Recall asks someone to generate an answer; recognition asks them to identify a previously encountered item. Neither measures every aspect of memory.
Encoding
The process of representing incoming information.
Retrieval
Accessing information for a present task.
Counterbalancing
Varying condition order so order is not always confounded with the condition.
01 · TRY
If an item cannot be recalled without a cue, must it have been completely lost?
Draw an encoding–retention–retrieval chain. Place the observed test scores at retrieval rather than directly labeling storage capacity. Give two explanations for the cue advantage, then design a counterbalanced comparison and calculate recognition hits and false alarms. Then use the context matrix in the explanation: compare both study rows, distinguish matching from a generally easier study or recall setting, and revise a claim using the original and replication source cards.
Complete the labeled table or trace with the supplied records. Explain each step in words before using the technical term.
Make a provisional claim
Loading saved evidence…
Inspect evidence and test an explanation
The investigation controls appear when your local evidence is ready.
04 · UNDERSTAND
Connect evidence and mechanism
Distinguish a storage claim from a retrieval result
Compare free recall, cues and recognition without treating memory as a perfect recording.
A question to investigate
If an item cannot be recalled without a cue, must it have been completely lost?
Build competing explanations for a failure to recall and test them with an additional retrieval condition.
Mechanism and evidence
Encoding, storage and retrieval are distinct memory processes. Encoding concerns representing information, storage concerns its retention, and retrieval concerns bringing it into use. Declarative memory includes facts and episodes, while procedural memory concerns skills. A failure on one retrieval task is not a direct observation that every trace is absent. Cues and task demands can change access. The linked reading introduces these concepts; the records here are newly invented exercises.
The cue condition yields 12/20 = 60% recall, compared with 6/20 = 30% without cues. The difference is 30 percentage points, not 30% relative improvement. Relative to the original six, the count doubles. Neither calculation establishes why it doubled. The original cue could aid access, but prior testing or order effects could also contribute in this setup.
Repair the study rather than choosing the most attractive story. Split equally difficult item lists and vary which retrieval condition comes first. Specify an assignment procedure and equal study time. Preserve the same scoring rule, and record whether a target appeared during earlier feedback. If answers are supplied before the second test, improved performance could reflect new learning instead of retrieval of the original study episode.
Recognition also needs two kinds of evidence. Its hit rate is 16/20 = 80%, and its false-alarm rate is 8/20 = 40%. Reporting only the 80% would omit the tendency to label new items as old. Do not equate recognition accuracy with a fixed personal ability, and do not infer a clinical condition from any of these teaching numbers.
For a more independent artifact, create a codebook with one row per item and columns for old/new status, cue condition, response and correctness. Explain how to retain missed responses rather than treating them as missing data. Your conclusion should identify the task and population narrowly. A model of stages organizes questions; it does not imply the brain stores exact files or that every memory follows a single straight line.
Inspect the relationships
Proposed explanation
Distinguishing observation to seek
Cue supports retrieval
Advantage persists when order is counterbalanced
Previous test improves performance
Performance changes with prior-test exposure
Answers were relearned
Improvement depends on answer feedback before retest
Your revision: Identify which comparison is absent from record A and qualify the conclusion accordingly.
Worked example
Free recall is 30%; cued recall is 60%. The 30-percentage-point difference is consistent with a retrieval benefit but is confounded with order. Recognition gives 80% hits and 40% false alarms.
Independent application
A new task has 15 hits among 25 old items and 5 false alarms among 10 new items. Which pair of rates is correct? Explain the denominator, process or rule before checking the response. Record a revision after feedback.
A learner recalls an item in one room but not another. This observation does not directly reveal whether a memory was stored, what cue was used, or which other differences mattered. Encoding specificity is the idea that a cue’s usefulness depends on its relation to how the material was encoded. Environmental context-dependent retrieval concerns the external study and retrieval settings. Internal-state dependence is a different comparison; a room-change result does not establish it.
Before forming a claim, separate two factors: where study happened and where recall happened. A comparison that changes both cannot identify which change mattered. The four-cell arrangement lets you hold study context fixed while comparing recall contexts, then repeat that comparison for the other study context.
Which comparison tests matching?
These are three invented sets of cell means, not either published study’s data. Each value is mean words recalled from a 20-word list. Treat the four cells equally for this descriptive exercise; no participant scores, sample sizes or uncertainty estimates are supplied.
Predict whether a high A-to-A score alone proves a matching advantage. Then choose a pattern and compare both study rows.
Rows: study context. Columns: recall context. Units: mean words recalled.
Study context
Recall A
Recall B
Study A
12 — match
8 — mismatch
Study B
7 — mismatch
11 — match
Study A held fixed: match minus mismatch = 12 − 8 = 4 words.
Study B held fixed: match minus mismatch = 11 − 7 = 4 words. The matching column changes with the study row.
Matched mean: 11.5 words. Mismatched mean: 7.5 words. Difference: 4 words. Each study row favors its matching recall context by 4 words. This describes a match pattern, not proof of a storage or retrieval mechanism.
Equal-cell means: study A 10, study B 9; recall A 9.5, recall B 9.5. The matching contrast averages the two within-study differences. It is not a percentage-point effect, an individual score or a significance test.
Record the matrix, both row comparisons and a qualified claim in the Experiment stage’s source notebook. Explain what assignment, order, list difficulty and participant-level evidence you would need before interpreting a real comparison causally. Changing the selector replaces the supplied pattern; it does not simulate learning or change any real study.
The selector resets on reload and is not included in saved captures. Your source annotations preserve the reasoning you choose to record. No personal memory test, timer or background work runs.
Read an original report alongside a replication
Godden and Baddeley (1975) reported better free recall when divers’ learning and recall environments matched. Their abstract also describes a follow-up addressing movement-related disruption. Read the source card as a reported finding under a particular design, not a guarantee that returning to a location restores every memory.
Murre’s 2021 replication reports sixteen divers, spoken lists, a four-minute delay and all four conditions for each diver. It did not find the same-context advantage. The source card supplies this bounded abstract-level comparison, not the full results or an estimated effect size.
Your source investigation: write one sentence accurately summarizing each report, then revise this claim: “The classic experiment proves that matching locations always improves memory.” Cite the original and replication source IDs. Add a table with three columns: possible explanation for the differing findings, evidence needed to examine it, and whether that evidence is supplied here. Possible questions include participant selection, material difficulty, ordering, delay and uncertainty. A plausible difference is a hypothesis to investigate, not an established explanation merely because the studies disagree.
Repeated measures mean the same people contribute to multiple conditions. Their scores are related; do not count four conditions as four independent samples. For an actual analysis, inspect participant-level matching, condition order, missing observations and uncertainty. Four means alone do not let you compute the uncertainty of a within-person contrast. Nor should you assign statistical significance to the invented workbench patterns.
A failure to reproduce an effect limits confidence in a simple universal claim. It is not proof that the original researchers were dishonest or that context never matters. Likewise, a supportive study does not establish the effect for every task or population. Different studies call for comparison of methods and the wider evidence, not a vote between authors.
Transfer: repair the room-memory headline
A separate invented summary gives mean recall out of 20: study in room X then recall in X 9, study X then recall Y 9, study Y then recall X 5, study Y then recall Y 5. A headline says, “Nine in the same room versus five in a different room proves matching rooms helps.” Reconstruct the matrix before opening feedback. Identify which factors the headline’s selected cells change, calculate both within-study matching comparisons, and write a narrower conclusion. Propose a nonclinical research design using supplied records; this activity does not ask you to conduct a study on classmates or recreate the diving procedure.
Check the room-context comparison
Holding study X fixed, match minus mismatch is 9−9=0. Holding study Y fixed, it is 5−5=0. The matched mean is (9+5)/2=7; the mismatched mean is also (9+5)/2=7. Selecting XX9 and YX5 holds recall X fixed but changes study context, so it cannot isolate matching. The supplied pattern favors study X regardless of recall room. It does not prove why: assignment, materials and participant differences remain unspecified. A defensible design varies study and recall context, balances materials and condition order, holds timing and scoring constant, and records participant-level results. It must distinguish a design proposal from completed evidence.
Record the first argument before viewing this response if you want to distinguish it from a guided revision. This disclosure is practice feedback, not an independently scored assessment.
Worked example
Free recall is 30%; cued recall is 60%. The 30-percentage-point difference is consistent with a retrieval benefit but is confounded with order. Recognition gives 80% hits and 40% false alarms.
