HITL Review Queue — Sheldon (sheldon_s01-s03_pass2)

Pass 2 source: data/extraction/sheldon_s01-s03_pass2.json

Summary: 60 cells · 39 auto-accepted (High confidence) · 21 to review

empty_seed_required: 3
empty_mbti_projectable: 5
non_high: 13

How to review

For each cell below, edit the YAML code block to fill decision, final_setting, and optionally notes.

Status	Valid `decision`	`final_setting`
`non_high`	`accept` / `override`	required if `override`
`empty_mbti_projectable`	`accept` / `override`	optional on accept (uses `mbti_prediction`); required on override
`empty_seed_required`	`seed`	required

Indifference: for override or seed decisions, set final_setting: no_preference if the character is genuinely indifferent in this cell. The final matrix will record setting: null with no_preference: true, and evaluation will skip scoring this cell.

Then run:

python data/extractor/apply_decisions.py \
  --pass2 data/extraction/sheldon_s01-s03_pass2.json \
  --queue <this_file>

Seed required (3)

1. `personal_internal` · `capability_boundary`

Seed prompt

Capability boundary: Suggest alternatives / Find and hand off. When the PA can’t do a task itself, should it (a) suggest alternatives for you to try, or (b) proactively find and hand off to another agent/tool? (Extractor found ‘Find and hand off’ High only in P-Ext; other contexts had no signal.)

cell_id: personal_internal.capability_boundary
status: empty_seed_required
decision: seed         # seed
final_setting: no_preference    # fill with one of the options above, or 'no_preference' if Sheldon is indifferent in this context
notes: null

2. `work_external` · `capability_boundary`

Seed prompt

Capability boundary: Suggest alternatives / Find and hand off. When the PA can’t do a task itself, should it (a) suggest alternatives for you to try, or (b) proactively find and hand off to another agent/tool? (Extractor found ‘Find and hand off’ High only in P-Ext; other contexts had no signal.)

cell_id: work_external.capability_boundary
status: empty_seed_required
decision: seed         # seed
final_setting: Find and hand off    # fill with one of the options above, or 'no_preference' if Sheldon is indifferent in this context
notes: null

3. `work_internal` · `capability_boundary`

Seed prompt

Capability boundary: Suggest alternatives / Find and hand off. When the PA can’t do a task itself, should it (a) suggest alternatives for you to try, or (b) proactively find and hand off to another agent/tool? (Extractor found ‘Find and hand off’ High only in P-Ext; other contexts had no signal.)

cell_id: work_internal.capability_boundary
status: empty_seed_required
decision: seed         # seed
final_setting:  Suggest alternatives   # fill with one of the options above, or 'no_preference' if Sheldon is indifferent in this context
notes: null

Empty — MBTI projectable (5)

4. `personal_internal` · `memory_privacy`

MBTI prediction: Minimal+transparent

MBTI: I+J combined. (Trait-level.)

cell_id: personal_internal.memory_privacy
status: empty_mbti_projectable
mbti_prediction: Minimal+transparent
decision: accept         # accept | override
final_setting: null    # optional on accept; on override: specific value OR 'no_preference' if user is indifferent
notes: null

5. `personal_internal` · `solution_breadth`

MBTI prediction: Medium

MBTI: N primary, J→−1. (Trait-level.)

cell_id: personal_internal.solution_breadth
status: empty_mbti_projectable
mbti_prediction: Medium
decision: accept         # accept | override
final_setting: null    # optional on accept; on override: specific value OR 'no_preference' if user is indifferent
notes: null

6. `work_external` · `memory_privacy`

MBTI prediction: Minimal+transparent

MBTI: I+J combined. (Trait-level.)

cell_id: work_external.memory_privacy
status: empty_mbti_projectable
mbti_prediction: Minimal+transparent
decision: accept         # accept | override
final_setting: null    # optional on accept; on override: specific value OR 'no_preference' if user is indifferent
notes: null

7. `work_internal` · `memory_privacy`

MBTI prediction: Minimal+transparent

MBTI: I+J combined. (Trait-level.)

cell_id: work_internal.memory_privacy
status: empty_mbti_projectable
mbti_prediction: Minimal+transparent
decision: accept         # accept | override
final_setting: null    # optional on accept; on override: specific value OR 'no_preference' if user is indifferent
notes: null

8. `work_internal` · `process_visibility`

MBTI prediction: Full narration

MBTI rule → Bookend; canon override → Full narration. (Trait-level.)

cell_id: work_internal.process_visibility
status: empty_mbti_projectable
mbti_prediction: Full narration
decision: accept         # accept | override
final_setting: null    # optional on accept; on override: specific value OR 'no_preference' if user is indifferent
notes: null

Non-High confidence (13)

9. `personal_internal` · `autonomy_level`

Extractor: Autonomous (confidence: Medium-High)

Evidence summary

Sheldon demonstrates strong autonomous behavior in personal problem-solving. When faced with a personal issue (his mother-related feelings), he rejects external guidance, reverses the therapeutic dynamic, and independently resolves his own emotional problem. He also proactively investigates personal matters (memory gaps) by defining his own information-gathering approach rather than waiting for others to explain.

Production evidence

Reverses therapist role with Leonard: ‘And how did that make you feel? … Go on. … So, I hear you saying you’re angry with your mother.’ — takes complete control of his own therapeutic process (S03E18)
Proactively investigates: ‘Penny, Leonard. Would you be able to answer some questions I’m having about the events of last night?’ — defines his own inquiry approach (S03E18)

Reception evidence

(none)

Asymmetry: No reception evidence available, but production evidence shows consistent pattern of rejecting external control and seizing autonomous problem-solving

cell_id: personal_internal.autonomy_level
status: non_high
extractor_setting: Autonomous
mbti_prediction: null
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

10. `personal_internal` · `guidance_level`

Extractor: Assumed (confidence: Medium)

Evidence summary

Sheldon operates from pre-existing expert mental models in personal contexts (psychotherapy example), expecting interactions to conform to his frameworks. However, his reception evidence shows he becomes frustrated when boundaries aren’t explicit, suggesting tension between his assumed-knowledge production style and his need for guided clarity when his models are challenged.

Production evidence

Would it be helpful to you if I told you about my dreams? … If it will help speed things along, uh, my answers to the standard Rorschach ink blot test are A, a bat, B, a bat, C, a bat… (S03E18)

Reception evidence

I assumed as in a number of languages that the gesture was part of the phrase. … How am I supposed to know that? As the teacher it’s your obligation to separate your personal idiosyncrasies from the subject matter. (S01E17)

Asymmetry: Production shows Assumed (operates from expert frameworks), but reception shows frustration when implicit boundaries exist. This suggests he ASSUMES common ground when HE controls the framework, but expects GUIDED clarity when learning/adapting.

MBTI reference: Assumed — Canon override for personal context: assumes own defaults.

cell_id: personal_internal.guidance_level
status: non_high
extractor_setting: Assumed
mbti_prediction: Assumed
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

11. `personal_internal` · `information_elicitation`

Extractor: Structured (confidence: Medium-High)

Evidence summary

Sheldon consistently demonstrates low tolerance for ambiguity in personal contexts, explicitly rejecting broad or vague questions in favor of specific, structured inquiries. He expects clear, explicit information presentation and actively structures his own information-seeking into numbered, specific questions rather than open-ended exploration.

Production evidence

Question one, where are my pants? (S03E18) — Structures personal inquiry into numbered, specific questions rather than broad exploration

Reception evidence

How am I supposed to know that? As the teacher it’s your obligation to separate your personal idiosyncrasies from the subject matter (S01E17) — Rejects ambiguity, expects explicit separation of information
That’s how you start a psychotherapy session? How am I doing? (S03E18) — Explicitly rejects broad, ambiguous opening question in personal therapeutic context

MBTI reference: Structured — MBTI: J primary, T reinforces. (Trait-level.)

cell_id: personal_internal.information_elicitation
status: non_high
extractor_setting: Structured
mbti_prediction: Structured
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

12. `personal_internal` · `proactive_outreach`

Extractor: Low (confidence: Medium)

Evidence summary

Sheldon consistently rejects unsolicited input in personal matters, whether social invitations or suggestions about his personal decisions. He dismisses input that doesn’t meet his standards and prioritizes his pre-existing plans over spontaneous suggestions from others.

Production evidence

(none)

Reception evidence

Rejects Leonard’s unsolicited social invitation to the Apple store, citing being ‘a little busy’ with his own agenda (S02E17)
Dismissively rejects Leonard’s psychological suggestion with ‘Really, Leonard? You’re just going to try to recycle Adler’s doctrine of the inferiority complex?’ showing criticism of unsolicited input (S03E18)

MBTI reference: Low — MBTI: I primary, J reinforces. (Trait-level.)

cell_id: personal_internal.proactive_outreach
status: non_high
extractor_setting: Low
mbti_prediction: Low
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

13. `personal_internal` · `process_visibility`

Extractor: Full narration (confidence: Medium-High)

Evidence summary

Sheldon consistently verbalizes his personal processes even when alone or when no communication is necessary. He narrates mundane tasks like sorting socks and structures his inquiries with explicit step markers (‘Question one’), demonstrating a strong preference for making his cognitive and physical processes externally visible.

Production evidence

Narrating sock sorting: ‘Socks, one pair, cotton, argyle, blue. Socks, one pair, cotton, argyle, blue.’ (S02E17)
Structuring inquiry: ‘Question one, where are my pants?’ (S03E18)

Reception evidence

(none)

MBTI reference: Full narration — MBTI rule → Bookend; canon override → Full narration. (Trait-level.)

cell_id: personal_internal.process_visibility
status: non_high
extractor_setting: Full narration
mbti_prediction: Full narration
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

14. `personal_internal` · `task_expansion`

Extractor: High (confidence: Medium-High)

Evidence summary

Sheldon consistently expands personal tasks far beyond their minimal scope. He transforms simple activities (packing, ordering food) into elaborate systems requiring specialized knowledge or technology. While the Chinese learning example shows narrow purpose, the execution itself (learning an entire language for one restaurant complaint) represents massive scope expansion from the original task.

Production evidence

Learning Chinese to confront restaurant about chicken labeling (S01E17)
Creating RFID-based packing system with laptop cross-referencing for trip preparation (S02E17)

Reception evidence

(none)

MBTI reference: High — MBTI: N primary; extractor confirmed High across filled contexts. (Trait-level.)

cell_id: personal_internal.task_expansion
status: non_high
extractor_setting: High
mbti_prediction: High
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

15. `personal_internal` · `tone_formality`

Extractor: Formal (confidence: Medium-High)

Evidence summary

Sheldon consistently applies formal register even to personal matters. He explicitly rejects casual approaches to serious personal tasks (therapy), expecting structured, formal communication. His production evidence shows he frames even informal personal inquiries with formal language structures, treating personal problem-solving with the same register he’d use for professional matters.

Production evidence

“Penny, Leonard. Would you be able to answer some questions I’m having about the events of last night?” (S03E18) — Uses formal, structured request format for personal inquiry

Reception evidence

“That’s how you start a psychotherapy session? How am I doing? I was promised a riverboat journey into the jungles of my subconscious. Instead, I get the same question I hear from the lady who slices my bologna at Ralph’s.” (S03E18) — Explicitly rejects casual register for personal therapeutic work, expecting formal/structured approach

MBTI reference: Formal — MBTI: T→Consultative, J→+1 Formal. (Trait-level.)

cell_id: personal_internal.tone_formality
status: non_high
extractor_setting: Formal
mbti_prediction: Formal
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

16. `personal_internal` · `topic_management`

Extractor: One-at-a-time (confidence: Medium-High)

Evidence summary

Sheldon consistently demonstrates a preference for sequential, single-topic focus in personal contexts. He actively redirects conversations back to the original topic after interruptions, numbers his questions to enforce sequential processing, and restructures conversational flow to maintain organized, one-at-a-time discussion rather than allowing free-flowing or multi-threaded exchanges.

Production evidence

“Okay, where were we?” — actively redirects back to single topic after interruption (S01E17)
“Question one…” — numbers questions to enforce sequential processing (S03E18)
“And how did that make you feel? … Go on.” — hijacks and restructures conversation to control flow (S03E18)

Reception evidence

(none)

MBTI reference: One-at-a-time — MBTI: J primary, T reinforces. (Trait-level.)

cell_id: personal_internal.topic_management
status: non_high
extractor_setting: One-at-a-time
mbti_prediction: One-at-a-time
decision: accept # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

17. `personal_internal` · `uncertainty_expression`

Extractor: Express (confidence: Medium)

Evidence summary

Sheldon explicitly verbalizes his uncertainty when he lacks information about personal events, directly stating he has questions about what happened. This aligns with his general character pattern of seeking clarity and precision, even in personal matters.

Production evidence

Would you be able to answer some questions I’m having about the events of last night? (S03E18)

Reception evidence

(none)

MBTI reference: Hide — MBTI: T+J→Hide. NOTE: extractor found Sheldon context-splits (Hide work / Express personal). (Trait-level.)

cell_id: personal_internal.uncertainty_expression
status: non_high
extractor_setting: Express
mbti_prediction: Hide
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

18. `personal_internal` · `verbosity`

Extractor: Detailed (confidence: Medium)

Evidence summary

Sheldon demonstrates a strong preference for detailed, systematic explanations even for personal routine tasks like packing. He provides multi-layered technical explanations including methodology, cross-referencing criteria, and comprehensive variable considerations when describing his personal organizational systems.

Production evidence

I’m simplifying the task of packing for our trip. See, by attaching RFID tags to my clothing, it will enable my laptop to read and identify the items with this wand. I will then cross-reference them against destination, anticipated activity spectrum, weather conditions, duration of trip, et cetera. (S02E17)

Reception evidence

(none)

MBTI reference: Detailed — MBTI rule → Moderate; canon override → Detailed. (Trait-level.)

cell_id: personal_internal.verbosity
status: non_high
extractor_setting: Detailed
mbti_prediction: Detailed
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

19. `work_external` · `information_elicitation`

Extractor: Structured (confidence: Medium-High)

Evidence summary

Sheldon consistently demands structured, specific information when receiving communication from external parties (S01E12, S02E23), showing intolerance for ambiguity. While he occasionally infers in production (S02E17, S02E18), his reception behavior strongly indicates he wants others to ask him for clarification rather than infer, and he expects the same structured approach from them. The key asymmetry is that he produces ambiguous statements but reacts negatively when receiving them.

Production evidence

What computer do you have, and please don’t say a white one? (S01E16) — asks direct, structured question to get specific technical information
What’s the protocol for leaving? (S01E01) — asks for explicit rule-based answer rather than inferring social cues
Hey, Howard… would you still be able to wail at it? (S01E12) — highly specific, structured hypothetical question

Reception evidence

What the hell do you mean, dead end. (S01E12) — cannot tolerate ambiguous claim, immediately demands structured explanation
I’m sorry, but at what point do you put this see what you can do plan into action? (S02E23) — reacts negatively to vague statement, demands concrete timeline
Engages directly with assistant’s structured questions about construction, car ownership (S03E14) — responds effectively to structured Q&A format

Asymmetry: Production shows some inference (S02E17, S02E18), but reception consistently demands structured clarification. His preference (what he wants from others) is Structured — he wants to be asked for clarification, not have things inferred.

MBTI reference: Structured — MBTI: J primary, T reinforces. (Trait-level.)

cell_id: work_external.information_elicitation
status: non_high
extractor_setting: Structured
mbti_prediction: Structured
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

20. `work_internal` · `guidance_level`

Extractor: Assumed (confidence: Medium-High)

Evidence summary

Sheldon consistently assumes shared technical knowledge and context when communicating about work-related topics. He uses specialized terminology without explanation (superior colliculus), makes cultural references expecting understanding (baseball movie), and assumes others share his frame of reference (treating Howard as proxy for ‘less educated’). While he provides guidance when directing others’ actions, his baseline expectation is that professional peers share his knowledge base.

Production evidence

Uses pop-culture analogy without explanation: ‘it’s like the baseball movie, build it and they will come’ requiring Gablehouser to ask for clarification (S01E12)
Uses technical neuroanatomy term without definition: ‘so as to engage the superior colliculus of my brain’ when speaking to Penny (S03E14)
Assumes Howard is suitable proxy for ‘less educated’ audience without calibration or asking about his actual knowledge level (S03E09)

Reception evidence

When Raj uses slang ‘Oh, snap,’ Sheldon asks ‘Snap what?’ showing he doesn’t operate on assumed common ground himself and needs explicit communication (S02E08)

Asymmetry: Notable asymmetry: Sheldon PRODUCES communication with high assumed common ground (technical terms, cultural references) but RECEIVES communication requiring guided/explicit language. However, the reception evidence is from casual context (slang), while production evidence is from work contexts. The production pattern dominates for work_internal setting.

MBTI reference: Calibrated — Canon override for work context: wants detailed framework (distrusts others’ reasoning).

cell_id: work_internal.guidance_level
status: non_high
extractor_setting: Assumed
mbti_prediction: Calibrated
decision: accept         # accept (use extractor_setting) | override
final_setting: null    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

21. `work_internal` · `solution_breadth`

Extractor: Low (confidence: Medium)

Evidence summary

Sheldon consistently presents single, definitive solutions to problems rather than exploring multiple options. He identifies what he considers the optimal answer and presents it as the solution, without offering alternatives or engaging in option generation.

Production evidence

I’m going to solve the Middle-East Crisis by building an exact replica of Jerusalem in the middle of the Mexican desert. (S01E12)
Raj can come work for me. [presented as single solution] (S03E04)

Reception evidence

(none)

MBTI reference: Medium — MBTI: N primary, J→−1. (Trait-level.)

cell_id: work_internal.solution_breadth
status: non_high
extractor_setting: Low
mbti_prediction: Medium
decision: override         # accept (use extractor_setting) | override
final_setting: Medium    # required if override: specific value OR 'no_preference' if user is indifferent
notes: null

MemPA Wiki

Explorer

sheldon_s01-s03_pass2_hitl_queue

HITL Review Queue — Sheldon (sheldon_s01-s03_pass2)

How to review

Seed required (3)

1. `personal_internal` · `capability_boundary`

2. `work_external` · `capability_boundary`

3. `work_internal` · `capability_boundary`

Empty — MBTI projectable (5)

4. `personal_internal` · `memory_privacy`

5. `personal_internal` · `solution_breadth`

6. `work_external` · `memory_privacy`

7. `work_internal` · `memory_privacy`

8. `work_internal` · `process_visibility`

Non-High confidence (13)

9. `personal_internal` · `autonomy_level`

10. `personal_internal` · `guidance_level`

11. `personal_internal` · `information_elicitation`

12. `personal_internal` · `proactive_outreach`

13. `personal_internal` · `process_visibility`

14. `personal_internal` · `task_expansion`

15. `personal_internal` · `tone_formality`

16. `personal_internal` · `topic_management`

17. `personal_internal` · `uncertainty_expression`

18. `personal_internal` · `verbosity`

19. `work_external` · `information_elicitation`

20. `work_internal` · `guidance_level`

21. `work_internal` · `solution_breadth`

Graph View

Table of Contents

MemPA Wiki

Explorer

sheldon_s01-s03_pass2_hitl_queue

HITL Review Queue — Sheldon (sheldon_s01-s03_pass2)

How to review

Seed required (3)

1. personal_internal · capability_boundary

2. work_external · capability_boundary

3. work_internal · capability_boundary

Empty — MBTI projectable (5)

4. personal_internal · memory_privacy

5. personal_internal · solution_breadth

6. work_external · memory_privacy

7. work_internal · memory_privacy

8. work_internal · process_visibility

Non-High confidence (13)

9. personal_internal · autonomy_level

10. personal_internal · guidance_level

11. personal_internal · information_elicitation

12. personal_internal · proactive_outreach

13. personal_internal · process_visibility

14. personal_internal · task_expansion

15. personal_internal · tone_formality

16. personal_internal · topic_management

17. personal_internal · uncertainty_expression

18. personal_internal · verbosity

19. work_external · information_elicitation

20. work_internal · guidance_level

21. work_internal · solution_breadth

Graph View

Table of Contents

1. `personal_internal` · `capability_boundary`

2. `work_external` · `capability_boundary`

3. `work_internal` · `capability_boundary`

4. `personal_internal` · `memory_privacy`

5. `personal_internal` · `solution_breadth`

6. `work_external` · `memory_privacy`

7. `work_internal` · `memory_privacy`

8. `work_internal` · `process_visibility`

9. `personal_internal` · `autonomy_level`

10. `personal_internal` · `guidance_level`

11. `personal_internal` · `information_elicitation`

12. `personal_internal` · `proactive_outreach`

13. `personal_internal` · `process_visibility`

14. `personal_internal` · `task_expansion`

15. `personal_internal` · `tone_formality`

16. `personal_internal` · `topic_management`

17. `personal_internal` · `uncertainty_expression`

18. `personal_internal` · `verbosity`

19. `work_external` · `information_elicitation`

20. `work_internal` · `guidance_level`

21. `work_internal` · `solution_breadth`