Cognitive Psychology
False Memory and Source Monitoring
Three short lists, each on a theme, each item labelled with where it came from. A few minutes later: did you see this word, how sure are you, and which source was it from? Those are three questions, and the answers do not always agree with one another.
Original word lists — not a memory test, and not a claim about anybody's reliability
Learning objective
By the end you should be able to treat recognition, confidence and source as three separate measures, and say what it means when they disagree.
About 25 minutes. Nothing you do here is saved or sent anywhere.
Before you start
Nothing here is a test of you. Whatever the results show, this is a demonstration of how memory works rather than an assessment of yours.
It works properly once. Do the task before reading on. A second theme set is available, and the worked example reaches every conclusion without studying anything.
- Answer the prediction below — it unlocks the study phase.
- Study 24 items, each labelled with the source it came from.
- At test, for each word: say whether you saw it and how sure you are.
- If you say you saw it, say which source it came from.
- Read the three measures separately — they will not tell the same story.
- Prefer not to perform it? Load the worked example: a simulated class dataset with a fixed seed.
First, a prediction
You are about to study three short lists. Each list is on a theme, and every item is labelled with one of two sources: the handout or the whiteboard. At test you will be shown single words and asked three things about each.
Study, then test
Twenty-four items, one at a time, each with its source shown beside it. Then twelve test words. Nothing at test is timed.
Key terms
- Recognition
- Deciding whether an item was in the study phase. It is one of the three measures reported here.
- Confidence
- How sure you were, recorded with each recognition decision and reported separately from whether you were right.
- Source
- Which of the labelled origins an item came from. You are asked for it only about items you say you saw.
- Source monitoring
- Deciding where a remembered item came from, which is a different judgement from deciding whether you have met it before.
What did you see, and where did it come from?
Results are withheld until the test finishes, because seeing them part-way through would change every answer after that.
- 1 Study
- 2 Test
- 3 Results
Status Answer the question above to unlock this.
Answer the question above to unlock the study phase.
Three measures, three answers
The result
Reading your pattern
| Kind of word | Tested | Called "seen" | Average confidence | Source |
|---|
Item by item — what each test word was, and what you said
| Word | What it was | True source | You said | Confidence | Source given |
|---|
Transfer challenge — change the conditions
Each row changes one thing about the study, and asks about one outcome. Read which outcome each question asks about. The same change can move one measure a great deal and leave another almost untouched.
What this demonstrates
Three questions, three different answers
"Did you see it", "how sure are you" and "where did it come from" measure different things. A word can feel strongly familiar and carry no source information at all. A word can be correctly recognised and confidently attributed to the wrong source. The interesting results in this literature are the places where the three come apart, which a single memory score would hide.
Gist is stored, and gist is useful
What survives from a themed list is not only the individual items but the theme they share. A test word that fits the theme therefore matches something real that was encoded. That is why it can feel familiar, and why it is often endorsed with genuine confidence rather than as a guess. Calling this a malfunction gets the design backwards. A system that abstracted nothing would remember lists beautifully and generalise not at all.
Source monitoring is a decision, not a retrieval
Source information is not usually stored as a tag attached to a memory. It is inferred at the moment of remembering, from the qualities of what comes to mind. Those qualities are perceptual detail, whether the memory feels thought rather than seen, and how plausible each candidate source is. That is why source judgements are systematically better when the sources were more distinguishable, and why they degrade faster than recognition does.
What follows for testimony, and what does not
It is a long way from a themed word list to an eyewitness account. The mechanisms invoked here — gist encoding, familiarity without recollection, source inference — do appear in applied work on misinformation and on suggestive questioning, and the applied literature is careful about the size of the leap. The intrusions here are highly constrained, not arbitrary. And most memory, most of the time, is accurate enough to act on.
For teaching elsewhere: take this activity as one self-contained block of HTML, on the clipboard or as a file. Either way it is styled so that it will not disturb the page you put it into.