Learning and memory
Separate immediate performance from delayed retention, explain a retrieval-practice comparison, and design a fair transfer test.
Start with: Percentages and experimental comparison. Retrieval means reconstructing an answer without seeing it.
01 · Commit to a prediction
Responses stay in this page only. Reloading or closing may discard them. Nothing is transmitted, saved or synchronized.
02 · Change an assumption
Before changing a control, say what should move and why. Start with the experiments below. Reset restores the starting model; it preserves your written responses.
Keep n = 100. Change only final-test occasion from immediate to delayed. Predict whether the ranking must stay fixed.
The synthetic example switches from restudy 80% versus retrieval 65% to restudy 40% versus retrieval 60%. These invented rates demonstrate a possible crossover.
At the delayed test, set n to 50, then to 200. What changes and what cannot change?
Expected recall rates remain 40% and 60%. Expected counts quadruple and the sampling standard error of the difference halves. More participants do not change the assumed effect.
10 to 200 · step 10
Calculated model output. The table gives the same values. Displayed values are rounded; calculations keep full precision.
03 · Connect the mechanism
Retrieval practice makes the learner reconstruct information; restudy presents it again. Both can contribute to learning. A delayed test asks which knowledge remains accessible after the original context is gone. Roediger and Karpicke’s prose-learning experiments found a reversal between immediate and delayed comparisons. The constructed rates below are a teaching scenario, not their data or a prediction about you. The experiment demonstrates how to evaluate a learning mechanism, not a neural model of memory.
p is the probability of one prespecified binary recall outcome for an independently sampled learner in a group. n is learners per group, 10 to 200 here. pᵣ and pₛ denote retrieval and restudy. Rates are fractions in formulas, percentages in the display. SE is in proportion units, displayed as percentage points. This deliberately simplified independent-groups design is not a reconstruction of the cited study.
In the synthetic delayed condition, n = 100 gives expected successes 60 and 40. The expected difference is 20 percentage points. Under independent Bernoulli outcomes, difference variance is 0.6 × 0.4/100 + 0.4 × 0.6/100 = 0.0048. Its square root is about 0.069282, or 6.9282 percentage points. These are expectations and sampling spread, not observed participants or a fitted confidence interval.
The outcome probabilities are assumed, not estimated. No curve predicts your forgetting rate, ideal schedule or mastery. The same person answering many questions does not create independent learners. A fair intervention comparison controls time, initial knowledge, materials, scoring and final-test conditions; random allocation reduces systematic group differences. A recall advantage does not by itself establish novel transfer or identify the neural mechanism. Desirable difficulty is not a rule that harder is always better.
The annotated sources distinguish established results from this lesson’s original examples.
04 · Follow the structure
Close the explanation, reconstruct it, then check the reasoning and correct omissions.
Boundary: Remembering the wording can coexist with inability to solve a changed problem.
Compare later unaided performance using the actual task and safe practice conditions.
Boundary: Recalling instructions is not evidence of reliable skilled execution under real constraints.
05 · Retrieve without hints
Write an answer before opening its feedback. Later, return directly here without rereading above. Recognition, explanation and transfer are separate outcomes. No page action or answer reveal measures mastery.
Recognition and explanation
It proves access to feedback, not successful retrieval. Evidence requires an attempt whose support level is known, and durable understanding needs later attempts.
Self-check: Separate exposure, supported performance and unaided retention.
Calculation
Retrieval: 30; restudy: 20. Their rates still differ by 20 percentage points. SE is √0.0096 ≈ 9.798 percentage points.
Self-check: Keep counts separate from rates and do not label expectations as observations.
Novel transfer
Randomly allocate comparable learners or counterbalance matched materials, equalize practice time, prespecify delayed novel problems, blind scoring if possible, and report variability. Self-selection, prior knowledge and different practice exposure can explain the original contrast.
Self-check: Name an alternative cause, control it through design and measure transfer rather than copying the practiced questions.
On a later day, try again and record actual evidence in the curriculum. A later unaided explanation and a fresh transfer problem give stronger evidence than immediate familiarity. No reminder is scheduled.
Sources & scope
Original teaching examples. Reference links need a connection; the lesson itself does not. Built 2026-10-11. Learner understanding is not assessed.