What adaptive learning actually means
“Adaptive” has been attached to enough products that it has nearly stopped meaning anything. So here is a definition strict enough to be useful:
Definition
Adaptive learning is an approach in which what a student is given next — the content, the difficulty and the sequence — is decided from evidence of what that student currently understands, and is revised continuously as new evidence arrives.
The test of whether a system is genuinely adaptive is simple: can a wrong answer change the subject of the next question, not just its difficulty?
That test rules a lot out. A quiz that gets harder when you are right and easier when you are wrong is not adaptive learning; it is a difficulty dial. It moves a student up and down one ladder made of the same material. If the reason they are failing is three rungs below the bottom of that ladder, no amount of moving along it will help.
Three things a system can adapt
It helps to separate them, because products described the same way often do very different amounts of work.
Difficulty
Easier or harder questions on the same topic. Useful, common, and the weakest form — it assumes the topic is the right one.
Pace and volume
More practice where a student is shaky, less where they are secure. Respects the student's time, but still can't change direction.
Direction
Which concept the student works on next — including going backwards, to an earlier idea the current one depends on. This is the level that changes outcomes.
Level 3 is the one that requires a concept map. You cannot send a student backwards to the right place unless the system knows which concepts the failed question depended on, and which earlier concepts those depend on in turn. Without that map, a system's only honest option is to make the next question a bit easier and hope.
Why one path cannot fit thirty students
The strongest argument for adaptive learning is not that personalisation is pleasant. It is that the same wrong answer does not mean the same thing for two different students.
Three students get the same question wrong — a jacket marked ₹800 sold at a 25% discount — and all three answer 775.
Needs more of the same
Fractions and percentage are solid; discounts are new. More practice on discounts is exactly right.
Needs a different skill
Each step is fine in isolation; holding two steps together is not. More discount questions won't touch the real obstacle.
Needs to go backwards
Fractions were never fully secured. Discount practice assumes the one thing they cannot do, and will make things worse.
Give all three the same remedial worksheet and one is helped, one wastes the hour, and one is actively set back. That is the cost of a single path — and it is paid quietly, by the students least able to afford it.
What adaptive learning requires
Direction-level adaptation is not a feature that can be added to a question bank. It needs four things working together.
- A concept map of the subjectEvery concept, and every dependency between concepts, written down — per subject, per class. This is editorial work, not engineering.
- Questions tagged to conceptsTo the ideas a question actually requires, not the chapter it sits in. One question usually depends on several concepts, which is what makes a response informative.
- Evidence weighed across many responsesA single wrong answer is weak evidence — it could be a slip or a misread. A concept is judged on the pattern across everything that touches it.
- Continuous reassessmentA concept isn't closed because an explanation was shown. It's closed when later answers show it holding up — and reopened if they don't.
That fourth point is where a lot of adaptive systems quietly stop. Diagnosing once and then following a fixed remediation plan is not adaptive; it is a branching worksheet. The loop has to keep running.
How Sublogy adapts
In Sublogy, the adaptation is done by CUBALE — the Concept Understanding Based Adaptive Learning Engine — and it runs as a loop rather than a one-off diagnosis.
| A traditional path | An adaptive path in Sublogy |
|---|---|
| Same content for everyone | Student responses are read as evidence about concepts |
| Same sequence for everyone | Concept understanding is assessed across many questions |
| Same practice for everyone | The areas needing attention are identified, including foundational ones |
| Test at the end | Targeted practice on the earliest concept that isn't secure |
| Move on regardless | Reassessment — and if it hasn't held, the path adapts again |
A few specifics that matter in practice:
- Questions are chosen to match what a student has already mastered, so the work stays hard enough to be interesting and easy enough to finish. A student who can't complete anything stops learning and starts guessing.
- The engine can send a student backwards. If the real gap is fractions, the student gets fractions — even though the question they failed was about discounts.
- Every question carries a worked example, so a wrong answer becomes a lesson rather than a mark.
- Help is available at the point of confusion. Sublogy Buddy explains the step on screen in simpler words when a student is stuck mid-problem.
- Progress is visible to the student. Belts, White through to Black, so adaptation doesn't feel like being moved around by an invisible hand.
- The teacher sees the same picture. Which concepts need attention, how many students each affects, and who is working on what.
What we don't claim. CUBALE identifies patterns and detects likely gaps, then recommends areas for reinforcement. It adapts on the evidence available rather than claiming to know every misconception perfectly — which is exactly why the teacher stays in the loop and every recommendation is shown as something to look at, not a verdict.
Questions about adaptive learning
What is adaptive learning?
Adaptive learning is an approach in which what a student is given next — the content, the difficulty and the sequence — is decided from evidence of what that student currently understands, and is revised continuously as new evidence arrives. Done properly it changes which concept a student works on, not just how hard the next question is.
Is adaptive learning the same as adjusting difficulty?
No. Adjusting difficulty moves a student up and down a single ladder of the same material. Genuine adaptive learning can change the subject of the work entirely — sending a student who fails a question on discounts to practise fractions, because that is where the actual gap is.
Does adaptive learning mean every student learns something different?
Not different content — the syllabus is still the syllabus. What differs is the order, the amount of practice, and which supporting concept gets strengthened along the way. Two students reach the same destination by routes shaped around what each of them already understands.
What does adaptive learning need in order to work?
Three things: a map of the concepts in the subject and how they depend on each other, questions tagged to the concepts they actually require, and continuous analysis of every response. Without the concept map, an adaptive system can only change difficulty, not direction.
Does adaptive learning replace the teacher?
No. An adaptive engine is good at the part a person cannot do — holding thirty students, hundreds of weekly responses and dozens of interlinked concepts in view at once. Deciding what to do about a student is still a teaching judgement, which is why Sublogy surfaces recommendations to a teacher rather than acting as the final word.
A path built around each student, at class scale
Sublogy is provided through schools, for classes 6 to 10.
Sublogy for schools How CUBALE works