01 The Provability Test
The Core Distinction
Inference questions ask: "Based on the text, what can most reasonably be inferred about…?" The key word is based on the text. The correct answer must be directly provable from evidence in the passage — not just logically possible, not just consistent with the passage's topic, and not dependent on outside knowledge.
The Provability Test — Apply to Every Inference Choice
"If I can point to specific text in the passage that directly supports this conclusion, and it cannot be reasonably denied based on what the passage says — it is provable."
If you cannot point to specific passage text that directly supports a choice, eliminate it — even if it seems logical, even if it is factually true in the real world, and even if three out of four choices seem possible. There is always exactly one answer that is provable from the text. Your job is to find it.
The Two Questions to Ask for Every Inference Choice
02 The Provability Spectrum

Inference choices range from clearly provable to clearly speculative. Most wrong answers sit in the "possible but not provable" zone — they feel right because they're reasonable, but they require you to go beyond what the text actually says.

✓ Provable ⚠ Possible — Not Provable ✗ Contradicted / Outside
✓ Provable (Correct)
Passage says a scientist "worked in isolation for three years before publishing." You can infer she did not collaborate with other researchers during that period — directly supported by "isolation."
⚠ Possible — Not Provable (Wrong)
Same passage. Choice: "She found working alone more productive than collaborating." Possible — but the passage never says this. "Isolation" describes her circumstance, not her preference or productivity.
✗ Outside / Contradicted (Wrong)
Same passage. Choice: "She was the only scientist in her field during that period." Nothing in the passage supports this — it requires outside reasoning about the field, not just the text.
03 Literary vs. Scientific Inferences

The provability test applies to both literary and scientific passages, but what counts as "directly supported" differs slightly by passage type.

TypeWhat the passage providesWhat a valid inference looks likeCommon wrong-answer trap
Literary Character action, dialogue, narrative detail, narrator observation, emotional description A conclusion about a character's emotional state, motivation, or relationship — directly supported by their specific words or actions in the text Inferring a character feels something that the text hints at but never shows — over-reading emotional detail
Scientific Experimental results, data, study findings, researcher conclusions, hedged claims A conclusion that follows directly from the stated results — using the same hedging language the passage uses (e.g. "suggests," "is associated with," not "proves" or "causes") Upgrading hedged conclusions to absolute claims — passage says "may reduce" but wrong answer says "eliminates"
Literary Contrast between two characters' reactions to the same event One character values X more than the other — if the text directly shows this through their differing responses Inferring deep personal history, past trauma, or future intentions not implied by the present scene
Scientific Results showing Group A outperformed Group B under condition X Under condition X, Group A shows better outcomes than Group B — same scope, same condition as the study Generalising results beyond the study's scope — e.g. from adolescents to all age groups, or from one condition to all conditions
04 The Outside Knowledge Trap
⚠ The Hardest Inference Trap
The most difficult wrong answers on inference questions are factually true statements about the real world that are consistent with the passage's topic — but are not actually supported by the passage text. These feel provable because you know they're true, but your knowledge is not evidence. Only the passage counts.
✗ Outside Knowledge Answer (Wrong)
Passage about coral bleaching: choice says "Ocean temperatures have increased due to the burning of fossil fuels."
Factually true — but if the passage doesn't say it, you cannot infer it from the passage. The passage might only say "rising ocean temperatures" without stating a cause.
✓ Text-Based Answer (Correct)
Same passage: choice says "Repeated bleaching events damage coral reefs' ability to support marine species."
Directly supported — the passage states that bleached corals lose food sources and that biodiversity depends on healthy reefs. This conclusion is forced by what the text says.
The Outside Knowledge Test
Before selecting an inference answer, ask: "Would I be able to reach this conclusion if I had never heard of this topic and only had the passage text to work with?" If the answer is no — if your real-world knowledge is doing part of the work — the choice requires outside knowledge and must be eliminated.
05 Worked Examples
Worked Example 1 Literary inference · Character motivation
Literature
Every year for eleven years, Daniel had submitted his manuscript to the same publisher, and every year he had received the same form rejection letter. When the acceptance call finally came, he thanked the editor politely, hung up, and sat very still at his kitchen table for a long time before he could bring himself to tell anyone.
Provability test applied: What does the passage directly show us about Daniel?
— He submitted every year for 11 years → persistent, deeply invested
— He sat "very still" for "a long time" before telling anyone → the news produced an overwhelming emotional response, not a simple reaction
— These two details together point to: the acceptance came after enormous sustained effort and was emotionally too large to immediately celebrate

Based on the text, what can most reasonably be inferred about Daniel?

AHe no longer wants his manuscript published after so many years of waiting.Contradicted
BHe suspects the acceptance call was made by mistake.Outside passage
CThe acceptance, after years of rejection, left him emotionally overwhelmed and needing time to absorb the news.✓ Provable
DHe is disappointed because the publisher took too long to respond.Contradicted

Key: C is directly supported by "sat very still" and "a long time before he could bring himself to tell anyone" — both signals of an overwhelming emotional response. A and D are directly contradicted by the text (he had been submitting repeatedly — he clearly wanted it published). B requires an assumption not in the text. Notice that all four options are possible explanations for his behaviour — but only C is forced by the specific details the passage provides.

Worked Example 2 Scientific inference · Hedging language trap
Natural Science
Researchers studying urban ecology found that city residents living within half a mile of a park reported lower average stress scores than those living farther away. The researchers noted, however, that their study was cross-sectional and could not establish whether proximity to parks reduced stress or whether people with lower stress levels were more likely to choose to live near parks.
Provability test applied: The passage establishes:
— Park-proximate residents reported lower stress scores (a correlation)
— The study is cross-sectional — cannot establish direction of causation
— The researchers explicitly said they cannot tell if parks reduce stress OR if less-stressed people choose park proximity
Therefore: any inference claiming parks "reduce" or "cause lower" stress goes beyond what the text supports.

Based on the text, what can most reasonably be inferred about the relationship between park proximity and stress?

ALiving near a park causes city residents to experience lower stress over time.Possible — not provable (causation claim)
BThe association between park proximity and lower stress scores may not reflect a causal relationship in either direction.✓ Provable
CLess-stressed individuals are more likely to choose to live near parks than more-stressed individuals.Possible — not provable (one of two directions stated)
DThe researchers concluded that parks have no effect on stress levels in urban residents.Contradicted (they drew no such conclusion)

Key: B is the only choice that accurately reflects what the passage proves — a correlation of uncertain direction. A claims causation that the researchers explicitly said they could not establish. C states one of the two possibilities the researchers mentioned — but the researchers said they couldn't determine which direction is correct, so neither direction can be inferred. D directly contradicts the passage. Notice that A, B, and C are all possible — but only B is forced by the passage's specific statement about the study's limitations.


Session 9 — The Three Rules
Ready to practice?

14 questions — 3 literary and 3 scientific in guided practice, escalating in timed section. Includes 2 questions where 3 of 4 choices are all possible but only 1 is provable from the text.

Open Session 9 Exercises →