MindOS Learning Manual: Claim–Evidence Reasoning State | A Fact Is Not Evidence Until You Can Explain What It Supports

Wait, What?

A fact does not become evidence just because it is true.

A student writes three accurate facts beneath a claim.

The teacher asks:

How do those facts support your conclusion?

The student replies:

Because they are evidence.

The problem is not necessarily missing information.

The missing operation is the warrant: the reasoning that explains why the information counts in favour of the claim.

MindOS therefore treats evidence-based reasoning as a construction problem, not a formatting exercise.

Quick Answer

Claim–Evidence Reasoning State is the learner operation of making a bounded claim, selecting evidence that is relevant and sufficient, explaining the principle or warrant that connects the evidence to the claim, testing alternative explanations or counterexamples, and qualifying the conclusion to match the strength of the evidence.

The core structure is:

CLAIM
↓
EVIDENCE
↓
WHY DOES THIS EVIDENCE COUNT?
↓
WARRANT / REASONING
↓
ALTERNATIVE / COUNTEREVIDENCE
↓
QUALIFIED CONCLUSION

The RFE is:

Can the learner show not only what they believe and what information they found, but why that information justifies the exact conclusion being made?

Owned Learning Operation

CLAIM–EVIDENCE REASONING STATE = define claim → select relevant evidence → test sufficiency → construct warrant → inspect alternatives → qualify conclusion → reconstruct without scaffold → transfer.

This is distinct from Explanation State. Explanation broadly owns why or how a relationship works. Claim–Evidence Reasoning owns a narrower epistemic relation: why this evidence supports this claim.

It is also distinct from Source Evaluation, which asks how much trust to place in a source, and Intertextual Integration, which combines claims across sources. A source can be highly credible and still provide evidence that is irrelevant to the learner’s claim.

Claim, Evidence and Warrant Are Different Jobs

Component	Question	Common failure
Claim	What am I concluding?	Too broad, vague or unfalsifiable
Evidence	What information bears on the claim?	True but irrelevant fact
Warrant / reasoning	Why does that evidence support the claim?	Missing logical or disciplinary link
Qualifier	How strong is the conclusion?	Evidence supports “may”, learner writes “proves”
Rebuttal / alternative	What else could explain the evidence?	One-sided argument

Students often collapse evidence and reasoning because school writing can reward a familiar structure without inspecting the logic inside it.

A True Fact Can Still Be Irrelevant

Claim:

The writer presents the character as isolated.

Fact:

The story takes place in winter.

The fact is true.

It is not yet evidence for isolation.

It becomes potentially relevant only if the learner can establish a warrant—for example, that the writer repeatedly uses the empty winter landscape and absence of people to mirror or intensify the character’s separation.

Without the link, the learner has information beside a claim, not evidence supporting a claim.

Five Claim–Evidence Failure States

1. Claim Without Evidence

The learner states a conclusion confidently but supplies no observable support.

2. Evidence Dump

The learner provides quotations, data or facts without explaining how they bear on the conclusion.

3. Circular Warrant

The learner says the evidence supports the claim “because it proves the claim”. Nothing has been explained.

4. Overclaim

The evidence supports a limited tendency, but the learner writes “always”, “causes”, “proves” or “everyone”.

5. Template Completion

The learner fills boxes labelled Claim, Evidence and Reasoning but does not test the relationship among them.

A scaffold can organise thinking. It can also become a form that hides weak thinking.

The MindOS Claim–Evidence Protocol

Step 1 — Make the Claim Small Enough to Test

Weak claim:

Exercise is good.

Stronger claim:

In this experiment, the treatment group showed a larger average increase in X than the comparison group.

The stronger claim makes the evidential burden visible.

Step 2 — Select Evidence That Could Change the Claim

Ask:

If this evidence were different, would my confidence in the claim change?

If not, the evidence may be decorative.

Step 3 — Test Relevance Before Quantity

Five irrelevant quotations are not stronger than one directly relevant observation.

For each piece of evidence, ask:

What part of the claim does this bear on?
Is it direct or indirect?
Does it support the claim or only provide background?
Does it introduce a condition or exception?

Step 4 — Construct the Warrant

Complete:

This evidence supports my claim because…

Then forbid the learner from repeating the claim.

The answer should contain the disciplinary principle, causal link, inferential rule or textual relation that makes the evidence count.

Step 5 — Ask What Else Could Explain the Evidence

A strong argument is not one that has never met an alternative.

Could another cause produce the same result?
Could the quotation support a different interpretation?
Could measurement error explain the difference?
Could the rule have an exception?
Could the evidence fit two competing models?

Step 6 — Match the Strength of the Conclusion to the Evidence

Use language such as:

shows;
supports;
is consistent with;
suggests;
does not rule out;
under these conditions;
in this sample;
probably;
cannot yet establish.

Qualification is not weakness. It is evidential accuracy.

Step 7 — Remove the CER Labels

Once the learner can build the relation, remove the boxes and sentence stems.

Real reasoning rarely arrives with headings saying CLAIM, EVIDENCE and REASONING.

Worked Example: English

Claim: The writer presents the crowd as threatening.

Evidence: The crowd is described as “pressing closer” and individual faces disappear into “one dark wall”.

Weak reasoning: This shows the crowd is threatening because it sounds threatening.

Stronger warrant: The verbs and collective image remove personal distance and make the crowd appear physically enclosing, so the language supports an interpretation of danger rather than neutral gathering.

Alternative: Could the crowd be excited rather than threatening? The learner returns to surrounding context and tests which interpretation better explains the language.

Worked Example: Science

Claim: Increasing light intensity increased the rate of photosynthesis in this part of the experiment.

Evidence: Oxygen production increased as the lamp moved closer across the tested range.

Warrant: Oxygen production is being used as an indicator of photosynthetic rate, and greater light availability can increase the rate while light is limiting.

Qualifier: The data support the claim only across the tested range and do not show that rate would increase indefinitely.

Worked Example: Mathematics

Claim: The proposed solution cannot be correct.

Evidence: Substituting the proposed value into the original equation gives unequal left- and right-hand sides.

Warrant: A solution to an equation must satisfy the original equality. Failure under substitution is therefore direct counterevidence to the claim that the value is a solution.

This is claim–evidence reasoning even though there is no essay.

CER Is a Useful Scaffold—Not the Whole of Argumentation

The Claim–Evidence–Reasoning framework is widely used because it makes three load-bearing components visible: a conclusion, supporting data and the logic connecting them.

But a mature argument may also need:

qualifiers;
rebuttals;
counterevidence;
source credibility;
uncertainty;
comparison among competing explanations.

A 2024 Science Education study explicitly examined moving middle-school students beyond basic CER through critical questions. That is an important boundary: the three-part scaffold is a starting architecture, not a complete theory of reasoning.

How Do We Know?

Claim–Evidence–Reasoning is a classroom simplification of a longer tradition of argument analysis associated with Toulmin. Educational research across Science, Chemistry, Biology and other domains repeatedly documents the difficulty students have not only selecting evidence but explaining why evidence supports a claim.

A 2024 study in Science Education highlights a limitation of basic CER instruction: students also need to critique arguments and ask critical questions, because merely supplying the three components does not guarantee robust evaluation.

Meta-analytic evidence on argumentation more broadly is encouraging but not perfectly uniform. A 2025 meta-analysis of 17 experimental datasets in science education reported a positive overall effect of argumentation-based teaching across outcomes including academic achievement, scientific-process skills and conceptual knowledge. However, older meta-analytic work in computer-supported collaborative learning found that interventions could improve argumentation while evidence for corresponding domain-knowledge gains remained equivocal.

That contradiction is useful. It prevents MindOS from making the lazy claim that “better arguments automatically mean more subject knowledge”. Argumentation is itself a capability; whether it also improves domain learning depends on design, task, prior knowledge and how reasoning interacts with content.

Evidence Boundary

Most CER research is situated in scientific explanation and argumentation. The structure can be useful in English, Mathematics and other subjects, but the standards for evidence and warrants differ by discipline.

A literary interpretation is not justified in the same way as an experimental causal claim. A mathematical proof is not evaluated in the same way as a historical source argument.

The safe inference is therefore:

Students benefit from learning to distinguish claims, evidence and the reasoning that connects them, but the scaffold must be adapted to disciplinary standards and extended when uncertainty, alternatives or rebuttals matter.

When Claim–Evidence Reasoning Is the Wrong Tool

When the learner does not yet understand the underlying concept.
When the task is simple factual retrieval rather than justification.
When the problem requires a formal proof whose standards exceed a generic CER scaffold.
When the source itself is unreliable and Source Evaluation must happen first.
When several sources conflict and Intertextual Integration is the primary problem.
When the learner can state the warrant but lacks enough evidence to justify the claim.
When a rigid template makes the writing formulaic and hides real reasoning.

Scaffold Fade

Stage 1: learner fills explicit Claim / Evidence / Reasoning boxes.
Stage 2: learner labels the components inside a complete answer.
Stage 3: learner generates the warrant without a sentence stem.
Stage 4: learner adds qualifiers, alternatives or rebuttals when the evidence requires them.
Stage 5: learner produces a disciplined argument without visible CER labels and can audit another argument for missing links.

The scaffold succeeds when the logic remains after the headings disappear.

Immediate, Delayed and Transfer Checks

Immediate: can the learner distinguish the claim from the evidence?
Warrant: can the learner explain why each piece of evidence bears on the claim?
Sufficiency: can the learner identify what additional evidence would strengthen or weaken the conclusion?
Qualification: can the learner reduce an overclaim when evidence is limited?
Delayed: can the learner reconstruct the logic later without the CER template?
Transfer: can the learner use the same reasoning discipline in a different subject where evidence standards change?

AI Boundary: AI Can Fill CER Boxes Without Owning the Argument

An AI can generate a plausible claim, find supporting quotations and write a polished reasoning sentence in seconds.

That proves the tool can construct an argument-like artifact. It does not prove the learner can judge evidential relevance.

A safer sequence is:

learner writes the claim first;
learner selects evidence;
learner writes the warrant;
AI challenges one link or proposes a counterexample;
learner decides whether the challenge is valid;
learner revises the conclusion;
AI closes;
learner constructs a fresh argument independently.

Better prose is not automatically better reasoning.

Teaching Guide for Parents, Tutors and Teachers

When a learner gives a fact after a claim, do not immediately say “add more evidence”. First ask whether the existing evidence is doing any work.

“What exactly are you claiming?”
“Which part of that claim does this evidence support?”
“Why does the evidence count?”
“What rule, principle or textual relationship connects them?”
“What else could explain the same evidence?”
“Is your conclusion stronger than your evidence?”
“Can you rewrite the claim so it says only what the evidence earns?”

These questions turn “PEEL”, “CER” or other writing structures back into thinking structures.

MindOS Direction

If the learner does not understand why the relationship works at all: route to Explanation State.

If the evidence comes from a questionable source: use Source-Evaluation State before increasing its evidential weight.

If several sources must be reconciled: use Intertextual-Integration State.

If a counterexample breaks the claimed rule: use Counterexample-Generation State.

If the learner can construct the argument only with labels and sentence stems: use Scaffold Fading.

If the learner can reason on one familiar topic but not another: test Transfer State.

MindOS rule: facts become evidence only in relation to a claim, and evidence becomes an argument only when the learner can explain the warrant, test alternatives and make the conclusion no stronger than the evidence allows.