What Comes Next? Harnessing the Power of Prediction for More Effective Learning

At the time of writing, 1 Singapore Dollar (SGD) could be traded in the foreign exchange market for 3.05 Malaysian Ringgit (MYR) (rounded of to 2 decimal places).[1] That is, “1 SGD = 3.05 MYR”. Suppose that, for some reason, Singaporeans experience a sudden decrease in preference for Malaysian goods. Will this cause the fgure of 3.05 in “1 SGD = 3.05 MYR” to increase, decrease or remain the same, assuming no government intervention? Even if you have little idea about how exchange rates work, try to draw on whatever you may know to predict which option is correct. Lock in an answer before reading on.

The answer is “increase” and the reasoning involves economic theory[2] which, while interesting, lies beyond the scope of this article. The important point is that you have just experienced an exercise in prediction, where you were forced to draw on familiar knowledge to make a guess in an area that may be unfamiliar to you (exchange rates are part of the syllabus in the introductory economics class ofered at the National University of Singapore[3]). As a student, what might surprise you is that prediction (or simply, “guessing”) can be a strategy for more efective learning, especially if your guess was nincorrect.[4] This article will examine the power of prediction in greater detail, as applied to enhancing student learning. It will briefy describe an experiment which illustrates the power of prediction, explain possible reasons why prediction enhances learning, and then suggest ways you can harness the power of prediction for more efective learning as part of your academic routines.

Illustrating the Power of Prediction

The experiment mentioned above was carried out by Yan et al. and involved thirty-four undergraduates.[5] It replicates, with some changes, an experiment by Kornell et al.[6] All participants were shown sixty word pairs one at a time; for each word pair, the frst word was related to the second word (for instance, “table:chair” or “olive:branch”).[7]Thirty of the sixty word pairs were presented under a “study-only condition”[8], where participants were presented with the full word pair “twice consecutively, for 8 [seconds] and 5 [seconds], respectively”[9]. In other words, no guessing was required. The other thirty word pairs were presented under a “guess-frst condition”[10], where participants were given a cue followed by a blank (for instance, table:___) and given 8 seconds to guess the target word and fll in the blank (they cannot leave the blank unflled).[11] The full word pair was then shown for 5 seconds right after they made their guess.[12] Figure 1, adapting a similar diagram in the paper by Kornell et al.[13], illustrates the diference between the “study-only condition” and the “guess-only condition”.

After a fve minute break, a “cued-recall”[14] test of all sixty word pairs was administered to participants.[15] For the “guess-frst condition”, the mean proportion of targets recalled was 0.79 (standard deviation = 0.11); for the “study-only condition”, the mean proportion of targets recalled was 0.59 (standard deviation = 0.18).[16] In the researchers’ words, “the guess-frst condition produced better later recall of the target response than did the study-only condition”.[17]For completeness, it should be noted that for the “guess-frst condition”, the researchers had only analyzed items where the initial guess was incorrect[18] to derive the mean proportion of targets recalled (0.79). This avoids, or at least makes negligible, what Kornell et. al refer to as the “item selection problem”[19]: if the initial guess for a given word pair was already correct, one cannot really determine whether guessing had an impact on subsequent learning.[20].

However, one limitation of the experiment is that the interval between guessing (or just studying) and testing was only fve minutes[21]; in reality, students must retain information over far longer periods of time. Thus, the researchers conducted a separate experiment where the interval between guessing (or just studying) and testing was raised to fortyeight hours; this experiment was conducted online[22]. While overall recall rates were indeed lower after the longer delay, targets under the “guess-frst condition” were, on average, still better recalled (mean proportion of targets recalled = 0.32, standard deviation = 0.16) than targets under the “study-only” condition (mean proportion of targets recalled = 0.23, standard deviation = 0.14)[23]

The experiments described above illustrate the benefts of prediction, even if guesses were incorrect. In fact, the notion of “incorrect guesses” is highly relevant because it would likely be associated with scenarios where one has not yet learnt the material[24], for instance in this article’s opening example of exchange rates. This renders the strategy of prediction potentially meaningful for learning new concepts, something we students – by default – do frequently

Why Prediction Works Why Prediction Works

Before examining how you can translate experimental results into everyday routines, it is important to briefy understand why prediction might enhance learning. Perhaps the easiest way for learners – particularly those of us who have taken the ALS1010 Learning to Learn Better course – to understand the power of prediction is to see it as a specifc form of questioning content material[25] to foster deeper understanding of the content. In the case of prediction, questions could take on the form of “what happens next?” (during prediction) or “why did I predict this incorrectly (or correctly)?” (after prediction). Besides being a specifc form of questioning content material, prediction also forces one to search for relevant information in one’s mind, triggering existing knowledge and readying the brain to better receive (and thus retain) the actual answer to the question requiring prediction[26], as James M. Lang explains. However, it should be noted that the authors
previously mentioned that the correct answer must be given soon after the prediction is made for prediction to be efective for learning.

From Experimental Results to Everyday Routines

As a student, what will ultimately beneft you most is not knowledge of the intricacies of experiments, or extensive discussions on why prediction works. Rather, it is knowing how you can translate these results and reasons into everyday routines. Building on previously mentioned literature, this section will suggest three habits of prediction which you can easily incorporate into your academic routines. These suggestions will, of course, be written from learners’ points of view. In doing so, this article also plugs the gap of the previously mentioned literature, which was written from researchers’ or educators’ points of view.

The three suggested learning habits ft in diferent scales from small to large: concept, chapter, curriculum. The smallest scale, focusing on a specifc concept, has already been illustrated in the opening example. In the opening example, the concept at hand was changes in exchange rates. Moving from example to generalization, when reading your textbook or lecture notes, you can consider stopping at certain points when a process is being illustrated, and ask yourself: what might happen next? After making the prediction, read on to reveal whether your answer was right or wrong, and refect on why that was the case.

The next largest scale is “chapter”. Just as you can make predictions as you learn concepts while reading through the chapter, you can engage in the act of prediction even before you begin reading through the chapter itself. One way to do so would be to attempt a simple end-of-chapter question (if available) even before beginning the chapter. You should try to answer the frst end-of-chapter question (which tends to be simple) by drawing on prior knowledge, even if you are not too familiar with some concepts. When you begin to read the textbook immediately after, you should begin to see, based on the concepts introduced, whether your guess was correct or wrong (assuming solutions are not provided). You can then refect on why that was the case.

Finally, the largest scale is “curriculum”. This next suggestion draws directly from a suggestion by James M. Lang, who recommended that teachers “close class by asking students to make predictions about material that will be covered in the next class section”.[27] As a student, you can do this proactively too; suppose that, in a chapter on “Exchange Rates and Macroeconomic Policy”[28], you learnt about exchange rates in one class, and you know you will be discussing its relation to macroeconomic policy in the next class. At the end of the class or lecture on exchange rates, you could start predicting what your teacher or lecturer may cover in the next lesson. You could make use of the “Overview” section (if available) of your textbook or lecture notes to help you. Besides the usual beneft of prediction, this suggestion also helps you here with making connections between sub-topics.

Conclusion Conclusion

In summary, this article has illustrated and explained the use of prediction for the purpose of enhancing learning and built on existing literature to show how you can harness the power of prediction for more efective learning. If you need more convincing, consider that it takes only a few more minutes of your time to generate questions and predict answers, or just ten minutes to complete a simple end-of-chapter question. Despite the low time cost, bountiful benefts await in terms of more efective learning and ultimately, better academic performance. In the quest for better learning habits, the strategy of prediction may just be what you are looking for.

Reference

1. XE, “Current and Historical Rate Tables,” XE Currency Table: SGD – Singapore Dollar, accessed June 20, 2020, https://www.xe.com/currencytables/?from=SGD&date=2020-06-20.
2. Robert E. Hall and Marc Lieberman, “Exchange Rates and Macroeconomic Policy,” in Economics: Principles & Applications (Australia: South-Western Cengage Learning, 2013), 889-894.
3. “EC1101E Introduction to Economic Analysis,” LumiNUS, accessed June 21, 2020, https://luminus.nus.edu.sg/module-search/82d7ab83-c527-46df-b285-d309182a0fd8/module-description.
4. Veronica X. Yan et al., “Why Does Guessing Incorrectly Enhance, Rather than Impair, Retention?,” Memory & Cognition 42, no. 8 (2014): 1373-1383, https://doi.org/10.3758/s13421-014-0454-6.
5. Veronica X. Yan et al., “Why Does Guessing Incorrectly Enhance, Rather than Impair, Retention?,” Memory & Cognition 42, no. 8 (2014): 1373-1383, https://doi.org/10.3758/s13421-014-0454-6.
6. Nate Kornell, Matthew Jensen Hays, and Robert A. Bjork, “Unsuccessful Retrieval Attempts Enhance Subsequent Learning.,” Journal of Experimental Psychology: Learning, Memory, and Cognition 35, no. 4 (2009): 994, https://doi.org/10.1037/a0015729.
7. Veronica X. Yan et al., “Why Does Guessing Incorrectly Enhance, Rather than Impair, Retention?,” Memory & Cognition 42, no. 8 (2014): 1375, https://doi.org/10.3758/s13421-014-0454-6.
8. Ibid, 1376.
9. Ibid.
10. Ibid.
11. Ibid.
12. Veronica X. Yan et al., “Why Does Guessing Incorrectly Enhance, Rather than Impair, Retention?,” Memory & Cognition 42, no. 8 (2014): 1376, https://doi.org/10.3758/s13421-014-0454-6.
13. Nate Kornell, Matthew Jensen Hays, and Robert A. Bjork, “Unsuccessful Retrieval Attempts Enhance Subsequent Learning.,” Journal of Experimental Psychology: Learning, Memory, and Cognition 35, no. 4 (2009): 991, https://doi.org/10.1037/a0015729.
14. Veronica X. Yan et al., “Why Does Guessing Incorrectly Enhance, Rather than Impair, Retention?,” Memory & Cognition 42, no. 8 (2014): 1376, https://doi.org/10.3758/s13421-014-0454-6.
15. Ibid.
16. Ibid., 1377.
17. Ibid., 1376.
18. Veronica X. Yan et al., “Why Does Guessing Incorrectly Enhance, Rather than Impair, Retention?,” Memory & Cognition 42, no. 8 (2014): 1376, https://doi.org/10.3758/s13421-014-0454-6.
19. Nate Kornell, Matthew Jensen Hays, and Robert A. Bjork, “Unsuccessful Retrieval Attempts Enhance Subsequent Learning.,” Journal of Experimental Psychology: Learning, Memory, and Cognition 35, no. 4 (2009): 990-991, https://doi.org/10.1037/a0015729.
20. Ibid.
21. Ibid., 1379.
22. Ibid.
23. Ibid.
24. James M. Lang, “Predicting,” in Small Teaching: Everyday Lessons from the Science of Learning (San Francisco, CA: Jossey-Bass & Pfeiffer, 2016), 45.
25. Fun Man Fung, “Connecting the Dots,” January 18, 2018, Youtube video, 8:05, accessed June 21, 2020, https://www.youtube.com/watch?v=Fs50rS6oeF8&feature=emb_logo.
26. James M. Lang, “Predicting,” in Small Teaching: Everyday Lessons from the Science of Learning (San Francisco, CA: Jossey-Bass & Pfeiffer, 2016), 49-50.
27. James M. Lang, “Predicting,” in Small Teaching: Everyday Lessons from the Science of Learning (San Francisco, CA: Jossey-Bass & Pfeiffer, 2016), 60.

28. Robert E. Hall and Marc Lieberman, “Exchange Rates and Macroeconomic Policy,” in Economics: Principles & Applications (Australia: South-Western Cengage Learning, 2013), 885-920

About the Author

Ng Chia Wee

Chia Wee majors in Philosophy, politics and Economics at NUS and contributes his commentaries to CNA (Channel News Asia) and The Straits Times.

Know more about Ng Chia Wee