READING COMPREHENSION · REASONING CHAIN

Map the argument.
Connect only what fits.

Trace how a scientific passage develops, join evidence without mixing conditions, and stop every inference at the strongest claim the text can actually carry.

Learn the three audits ↓Open the passage lab →
1original passage
9paragraph jobs
6evidence bridges
90option proofs

THE REASONING LOOP

Job. Bridge. Bound.

  1. 01

    Assign the job

    Describe what each paragraph does in the passage—not merely what subject it mentions.

  2. 02

    Build the bridge

    Join only spans whose population, condition, measurement, and logical direction remain compatible.

  3. 03

    Cap the claim

    Take the smallest warranted step, then reject certainty, causation, or outside knowledge the text never supplies.

THREE DEPENDENT SKILLS

Structure tells you
what may connect.

This route builds on location and contextual meaning. It adds organization, separated-evidence integration, and bounded inference; author analysis remains a separate route.

01

RC-SCI-ORG-01

Trace paragraph jobs

Function is a verb

A topic label says “radiofrequency.” A structural description says “uses a treatment-by-task contrast to separate two candidate mechanisms.” Ask what would break if the paragraph disappeared.

  1. P01
    Split the problem

    Distinguishes position from direction and rejects an over-simple GPS analogy.

    opens
  2. P02
    Define the signal

    Introduces field features, limits the signature metaphor, and explains virtual displacement.

    defines
  3. P03
    Test regional markers

    Presents inherited, route-appropriate responses and states what the assay cannot establish.

    tests
  4. P04
    Add a second coordinate

    Extends the regional evidence beyond a simple north-south gradient.

    extends
  5. P05
    Change the question

    Shifts from inherited headings to learned, nondirectional place recognition.

    shifts
  6. P06
    Report and bound learning

    Combines the conditioning results with memory, feature-matching, and external-validity limits.

    qualifies
  7. P07
    Separate mechanisms

    Uses a treatment-by-task contrast to test whether map and compass share one mechanism.

    tests
  8. P08
    Move to population evidence

    Tests a field-derived natal-homing prediction in a long nesting record and preserves observational limits.

    shifts
  9. P09
    Assemble the chain

    Integrates inherited, learned, mechanistic, and population evidence while naming open links.

    integrates
02

RC-SCI-INT-01

Integrate compatible evidence

Every bridge needs two rails

Both cited spans must support the relationship. Shared vocabulary is not enough when ages, tasks, conditions, or levels of evidence differ.

Alignment auditPopulation · condition · measure · direction · degree

01
P02 + P03

Generated fields can isolate magnetic information without moving a turtle to the represented region.

Do not cross this boundary

The assay does not recreate the region's currents, odors, landmarks, or complete ecology.

02
P03 + P04

Regional responses include information beyond a simple north-south gradient.

Do not cross this boundary

The studies do not establish pixel-level position or the system's spatial resolution.

03
P05 + P06

The nondirectional assay demonstrates learned recognition and four-month retention of a rewarded signature.

Do not cross this boundary

A conditioned food response is not a complete natural migration or evidence of lifelong memory.

04
P01 + P07

Map recognition and compass orientation are functionally and experimentally distinguishable.

Do not cross this boundary

A differential radiofrequency effect does not identify either receptor.

05
P02 + P08

Slow field change makes geomagnetic imprinting generate a testable prediction about nesting distributions.

Do not cross this boundary

Matching a population prediction does not directly observe an individual's remembered field.

06
P03 + P06 + P08

Inherited responses, learned signatures, and population patterns provide complementary evidence at different levels.

Do not cross this boundary

The evidence cannot be collapsed into one experiment that tracks a turtle through an entire journey.

03

RC-SCI-INF-01

Take the smallest warranted step

Inference is controlled extension

A valid inference may not repeat one sentence word for word, but every part of it must be necessary or strongly supported by the cited pattern. Plausible science cannot fill a missing premise.

  • Do not turn association into causation.
  • Do not turn one interval into forever.
  • Do not turn an unidentified mechanism into a named receptor.
  1. 01
    Text

    Radiofrequency treatment spared learned map recognition but disrupted compass orientation.

  2. 02
    Warranted step

    The two tasks probably do not rely on one identical magnetoreception mechanism.

  3. 03
    Forbidden leap

    The map receptor must be magnetite in a particular organ.

DISTINCT PASSAGE · FROZEN TRAIL

Original prose.
Five traceable sources.

The passage is original DAT TRAIN writing. Its 20 scientific claims map by paragraph to NOAA geomagnetism guidance and four empirical sea-turtle studies.

RIGHTS RECORD
Basis
original
Rights holder
DAT TRAIN
Status
draft · human review open
Simulation-ready
no

Original DAT TRAIN passage independently synthesized from the frozen scientific source trail. No source wording, official examination content, memorized test material, or proprietary practice content was reproduced.

  1. 01
  2. 02
  3. 03
  4. 04
  5. 05

ONE PASSAGE · EIGHTEEN REASONING AUDITS

Build the claim beside its proof.

The passage and function map remain beside every question. After checking, each option exposes its classification, rationale, exact quotation, paragraph anchor, and frozen offsets.

18 PRACTICE QUESTIONS

Retrieve before you review.

Question order and all five answer options are shuffled when you begin. The correct answer stays attached to the same underlying choice.

SOURCE-SET BOUNDARY

Two source sets are not a Reading section.

These 18 original questions cover only organization, integration, and bounded inference. Their 5/6/7 objective split is a construction choice, not a claim about official frequency.

The passage and questions remain draft and uncalibrated. This learner route is open as a public beta, and no raw result is converted into an official score.

The next route adds a third distinct passage, author analysis, and the three-passage, 50-question source form without changing the existing sets.

Continue to the author-analysis lab →