Читать книгу Small Teaching - James M. Lang - Страница 12

To understand how and why predictive activities support learning, consider first an elegant series of experiments conducted by three researchers at UCLA (Kornell, Jenson Hayes, and Bjork 2009). The authors asked subjects in one of these experiments to memorize a series of loosely connected word pairs, such as whale–mammal. One group of the subjects was given 13 seconds to study each word pair; the other group was given 8 seconds to see only the first word and to make a prediction about the second, after which they had 5 seconds to see the full word pair. Since the second word of the pair was linked to the first, but only one of many possibilities (since you might well guess sea or ocean or large if you saw only the word whale), participants typically guessed the second half of the pair incorrectly. Note also that the subjects in the second (prediction) group had only 5 seconds to view the correct answer, so 8 seconds less than those in the first (nonprediction) group. Yet in spite of this shorter study time, and in spite of the fact that subjects in the second group frequently predicted the second half of the word pair incorrectly, the subjects in the second group performed significantly better than those in the first group when they were asked to recollect the word pairs on a subsequent exam: 67% accuracy in the second group versus 55% in the first. The unsuccessful prediction attempts in the experimental group, the authors explain, “were, remarkably, more effective than was spending the same time studying the answer to be recalled later” (p. 994). In other words, taking a few seconds to predict the answer before learning it, even when the prediction was incorrect, seemed to increase subsequent retention of learned material. This was true even when that prediction time substituted for—rather than supplemented—more conventional forms of studying.

The results of laboratory experiments like this one prompted another group of researchers, led by cognitive psychologist Elizabeth Bjork, to see whether they could reproduce the positive learning effect of prediction in an actual classroom (Carey 2014b). The researchers gave students in Bjork's introductory psychology class short multiple-choice pretests before some of the lectures in her course. Since the pretest questions asked them about material that had not yet been covered, the students performed about as well on the pretests as they would have from guessing randomly—again, as in the laboratory experiment, they made plenty of wrong predictions. Lectures on the subject matter followed immediately after the pretests, so the students received quick feedback on their answers. At the end of the term, the students took a final exam that contained multiple-choice questions similar to the ones on the pretests. The results paralleled the results of the laboratory experiment almost exactly: students performed around 10% better on questions from the subject areas in which they had been pretested than on those on which they had not. Bjork concluded from this experiment that “giving students a pretest on topics to be covered in a lecture improves their ability to answer related questions about those topics on a later final exam” (Carey 2014b). Note, of course, that even though the vocabulary has changed slightly here—from prediction to pretesting—the cognitive activity is similar: asking learners to give answers to questions or anticipate outcomes about which they do not yet have sufficient information or understanding. They are trying before they are ready.

Before we explore the reasons that prediction boosts learning, consider one final example of prediction in higher education, this one from an online environment (Ogan, Aleven, and Jones 2009). Three researchers from Carnegie Mellon University developed an online tutoring program that demonstrated the power of predicting in helping students improve their intercultural understanding in hybrid language courses. The two French courses described in the experiment each met once a week in a face-to-face environment, but otherwise the students did their course work online. Part of the goal for these courses was to help students develop what the authors called “intercultural competence, that is, the ability to think and act in culturally appropriate ways” (p. 268). This can be an extremely difficult skill to develop, as anyone who has ever traveled in a foreign country can likely attest. The ability to speak the language of a foreign country does not necessarily guarantee your ability to understand how to hail a cab, tip in a restaurant, or approach a stranger in the Paris Metro to ask which train will take you to the airport in time to catch your flight home (as I once discovered, to my great sorrow). So, in this experiment we are moving beyond the realm of simply knowing and retaining information into the broader realm of comprehension—that is, understanding how to use and apply in other contexts the information you have learned. The intercultural competence sought by these instructors requires learners to think and act with their knowledge, not just report it back.

To help students acquire this type of deeper comprehension, two experts in computer-assisted learning worked with a language professor to develop an online tutoring program based on the use of film clips. In the control condition of this experiment, students were shown film clips highlighting cultural attitudes or behaviors that are normally taught in introductory French classes. As the students watched the short film clips, they had the opportunity to take notes on what they saw. The students in the experimental group, by contrast, were given the opportunity to use the power of prediction to improve their learning. Their film clips would pause at key moments, ask them to make a prediction about what was about to unfold, and then require them to ponder what actually happened once the clip had finish: the authors described their three-part sequence with the catchy phrase pause–predict–ponder. The prediction the students had to make actually came from a drop-down menu of choices, but then they had open text boxes to explain why they made that prediction. After they had watched the remainder of the clip, they had to answer a simple question about whether or not their prediction was correct and then respond to prompts to help them reflect on their prediction, such as: “If so [i.e., if your prediction was correct], did you see anything you didn't expect about the French culture? If not, what happened that you didn't predict?” Students in both conditions concluded their viewing of the film clips with required postings to a discussion board to allow them to process and review what they had seen.

After the class, the researchers looked at two different measures to see whether the pause–predict–ponder exercises had improved student mastery of intercultural competence: student scores on tests of cultural knowledge, and their more general cultural thinking or reasoning skills through their discussion board posts. The students who had the opportunity to make predictions outscored their peers on the first exam by about that same 10% margin that we saw in the earlier experiments, with some diminishing returns on the subsequent assignments—which might tell us that prediction, like many of the active learning interventions we will consider in this book, especially helps new learners. Ratings of the posts in the discussion board also showed the students in the experimental condition performing significantly higher on assessments of intercultural competence than those in the control condition. In the discussion of their results, the researchers note that students made correct predictions only about 40% of the time, another point in favor of the notion (to be qualified shortly) that wrong predictions do no harm. They also point to an interesting side benefit they witnessed: students in the experimental condition posted more frequently on the discussion boards, and more frequently on target, than those in the control condition: “The experimental group showed a better ability to maintain a productive discussion compared to the control group (p. 283).” It seems that in this case the prediction activity also helped engage the students more thoroughly in the material.

Researchers who study the brain can help clarify the mechanics that underpin the results of all of these experiments. Neuroscientists are increasingly demonstrating that our brains are prediction-making machines, and that our learning stems most fundamentally from the cycle of making predictions and then adjusting our thinking in light of the accuracy of those predictions. Stanislas Dehaene is a professor of experimental cognitive psychology at the College de France, and the Director of the Neurospin Brain Imaging Center. In his book How We Learn: Why Brains Learn Better Than Any Machine … for Now, he offers a detailed but accessible tour of learning and the brain, and concludes that “Generating a prediction, detecting one's error, and correcting oneself are the very foundation of effective learning” (Dehaene 2020, p. 209). Our brains continuously create models of the world around us, use those models to predict how our experiences will unfold, and engage in corrective re-modeling in light of what actually happens. That corrective re-modeling is what we call learning. A new driver's brain makes continuous predictions about how the car will behave as she drives—she expects the car to slow down and come to a stop when she brakes. The first time she drives in the snow, applies the brake, and finds herself skidding, her brain notes the failure of her existing mental model—applying the brakes doesn't always stop the car with the same efficacy. She has to take weather conditions into account. Her mental model of driving expands, her predictive abilities as a driver improve. She has learned.

In a classroom setting, predictive activities reveal to students the gaps and problems in their existing knowledge of the course subject matter and provoke them to fill and repair their understanding. When I am asked to make a prediction or try a new skill, I am forced to surface whatever knowledge and skills I currently possess and use them in service of the task in front of me. In some cases, I will see immediately that I don't have what I need—I have no idea how to do this, I might think to myself. In a classroom setting, that recognition should push me toward learning: If I want to succeed in here, I need more information. But I might equally well think I do have enough information or skill to get the job done, in which case I will make a confident prediction, assuming that I already know everything I need to know. When I discover that my prediction was incorrect, I am put back on my heels again. What went wrong? What information was I missing? What would I have needed to know in order to get it right? Since very few students, if any, will come into our classes knowing everything they need to know already, the predictive activities we design are likely to reveal at least some errors in their thinking, showing them the gaps in their knowledge and skills. Ideally, your course material then gives them precisely what they need in order to fill those gaps.

Predictive activities can also give students a clearer understanding of what and how they need to study and learn in any particular course. As Elizabeth Bjork points out in relation to the practice tests in her experiments, “Taking a practice test and getting answers wrong seems to improve subsequent study, because the test adjusts our thinking in some way to the kind of material we need to know” (quoted in Carey 2014b). Well-planned pretests or predictive activities alert the students to essential course content and the testing style of the instructor. Envision a student sitting in a chemistry course in the fall semester of her first year of college. She had a year of chemistry in high school, and her high school teacher focused entirely on having students memorize facts and formulae, knowledge of which he assessed exclusively through multiple-choice tests. She has been conditioned by that teacher to think about and learn chemistry through rote learning, a memorization technique she will carry into her new course—unless, of course, her college professor opens the semester by asking the class to try to answer some conceptual questions, and mentions that these problems are similar to ones that will appear on the final exam. When that first-year student finds herself scratching her head and unable to come up with the answers, she will immediately see that she has to approach this course, and her learning, in a new way. She has to focus on conceptual understanding instead of the memorization of facts. If she hadn't encountered that first-day predictive activity, she might have spent the first five weeks of the semester, before the first major exam, focused entirely on repeating her high school study practices.

Ultimately and perhaps most simply, predictive activities mimic something we normally ask of learners who are attempting to master a skill: requiring them to try before they are ready. We can all likely draw from our experiences with attempts to master skills of one sort or another, and we know full well that however much one might read in advance about throwing a football or painting a portrait or giving a speech, the real learning happens after we have thrown ourselves into the situation and made that first (unsuccessful) attempt. When I took a class to become licensed in scuba diving, we spent the first half of every session in a classroom taking notes on some skill we would have to practice in the pool. I typically jumped in the pool for the second half of class thinking I had that skill mastered, but within a few minutes the gaps in my knowledge were revealed, and I floundered around for a while, doing it completely wrong until the instructor swam over and gave me the help I needed, at which point the real learning began.

We facilitate this type of learning in many academic contexts, asking students to try out cognitive skills before they are ready. I don't spend the entire semester lecturing to my freshman composition students about all of the writing techniques they will need to write a perfect academic essay and then give them one final assignment to show me how they have mastered those skills. I assign essays from the beginning of the semester, even though some of what they need to write great academic essays won't be covered for another 4 or 8 or 12 weeks. Asking students to make predictions before learning new material just represents another version of this common teaching approach.