Studying to the HILT: Why Learning Should Look More Like Exercise

Recently my wife and I committed to getting more exercise. And in researching possible exercise regimens we stumbled upon a popular new workout phenomena known as high intensity interval training (HIIT). If you’re interested, I would encourage you to check out some examples on YouTube. Personally, I get tired just watching the videos!

HIIT is based on evidence showing that short, high-effort, workout sessions can produce calorie-burning value equivalent to longer, moderately challenging, exercise routines. It seems brief episodes of pushing oneself to the limit — even if one is unable to perform flawlessly and experiences significant struggle — is more efficient for burning calories than lengthy and mildly strenuous exercise.

As it turns out, current research on memory and learning suggests a similar approach may also be ideal for training the mind.

Short bursts of high mental difficulty — effort likely to produce frequent errors and engender significant cognitive strain — is a very effective way to learn, particularly for low-complexity knowledge components (e.g., facts, categories, concepts, relationships, definitions, etc.) (see, Koedinger et al., 2012). Much like HIIT workouts for the body, the goal is to engender high levels of mental effort during study. Just as repeated reps with a light dumbbell lead to minimal muscular benefits, easy retrievals of information lead to minimal learning (Bjork, 2011). And the more effort involved in trying to recall information from memory, the more etched the information becomes in our minds (Pyc & Rawson, 2009). Yet evidence suggests students typically believe the opposite, thinking if information is easily and quickly retrieved then it is deeply learned (Benjamin et al., 1998).

Unfortunately, most adaptive studying tools fail to incorporate these research insights in their design. Adaptive studying products often encourage suboptimal practice insofar as they:

• Erroneously focus on a student’s current performance as indicative of long-term learning gains
• Sequence the rehearsal of study content in a misguided effort to forestall student forgetting/errors
• Focus primarily on question difficulty/accuracy rather than retrieval difficulty in recommending study questions
• Ignore the highly contextual and cue-dependent nature of human memory

As a result, adaptive study tools enable students to quickly achieve high levels of performance, but ultimately produce inefficient, non-transferable, and short-lived learning gains. It’s like working out with a light dumbbell and expecting massive improvement in muscle mass. And because adaptive study tools rely on a faulty mental model of learning, they encourage inefficient practice while grossly overestimating student learning and knowledge stability across contexts/settings.

So how might we improve the learning design of an adaptive studying product?

The first step is challenging two assumptions about memory and learning that hinder current thinking about how to best support student study.

Two Problematic Assumptions About Memory & Learning

Most adaptive study products exhibit two fundamental misunderstandings about human memory.

First, it is assumed that forgetting is the antithesis of learning. While many people recognize that allowing some attenuation of memory is valuable, it is widely believed that actually forgetting information is undesirable. For instance, claims like “The best time to practice is when you are on the verge of forgetting…” reveal this common assumption that we should get to the brink of forgetting but not cross it, which ostensibly entails an instructional failure. That is, rather than timely reinforcement of an idea on the verge of expiring, we are now in the position of needing to relearn the material and have lost memorial ground.

(An aside: This idea is often associated with mythological ideas about “the” forgetting curve. But despite ubiquitous visualizations to the contrary, there is no predefined rate at which humans lose access to information as this depends on many interacting factors (e.g., student prior knowledge, meaningfulness of material, learning strategy, contextual cues, emotional salience, interest, etc.).

But recent research suggests this view is mistaken. Learning occurs even when we are unable to successfully recall information and retrieval failures are often an important indicator that robust learning is happening (Kornell et al., 2015).

As the educational psychologists Kornell and Vaughn write,

“Retrieval attempts do enhance learning even when they are not successful. Students and teachers are prone to actively seek strategies that safeguard retrieval success, or to avoid strategies that might stimulate retrieval failure. We believe these efforts are often misguided” (2016, p. 36).

In fact, growing research into the “errorful generation” effect reveals that even when a student gets every question wrong on an assessment — because they haven’t even studied the content yet — they still benefit more after corrective feedback than a student who spends this time studying the correct answers, a result that students predictably do not recognize (see, Little & Bjork, 2016; Potts & Shanks, 2014; Yang et al., 2017).

Furthermore, as Storm and colleagues note, by spacing retrieval attempts to ensure success (i.e., attempting to thwart “the forgetting curve”), we undermine the memorial value that comes from more difficult practice,

“One of the key benefits of expanding retrieval practice is that it is designed to keep as many items as possible above the recall threshold, thus ensuring that each item benefits from every test opportunity. Ensuring success, however, may come at a cost; namely, each of the tests must be relatively easy, thus preventing the potentially large benefits that might have been accrued from more difficult tests.” (2014, 122).

So instead of worrying about success of retrieval, we should prioritize maximizing the amount of mental effort students expend in attempting to retrieve target information.

For example, a student who racks her brain for a piece of information, but fails to retrieve it successfully, is likely to benefit more from a learning trial after receiving feedback than a student who expends minimal cognitive effort but retrieves the correct answer (Hays et al., 2013). That’s right, retrieval success is likely to lead to less learning than retrieval failure.

“Assuming one will be given feedback, the best time to attempt retrieval may be long after an item has ceased to be retrievable” (Kornell, Klein, & Rawson, 2014, 9).

Thus high failure rates are an indication that students are benefiting from learning trials insofar as this is correlated with greater retrieval difficulty. Kornell and Vaughn again, “ Instead of worrying about retrieval success, students and teachers should embrace errors as a path to knowledge” (2016, p. 36).

The second mistaken assumption often exhibited by adaptive studying products is that forgetting is an erosion of information in memory, what Thorndike referred to as the “law of disuse” (1914). Forgetting is incorrectly perceived as a loss of information in memory rather than a loss of access to information.

Current research suggests that information in long-term memory remains stable and it is only our ability to retrieve information that fluctuates due to the cue-dependent nature of human memory. According to the influential New Theory of Disuse, retrieval strength is defined as ease of current access to information in memory while storage strength is a measure of how entrenched, enduring, and transferable knowledge is (Bjork & Bjork, 1992). And substantial experimental evidence suggests that gains in storage strength are greatest when retrieval strength is lowest — i.e., when students experience greatest difficulty in accessing target information while studying. This partially explains the benefit that forgetting provides to learning as described above.

The most widely recognized strategy for reducing student retrieval strength is through the effect of spacing. That is, waiting for students to sufficiently forget (i.e., lose access to) previously recalled information by refraining from retrieving it for an extended period of time.

However, when we understand that the benefit of spacing is fundamentally about increasing retrieval difficulty rather than obviating memorial decay, we gain the insight that spacing is only one strategy for achieving the goal of increasing retrieval difficulty for students (i.e., reducing access to target information). For instance, reviews of current research reveal that a student’s current retrieval strength for information is a function of numerous non-temporal factors, including: question context, available cues, feedback timing, interference effects, modality, and generative task requirements (see Healy et al., 2014; Soderstrom & Bjork, 2015).

All of this suggests that relying solely on spacing to reduce student retrieval strength during practice is inefficient and ignores many other available strategies for maintaining high retrieval difficulty for learners during study.

So, given these two considerations — that forgetting is a valuable and desirable part of the learning process and that retrieval difficulty is a function of both temporal and non-temporal factors — how might we imagine designing an innovative adaptive study product that better reflects students’ cognitive architecture?

Improving Study With High Intensity Learning Trials (HILT)

Imagine an adaptive study application that counterintuitively optimizes the efficiency of student learning trials by intentionally inducing desirable student retrieval difficulty through adaptive question selection. Analogous to HIIT workouts for the body, this application would engage students in high intensity learning trials (HILT) that maximize the impact and efficiency of study practice by ensuring learning sessions involve consistently high mental effort.

HILT would be designed to predict a student’s likely retrieval strength for a possible question based on her previous responses, and select future study questions on the basis of their likelihood of being challenging for the student to retrieve. Rather than seeing quick performance gains as a positive, high failure rates are expected and desirable.

HILT would embody the key cognitive insight that the path to enduring knowledge is often sinuous and errorful.

So how can we maintain consistent student mental effort?

As mentioned previously, any given study question is characterized by a high-dimensional web of memorially influential features, features that can be manipulated in an attempt to maintain consistent retrieval effort for learners. Each question has an associated assessment grammar, if you will. See figure.

This assessment grammar may include temporal features (e.g., time since last exposure), structural features (e.g., available cues/question structure), functional features (e.g., what generative activity is required), relational features (e.g., what questions preceded or followed [interleaving] and how knowledge components are connected), modality features (e.g., visual or auditory presentation), encoding features (e.g., depth of processing required), episodic features, fluency features (e.g., font type/readability), and informational features (e.g., what feedback/hints/scaffolds are given and when).

By leveraging this assessment grammar to manipulate the sequence and features of study questions, we can maintain high retrieval challenge for students.

An additional benefit of this approach is greater transfer of learning gains across contexts and situations. Typically studying occurs in a consistent setting/format/interface with little modification — repeated series of nearly identical learning trials (flashcards anyone?). However, by constantly varying the conditions of practice, e.g., removing cues and changing question format, we can increase mental strain while also engendering more transferable and enduring knowledge (Whiffen & Karpicke, 2017). Additionally, this approach would enable the identification of student weakness and strengths across dimensions, enabling more personalized practice and recommendations to optimize learning outcomes.

What About Student Motivation?

One of the biggest challenges with the proposed approach is that it will significantly increase student retrieval failure and cognitive struggle during study. But we must remember that errorless retrieval is often a sign that little learning is happening, not that students have successfully learned the material! Thus cognitive strain should not be viewed as something to avoid, and we must embrace the idea that we learn more from effortful failure than easy success.

However, it is clear that frequent errors and mental struggle may be discouraging and exhausting for students, and student motivation is just as important in the success of an educational product as the implementation of effective cognitive study techniques.

To tackle this challenge we might bring this article full circle and consider adopting the metaphor of a fitness workout. HILT might be packaged to students as a ‘mental workout’ tool that provides high-intensity study practice and measures student brain burn(i.e., overall retrieval effort and gains in storage strength) rather than focusing on misleading indicators of learning like accuracy, frequency, fluency, or speed. Just as HIIT workouts are designed to elicit maximum physical exertion, HILT adaptively introduces desirable learning difficulties to keep learner mental effort high.

An additional value of this approach is that it focuses student attention on learning improvement rather than performance. Students often focus on performance when making judgments of learning, succumbing to various biases that include expecting future performance will match current performance and that current performance will remain stable across time and space. Yet growing body of research suggests that the accuracy of a student’s beliefs about the stability of her current knowledge, and confidence in her ability to improve through additional practice, is greatly aided by making latent learning gains more salient (Koriat et al., 2004; Kornell & Hausman, 2017).

Thus a visual brain burn indicator may help students maintain the motivation to continue expending mental effort, even in the face of initial poor performance, and recognize that additional effort will pay off in further learning gains even if progress appears slow. The goal is to create “a space that encourages and embraces making mistakes and residing in states of not knowing” (Overoye & Storm, 2015, p.145).

Finally, the brief, but intense, study workouts envisioned by HILT align with recent research on the impact of working memory depletion and its possible effects on learning (see Chen et al., 2017). Put simply, by emphasizing short study sessions, HILT ensures sufficient time for the brain’s ‘muscles’ to recover from the high mental exertion robust learning requires.

Ultimately, the success of an idea like this will be judged primarily by the extent to which both students and instructors move from viewing errorless and undifferentiated retrieval practice as an efficacious study approach and instead embrace the idea that a study session was a “Great Workout” because it was mentally challenging and involved pushing students to their cognitive limit. That is, whether it encouraged students to study to the HILT!

So what do you think? Would you study with HILT?

References

Benjamin, A.S., Bjork, R.A.,& Schwartz, B.L. (1998). The mismeasure of memory: When retrieval fluency is misleading as a metamnemonic index. Journal of Experimental Psychology: General, 127, 55–68.

Bjork, R. A. (2011). On the symbiosis of remembering, forgetting, and learning. Successful remembering and successful forgetting: A Festschrift in honor of Robert A. Bjork, 1–22.

Bjork, R. A., & Bjork, E. L. (1992). A new theory of disuse and an old theory of stimulus fluctuation. In A. Healy, S. Kosslyn, & R. Shiffrin (Eds.), From learning processes to cognitive processes: Essays in honor of William K. Estes (Vol. 2, pp. 35–67). Hillsdale, NJ: Erlbaum.

Chen, O., Castro-Alonso, J. C., Paas, F., & Sweller, J. (2017). Extending Cognitive Load Theory to Incorporate Working Memory Resource Depletion: Evidence from the Spacing Effect. Educational Psychology Review, 1–19.

Hays, M. J., Kornell, N., & Bjork, R. a. (2013). When and Why a Failed Test Potentiates the Effectiveness of Subsequent Study. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39(1), 290–296.

Healy, A. F., Kole, J. A., & Bourne, L. E. (2014). Training principles to advance expertise. Frontiers in Psychology, 5, 2012–2015.

Koedinger, K. R., Corbett, A. T., & Perfetti, C. (2012). The Knowledge-Learning-Instruction Framework: Bridging the Science-Practice Chasm to Enhance Robust Student Learning. Cognitive Science, 36(5), 757–798.

Koriat, A., Bjork, R. A., Sheffer, L.,& Bar, S. K. (2004). Predicting one’s own forgetting: The role of experience-based and theory-based processes. Journal of Experimental Psychology: General, 133(4), 643– 656.

Kornell, N., & Hausman, H. (2017). Performance bias: Why judgments of learning are not affected by learning. Memory & Cognition, 1–11.

Kornell, N., Klein, P. J., & Rawson, K. A. (2015). Retrieval attempts enhance learning, but retrieval success (versus failure) does not matter. Journal of Experimental Psychology: Learning, Memory, and Cognition, 41(1), 283.

Kornell, N., & Vaughn, K. E. (2016). How Retrieval Attempts Affect Learning: A Review and Synthesis. Psychology of Learning and Motivation, 65, 183–215.

Little, J. L., & Bjork, E. L. (2016). Multiple-choice pretesting potentiates learning of related information. Memory & Cognition, 44(7), 1085–1101.

Overoye, A. L., & Storm, B. C. (2015). Harnessing the power of uncertainty to enhance learning. Translational Issues in Psychological Science, 1(2), 140–148.

Potts, R., & Shanks, D. (2014). The benefit of generating errors during learning. Journal of Experimental Psychology, 143, 644–667.

Pyc, M. A., & Rawson, K. A. (2009). Testing the retrieval effort hypothesis: Does greater difficulty correctly recalling information lead to higher levels of memory? Journal of Memory and Language, 60(4), 437–447.

Soderstrom, N. C., & Bjork, R. A. (2015). Learning Versus Performance: An Integrative Review. Perspectives on Psychological Science, 10(2), 176–199.

Storm, B. C., Friedman, M. C., Murayama, K., & Bjork, R. A. (2014). On the transfer of prior tests or study events to subsequent study. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40(1), 115–24.

Thorndike, E. L. (1914). The psychology of learning. New York, NY: Teachers College Press.

Whiffen, J. W., & Karpicke, J. D. (2017). The Role of Episodic Context in Retrieval Practice Effects. Journal of experimental psychology. Learning, memory, and cognition.

Yang, C., Potts, R., & Shanks, D. R. (2017). Metacognitive Unawareness of the Errorful Generation Benefit and Its Effects on Self-Regulated Learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43(7), 1073–1092.