NNAT3 Level D cheat sheet

A 2x2 or 3x3 grid is a matrix: the missing cell must satisfy every row rule and every column rule simultaneously, not just one of them.

Row rule and column rule are usually the same transformation applied along both axes; confirm a candidate by checking it against the row neighbor and the column neighbor separately.

Common transformations: color inversion (black↔white), rotation by 90/180/270 degrees, mirror flip across vertical or horizontal axis, size change, and count change.

When the grid shows a shape in cells 1 and 2 and a modified shape in cell 3, the rule is 'cell 3 = cell 2 transformed'; the missing cell repeats that same transformation on its predecessor.

Check the frame or border as a separate attribute from the interior fill; the correct option must match both attributes, and distractors typically match one only.

Attribute checklist for each candidate: outer shape, inner shape, fill pattern, orientation, number of elements, and position within the cell.

In 3x3 matrices, diagonal consistency is a strong confirmation signal but is not required; rely on row and column rules first.

If two options survive the row rule, apply the column rule to break the tie; only one option satisfies both.

Format is A : B :: C : ? - the relation from A to B must be applied identically to C to produce the answer.

State the A→B relation in words before looking at the options; this prevents being pulled toward visually attractive distractors.

Relation types: rotation, reflection, color inversion, size scaling, element addition or removal, shape substitution, and position shift.

Direction matters: A→B is not the same as B→A. If A grows and B shrinks, then C shrinking means the answer grows.

Check whether the relation is applied to the whole figure or to one component; mixed figures often transform only the inner or only the outer shape.

Distractor families to reject: the inverse relation (B→A applied to C), the identity (C repeated unchanged), and a partial transformation (only one attribute changed).

When both figures in a pair contain two shapes, verify that the relation holds for each shape independently before accepting the option.

A valid answer must be unique under the stated relation; if two options fit, the relation was described too loosely - restate it more precisely.

A series is a sequence of 3-5 figures with one missing; the rule advances by a constant step, so the missing figure is determined by its position in the sequence.

Quantify the step: count elements, count sides, count dots, or measure rotation angle between consecutive figures, then extrapolate.

Rotation series typically step by 45 or 90 degrees per figure; the total rotation across n steps is n times the step, so reduce modulo 360.

Alternating series: two interleaved rules (odd positions follow one rule, even positions another); identify which subsequence the missing figure belongs to.

Progressive series: the step itself changes (for example +1, +2, +3 elements); compute the second difference to confirm an arithmetic progression.

Shading cycles are common: a fill pattern repeats with period 2 or 3, so the missing figure's fill is fixed by position modulo the period.

Position cycles: an element moving clockwise around the corners of the cell returns to start after 4 steps; track it as a modular counter.

Verify the rule against every consecutive pair in the series, not just the pair adjacent to the blank.

Mental rotation preserves shape, size, and chirality; a rotated figure never becomes its mirror image.

Chirality test: an L-shape rotated stays an L; if the option is a mirror (reversed handedness), it is wrong regardless of orientation.

Folding problems: track which edges meet, count faces, and remember that a fold along a line reverses the side of the paper beyond that line.

Paper folding with punches: unfold step by step in reverse order, reflecting each hole across the fold line; the number of holes doubles at each unfold unless a hole lies on the fold line.

Cube nets: opposite faces never share an edge; faces separated by exactly one face in a straight line of the net are opposite.

Transparent vs opaque overlays: overlapping regions in transparent figures show the combined fill; in opaque figures the top layer hides the bottom.

Counting hidden blocks: a stack's height equals the number of visible top faces in that column; total blocks sum the column heights.

When comparing an option to the target, check three independent properties - outline, internal lines, and fill - before deciding.