A Lab for One Skill at a Time

The Lab is organised into categories, and the unit of practice is a single named skill: working memory, perceptual speed, deductive reasoning, error detection and so on. There is no composite number at the end, and that is a refusal rather than a missing feature. A single figure combining twelve different things tells you nothing you can act on.

What you get instead is per-skill: attempts, your latest and best accuracy, and a practice band describing where you currently are. It is feedback on a specific thing you can go and practise.

Each lab explains what a question of that type is actually asking before it gives you one. In this kind of practice a large share of wrong answers come from misreading the task rather than from failing at it, and two minutes of reading removes most of them.

That ordering also changes what the practice is for. A drill that starts with questions measures you; a drill that starts with an explanation teaches you, and then measures whether the teaching took.

It is a slower start and it produces a better second session. Somebody who has read what a working-memory item is asking arrives at the next set with a strategy rather than with a memory of having found it hard, and strategy is the thing that transfers. Familiarity with a question format is worth very little on its own; knowing what the format is testing is worth a great deal.

The skills are grouped into categories covering reasoning, memory, attention, speed and the strategies people use to learn. Those groups are a navigational convenience rather than a claim that everything inside one of them rises and falls together, and the Lab does not report a category score for exactly that reason.

Some of these are also much more trainable than others within a few weeks of practice, and the honest position is that the evidence for transfer between them is contested. What is not contested is that being familiar with a question type improves your performance on that question type. That is what this section offers, stated at the size it actually is.

This is educational practice. It is not an assessment of you, it does not produce a score you can put on anything, and nothing in it is normed against a population. A percentile requires a representative norm sample and published reliability evidence, and inventing one from the people who happened to visit a website would be worse than useless.

So the only comparison the Lab offers is to your own earlier attempts, which is the comparison that is actually available and actually useful.

• Look at the direction across several attempts, not at the last number.
• One attempt is mostly noise: mood, time of day and how much of the instruction you read.
• A plateau on one skill while another improves is information, not failure.
• A drop after a long gap is normal and is not evidence about you.

There is no schedule imposed on you and no timer on the practice. Work a lab as often as you like, and stop when you stop learning from it. The absence of a clock is deliberate: speed under pressure is a different skill from the one being taught, and mixing them means neither is being practised.

That also makes the Lab the untimed route for anyone for whom timed conditions are a barrier rather than a challenge.

The band beside a skill describes where your recent attempts sit, in words rather than as a number with a decimal point. That is a deliberate choice of resolution. A percentage implies a precision the underlying data does not have, and it invites comparison between skills that are not on a common scale in the first place.

A band is also easier to act on. Knowing that your working memory practice is early-stage and your deductive reasoning is consolidating tells you where to spend the next half hour. Knowing that one is 63 per cent and the other 71 per cent tells you that one number is bigger than another number, which is a fact about the numbers.

Your history is in this browser. Clearing site data, private browsing, storage eviction or a different profile removes it, and there is no account copy to restore from. The downloads page exports the history as a paginated PDF, a spreadsheet you can filter, plain text, or JSON.

The spreadsheet is the format worth taking if you are practising over weeks: it drops straight into whatever you already use to track anything else, and it outlives the browser that produced it.