“temporary tattoo on the brain versus a permanent one”
Bjork’s claim goes against instinct on purpose. What you underlined is the whole trap in the author’s own words: “the facilitating effect of apparent impediments to learning,” the whole thing “going against one’s intuitions.” The easy, fluent pass feels productive and evaporates. The hard, effortful pass feels like failure and it’s the one that holds. The difficulty is not in the way of the learning. It is the learning.
Underneath sit two separate quantities the paper keeps apart: retrieval strength (how accessible a memory is right this second) and storage strength (how deeply it is actually learned). Performance measures the first and everyone mistakes it for the second. Your two images caught the split cold — the temporary tattoo and the receding wave are retrieval strength; the permanent tattoo and the reshaped sand are storage strength.
New Terms — 1) Higher order representation 2) Reloading memories (#!)The (#!) is your own capture command — it fires both terms straight to the epochlearning board. Sharp catch: interleaving forces the mind to reload a half-receded memory instead of coasting on one already open, and each reload is a retrieval event that raises storage strength. One honest flag on the standing craft law: “higher-order representation” appears in the text exactly once, undefined. Before it ships past the queue as vocabulary, it owes one sentence in your own words — or it becomes jargon on the board.
Four margins, four footholds. Each one drags an abstract mechanism down into something you actually live — a grade, a full-RAM brain, a familiar page, a paper written in the wrong room. Read Bjork through these and the science stops being science and starts being a description of how your own head works.
“doesn’t matter what you know if you are tested w/ RAM full”
Your own idiom for the paper’s core warning: a test score is a snapshot of retrieval conditions, not an inventory of what’s stored. Full RAM at the moment of the test hides everything on the drive. Performance and learning are two different numbers.
“If you get a C or D you still know 70% of content”
Same move, aimed at the gradebook. A grade is a moment-of-retrieval reading, not a knowledge total. The 70% didn’t vanish; the test just caught it under bad conditions. This is the whole storage-versus-retrieval split in one report card.
“Definitely relate to the lazy/familiar brain issue”
You named the exact illusion the paper warns about: a familiar page feels known, so the brain calls it learned and moves on. Fluency masquerading as mastery. Catching it in yourself is the first defense against it.
“Paper writing in library vs @ home”
That’s the classic varied-conditions study, caught with your own example before the paper handed you one. Study the same thing in different places and the memory stops being welded to a single context. You reached for the textbook case on instinct.
the facilitating effect of apparent impediments to learning · going against one’s intuitionsTwo lines you pulled to the surface with your pen. They are the paper’s thesis in the authors’ own words, and the fact that these are the two you underlined is the read: you didn’t highlight the findings, you highlighted the counterintuition — the reason nobody does this on their own. The whole method lives in that gap between what feels productive and what is.
Anyone can summarize a learning-science paper. What makes this a read and not a book report is that you left the last page holding two product specs, not a paraphrase. A paper about retrieval practice made you spec a retrieval tool. That is the deepest form of comprehension there is — theory converted straight into something buildable.
The two on the left are verbatim ink: your final-page margin “E.G. Claude comprehension ck (#!)” is the seed of /quiz-me, and the guided-lesson voice recordings are the second spec. The two on the right are the same principle running through your standing habits — annotation is difficulty you add on purpose so the reading has to go through you instead of past you, and the freight article you walked Justin through is the negative case: SparkNotes strips out the exact difficulty that would have made it stay. One paper, four live applications, all in your own machine.
“E.G. Claude comprehension ck (#!)”
Read closed 07.15.26, 8:51 A.M. (started 07.12.26, 7:55 P.M.) — eight pages, every one annotated, banked to the Finished Collection vault next to Turing 1950. At the Done mark you didn’t write a summary sentence; you wrote a build order. And the loop closed on its own schedule: the comprehension check you specced from the paper ran one day later, on 07.16, and graded you on the paper — the retrieval-practice text generated the retrieval tool that tested whether you retained it. The paper worked on itself. One honest note, standing law: the Done star still has no own-words sentence for why the difficulties help — storage versus retrieval strength, forgetting as the thing that makes re-learning stick. You own every application; the engine under them is the one line this read still owes. Say it into the recorder and it’s finished.