SwapCode
SwapCode shows a row of shapes going in and the same shapes coming out in a new order. Your job is to name the rearrangement as a code of digits. There is nothing to guess: the answer is fully determined by what you see. What decides your score is whether you read the code the right way round, and whether you can keep a hidden middle row straight when two codes run one after the other.
What you see
The board has three rows. At the top is the input row: three to six shapes, with their positions numbered 1 to n above them. Below it is the output row: the same shapes, rearranged, with their slots numbered underneath. Under that sits the code: one digit per output slot. Each digit of your code is drawn as a wire from its input position down to its output slot, so you can see at a glance whether the shapes line up.
- The code
Digit j of the code names the input position whose shape ends up in output slot j. A code of 3 1 4 2 means: slot 1 takes the shape from input position 3, slot 2 takes input 1, slot 3 takes input 4, and slot 4 takes input 2.
The shapes in one trial are always different from each other (circles, squares, triangles, diamonds, stars and hexagons in outline, striped and solid), except in one deliberate case at level 8 described below. The generator never gives you a code that leaves every shape in place.
There are two ways to answer.
- Choose (levels 1 and 2). Three or four candidate codes are listed; pick one. The wrong candidates are built to catch specific mistakes: the code read backwards (its inverse), the right code with two digits swapped, and the right code shifted along by one place.
- Build (level 3 upwards). The code row starts empty, with slot 1 selected. Click the input shape that fills the selected slot, or press its position number, and the selection moves to the next empty slot. You can click any slot to go back to it, and Clear (C) empties the whole code.
From level 6, some trials chain two codes in series. The input passes through the first code, producing a middle row, and the middle row passes through the second code to become the output. One code is given, printed above its wires; you deduce the other. The middle row's positions are shown as dots, but its shapes never are. You have to work them out.
At level 8 there are two examples that share the same code. In the first, one shape appears twice in the input, so that example alone cannot tell you which copy went where. The second example uses all-different shapes and settles it.
| Level | Shapes | How you answer | What makes it harder |
|---|---|---|---|
| 1 | 3 | choose from 3 codes | nothing extra |
| 2 | 4 | choose from 4 codes | at most 2 shapes stay put |
| 3 | 4 | build the code | at most 2 shapes stay put |
| 4 | 4 | build | every shape moves |
| 5 | 5 | build | at most 1 shape stays put |
| 6 | 4 | build | two codes, the first given, find the second |
| 7 | 4 | build | two codes, the second given, find the first |
| 8 | 5 | build | two examples, one with a repeated shape; every shape moves |
| 9 | 6 | build | two codes, the first given; every shape moves |
| 10 | 6 | build | two codes, the second given; every shape moves |
A shape that stays put is a free digit: if the star sits in position 3 in both rows, slot 3's digit is 3. Levels 4, 8, 9 and 10 remove those gifts entirely.
In practice mode there is no countdown. A clock counts up in the side rail, sessions start at level 3, and you press Check answer (or Enter) when the code is complete.
What it measures
SwapCode is a deductive task: unlike induction, there is no rule to discover and no ambiguity to resolve. Every trial has exactly one correct code, and it follows with certainty from the input and output rows. What varies is how many relationships you must hold and combine at once. On our profile it feeds the Rule learning facet of Reasoning and carries 9 percent of the Reasoning dimension score. Internally the game is weighted 60 percent deductive reasoning and 40 percent working memory.
The best research lens for SwapCode is relational complexity, from Graeme Halford and colleagues. Their argument is that the limit on reasoning is not how many items you can hold but how many related dimensions you must process together. They propose a soft limit of about one quaternary relation (four related dimensions) processed in parallel, and they describe two ways people get past it: segmentation, breaking a task into steps that each fit within capacity, and conceptual chunking, collapsing several dimensions into one.
Halford, Cowan and Andrews later argued that working memory and reasoning share this capacity limit, based on how many bindings between elements can be kept active at once. That is why SwapCode leans on working memory even though it looks like pure logic: every digit you enter is a binding between a slot and a position, and losing one mid-trace produces a transposition error.
How it is scored here
The evaluator compares your code with the answer slot by slot.
- Choose mode: all or nothing.
- Build mode: an exact code is Correct. Otherwise you earn partial credit equal to the share of slots you got right. Four of five slots right is 80 percent. A code must use each position exactly once: one that repeats a digit is not a code and scores zero.
The evaluator also classifies wrong codes: the inverse (read backwards), a transposition (the answer with exactly two digits swapped), a cyclic shift, or a code that repeats a digit. Your session report shows Positions deduced (the share of slots right) and Distance from the right order, which counts how many out-of-order pairs your code contains, a guide to how many neighbouring swaps would have fixed it.
Difficulty adapts after each trial. An exact code moves you up a level. A wrong pick in choose mode, or a built code with no slots right, moves you down. A Partly correct built code holds you at the same level. Your 0–100 SwapCode score is centred so that 50 means what the difficulty you work at predicts; the difficulty credited to a trial rises with the length of the code. Time is not part of the score, so careful tracing costs you nothing. A score appears after at least four trials, and recent sessions weigh more than old ones.
Where you will meet it
SwapCode belongs to a family of code-deduction tasks that is common in Indian campus hiring, because it appears in game-based aptitude rounds.
Publicly reported uses. Formats change, so check the latest invite you receive.
- AonswitchChallenge (deductive reasoning): find the code that changes the input row of shapes into the output row; harder items chain codes across rows, with the output of the first code feeding the next.
- CapgeminiCandidates report a Switch Challenge among the four games drawn from a pool of 24 in the game-based aptitude round.
Aon's own practice guide describes the same reading of the code that SwapCode uses. Its worked example shows four shapes, a code of 3-2-4-1, and an output whose first slot holds the shape from input position 3. It notes that a code of 1-2-3-4 would mean nothing moved, that there is no set number of questions, and that in harder items the output of the first code is the input for the next. The guide says the real assessment is timed with an open-ended number of questions, so the aim is to answer as many as you can accurately; its practice items each offer three candidate codes.
The method
- Pull, never push
Start from the output. For output slot 1, ask: which input position did this shape come from? Write that number. Then slot 2, and so on. Never start from an input shape and ask where it went; that produces the inverse code.
- Take the free digits first
Scan for shapes that sit in the same position in both rows. Their digit equals their own position. At levels 4, 8, 9 and 10 there are none, which is itself useful: if you have written a digit equal to its own slot there, you have made an error.
- Check it is a permutation
When the code is full, every digit from 1 to n must appear exactly once. A repeated digit means two slots claim the same input shape; one of them is wrong.
- Chained trials: write the middle row
If the first code is given, apply it to the input to get the middle row, then solve the second code from middle to output. If the second code is given, run it backwards: for each output slot, the second code's digit tells you which middle position that shape occupied. Fill the middle row that way, then solve the first code from input to middle.
- Repeated shapes: use the clean example
At level 8, solve every slot you can from the first example, then settle the repeated shape's slots from the second example, where every shape is different.
- In choose mode, test one slot
Read the first slot of the output, find its input position, and cross out every candidate that does not start with that digit. Usually one or two slots are enough.
Worked example 1: a single code
Show the solution
Pull, slot by slot.
- Output slot 1 is the triangle. The triangle is in input position 3. Digit 3.
- Slot 2 is the circle, from input position 1. Digit 1.
- Slot 3 is the star, from input position 4. Digit 4.
- Slot 4 is the square, from input position 2. Digit 2.
The code is 3 1 4 2. Every digit from 1 to 4 appears once, and no shape stays put.
If you had pushed instead (the circle went to slot 2, the square to slot 4, the triangle to slot 1, the star to slot 3) you would have written 2 4 1 3. That is the inverse, and in choose mode it will be sitting in the list waiting for you.
Worked example 2: the first code given
Show the solution
First build the middle row by applying the first code (2 1 4 3) to the input, pulling as usual.
- Middle slot 1 takes input 2: square.
- Middle slot 2 takes input 1: circle.
- Middle slot 3 takes input 4: star.
- Middle slot 4 takes input 3: triangle.
Middle row: square, circle, star, triangle. Now solve the second code from middle to output.
- Output slot 1 is the triangle: middle position 4.
- Slot 2 is the star: middle position 3.
- Slot 3 is the square: middle position 1.
- Slot 4 is the circle: middle position 2.
The second code is 4 3 1 2.
The trap is to skip the middle row and read the output straight against the input. That gives 3 4 2 1, which is the combined effect of both codes, not the second code.
Worked example 3: the second code given
Show the solution
The second code is known, so run it backwards to recover the middle row. The second code's digit for each output slot says which middle position that shape came from.
- Output slot 1 (star) came from middle position 2, so middle 2 is the star.
- Output slot 2 (circle) came from middle 3, so middle 3 is the circle.
- Output slot 3 (square) came from middle 4, so middle 4 is the square.
- Output slot 4 (triangle) came from middle 1, so middle 1 is the triangle.
Middle row: triangle, star, circle, square. Now solve the first code from input to middle: triangle is input 3, star is input 4, circle is input 1, square is input 2.
The first code is 3 4 1 2. Check: applying 3 4 1 2 and then 2 3 4 1 to circle, square, triangle, star gives star, circle, square, triangle, the output shown.
Tips and traps
Tracing by feel
- Follows where each input shape went
- Maps input straight to output on chained trials
- Fills slots in whatever order catches the eye
- Submits without checking the digits
Tracing by procedure
- Pulls: for each output slot, finds its source
- Writes out the hidden middle row, then solves one code at a time
- Works left to right and lets the selection advance
- Checks every digit 1 to n appears exactly once
Levelling up
Three things grow as the levels rise: the number of shapes (three to six), the loss of free digits (from level 4, and completely at 8, 9 and 10), and chaining (from level 6). Six-shape chained codes at levels 9 and 10 are the hardest trials: six middle positions to hold, no fixed points, and two codes to keep apart.
- At five and six shapes, say each digit as you enter it: slot four, from two. Speaking the binding protects it.
- At chained levels, always build the middle row explicitly, even when you feel you could skip it. The time it costs is smaller than the time a wrong answer costs you in level.
- When the second code is given, practise the backwards step on its own until it is fluent. It is the least familiar operation in the game and the most common source of errors at level 7 and above.
- Week 1
- Three sessions of 12 to 15 trials, saying the pull question aloud for every slot
- Before checking each answer, confirm every digit 1 to n appears once
- Note any inverse errors in the session report and rework them by hand
- Week 2
- Three sessions aiming to reach level 6
- On every chained trial, write the middle row on paper before touching the code
- Hand-solve two second-code-given puzzles a day using worked example 3 as a template
- Week 3
- Three sessions holding level 7 or above
- Try one session writing the middle row in your head instead of on paper
- Finish each session with two quick choose-style drills: check the first slot of each candidate only
Check yourself
- Input: circle, square, triangle. Output: square, triangle, circle. What is the code?
- In a build-mode trial with five shapes, you get three slots right and two wrong. What happens?
- The first code is 3 1 2 and the input is circle, square, triangle. What is the middle row?
and 2 more
Sources
- Aon. Practice Tasks: Deductive Logical Reasoning (switchChallenge). PDF
- Aon. Prepare for your online assessment: Deductive Reasoning (switchChallenge). aon.com
- Halford, G. S., Wilson, W. H., & Phillips, S. (1998). Processing capacity defined by relational complexity: Implications for comparative, developmental, and cognitive psychology. Behavioral and Brain Sciences, 21(6), 803–831. PubMed
- Halford, G. S., Cowan, N., & Andrews, G. (2007). Separating cognitive capacity from knowledge: A new hypothesis. Trends in Cognitive Sciences, 11(6), 236–242. PubMed
- PrepInsta. Capgemini Exceller game-based aptitude test questions (candidate-reported). prepinsta.com
Put it into practice: SwapCodeSign in to try it