Reconstructive memory: wording, source and confidence
Distinguish event information from later suggestions and evaluate false reports without assuming all memories are inaccurate.
Question and prediction
Can a question change what is later reported about an event?
Predict why a source label might matter when the same detail appears in a later narrative rather than the original scene.
Mechanism and evidence
Memory is reconstructive: retrieval draws on stored information, knowledge and cues rather than replaying an exact recording. Schemas organize expectations and can support comprehension while also shaping errors. Source monitoring concerns judging where information came from—seeing it, hearing it, imagining it or reading about it later. Familiarity with a detail does not establish its source.
Loftus and Palmer’s 1974 experiments varied wording after participants watched vehicle-collision films. Wording affected speed estimates; in a later test, one wording condition increased reports of broken glass that was not shown. This supports examining post-event information and later reporting. It does not show that every witness is wrong, that any single remembered event is false, or that the experiment uniquely settles every underlying mechanism.
In the fictional harmless scene, an original display contains a blue folder. A later account mentions a red folder. On a final test, some learners select red and identify the later account as their source; others select red and claim they saw it. Both are wrong about the scene, but the source judgments distinguish reliance on a later report from source confusion. Keep accuracy and source attribution separate.
Confidence and accuracy are related under some conditions but are not interchangeable measures. A study must define the timing and conditions of each confidence report. To investigate reconstruction, use a known-ground-truth neutral scene and compare suggestion with no-suggestion conditions. This task uses supplied records; it does not solicit traumatic memories, challenge personal experiences or attempt to implant memories.
Inspect the relationships
| Response | Claimed source | What it establishes in the toy case |
|---|
| Red | Later narrative | Incorrect scene answer with source identified |
| Red | Original scene | Incorrect answer plus source misattribution |
| Blue | Original scene | Correct content/source combination |
Worked example
Red responses occur in 8/20=40% versus 2/20=10%, a 30-point difference. Within the suggestion group’s eight red responses, five misattribute the source to the scene. The error rate and source-misattribution count answer different questions.
Your investigation
Make an event/narrative/test timeline and a source-attribution table. Compare the toy mechanism with the published wording study. Write a 200-word interview-protocol critique proposing neutral questions and distinguishing inaccurate content, source confusion and intentional deception.
Unfamiliar transfer
A person gives an inaccurate detail confidently. What cannot be concluded from that alone? Explain your reasoning before checking feedback.
Reading
Loftus and Palmer (1974): Reconstruction of automobile destruction. The records below are original fictional teaching cases, not research results from this source.
Fictional source sequence Original neutral scene: blue folder. Later narrative: red folder. Of 20 records, 8 select red; 5 of those say they saw it in the scene, while 3 identify the later narrative as the source.
Fictional comparison In a separate comparable no-suggestion group, 2 of 20 select red. Assignment is random in the toy design; source confidence and attention are not measured.
Authentic evidence and its limits
Published broken-glass responses — Elizabeth F. Loftus and John C. Palmer (1974). Experiment 2, Table 2 reports 16 of 50 affirmative broken-glass responses after the ‘smashed’ wording, 7 of 50 after ‘hit’, and 6 of 50 in the control condition. No broken glass appeared in the film.
Attributed original counts. The response occurred one week later. These are not the fictional folder data and do not establish intentional lying or settle every mechanism of memory/reporting change.
Read the original source. Original LearnAI paraphrase and attributed factual results; no reproduced figure or substantial verbatim passage. Original publication retains its rights.
Original-data task: Calculate affirmative-report rates for the three published conditions. They are 32%, 14% and 12%. The smashed-versus-hit difference is 18 percentage points. Explain why these counts do not by themselves establish deliberate deception.
