Can Explanations Make Learning More Difficult?

The act of generating explanations will always be at the heart of human learning. For example, as a toddler you learn to stop touching pointy things by generating the “explanation” that pointy things cause pain. Similarly, when economists try to learn what causes recessions, they are essentially seeking an explanation for why past recessions have occurred. The agreed upon explanation eventually becomes new knowledge — we have then “learned” why recessions happen (to the best of our ability).

Although generating explanations is a powerful learning strategy, it’s worth asking whether it can lead us astray. That was the question that prompted a pair of experiments (pdf) led by Berkeley’s Joseph Jay Williams and Tania Lombrozo. In their initial experiment participants were presented with facts about a car (e.g. color, transmission type, ideal climate, etc.) and then asked to guess which of two fictional categories the car belonged to (i.e. Is it a “Dax” or a “Kez”?) After guessing, participants were told the correct answer. This was repeated for all ten cars in the experiment, a process that constituted one “learning block.” In order to learn which categories the cars belonged to, participants were allowed to go trough a maximum of 15 blocks, although they could stop once they had correctly classified all 10 vehicles. The researchers were interested in the speed at which participants learned to categorize the cars.

Now for the interesting part. Before beginning the experiment, one group of participants were instructed to explain why they thought each car was a “Dax” or a “Kez” throughout the course of the experiment (“explain condition”). The other group was merely told to say out loud what they were thinking (“think aloud condition”). In addition, half of each group was presented with 10 vehicles for which there was a perfect pattern of categorization — for example, all cars designed for cold weather were a “Dax” and all cars designed for warm weather were a “Kez.” The other participants were presented with 10 vehicles for which an exception created a misleading pattern — for example, one cold weather car was actually a “Kez” and one warm weather car was actually a “Dax.” This created four experimental groups — explain/perfect pattern, explain/misleading pattern, think aloud/perfect pattern, and think aloud/misleading pattern.

When the researchers looked at how quickly participants learned, they found that when the pattern was perfect, subjects who generated explanations learned faster. However, when an exception created a pattern that had the potential to be misleading, attempts to generate an explanation led to slower learning. It appeared that the desire to generate a broad explanation made it harder for participants to deal with a unique circumstance. A follow up experiment that involved categorizing people as likely or unlikely to give to charity — a somewhat more ecological scenario —  replicated the initial findings. When a person violated the apparent pattern (e.g. young people are unlikely to donate), the desire to explain led people to learn about donation tendencies more slowly.

It’s important to note that the experiment involved short term categorization, an extremely simple task compared to figuring out what causes recessions, so it’s not as if the results mean that explaining isn’t a good way to learn how to solve society’s most complex problems. But the study does suggest that, at the margin, people should be conscious of unique circumstances and not try to jam every new occurrence into an existing pattern.

The study is also a good reminder of how much variance there is in the act of learning. Making sense of information is a complex process that can be enhanced or destroyed by small and unique personal, motivational, and contextual differences. And so when it comes to improving our education system there probably is no silver bullet. People are too different. Just as we can’t find a single cure for cancer because it has so many variations, human differences are likely to prevent schools from finding a quick and easy universal cure for poor learning. (Of course that doesn’t mean we shouldn’t still try.)
Williams, J., Lombrozo, T., & Rehder, B. (2013). The Hazards of Explanation: Overgeneralization in the Face of Exceptions. Journal of Experimental Psychology: General DOI: 10.1037/a0030996

One Response to Can Explanations Make Learning More Difficult?

  1. Your point about variance in learning is very true. We all have different and constantly adapting cognitive models through which we interpret information and this is often forgotten in educational settings. I think we need more educators to understand this point though, since that will provide more impetus for designing future intelligent systems that customise teaching based on what is known/has been learned about each given individual. This is especially important given that most education and teaching methodologies are still shockingly old fashioned.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s