Cog 8 Eng

PrepMaster · Cognitive Traps v2 — 7 + 1
PrepMaster · Cognitive Trap Taxonomy v2

7 Item-Level Traps
+ 1 Meta-State

A pure cognitive classification, stripped of section and question-type labels. v2 separates the layers, preserves a remediation branch beneath each trap, and rebalances the four NOQ axes.

7 + 1
item-level × 7 · meta-state × 1
CT-1 through CT-7 are traps detected at the item level; CT-8 is a session-level state. They are not placed on the same plane — CT-8 is labeled as a separate layer to keep the classification coherent. Diagnosis and reporting run on the 8 codes; remediation branches on the PM sub-codes beneath them.
Mapping · Rebalanced

NOQ Four-Axis MappingThe Rhetorical axis previously rested on a single signal source (CT-6). v2 adds the tone facet of CT-5 and the claim-boundary facet of CT-3 as secondary sources, so every axis now draws on two or more.

The three traps clustering on the Calibration axis are framed as a finding, not a flaw — the claim that the real bottleneck for these students is cognitive calibration (haste, discrimination, session collapse) rather than vocabulary is a stronger story than a tidy split. (Solid = primary signal source · dashed = secondary.)

Full Specifications

8 Traps · Four Layers + Remediation BranchCognitive Root → Trap Structure → Detection Logic (Supabase) → Socratic Output. Beneath each, a PM sub-code splits the distinct remediations inside the same trap.

1
Premature Closure
CT-1
ITEM-LEVEL
Cognitive Calibration
L1 · Cognitive Root

The fluency of System 1 is mistaken for truth. A "this feels familiar = correct" signal shuts down verification. The moment certainty arrives is the moment to stop and check.

L2 · Trap Structure

A correct-looking cue is placed at the front or surface of a choice, leading the student to stop before reading the rest.

L3 · Detection Logic
time_on_choice < 2000ms AND is_correct = false AND revision_count = 0 → CT-1
L4 · Socratic Output

"It felt familiar, so you were glad to see it — but did you read it to the end? The moment you feel certain is the moment to slow down and check."

Sub-patterns · Remediation Branch
PM-R-01hasty confirmation
A 1-second "why is this right?" check before selecting. Re-read the final word.
PM-V-S-04partial-root overconfidence
Whole word first; anchor a root only after three real example words.
2
Direction Error
CT-2 · Polarity / Direction
ITEM-LEVEL
Conceptual Leap
L1 · Cognitive Root

The semantic or logical field is recognized correctly, but its sign (+/−) or direction is reversed. The right neighborhood, the wrong-pointing arrow.

L2 · Trap Structure

A same-field word of opposite direction, or a choice with cause and effect swapped, is placed as the distractor.

"abate" → selects "intensify." Both relate to magnitude — direction is reversed.

L3 · Detection Logic
distractor_relation IN ('opposite_polarity','reversed_causality') AND is_correct = false → CT-2
L4 · Socratic Output

"Everything but the direction was perfect. Is this getting bigger or smaller? Let's draw the arrow again."

Sub-patterns · Remediation Branch
PM-V-S-01antonym confusion
Train an ↑/↓ direction tag on each word.
causality reversal
Mark "which comes first" on a time-ordered arrow.
logic-direction misread
Flag the direction of connectives (however · therefore) before reading on.
3
Scope Distortion
CT-3
ITEM-LEVEL
Conceptual Leap+ Rhetorical
L1 · Cognitive Root

The boundary of the claim the text guarantees is stretched (over-) or shrunk. Low sensitivity to quantifiers (some / most / all); partial evidence is scaled to the whole, or inference runs past the text's edge.

L2 · Trap Structure

The passage says "some" while the choice says "all," or a step the text never guarantees is dressed as the answer. Modifier shifts (most → all).

L3 · Detection Logic
distractor_scope > passage_scope AND is_correct = false → CT-3 -- compare pre-tagged quantifiers
L4 · Socratic Output

"How far did the passage actually go? Between 'some' and 'all,' where should you stop? Let's see if you took a step the text doesn't guarantee."

Sub-patterns · Remediation Branch
PM-R-02scope inflation
Circle the passage quantifier → match it 1:1 against the choice.
PM-R-09evidence-scope error
Count "how many cases?" to block part-to-whole jumps.
PM-R-11over-inference
Require a passage line number per answer — out of bounds = eliminated.
4
Surface Capture
CT-4
ITEM-LEVEL
Lexical Noesis
L1 · Cognitive Root

Attraction to surface form — same word, similar spelling, prominent position — instead of meaning. When the answer is paraphrased, it "doesn't look like a match" and gets missed.

L2 · Trap Structure

A distractor copies a passage word verbatim (lexical magnet) while the real answer is disguised as a paraphrase.

L3 · Detection Logic
distractor_lexical_overlap = 'high' AND correct_answer_is_paraphrase = true AND is_correct = false → CT-4
L4 · Socratic Output

"The same word from the passage pulled you in, didn't it? The same word doesn't guarantee the same meaning. Let's match by meaning, not by word."

Sub-patterns · Remediation Branch (all three differ)
lexical magnet
Forbid direct word-to-word matching; allow meaning-matching only.
PM-R-10paraphrase blindness
Generate a "say it another way" first, then match.
PM-R-06visual magnet
Defeat pull of position/formatting by exposing shuffled choice order.
5
Intensity / Register Miscalibration
CT-5
ITEM-LEVEL
Lexical Noesis+ Rhetorical
L1 · Cognitive Root

The concept is right but the magnitude (intensity) or formality (register) is off. The gradient within a synonym cluster goes unresolved — common in learners who acquired English from casual media.

L2 · Trap Structure

A distractor with the same meaning but too weak, too strong, or too informal is placed beside the answer.

"loquacious" → selects "chatty." Meaning correct; register too informal.

L3 · Detection Logic
(distractor_register = 'informal' OR intensity_delta > 1) AND is_correct = false → CT-5
L4 · Socratic Output

"Your answer means the same thing — but would you use it in a formal essay, or a text to a friend? Let's recalibrate the intensity and register by one notch."

Sub-patterns · Remediation Branch
intensity misread
Sort synonyms on a weak→strong ladder.
PM-V-S-05register mismatch
Sort into informal / neutral / formal tiers + academic reading.
PM-R-03tone distortion (→Rhetorical)
Track the author's attitude cues (adjectives, adverbs) for tone match.
6
External Override
CT-6
ITEM-LEVEL
Rhetorical Architecture
L1 · Cognitive Root

Background knowledge, common sense, or intuition wins over the given text. Top-down overrides bottom-up evidence, and the author's intent is replaced with the reader's own.

L2 · Trap Structure

A choice that is "plausible" by common sense but has no support in the passage — a trap that flatters received wisdom.

L3 · Detection Logic
distractor_matches_common_belief = true AND passage_support = false AND is_correct = false → CT-6
L4 · Socratic Output

"It might even be true — but did this passage say it? Put your finger on the evidence the text gave you, not on what you already believe."

Sub-patterns · Remediation Branch
PM-R-04background-knowledge backfire
Force a "which line?" evidence label on every answer.
PM-R-08author-intent misread
Infer intent from text tone cues, not from one's own view.
PM-R-07intuitive ranking
A ranking holds only if a comparison sentence exists in the text.
7
Best-Answer Discrimination
CT-7
ITEM-LEVEL
Cognitive Calibration
L1 · Cognitive Root

"Almost right" is not separated from "most right." The student stops at the first plausible answer and evaluates absolutely instead of comparing relatively.

L2 · Trap Structure

A choice that matches 80–90% but flips in the final 2–3 words (half-truth), or a near-pair where both look correct.

L3 · Detection Logic
(hover_count >= 2 AND is_correct = false) OR distractor_tag = 'half_truth' → CT-7
L4 · Socratic Output

"This one's right? Then did you read the one next to it too? When two answers are 'right,' which is more precise? Set them side by side and compare to the last word."

Sub-patterns · Remediation Branch
PM-R-01 / CB-01half-truth
Read to the final word and check for a meaning flip.
PM-R-12relative-accuracy blind spot
A forced 1:1 runner-up comparison before locking an answer.
8
Cognitive Calibration Collapse
CT-8
META-STATE · session level
Cognitive Calibration
L1 · Cognitive Root

A failure to monitor one's own state. Under time pressure, a run of errors, or miscalibrated confidence, thinking quality degrades mid-session. This is a state that runs across the whole session, not a single item.

L2 · Trap Structure

Accuracy drops sharply in the back half or under pressure, recovery fails, and the gap between confidence and accuracy widens.

L3 · Detection Logic
-- session-end trigger · session_summary only recent_5_accuracy sharp drop OR |confidence_self_report − is_correct| gap OR no recovery after streak_fracture → CT-8
L4 · Socratic Output

"You wobbled after one wrong answer, didn't you? Your score didn't stop there — it fell from the next question on. The miss is over. Look only at the next one."

Sub-patterns · Remediation Branch (session-state signals)
PM-V-S-09~12session meta-patterns
A reset routine after a miss (3-second breath · next item only).
time-pressure collapse
A pacing alarm in the back half · speed adjusted to items remaining.
confidence miscalibration
1-second confidence input after each choice → confidence-accuracy feedback loop.
PrepMaster Inc. · Cognitive Trap Taxonomy v2 (EN) · 7 item-level traps + 1 meta-state · 2-layer diagnosis / remediation · NOQ four-axis rebalanced · internal master spec