MindOS Learning Manual: Intertextual-Integration State | Reading Four Sources Is Not the Same as Building One Coherent Understanding

Wait, What?

You can read four sources correctly and still fail to understand the topic.

A student opens a textbook chapter, a teacher handout, an article and an AI explanation.

They take notes from all four.

Nothing is obviously wrong.

But the notes remain four separate piles.

The learner has performed multiple-source reading without performing multiple-source integration.

Quick Answer

Intertextual-Integration State is the learner operation of selecting relevant information from several documents, organising relationships among those documents, corroborating or distinguishing their claims, and synthesising them into one coherent representation of the topic or problem.

The key distinction is simple:

Collecting information from several sources is not the same as connecting what those sources mean together.

The final learning object is not Source A + Source B + Source C + Source D.

It is the model the learner can construct across them.

Owned Learning Operation

INTERTEXTUAL-INTEGRATION STATE = define question → extract relevant claims → preserve source identity → compare relations → corroborate/conflict-check → synthesise → qualify uncertainty → reconstruct without source set → transfer.

This page does not own citation mechanics, bibliography formatting, source storage, version control or keeping track of where a fact came from. Those are Student/Studying Interface jobs.

MindOS owns what happens cognitively after the learner has access to the sources:

How do these documents become one coherent, qualified understanding?

Why Multiple Sources Are Harder Than One Source

One document usually gives the learner one local structure. Several documents create extra work.

They may use different vocabulary for the same idea.
They may emphasise different aspects.
They may repeat the same claim using different evidence.
They may disagree.
One may be more specific while another is more general.
One may describe mechanism while another describes outcome.
One may be current while another reflects older understanding.
The learner must remember not only what was said, but which source said it.

This creates a different learner job from ordinary comprehension.

Four Multiple-Source Failure States

1. Source Stacking

The learner writes one paragraph from Source A, then one from Source B, then one from Source C.

All sources are present. Integration is absent.

2. Agreement Collapse

Two sources use different wording, so the learner assumes they disagree when they actually describe the same relationship at different levels.

3. Conflict Collapse

Two sources genuinely disagree, but the learner merges them into one smooth summary and erases the disagreement.

4. Provenance Collapse

The learner remembers the claim but no longer knows which source supported it, qualified it or contradicted it.

This matters because evidence cannot be weighed intelligently once its origin disappears.

The Source-Claim Matrix

A useful early scaffold is a simple source-claim matrix.

Source	Main claim	Evidence / basis	Agrees with	Differs from	What it adds
A	Claim	Evidence	B	C	Mechanism
B	Claim	Evidence	A	—	Example
C	Qualified claim	Different evidence	—	A	Boundary

The table is not the final product.

Its purpose is to make relationships visible long enough for the learner to build a synthesis.

The MindOS Intertextual-Integration Protocol

Step 1 — Define the Question Before Reading Across Sources

Without a question, the learner cannot judge which information from each source deserves integration.

“Learn climate change” is too broad.

“What evidence supports the claim that recent warming is primarily driven by human activity?” creates a usable integration target.

Step 2 — Extract Claims, Not Just Facts

For each source, ask:

What is this source actually claiming?
What evidence or reasoning supports the claim?
What uncertainty or condition does it include?

A sentence copied from a source is not yet an integrated unit.

Step 3 — Preserve Source Identity

Keep enough provenance visible that the learner can later say:

Source A explains the mechanism; Source B provides historical evidence; Source C qualifies the effect under a different condition.

This is not citation formatting. It is epistemic bookkeeping.

Step 4 — Compare Across Sources

For each pair of relevant claims, classify the relation:

same claim, different wording;
same claim, different evidence;
compatible but different levels;
complementary;
conditional difference;
direct conflict;
not actually comparable.

This prevents the learner from treating all difference as disagreement and all overlap as redundancy.

Step 5 — Corroborate Without Counting Votes

Three sources saying the same thing is not automatically stronger than one better source.

Ask:

Are the sources independent?
Do they rely on the same underlying evidence?
Does one source merely repeat another?
Do they agree for the same reason?
Is one source more authoritative for this exact claim?

This keeps corroboration from collapsing into source counting.

Step 6 — Build the Synthesis Sentence

Try to write one sentence that none of the individual sources states exactly, but that is justified by the relationship among them.

For example:

Taken together, the sources support X under conditions Y and Z, while leaving uncertainty about Q.

This is a genuine integration move because the learner is constructing a new representation across documents.

Step 7 — Reconstruct Without the Source Set

Close the documents.

Can the learner explain:

the central conclusion;
which sources support which part;
where disagreement remains;
what uncertainty should not be erased?

If not, the integration may still live in the external notes.

Worked Examples Across Subjects

English: compare two literary-critical interpretations of the same character. One emphasises motive; another emphasises social context. The learner should decide whether these claims conflict, complement one another or operate at different levels, then construct a justified synthesis.

Mathematics: a student studies two solution methods for the same problem. One source uses algebra; another uses a graphical argument. Intertextual integration asks what invariant relationship both methods preserve, where one is more efficient, and what each representation exposes.

Science: one source explains a biological mechanism, another reports experimental findings, and a third describes limitations. The learner should not paste all three together. They should build a model in which mechanism, evidence and uncertainty occupy distinct but connected roles.

When Sources Disagree

Disagreement is not automatically a failure of integration.

Sometimes the correct synthesis is:

The evidence remains contested.

MindOS does not force false harmony.

When sources conflict, ask:

Are they answering exactly the same question?
Are they using the same definitions?
Do they study different populations or conditions?
Are they drawing different conclusions from the same evidence?
Is one source newer or methodologically stronger?
Can both claims be true under different boundaries?

The learner should preserve uncertainty until the evidence justifies resolution.

Competing Explanations When Integration Fails

Single-document comprehension may be weak.
The learner may lose source identity while reading.
Working-memory demands may be too high.
Topic knowledge may be insufficient to recognise relationships.
The learner may lack strategies for handling disagreement.
The task may not specify what kind of synthesis is required.
The documents may themselves be poorly matched or too complex.

Do not diagnose “poor synthesis” until these alternatives are checked.

How Do We Know?

A 2024 systematic review in Educational Psychology Review focused specifically on K–12 students’ intertextual integration. It defines intertextual integration as selecting, organising, corroborating and synthesising information from multiple documents to construct a coherent representation across them.

The review identified 25 studies and found substantial variation in learners, tasks and document types, with limited systematic replication. Even so, literacy, cognition, metacognition, knowledge, beliefs and motivation generally showed positive small-to-moderate associations with intertextual integration.

The review is especially valuable because it separates intertextual integration from nearby multiple-document jobs such as search, selection and source evaluation. That supports the MindOS ownership boundary used here.

Espinas & Chandler (2024), Correlates of K–12 Students’ Intertextual Integration

Evidence Boundary

The 2024 review is primarily correlational. It identifies factors associated with stronger intertextual integration; it does not prove that changing any one factor will necessarily cause better synthesis.

The studies also used different document sets, measures and age groups. That heterogeneity limits precise universal prescriptions.

The safe inference is narrower: integrating multiple documents is a distinct and cognitively demanding learner operation that depends on more than reading each source correctly; learners must preserve source relations and construct a coherent cross-document representation.

When Intertextual Integration Is the Wrong Tool

When the learner cannot yet comprehend the individual documents.
When the task requires one authoritative source rather than synthesis.
When the sources are redundant and add no meaningful comparison.
When provenance has been lost and claims cannot be traced safely.
When the learner is still trying to learn the basic concept for the first time and multiple sources would add unnecessary load.
When the actual job is source evaluation, citation or file organisation rather than integration.

Scaffold Fade

Stage 1: tutor provides a source-claim matrix.
Stage 2: learner fills claim/evidence/agreement columns independently.
Stage 3: learner creates the comparison structure without a template.
Stage 4: learner writes synthesis while preserving disagreements and uncertainty.
Stage 5: learner decides independently which sources need integration and which should remain separate.

The external matrix should eventually disappear. The learner should carry the relationship structure internally.

Immediate, Delayed and Transfer Checks

Immediate: can the learner state the central cross-source conclusion?
Provenance: can the learner remember which source supports or qualifies which claim?
Conflict: can the learner preserve a real disagreement rather than averaging it away?
Delayed: can the learner reconstruct the synthesis later without reopening every source?
Transfer: can the learner integrate a new source that changes, strengthens or challenges the existing model?

Teaching Guide for Parents, Tutors and Teachers

When a learner has several sources, do not ask only “Have you read all of them?” Ask:

“What claim does each source make?”
“Which two sources agree—and do they agree for the same reason?”
“Where is the real disagreement?”
“Which source adds something nobody else adds?”
“Can you write one sentence that combines the sources without pretending they are identical?”
“What uncertainty remains?”
“Now close the documents. What is your model?”

This turns multi-source work from information collection into synthesis.

MindOS Direction

If the learner cannot understand the individual documents: repair single-source comprehension first.

If sources are understood but claims are not compared: use Comparison State.

If cross-source relations are hard to hold: use Concept Mapping as a temporary relational scaffold.

If the learner has lost where claims came from: route to the Student/Studying Interface Source-Trail or Citation owner rather than duplicating that machinery here.

If sources contradict the learner’s prior model: use Refutation State or Prior-Knowledge Activation depending on the mechanism.

If the synthesis works only with this source set: test Transfer State.

MindOS rule: multiple sources become learning only when the learner can preserve who said what, understand how the claims relate, and build a coherent model that survives after the tabs are closed.