That rubber-ducking rubber-ducker!

Tuesday, October 16th, 2018

Anyone who has ever tried to solve a problem knows that the surest way to solve it is to call someone over and then explain how it just doesn’t make sense. That someone doesn’t even have to be a real person. (The Pragmatic Programmer calls this rubber-ducking, since explaining all your problems to a cute little toy works just fine.)

A group of researchers studied how to maximize the talking-aloud effect:

109 participants were tasked with solving different variations of the Tower of Hanoi puzzle (try it yourself right here) in the fewest number of moves, before being given a final test on the most challenging variation (to see how effectively they could transfer what they’ve learned to a new problem).

Participants were randomly assigned to one of five groups, each of which was designed to test a different kind of thinking aloud.

Before each move, the “metacognitive” group was asked to answer questions like “How are you deciding which disk to move next?” or “How do you know that this is a good move?” The idea was to get them to adopt a higher-level process focus, by thinking about what they were doing (consciously monitoring performance) and how they were doing — i.e. whether the move was a good one or not (evaluating success/failure/effectiveness).

The “if-then” group’s instructions were a little more rigidly structured, but similarly intended to get them focused on the problem-solving process: “Before each move, I want you to tell me where you are going to move each disk, and why. Specifically, I want you to state this in an ‘if-then’ statement, for example, ‘if I move this disk to this peg, then this will happen’.”

The “problem-focused” group was asked to answer questions like “What is the goal of the problem?” or “What are the rules of the problem?” before each move. The idea was to give them some structure, but not at the higher process level of the other two groups.

The “think-aloud” control group was given no real structure to guide their thinking, but simply told to “think out loud while you are solving this problem. Try to keep talking as much as you can so that I can hear what you are thinking about as you solve the problem.”

The “silent” control group was given no additional instructions beyond the standard instructions for the puzzle, so did no verbalizing of their thoughts.

[...]

On average, the control groups (silent and think-aloud) made more mistakes than the two process-focused (metacognitive and if-then) groups. This was true for every variation of the puzzle during the practice trials — from the easiest 2-disk version to the more complex 5-disk version.

Then, when the participants were tested on their ability to solve the most challenging 6-disk puzzle (to see how effectively they could transfer what they learned from the practice puzzles), the control groups made an average of 2.5 error moves for every correct move vs. just 1 error move for the process-focused groups.

The problem-focused group fared somewhere in the middle. Better than the control groups, but not as good as the process-focused groups.

[...]

1. Unless we are guided, we tend not to focus on or engage in process-level thinking. It’s more natural for us to simply execute a skill, stop, and repeat the skill on “autocorrect” mode until the problem seems to go away. Like playing a passage over and over until it sounds better. Hitting forehand volleys over and over until we get into a groove and everything seems peachy.

Except that in “solving” problems on this implicit level, while we may be able to work ourselves up to a pretty high level of performance in the short term, it involves making more mistakes during the process, and we don’t actually figure out what the solution is, so therefore can’t apply it very effectively to future problems that we might encounter.

2. When, on the other hand, we focus on what we are doing and why we are doing it (whether we are verbalizing these out loud or not), we can not only solve problems more efficiently, but transfer those solution to similar new problems we might encounter later.

Comments

  1. Aretae says:

    One of the great programming books.

  2. CVLR says:

    Don’t bother talking to an inanimate avian bath toy; write a lengthy explanatory comment instead.

    Future generations will praise your name.

Leave a Reply