by Thinking

Danger: Hubris
Creative Commons License photo credit: toolmantim

Seth Roberts talks about his graduate school days and how he got into self-experimentation by way of the axiom that, “The best way to learn is to do:”

And then I was in the library and I came across an article about teaching mathematics and the article began, “The best way to learn is to do.” And I thought “Huh well that makes a lot of sense.” And I realized you know that it was a funny thing that that’s what I wasn’t doing: I was thinking. And I also thought to myself well I want to learn how to do experiments. And if the best way to learn is to do then I should just do as many experiments as possible as opposed to trying to think of which ones to do. And that was really a vast breakthrough in my graduate training and everything changed after that.

Quoted from a 10 minute presentation by Seth Roberts

Roberts’ goes on to apply these ideas to graduate students and professors, when he notes, “Grad students … worry too much about what to do. Professors often … do something more complex than necessary.”

This is a simple, but enormously important idea: we learn more by doing first than by thinking first. The idea lies at the foundation of empiricism. Despite its simple power, it seems that across all aspects of life, we show a clear preference for thinking over doing. Why?

One reason may be that we overestimate our ability to “figure things out” by thought alone—generally, this is hubris. Another reason for this preference for thinking is poor assumptions. For the novice, the presumption is that less experience, or fewer trials under your belt, must be supplanted with more reasoning and thought. Alternatively for the expert, the problem is reliance on accumulated experience to create a basis for reliable reasoning and thought.

Regardless of the “Why?,” thinking before doing causes problems. Specifically:

  • Thinking results in unnecessary complexity, which obfuscates our ability to interpret results,
  • Thinking sets expectations, biasing analysis towards certain results, and
  • Thinking is time-intensive, reducing resources that could be used doing.

These problems hinder our ability to learn—not just in scientific experiments, but in virtually every aspect of our lives. Here are just a few examples where I’ve seen the problem manifested:

  • Our education system is founded on thinking over doing. School boards think through what subjects students should learn. Even when choice is introduced such as in college, there are enormous costs to trying a lot of disparate subjects. Not surprisingly, students get locked into fields of study only to learn when it’s too expensive to do anything about it that they don’t particularly enjoy their chosen major. Conversely, look at blogging, which seems to result in a great deal of knowledge gathering, but is driven heavily by random curiosity.
  • More on blogging. Perhaps the triumph of blogging lies in the thoughtlessness of it. Sure, you put plenty of thought into a blog post as you are writing it, but unlike writing a book, the blog post is so much cheaper that you end up having many more iterations and much less “thinking” behind each individual post. Contrast blogging to the editorial thinking that is put into mainstream journalism. This thinking results in a lot of censorship — not in the classic “You can’t use that word” sense, but in the “I think your idea should be altered in X, Y, Z ways.” The result? More complex ideas. Fewer ideas. Bad ideas. (Added 4/17/09)
  • The same problem is seen with career choices. We think our way into a certain career versus learning what works and what doesn’t work by simply trying out different types of work. We try to think our way into figuring out our passions. It just doesn’t work.
  • Or apply the idea to William Glasser’s Control Theory. Glasser argues that it is difficult to impossible to change what we think or feel about something that happens to us. Thus, our best course of action is to simply do something.
  • Or consider another book: Daniel Gilbert’s Stumbling on Happiness. Gilbert makes the point that, “We insist on steering our [lives] because we think we have a pretty good idea of where we should go, but the truth is that much of tour steering is in vain … because the future is fundamentally different than it appears through the prospectiscope.” Thinking through what we want is something we all do, yet it rarely is effective at leading to happiness. How often do we finally get what we want only to realize that the experience is not what we expected? This is a failure of thought.
  • Nassim Taleb harps on overreliance on thinking all the time. The Black Swan is essentially a book about hubris and the misguided belief that we can think through everything. As another example, Taleb doesn’t read the news because it formalizes thought, effectively handicapping our cognitive function by creating bias. In Nassim Taleb’s recent interview on EconTalk, he talks about “tinkering,” which is more or less just trying different stuff out and seeing what works, as a means to learn.
  • Or look at thinking over doing as it pertains to governments and political debate. Was there ever such an embodiment of preference for thinking over doing? Every government generally and every government program specifically is a thought-out experiment tested on a massive scale. Should it come as a surprise that governments and government programs are so dysfunctional? Observe how political philosophers consistently prefer thought to action, a la Folk Activism, dismissing attempts at trial and error or ignoring the importance of seeking new frontiers for experimentation, while arguing, “We’ve yet to see pure [ socialism | capitalism ]; therefore, you can’t say it wouldn’t work!”
  • I haven’t read Arnold Kling’s Selfish Reasons to Have More Kids (Preface), but that’s at least partially because it hasn’t been published yet! In thinking about the question of children, two thoughts come to mind in relation to the doing/thinking problem (And both relate to Kling’s review of a study about how “Almost no one regrets having kids“:
    • Couples who choose not to have kids have overthought the problem and will almost certainly regret their decision not to have kids.
    • Parents who think they should only have two kids (for example) will likely end up wishing they had had more—it seems parents tend to think they should have more kids than they end up having!

    Having kids isn’t like waking up and making an omelette, so I realize that this one fits into the doing-vs-thinking paradigm a bit loosely, but nonetheless, it’s just another example of how thinking fails. (Added 4/17/09)

  • Life is the result of trial and error performed on a massive scale and is ongoing. As complex as a DNA molecule may be, the individual building blocks are simple. So here’s an example of doing (DNA replication) and simplicity leading to unfathomable complexity—life. Evolution is the triumph of doing and is clearly a thoughtless process. (Added 4/17/09)

As Seth Roberts realized in his graduate days, “I should just do as many experiments as possible as opposed to trying to think of which ones to do.” But why does doing first work better than thinking first? Perhaps it is because doing is fundamentally an iterative process: doing is trial. The idea of trial and error as a method of learning means making mistakes and learning from them. Making mistakes and figuring out what doesn’t work can also be desirable as evidence of absence. I further wonder if it is the sheer number of trials that spur the creation of knowledge. Could it be that the more experiments/trials/iterations, the greater the chance of winning the lottery and learning something truly worthwhile? Maybe so.

I can’t help but conclude that, regardless of the reasons, thinking should almost always be put on hold in favor of action. Stop thinking and start doing. Follow whims, opportunities, gut instincts, and curiosities. Observe as much as possible. Expect failure and realize that it is through innumerable failed attempts that one can stumble on success.

Standing can improve sleep?

Stumbled on a strange idea just weird enough to warrant self-experimentation: standing to improve sleep. I found it thanks to a shared Google reader item from Patri Friedman. It was part of some notes taken from a meetup meeting:

Change in breakfast caused early awakening to get worse. This was the first thing that had made a difference. Led to discovery that any breakfast hurts; supported by rat research. Later found that standing a lot improved sleep. Had to stand at least 8 hr to get effect; no effect of 6 hr of standing. Great sleep when stood 10 hr but really hard to do. More recently discovered that standing on one leg to exhaustion helps. Do twice in one day. Same effect …

Told you it was a bit different. I’m working on hacking sleep a bit, and am immensely curious as to if this experiment would prove out for me. I sit at a desk and start at three computer screens all day, so the one-legged standing to exhaustion will be the method I try.

Stay tuned.

Sidenote: I already skip breakfast most days, preferring to wait until lunch before I break my nighttime fast.

Update: Seth kindly responded to my request for additional information on what he means by standing to exhaustion:

Justin, what I mean by “standing on one leg to exhaustion” has nothing to do with balance ? I usually touch something to make it easy to balance. I mean I stand on one leg until it becomes difficult due to muscles getting tired. I never actually endure pain or even discomfort but I go right up to the point where I would start to if I continued. In the beginning I could only do this for 2-3 minutes; now it’s up to about 11 minutes. At the moment the “whole regimen” is what you call one set: standing on each leg to exhaustion once. In other words, left leg once, right leg once.

Got it – and so the experiment can start!

Update 2: Finally got around to trying this out this evening. I still had some standing stamina at 10 minutes, so I just stopped there (per leg). We’ll see how the sleep goes tonight.