When we talk about creating durable knowledge, we often focus on the strategies used to initially put memories into our minds. This process is called encoding.
However, the ability to remember something is not just a function of how the memory is put in, but also how it’s taken out. As it turns out, those two processes are closely related. Memories aren’t stored and checked out like DVDs at Blockbuster (RIP). Each memory is actually reconstructed every time we “remember” it.¹ This return to the original memory state is like the mind has another chance to experience the memory and encode it. It partially explains why retrieval practice (i.e. effortful recall from memory) is so powerful. Each time you reconstruct a memory, you are essentially encoding it again.
So how do our minds work to reconstruct memories at retrieval, and how we can we take advantage of that mechanism to learn better?
Studies show that we’re more likely to remember something if we’re thinking the same way at retrieval as we were at the time of memory formation. In other words, successful retrieval is more likely when there’s a match between cognitive processes at retrieval and encoding. This concept is called transfer appropriate processing or TAP.
Here’s an example of TAP. First, you’re assigned one of two encoding processes to apply to a list of words. I’ll illustrate the options with the word “dungeon”:
Think of a word that rhymes with “dungeon.”
Create a sentence with the word “dungeon.”
At retrieval, you’re asked to recall these words. One of the cues is “a place used to imprison people.” You might be more likely to remember “dungeon” if you used the second encoding process. The semantic use of the word at encoding matches the semantic use of the word at retrieval. This is TAP.
However, if I asked you remember a word from the list that rhymes with “luncheon,” which encoding process do you think will be more useful to this retrieval process? Consistent with TAP, the first encoding strategy works better for a retrieval task that relies on rhyming.²
In practice, these results mean that we should try to match our encoding processes to how we will be retrieving the information. If my goal task is passing Step 1, then learning through multiple choice QBanks would lead to TAP at retrieval—and correlates with better scores.³⁻⁵ If my goal is forming a strong differential diagnosis, then learning through an app like Human Dx (a clinical case simulator) would would lead to TAP at retrieval. If my goal is using different suture knots, then learning with physical suture kits would lead to TAP at retrieval. While TAP isn’t sufficient for good learning, it does make learning more efficient. As you can see, trying to maintain TAP means that I’ve had to vary my learning methods throughout my education as my goals continue to evolve.
Lastly, I used to wonder how where memory palaces, reading ahead of class, going to lecture, etc., might fit in with TAP. The TAP match that we want between retrieval and encoding doesn’t specifically mean a match between retrieval and first-time encoding, but between retrieval and retrieval-induced encoding. Meanwhile, mnemonic strategies, such as the memory palace, facilitate retrieval success—also important for effective retrieval practice, but a story for another day!
For example:
Transfer appropriate processing
Initial encoding: Memory palace
Retrieval-induced encoding: Multiple choice questions in Step 1-style vignettes
Retrieval goal: Step 1 multiple choice questions
Transfer inappropriate processing
Initial encoding: Memory palace
Retrieval-induced encoding: Multiple choice questions about loci in the palace
Retrieval goal: Step 1 multiple choice questions
To put it all together, I want to practice retrieval in a transfer appropriate manner, such that my retrieval-induced learning is encoded in a goal-relevant way. As we so often conclude, effortful recall is critical for efficient learning.
1. Schacter, D. L. & Addis, D. R. The cognitive neuroscience of constructive memory: remembering the past and imagining the future. Philos Trans R Soc Lond B Biol Sci 362, 773–786 (2007).
2. Morris, C. D., Bransford, J. D. & Franks, J. J. Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior 16, 519–533 (1977).
3. Deng, F., Gluckstein, J. A. & Larsen, D. P. Student-directed retrieval practice is a predictor of medical licensing examination performance. Perspect Med Educ 4, 308–313 (2015).
4. Bonasso, P., Lucke-Wold, B., Reed, Z., Bozek, J. & Cottrell, S. Investigating the Impact of Preparation Strategies on USMLE Step 1 Performance. MedEdPublish 4, (2015).
5. Burk-Rafel, J., Santen, S. A. & Purkiss, J. Study Behaviors and USMLE Step 1 Performance: Implications of a Student Self-Directed Parallel Curriculum. Acad Med 92, S67–S74 (2017).
Note: We’ve gotten a few questions over the years about using memory palaces to learn some sort of physical skill, such as karate or playing the guitar or checkers. If my goal were to ace a written test on music theory, I might make a palace to organize the concepts and encode that information using Anki. If my goal were to play the guitar, however, I might jump right in and learn with hands on the strings in a transfer appropriate way.