Before, or after?
An assumption sits **before** a claim: it is the thing that must already be true for the claim to work at all. An inference sits **after**: it is what you may conclude once the claim is granted. They point in opposite directions, they are routinely confused, and a question that asks for one will always offer you a good example of the other.
The direction test
Take a claim: the new studio will improve student work, because it has better light.
An assumption — what must already be true? That light affects the quality of work. If that is false, the whole claim collapses. Assumptions are load-bearing: remove one and the argument falls over.
An inference — what follows if we grant it? That the old studio's light was worse. That is a consequence, not a foundation. Removing it hurts nothing.
So the test is direction, and it is mechanical: try denying the option. If denying it destroys the argument, it is an assumption. If denying it merely contradicts a consequence, it is an inference.
That single move — the denial test — separates the two reliably and takes about three seconds. It is far more dependable than asking yourself which one feels like an assumption, because both feel closely related to the passage. That is the point of the trap.
The other three traps, briefly
Scope. The passage discusses one studio; the option concludes something about all studios. Watch for quantifiers appearing that were not in the passage — every, always, never. A conclusion wider than its evidence is wrong however sensible it sounds.
Causation from correlation. The passage says two things happened together; the option says one caused the other. Extremely tempting, almost never supported. Two things co-occurring is compatible with either causing the other, or with a third thing causing both.
Outside knowledge. The option is true — you happen to know it is true — but the passage never said it. This is the cruellest one, because being informed works against you. The question is what the passage supports, not what is the case. A well-read candidate is more vulnerable here than an ignorant one, which is worth knowing about yourself.
All three are versions of the same failure: concluding more than you were given. That is this entire module in five words.
The feeling of obviousness is a signal
Here is the habit that ties the module together, and it is slightly uncomfortable.
In this area, the option that feels obviously right is disproportionately likely to be the trap. That is not cynicism; it is design. Setters build distractors from exactly the intuitions that usually serve you well — probability, causation, sensible scope, common knowledge. A well-built distractor has to feel right, or it would not distract anybody.
So train the flinch. When an option feels immediately, comfortably correct in a reasoning question, spend three extra seconds asking why it feels that way. If the answer is because it is probably true or because that is how the world works rather than because the passage forces it, you have found the trap rather than the answer.
And the reverse holds. The correct answer in these questions is often the dull one — the narrow, hedged, almost trivial option that says less than you want it to. Boring and forced beats interesting and probable, every time, in this one area.
The rules behind this
Sourced to the official brochure rather than restated here, so there is one place to correct when the Council revises it.
Part B examines six named areas: Visual Reasoning, Logical Derivation, General Knowledge/Architecture and Design, Language Interpretation, Design Sensitivity and Thinking, and Numerical Ability.
Visual Reasoning — understanding and reconstructing 2D and 3D composition. Logical Derivation — decoding a situation or context and drawing conclusions. General Knowledge, Architecture and Design — current issues, important buildings, historical progression, innovation in materials and construction. Language Interpretation — meaning of words and sentences, English grammar. Design Sensitivity and Thinking — observing and analysing people, space, product, environment; semantics, metaphor, problem identification. Numerical Ability — basic mathematics and its association with creative thinking; unfolding space using geometry.
Source · verified 2026-07-16
Part B allows 108 seconds per question, presented one after another, on an adaptive engine.
90 minutes across 50 questions. The adaptive structure dates to NATA 2025 per the President's foreword in V2.0, which states that NATA 2026 continues it.
Source · verified 2026-07-16
Part B questions appear one after another at 108 seconds each. There is no evidence of a review screen.
The skip-flag-and-return habit that works in JEE does not transfer. Budget the 108 seconds and commit.
Source · verified 2026-07-16
What almost everyone believes
“If I know the option is true, it must be a valid inference from the passage.”
The question is what the PASSAGE supports, not what is true in the world. Outside knowledge is a trap, not evidence.
This is the cruellest distractor because being informed works against you: a well-read candidate recognises a true statement and accepts it, while the passage never supported it. The exam is testing whether you can confine yourself to the given evidence — which is exactly the discipline of not concluding more than you were given, the failure underlying every trap in this area. When an option feels obviously right, ask why: if the answer is 'because I know it is true' rather than 'because the passage forces it', you have found the trap.
Depending on how long you have
Foundation
Understand the skill. Months out, or starting from zero.
Practise the denial test on real arguments — adverts, opinion columns, anything that argues. Ask what must already be true for this to work. It is a genuinely useful life skill and it happens to be examinable, which is a rare combination.
Drill
The practice protocol. What to repeat, how often, how to score it.
Sort every option before answering: assumption (before) or inference (after)? Then check scope, causation and outside knowledge. Track which trap catches you most — for most people it is one specific type, repeatedly, and knowing which one is worth more than general practice.
Exam-Day
What to actually do under the constraint — 108 seconds, no instruments, one pass.
Read the stem for the direction word — assumption or inference — and note that the options will contain both. Deny the option to test it. Distrust anything that feels obviously right, and prefer the dull hedged option over the interesting one. Boring and forced beats interesting and probable.
Try it
Fifteen minutes. The denial test is the whole exercise.
- 01Take the claim: 'The new studio will improve student work, because it has better light.'
- 02Write three options: (a) light affects work quality, (b) the old studio had worse light, (c) all studios should be rebuilt.
- 03Deny each one. Which denial destroys the argument? That one is the assumption. (It is (a).)
- 04Which is merely a consequence? (b) — an inference. Which is out of scope? (c) — the passage discussed one studio.
- 05Now do this with three adverts or opinion columns. Ask each time what must already be true. That habit outlives the exam.
The short version
An assumption sits before a claim and is load-bearing; an inference sits after and is a consequence. Deny the option: if the argument collapses, it was an assumption. The other traps — scope creep, causation from correlation, outside knowledge — are all one failure, concluding more than you were given. And treat obviousness as a signal: distractors are built from the intuitions that usually serve you, so the correct answer in this area is often the dull, narrow, hedged one. Boring and forced beats interesting and probable.
That completes Logical Derivation. Take it to the mock, where the temptation to conclude too much has a clock on it.
Questions people actually ask
- What is the difference between an assumption and an inference?
- Direction. An assumption sits before a claim — it must already be true for the claim to work, and denying it collapses the argument. An inference sits after — it is what follows once the claim is granted, and denying it destroys nothing. Deny the option to test which you have; it takes about three seconds.
- Why do I keep choosing the wrong option in reasoning questions?
- Probably because the distractors are built from intuitions that normally serve you well — probability, causation, sensible scope, common knowledge. A good distractor has to feel right or it would not distract. When an option feels obviously correct, ask why: if the reason is that it is probably true rather than that the passage forces it, that is the trap.
- Can I use outside knowledge in a NATA reasoning question?
- No, and it actively hurts. The question asks what the passage supports, not what is true. Being well-read makes you more vulnerable here, because you recognise a true statement and accept it when the passage never supported it. Confine yourself to the given evidence.