SCIENCE
PUBLIC EXPERIMENT / 01
Familiarity can feel like evidence.
A short, harmless test of what happens when a sentence passes through your attention once before you judge it.
INSTRUMENT STATUS
READY
01 / EXPOSE
02 / SETTLE
03 / JUDGE
The condition stays hidden until the ledger opens.
01Before the instrument starts
Read 24 low-stakes statements. Rate how true they feel.
Some statements will appear once before you rate them. Others will be new. You will not be told which is which until the end.
The run takes about three minutes. It contains harmless science and natural-history claims, including some false ones. There are no trick questions and no account to create.
LIVE RUN
Keep your first impression.
EXPOSURE01 / 12
Read each statement once.
STATEMENT 01
Read it once. Nothing to answer yet.
INTERVAL / 08 SECONDS
Let the ink settle.
A short neutral interval separates reading from judgement. The next page will open automatically.
08
STATEMENT 01 / 24
SCIENCE
Use your first impression. Looking things up can wait until the reveal.
02The ledger opens
DEBRIEF / YOUR RUN
Here is what happened in this particular run.
The useful question is whether the repeated statements received different ratings from the novel ones. This number is a description of this run, not a label for you.
REPEATED − NOVEL
+0.00
Average truth rating difference on the six-point scale.
REPEATED MEAN
0.00
12 statements shown during exposure.
NOVEL MEAN
0.00
12 statements shown for the first time.
YOUR RESPONSESSource-backed correction
PUBLIC RESULTS
What happens across runs?
Loading the public ledger…
No aggregate data has been loaded yet.
03Method, limits, sources
READ THE INSTRUMENT
Repetition is one route by which a sentence can become easier to process.
The illusory truth effect is the laboratory finding that repeated statements can receive higher truth ratings than otherwise comparable new statements. Fluency is one proposed route, not a complete explanation. Accuracy prompts, plausibility, knowledge, delay, and context can change the pattern.
This version uses a balanced, harmless pool so the task stays interesting without asking anyone to practise consequential misinformation. The public result is a demonstration with self-selected visitors, not a representative sample or a peer-reviewed human-subjects study.
A
Balanced by design
Every run contains 12 true and 12 false items. Eight matched pairs are mutually exclusive so their altered versions do not appear together.
B
Anonymous by default
The server stores a random session ID, the assignment, ratings, coarse timing fields, and a UTC day for aggregate counts. No identity or IP address is retained.
C
Small result, honest language
The personal number describes one run. Aggregate results show N, central tendency, uncertainty, and exclusions. Empty or low-N states stay empty or caveated.
EVIDENCE BASE
Start with the papers.
Why is Romanian not included yet?
Translation can change familiarity, plausibility, length, and the availability of a fact. Romanian support is planned only when a separate Romanian item set can be calibrated and reported as its own language condition. Validity comes before a language checkbox.