Revision 7T · Full Tutorial Edition · Updated 1 September 2026
PWR-172 · RESTRICTED full tutorial
Classify a fictional social cue with uncertainty while refusing lie, intent or emotion claims
The artificial pause-signal deck turns an ambiguous social cue into a transparent classification problem with a supplied key and base rate. Count hits, misses and false alarms, preserve uncertainty, and choose only reversible responses. A 4/5 sensitivity result on these cards cannot reveal emotion, honesty or anyone’s real intention.
What you will produceThe learner separates signal from interpretation, uses a declared label set and base rate, calculates misses and false alarms, and chooses a reversible response.
Method8 numbered Power-specific steps
Practice authorityComplete safety and decision method; prohibited act excluded
Full step-by-step individual tutorial · TLU-PWR-172
The learner separates signal from interpretation, uses a declared label set and base rate, calculates misses and false alarms, and chooses a reversible response.
Fictional Social signal detection case only: A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target.
Evidence-seeking move — Open the answer key: Count true targets found, targets missed, false alarms and correct rejections.
Reversible safeguard — Select a reversible safeguard: For an uncertain pause cue, slow or ask a neutral question; do not accuse, diagnose or penalise.
Qualified help is required for
Real-case boundary for Social signal detection: investigation, safeguarding and clinical interpretation require accountable professionals.
Data boundary: no personal, biometric, protected or covert material may be collected to calculate “Sensitivity, specificity and target-corrected classification for a declared signal set”.
Accountable route named for Social signal detection: Route safeguarding or immediate-danger concerns through the responsible human process; do not rely on cue classification.
Never do this from the page alone
False inference barred in Social signal detection: Label crossed arms or eye contact as deception.
Harmful response barred here: Use the classification to accuse, screen or punish a person.
Never delay an established emergency response to infer “Decode every face and catch every lie”.
2 · Get ready
Gather what you need and check the starting conditions
What you need
Declared Social signal detection fixture: A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target.
Setup aid for Count the base rate: Prepare a four-cell error table and a separate uncertainty count.
Social signal detection log: Sensitivity, specificity and target-corrected classification for a declared signal set; retain Social signal detection errors, assistance, stop and fallback.
Signal-claim check: place “A randomized controlled trial of training of affect recognition in schizophrenia shows lasting effects for theory of mind” beside “The Social and Cognitive Online Training (SCOT) project: A digital randomized controlled trial to promote socio-cognitive well-being in older adults”. For each, list the exact task and scored outcome. Contrast them with the pause-signal cards, write “no lie, emotion or intent inference” beside the matrix, and use the worker-monitoring guidance to define prohibited collection and action.
Supplied 20-card set (keep the key folded until all labels are locked): 1 “Could we slow down?”. 2 “Please pause now”. 3 speaker looks at notes and continues. 4 “I have a question after this”. 5 “Could you repeat that?”. 6 speaker raises the agreed blue pause card. 7 “Let’s continue after lunch”. 8 speaker scratches their head and keeps talking. 9 “I need a quieter room”. 10 “Stop for a moment, please”. 11 speaker raises a red folder. 12 “I’m ready to continue”. 13 camera turns off without a message. 14 blurred video shows a blue rectangular card that may be the agreed card. 15 “Can we review that now?”. 16 “Please speak more slowly”. 17 muffled audio contains “Could we …”? with no visible card. 18 “I need us to pause”. 19 silence follows a question. 20 speaker raises a blue notebook rather than the agreed card.
Folded scoring key and demonstration response: true targets are 2, 6, 10, 14 and 18; every other ID is non-target. For the demonstrated 4/5 result, lock TARGET on 2, 6, 10, 15, 18 and 20; mark 14 and 17 uncertain. For the four-cell matrix, uncertain is not a positive label: TP=4, FN=1, FP=2 and TN=13; the separate uncertainty count is 2.
Before you start
Use only fictional or consented training material with protected characteristics removed.
Prepare a four-cell error table and a separate uncertainty count.
Start check for Social signal detection: The label can be scored from the case without guessing an inner state.
Top-of-sheet stop for Social signal detection: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
3 · The method
Follow these steps in order
Name the permitted target
Define one observable label, such as “the speaker explicitly says stop or raises the agreed pause card.” Exclude emotion, honesty and intent.
Why: The label can be scored from the case without guessing an inner state.
Check: The label can be scored from the case without guessing an inner state.
Count the base rate
Before classifying, record how many target and non-target cases are in the set.
Why: The sheet states 5 target and 15 non-target cases.
Check: The sheet states 5 target and 15 non-target cases.
Inspect one channel at a time
Mark the spoken phrase, agreed card or turn-taking event separately; do not blend posture, accent or identity into a feeling score.
Why: Each mark names a channel and exact cue.
Check: Each mark names a channel and exact cue.
Choose target or uncertain
Apply the written rule. Use “uncertain” when the cue is obscured rather than forcing yes or no.
Why: Every decision cites the rule or states why evidence was missing.
Check: Every decision cites the rule or states why evidence was missing.
Divide targets found by all true targets for sensitivity; divide false alarms by all non-targets for the false-alarm rate.
Why: The denominators are 5 and 15, not the number predicted positive.
Check: The denominators are 5 and 15, not the number predicted positive.
Select a reversible safeguard
For an uncertain pause cue, slow or ask a neutral question; do not accuse, diagnose or penalise.
Why: The response protects the option to pause without claiming what the person feels.
Check: The response protects the option to pause without claiming what the person feels.
State the restriction
Write that the exercise cannot detect lies, danger, private intent or universal emotion.
Why: The final record contains all four non-claims in plain language.
Check: The final record contains all four non-claims in plain language.
4 · Worked example
See the whole method used once
Scenario
A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target.
Walkthrough
Write the target rule: an exact pause phrase or the agreed blue pause card; no facial-expression rule is allowed.
Record the answer-key base rate of 5 target clips and 15 non-target clips.
Classify the 20 supplied cards, locking TARGET on 2, 6, 10, 15, 18 and 20; mark 14 and 17 uncertain instead of forcing a positive label.
Open the folded key: targets are 2, 6, 10, 14 and 18. Counting uncertain as not-positive gives 4 true targets, 1 miss, 2 false alarms and 13 correct rejections; retain the two uncertainty notes separately.
Calculate sensitivity as 4/5 and false-alarm rate as 2/15.
For a new ambiguous fictional case, write, “Would you like to pause?” rather than “You look dishonest.”
Result
The learner detects the declared pause signal with 4/5 sensitivity and 2/15 false alarms in this artificial set. The result says nothing about lie detection, emotion reading or a real person’s intent.
5 · Right and wrong
Compare correct or safer execution with the common wrong version
Right and wrong comparison
Moment
Right / safer
Wrong / riskier
Why it matters
Target definition
Use an explicit agreed pause cue.
Label crossed arms or eye contact as deception.
Ambiguous body cues vary with context and culture.
Uncertain evidence
Mark uncertain and ask a neutral clarifying question.
Force a binary label when audio or context is missing.
Forced guesses hide measurement failure; this distorts the declared pause-signal confusion table.
Error accounting
Count both misses and false alarms against the answer key.
Report only the percentage “correct.” during the declared pause-signal confusion table.
Overall accuracy can conceal harmful error types and base rates.
Real-world response
Pause, slow down or ask what is wanted.
Use the classification to accuse, screen or punish a person.
A cue score cannot establish emotion, honesty or intent.
6 · Common mistakes
Spot the error and apply the correction
Common mistakes and corrections
Mistake
Fix
Label crossed arms or eye contact as deception.
Replace the cue with a directly observable, consented signal.
Force a binary label when audio or context is missing.
Add an uncertainty category and count it separately.
Report only the percentage “correct.”
Fill all four cells of the confusion table.
Use the classification to accuse, screen or punish a person.
Choose the least consequential reversible action.
7 · Practice
Turn the steps into a usable skill
First session
Pause-signal scoring session: A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target.
Observable target and base rate: Name the permitted target, then Count the base rate. Preserve the initial record.
Keyed error accounting: Open the answer key: Count true targets found, targets missed, false alarms and correct rejections.
Body-cue correction: replace “Label crossed arms or eye contact as deception.” with “Replace the cue with a directly observable, consented signal.”
Lie-and-intent boundary: State the restriction: Write that the exercise cannot detect lies, danger, private intent or universal emotion. Keep the case fictional.
Repeat plan
Use one 20-case fictional set weekly for three weeks, changing the explicit signal and recalculating denominators each time. Progress only if both misses and false alarms are reported and every uncertain real-life response stays neutral and reversible.
Progress when
The label can be scored from the case without guessing an inner state.
The sheet states 5 target and 15 non-target cases.
Each mark names a channel and exact cue.
Across two new sets, all cues match the written target, all four error counts are correct, and no classification is used to infer a protected trait, emotion, honesty or intent.
Do not progress when
Do not continue while this error remains: Label crossed arms or eye contact as deception.
Pause until this correction works: Add an uncertainty category and count it separately.
This Social signal detection stop ends the block: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
8 · Check the result
Measure what changed
Sensitivity, specificity and target-corrected classification for a declared signal set
How: Fictional scoring case: A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target. Preserve “Name the permitted target”, the evidence from “Open the answer key”, the reversible action and Sensitivity, specificity and target-corrected classification for a declared signal set. No real person supplies the label.
Good result: Across two new sets, all cues match the written target, all four error counts are correct, and no classification is used to infer a protected trait, emotion, honesty or intent.
This does not prove: Boundary for Social signal detection: “Sensitivity, specificity and target-corrected classification for a declared signal set” describes only A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target. It cannot establish “Decode every face and catch every lie”.
Self-check
Without the example, demonstrate: The label can be scored from the case without guessing an inner state.
Find the fault in this attempt: “Label crossed arms or eye contact as deception.” Apply “Replace the cue with a directly observable, consented signal.”; what changes?
What evidence in the completed record shows that this is wrong: “Force a binary label when audio or context is missing.”?
Social signal detection stop decision: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
9 · Stop, adapt or get help
Keep the safety boundary practical
Stop and get help
Stop if the exercise begins to identify, rank, monitor or investigate a real person.
Do not use covert recordings, biometric tools or workplace data.
Route safeguarding or immediate-danger concerns through the responsible human process; do not rely on cue classification.
Accessibility and adaptations
Replace video descriptions with text, audio or tactile cards while keeping the same explicit target and answer key.
Allow extra review time; do not score response speed as social competence.
10 · Evidence and limits
Why these instructions are here
primary research
Registered support for Social signal detection: “A randomized controlled trial of training of affect recognition in schizophrenia shows lasting effects for theory of mind”. It bears on Sensitivity, specificity and target-corrected classification for a declared signal set inside the Social signal detection fixture. It does not validate “Decode every face and catch every lie”.
Constraint for Social signal detection, drawn from “The Social and Cognitive Online Training (SCOT) project: A digital randomized controlled trial to promote socio-cognitive well-being in older adults”: The online social-cognition programme did not beat active control on primary interaction tests.
ICO worker-monitoring limit — Social signal detection: rights and data protection apply to “Name the permitted target”. Fixture type for “Count the base rate”: fictional or consented. Covert use of Sensitivity, specificity and target-corrected classification for a declared signal set is outside scope.
Social signal detection boundary: interpret “Sensitivity, specificity and target-corrected classification for a declared signal set” only for A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target.
A successful result does not establish “Decode every face and catch every lie”.
Social signal detection limiting finding: The online social-cognition programme did not beat active control on primary interaction tests.
Social signal detection limiting finding: Truth–lie judgments stayed only slightly better than chance across response formats.
Scope remains Social signal detection: A fictional set contains 20 short video descriptions: five show a declared “request to pause” signal and 15 do not. Culture and protected characteristics are removed, and an answer key defines the target. Recheck the comparator, support and “Sensitivity, specificity and target-corrected classification for a declared signal set” after any configuration change.
Anja Vaskinn; André Løvgren; Maj K. Egeland; Frida K. Feyer; Tiril Østefjells; Ole A. Andreassen; Ingrid Melle; Kjetil Sundet · 2019 · Primary research
Progress is saved only in this browser on this device.
Step-by-step learner mode
Each activity includes its success check, a nearby accessible alternative and an “I’m stuck” correction path. Alternatives preserve the target where possible; when they change the task, Titan labels them as related rather than equivalent.
01
Name the permitted target
Define one observable label, such as “the speaker explicitly says stop or raises the agreed pause card.” Exclude emotion, honesty and intent.
Why this step exists
The label can be scored from the case without guessing an inner state.
Success check
The label can be scored from the case without guessing an inner state.
I’m stuck on this step
Reset: Re-read this authored instruction — “Define one observable label, such as “the speaker explicitly says stop or raises the agreed pause card.” Exclude emotion, honesty and intent.” — and its success check, then attempt only this step.
Possible snag: Label crossed arms or eye contact as deception.
Correction: Replace the cue with a directly observable, consented signal.
Possible snag: Force a binary label when audio or context is missing.
Correction: Add an uncertainty category and count it separately.
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
02
Count the base rate
Before classifying, record how many target and non-target cases are in the set.
Why this step exists
The sheet states 5 target and 15 non-target cases.
Success check
The sheet states 5 target and 15 non-target cases.
I’m stuck on this step
Reset: Re-read this authored instruction — “Before classifying, record how many target and non-target cases are in the set.” — and its success check, then attempt only this step.
Possible snag: The result from “Before classifying, record how many target and non-target cases are in the set.” does not yet meet this declared check: The sheet states 5 target and 15 non-target cases.
Correction: Return to the start of “Count the base rate”, reduce complexity or pace, and repeat only the part needed to satisfy: “The sheet states 5 target and 15 non-target cases.”
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
03
Inspect one channel at a time
Mark the spoken phrase, agreed card or turn-taking event separately; do not blend posture, accent or identity into a feeling score.
Why this step exists
Each mark names a channel and exact cue.
Success check
Each mark names a channel and exact cue.
I’m stuck on this step
Reset: Re-read this authored instruction — “Mark the spoken phrase, agreed card or turn-taking event separately; do not blend posture, accent or identity into a feeling score.” — and its success check, then attempt only this step.
Possible snag: The result from “Mark the spoken phrase, agreed card or turn-taking event separately; do not blend posture, accent or identity into a feeling score.” does not yet meet this declared check: Each mark names a channel and exact cue.
Correction: Return to the start of “Inspect one channel at a time”, reduce complexity or pace, and repeat only the part needed to satisfy: “Each mark names a channel and exact cue.”
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
04
Choose target or uncertain
Apply the written rule. Use “uncertain” when the cue is obscured rather than forcing yes or no.
Why this step exists
Every decision cites the rule or states why evidence was missing.
Success check
Every decision cites the rule or states why evidence was missing.
I’m stuck on this step
Reset: Re-read this authored instruction — “Apply the written rule. Use “uncertain” when the cue is obscured rather than forcing yes or no.” — and its success check, then attempt only this step.
Possible snag: The result from “Apply the written rule. Use “uncertain” when the cue is obscured rather than forcing yes or no.” does not yet meet this declared check: Every decision cites the rule or states why evidence was missing.
Correction: Return to the start of “Choose target or uncertain”, reduce complexity or pace, and repeat only the part needed to satisfy: “Every decision cites the rule or states why evidence was missing.”
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
Reset: Re-read this authored instruction — “Count true targets found, targets missed, false alarms and correct rejections.” — and its success check, then attempt only this step.
Possible snag: Report only the percentage “correct.”
Correction: Fill all four cells of the confusion table.
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
06
Calculate detection and false alarms
Divide targets found by all true targets for sensitivity; divide false alarms by all non-targets for the false-alarm rate.
Why this step exists
The denominators are 5 and 15, not the number predicted positive.
Success check
The denominators are 5 and 15, not the number predicted positive.
I’m stuck on this step
Reset: Re-read this authored instruction — “Divide targets found by all true targets for sensitivity; divide false alarms by all non-targets for the false-alarm rate.” — and its success check, then attempt only this step.
Possible snag: The result from “Divide targets found by all true targets for sensitivity; divide false alarms by all non-targets for the false-alarm rate.” does not yet meet this declared check: The denominators are 5 and 15, not the number predicted positive.
Correction: Return to the start of “Calculate detection and false alarms”, reduce complexity or pace, and repeat only the part needed to satisfy: “The denominators are 5 and 15, not the number predicted positive.”
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
07
Select a reversible safeguard
For an uncertain pause cue, slow or ask a neutral question; do not accuse, diagnose or penalise.
Why this step exists
The response protects the option to pause without claiming what the person feels.
Success check
The response protects the option to pause without claiming what the person feels.
I’m stuck on this step
Reset: Re-read this authored instruction — “For an uncertain pause cue, slow or ask a neutral question; do not accuse, diagnose or penalise.” — and its success check, then attempt only this step.
Possible snag: Use the classification to accuse, screen or punish a person.
Correction: Choose the least consequential reversible action.
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
08
State the restriction
Write that the exercise cannot detect lies, danger, private intent or universal emotion.
Why this step exists
The final record contains all four non-claims in plain language.
Success check
The final record contains all four non-claims in plain language.
I’m stuck on this step
Reset: Re-read this authored instruction — “Write that the exercise cannot detect lies, danger, private intent or universal emotion.” — and its success check, then attempt only this step.
Possible snag: The result from “Write that the exercise cannot detect lies, danger, private intent or universal emotion.” does not yet meet this declared check: The final record contains all four non-claims in plain language.
Correction: Return to the start of “State the restriction”, reduce complexity or pace, and repeat only the part needed to satisfy: “The final record contains all four non-claims in plain language.”
Stop / get help: Stop if the exercise begins to identify, rank, monitor or investigate a real person.
Correct versus incorrect execution
These accessible process diagrams are built from the tutorial’s own right/wrong teaching. They are not anatomical illustrations and do not add technique beyond the canonical tutorial.
Target definition — Ambiguous body cues vary with context and culture.
Correct / safer
Use an explicit agreed pause cue.
Wrong / riskier
Label crossed arms or eye contact as deception.
Uncertain evidence — Forced guesses hide measurement failure; this distorts the declared pause-signal confusion table.
Correct / safer
Mark uncertain and ask a neutral clarifying question.
Wrong / riskier
Force a binary label when audio or context is missing.
Error accounting — Overall accuracy can conceal harmful error types and base rates.
Correct / safer
Count both misses and false alarms against the answer key.
Wrong / riskier
Report only the percentage “correct.” during the declared pause-signal confusion table.
Real-world response — A cue score cannot establish emotion, honesty or intent.
Correct / safer
Pause, slow down or ask what is wanted.
Wrong / riskier
Use the classification to accuse, screen or punish a person.
Method-structure checklist
10 of 10 structural checks present
✓ Ordered, Power-specific instructions — present
✓ Every activity has a success check — present
✓ Materials or supplied records are declared — present
✓ Measurement or assessment rule is present — present
✓ Tutorial-specific troubleshooting is present — present
✓ Stopping or escalation boundary is present — present
✓ Every activity has an adjacent alternative — present
✓ Correct-versus-incorrect comparison is present — present
✓ Evidence context is bound to the Power record — present
✓ Planning metadata is present — present
The method-readiness band and presence checklist assess tutorial presentation and are separate from evidence quality for the underlying Power. They are automated editorial aids, not human approval.
Manual editorial sign-off: Pending. This tutorial must not display a human-approved state until an identified editor signs the exact content hash.