What is a reasoning pattern, and why it matters more than the right answer
By The Sophoz Curriculum Team
The short answer. A reasoning pattern is the recurring way a child approaches a kind of problem, independent of whether the answer came out right. Two children can both score 38 out of 60 and have completely different reasoning patterns, which means completely different things need to happen next. The reason patterns matter more than scores is that a pattern travels: a child who hunts for keywords instead of reading the structure will do it in fractions, in ratio, in geometry and in the Achievers Section. Fix the topic and you have fixed one topic. Fix the pattern and you have fixed forty.
The term comes up increasingly in adaptive learning marketing without ever being defined. This is the definition, with examples you can recognise on a real marked paper.
The definition, and why the right answer hides it
A score records outcomes. A reasoning pattern records method.
Consider two children answering the same question: a recipe for four people needs 300 g of flour, how much for six people.
Child A writes 450 g. They thought: six is one and a half times four, so one and a half times 300.
Child B also writes 450 g. They thought: the difference between four and six is two, and two is half of four, so add half of 300.
Both are right. Both get one mark. But Child B's method works here by coincidence and will fail the moment the numbers are eight and twelve, where the difference is four and the ratio is still one and a half. Child B has a reasoning pattern that produces correct answers on a subset of cases and will collapse without warning.
Now reverse it. Two children both answer 500 g, which is wrong.
Child C used the correct method and made an arithmetic slip.
Child D added 200 because they saw two extra people and two hundred felt proportionate.
Same mark. Child C needs nothing except more care. Child D needs the whole idea of proportion.
This is the central point. The mark is identical in each pair and the required response is completely different. A system, a tutor or a parent who works from scores is working from the least informative thing on the page.
Six reasoning patterns you will recognise
These are patterns rather than topic errors, which means each one shows up across many topics at once.
1. The keyword hunter
What it looks like. The child scans the question for a trigger word and acts on it. "Altogether" means add. "Left" means subtract. "Each" means divide. They never read the structure of the situation.
Where it shows up. Everywhere in word problems, and it is the single most common pattern in classes 3 to 6.
Why it survives so long. It works on most school questions, because school questions are written to be clear. Olympiad questions are written to be unfamiliar, and they routinely include the word "left" in a question that requires multiplication.
How to spot it. Give a question containing a misleading keyword. If the child produces the operation the keyword suggests rather than the one the situation requires, that is the pattern.
The fix. Before any working, the child says what is happening in the situation in one sentence without using any numbers. "Some people shared sweets and then more people arrived." The retelling defeats keyword matching, because keywords do not survive paraphrase.
2. The formula-first
What it looks like. A formula is reached for before the situation is checked. Length times breadth applied to a shape that is not a rectangle. The average formula applied to groups of different sizes.
Why it happens. The formula was learned as a procedure rather than as a statement about a situation. A child who knows that the area of a rectangle is length times breadth, but not why, has no way to judge whether this shape qualifies.
How to spot it. Ask what the formula requires before they use it. A child with this pattern cannot answer, or answers with the formula again.
The fix. A four-second check spoken aloud before every formula. "This works when all four corners are right angles. Does this shape do that?" The check is the skill.
3. The one-pass solver
What it looks like. Every question is attempted once, forwards, and whatever emerges is the answer. Nothing is checked, nothing is worked backwards, no estimate is made first.
Where it shows up. As scattered, unpredictable errors that look careless and are not. It is a structural absence of a step, not a lapse.
Why it matters more than it seems. Schoenfeld's research on mathematical problem solving argues that success depends on four things: knowledge, heuristics, control, meaning the decisions about what to pursue and when to abandon it, and beliefs. His claim is that control is what most distinguishes competent from incompetent problem solvers, not the stock of knowledge. The one-pass solver has knowledge and no control.
The fix. One compulsory habit: before writing the final answer, say whether it is roughly the right size. "About 450, and 450 is bigger than 300, which makes sense because there are more people." Ten seconds, and it catches a large share of everything.
4. The two-point pattern-guesser
What it looks like. In a sequence 1, 4, 7, 12, 19 the child sees two gaps of 3 and answers as though the rule is add 3.
Why it happens. Two data points support infinitely many rules, and school sequences almost always use the simplest, so the habit of verifying never forms.
The fix. A rule is not a rule until it has been tested against a term that was not used to build it. Find the rule from three terms, check it on the fourth.
5. The big-number freezer
What it looks like. A blank space on any question mentioning the 100th term, a ten-digit number, or a large arrangement.
What is actually happening. The child correctly works out that writing a hundred terms is not viable, and then stops, because nobody has taught them that the move is to write out the first five and look.
Why it is worth naming as a pattern. It is not a gap in knowledge. The child often has everything needed. It is a missing heuristic, and it converts a whole class of olympiad questions from impossible to routine once installed.
The fix. An explicit rule: when the number in the question is too big to handle, do the same question with a small number, then look at what happened.
6. The answer-recogniser
What it looks like. A child who has been through fifty worked solutions, follows every one, agrees with every one, and cannot start question one unaided.
What is happening. Reading a solution produces a strong feeling of understanding, and that feeling is recognition. Recognition is not what an exam tests.
The evidence. Students who read a passage then tested themselves recalled 56 % a week later; students who read it four times recalled 42 %, and predicted they would do better (Roediger and Karpicke, 2006). In the largest review of study techniques, rereading landed in the low-utility tier while practice testing was rated high (Dunlosky et al., 2013).
The fix. Read the solution once, close it, write it from scratch. Three solutions done this way beats fifteen read.
Why a pattern is worth more than a topic
Here is the argument in one line: topics are many and patterns are few.
A class 6 syllabus has perhaps forty topics. The six patterns above, plus a handful of others, account for a large share of the errors across all forty. A child with the keyword-hunter pattern loses marks in fractions, ratio, time, money, measurement and geometry, and a tutor working topic by topic will fix each of those separately, slowly, and watch the pattern reappear in the next topic.
This is also why a pattern is the thing worth reporting to a parent. "68 % in fractions" is a fact about one topic in one week. "She reads for keywords rather than structure, which is why word problems go wrong even when the arithmetic is fine" is a fact about how she works, and it stays true next month.
Patterns are visible in the data, if anyone looks
A worked example from real exam data, since this usually stays theoretical.
SOF publishes the section-wise marks of students sitting exactly on the Level 2 qualifying boundary. In the IMO 2025-26 data, those boundary students scored 12 to 15 out of 15 on the Achievers Section, the hardest part of the paper, while scoring as low as 5 out of 10 on Everyday Mathematics, the most routine part.
Strong on the hardest section, weak on the easiest. That is a reasoning pattern visible at population scale: these are students who engage well with unfamiliar structure and lose marks on familiar, applied, multi-step arithmetic where care rather than insight is required. A cohort that looks like this does not need harder problems. It needs the one-pass-solver fix.
No preparation material we have seen reads the published data this way, and the data is free.
How to find your child's patterns this week
You do not need software for this. You need one marked paper and twenty minutes.
Step 1. Sort every wrong answer into four buckets: misread the question, arithmetic slip, wrong method, genuinely did not know. Have the child do the sorting out loud, before seeing the score.
Step 2. Look for repeats. One misread is a lapse. Five misreads is the keyword hunter. One arithmetic slip is a slip. Five slips spread across easy questions is the one-pass solver, because someone who checks catches most of their own slips.
Step 3. Ask "how did you get that" on three questions they got right. This is the step people skip and it is where the coincidence-correct answers surface. Child B above is invisible on any marked paper.
Step 4. Name the pattern out loud, without blame. "You go straight for the operation the word suggests. Let's read the situation first." A named pattern is a thing to work on. An unnamed one is just a feeling of being bad at maths, and that is expensive: research finds maths anxiety consumes the working memory a child needs to do mathematics, with the effect strongest among children with higher working memory (Ramirez, Gunderson, Levine and Beilock, 2013). Naming the pattern is what stops a score becoming a self-description.
Where Sophoz fits, for transparency
Sophoz is built around this idea, which is why we have a stake in the definition and you should read accordingly.
Practically, it means three things in the product. Assessment looks at reasoning and time taken rather than only at correctness, because a right answer produced in four seconds and one produced in forty are not the same event. Progress is tracked across six levels of mastery, from Recall through Explain, Apply, Connect and Generalise to Teach, because being able to state a rule and being able to explain it to somebody else are different things, and the gap between them is usually where a pattern hides. And weekly parent reports quote what the child actually wrote and name the misconception in a sentence, rather than reporting a percentage.
None of which is required to do any of it yourself. The four-step sort above is free and works on any marked paper.
Frequently asked questions
What is a reasoning pattern in learning? The recurring way a child approaches a type of problem, independent of whether the answer was right. Two children can score identically with completely different methods, one of which will fail on the next question and one of which will not. A pattern describes the method, which is why it predicts future performance better than a score does.
Why does the right answer not tell you enough? Because a correct answer can come from a method that works by coincidence. A child who finds 300 g for four people becomes 450 g for six by adding half of 300 is right here and will be wrong when the numbers are eight and twelve. The mark is the same; the understanding is not.
What are common reasoning patterns in maths? Six that recur across topics: keyword hunting instead of reading structure, reaching for a formula before checking whether it applies, solving in one pass with no checking, guessing a sequence rule from two terms, freezing on large numbers instead of trying a smaller case, and recognising worked solutions without being able to reproduce them.
How do I find my child's reasoning patterns? Take one marked paper and sort every wrong answer into four buckets: misread, arithmetic slip, wrong method, did not know. Repeats within a bucket indicate a pattern rather than a lapse. Then ask "how did you get that" on three questions they got right, which is where coincidence-correct methods surface.
Why do reasoning patterns matter more than topic scores? Because topics are many and patterns are few. A child who hunts keywords loses marks in fractions, ratio, time, money and geometry. Fixing each topic separately is slow and the pattern reappears in the next topic; fixing the pattern improves all of them at once.
Can an app detect reasoning patterns? Some can. The test is what happens after a wrong answer: a system that serves another similar question has recorded an outcome, while one that checks an underlying prerequisite or names why that particular wrong answer was chosen has identified something about the method. Six concrete tests for this are in personalised learning app or just a recommendation engine.
SOF cutoff data verified 21 September 2026 at sofworld.org. Research citations are linked in full in what the research says about olympiad preparation.