Interlaza

Reading the data

A criterion was reached. But was it learned?

How Interlaza separates practice, criterion, generalization, transfer and maintenance so a progress label never carries more meaning than the evidence.

For: Families

By INTERLAZA 4 min read Updated

Families deserve more than a green badge. They deserve to know what the badge means.

Suppose a child repeatedly matches a picture of one red cup to the same picture. That performance may meet the activity’s criterion. It does not yet tell us whether the child will identify a different cup, respond to the spoken word, use the relation at home or retain it next month.

The sentence the data supports is: the criterion for this task was reached. Then the program can ask the next question.

Figure — A criterion is a milestone, not the finish line

  1. Criterion reached

    The defined performance threshold was met in this task.

  2. Generalization

    The response appears with new examples, people or settings.

  3. Transfer

    The relation supports a different, relevant performance.

  4. Maintenance

    The performance remains after time has passed.

Each step answers a different question. Interlaza can document evidence at every step, but reaching the first does not prove the other three.

Practice is evidence of teaching, not always evidence of criterion

Prompts, fading and errorless procedures can make practice successful. That is exactly what they are for. But a prompted response should not silently become an independent one in the progress record.

Interlaza records the support attached to a response so an Instructor can distinguish two useful facts:

  1. the learner completed the teaching trial with a certain level of help; and
  2. the learner later responded independently under the scoring conditions.

Neither fact makes the other disappear. Together they show a path from support toward independence.

What the estimate contributes

For scored trial-based activities, Interlaza can maintain a probability estimate that updates as eligible independent evidence arrives. It is more informative than declaring success after a short streak because it carries the history of the target forward.

It is still a model, not a mind reader. Crossing a configured threshold means the model’s criterion was met for that target under those conditions. It does not certify language, understanding or everyday use.

That is why the most useful next step is often a probe: a few trials with no help and no telling the child whether they were right, changing one thing while the relation itself stays the same — a new picture, another voice, a physical object, a different room or a later date. Testing a different modality, like the spoken word after a purely visual match, is not always this kind of check: word-to-picture is its own relation in the app’s activity types, and it may need its own teaching before it can be probed this way rather than simply generalized to.

Some of these checks Interlaza can schedule on its own once configured — an unprompted transfer check after a supported response, or a comparison held back from teaching so it stays genuinely untrained. Changing the room, the physical object or the moment is still something an Instructor has to arrange; the app organizes and records the check, it does not stage the room.

What families should look for

Do not judge progress by speed alone. Look for a widening set of observations:

  • less help across comparable practice;
  • independent performance across more than one session;
  • success with examples that were not used in teaching;
  • the relation appearing in a useful context;
  • the child doing it again after some time has passed.

A chart only summarises what happened in the activities that were recorded. It does not show tiredness, the setting, or what the child does outside them. The value of Interlaza is that these questions can live in one record, with the limits of each result still visible.