MindFrame

How it works

The feedback loop that changes how you think

Metacognition is only trained through precise, immediate feedback on the gap between what you believed and what turned out to be true. MindFrame is built around five feedback loops — one for each dimension of metacognitive skill.

4–5-minute sessions
5 scored skills
AI-scored reasoning

The session flow

Every MindFrame session follows this exact loop. Each step is a deliberate metacognitive act.

01

Choose a mode and answer a challenge

Metacognitive planning

You decide where to train your thinking today. Choosing Bias Hunter signals intent to work on spotting distortions; choosing Calibration Lab means you are working on confidence accuracy. The choice is deliberate, and it shapes what the session measures.

02

Rate your confidence before you see the result

Self-monitoring

Before seeing any result you state how certain you are, from 50% (a guess) to 99% (near certain). Committing before feedback can colour it gives a clean signal of what you actually believed when you decided — the data point most training tools never collect.

Confidence scale

50%60%70%80%90%99%
← pure guess · near certain →
03

Your reasoning is scored, not just your answer

Reasoning evaluation

Getting the right answer is the floor, not the ceiling. When you explain your reasoning in depth, Claude scores the argument itself — gaps, unsupported jumps, sound inference. Shorter or routine answers are scored by a published formula, and your results say which one was used. Two people can reach the same answer with very different reasoning; MindFrame tracks both.

04

See your calibration gap

Calibration feedback

The gap between your confidence and your accuracy is your Calibration Error. Say 85% and be right 65% of the time on questions like that, and the gap is 20 points. Most people have never seen this number about themselves, and shrinking it is the main thing MindFrame trains.

Stated confidence

85%

Actual accuracy

65%

Overconfidence gap

20 pts

05

Get one clear next move

Metacognitive regulation

Your results name the skill that cost you most this session, the pattern behind your misses, and the one thing to work on next — with a recommended mode. Members also get an AI-written debrief that reads every answer, confidence rating and reasoning trace in the session.

The scoring system

Three readings, not one

Many training tools score one thing: whether you got it right. MindFrame reads three things from every scored answer.

Layer 1

Accuracy

Were you right?

Measured asShare of graded answers correct

Whether each answer matched an answer key written before the challenge was published.

Layer 2

Reasoning quality

Was your reasoning sound?

Measured asScored 0–100, AI or formula

Your explanation is scored separately from the answer: by Claude when you explain in depth, by a published formula otherwise. Your results say which one was used.

Layer 3

Calibration

Did you know how sure to be?

Measured asStated confidence against results

How closely the confidence you stated before seeing the result matched how often you were right.

Example session result

Outcome Score

78%

7 of 9 correct

Reasoning Score

64 / 100

AI-scored

Self-awareness

71 / 100

Calibration accuracy

The calibration metric

What is the Brier Score?

The Brier Score is the scientific standard for measuring probability calibration. Here is an analogy that makes it concrete.

🌦

Forecaster A

The overconfident one

Says “60% chance of rain” every single day. Accurate roughly 60% of the time. Sounds reasonable — but their forecasts contain almost no information. They never distinguish between a clear summer day and a building storm front.

High Brier Score — poor calibration. Predictions don't distinguish between certain and uncertain outcomes.

🌤

Forecaster B

The calibrated one

Says “95% chance of rain” when a storm is imminent, “5% chance” on a cloudless day. Both forecasters might be right the same number of times — but Forecaster B tells you something useful every time they speak.

Low Brier Score — excellent calibration. Predictions carry real information and update with evidence.

In MindFrame: your Brier Score is computed across every confidence rating you make in a session. A Brier Score of 0 is perfect calibration. A score of 1.0 means you were maximally wrong about how certain you were. Most untrained people start between 0.15 and 0.30. MindFrame tracks your improvement across sessions.

The long game

How MindFrame compounds over time

A single session shows you the gap. Repeated sessions reveal the pattern. Consistent practice is designed to sharpen how accurately you know what you know — and MindFrame shows you whether it is happening for you.

Session1

You see the gap

Your first calibration score: how far your stated confidence was from how often you were right. This number is the baseline everything else is measured against.

Session5

You see the pattern

Five sessions generates enough data to surface a pattern — which types of questions you systematically overestimate, which modes expose your weakest dimensions, where your reasoning breaks down.

Session20

You have a cognitive fingerprint

Your session scores roll up into an eight-axis Cognitive Fingerprint — shareable and specific to you — and you can compare your calibration error today with session 1.

Five scored skills

What MindFrame actually trains

Each dimension is independently measured, independently scored, and independently coached — because they can diverge. You can be highly calibrated but reason poorly. You can reason well but fail to update beliefs.

◎

Calibration

Matching confidence to reality

The ability to know how likely you are to be right before you get the answer. High calibration means your 80% feels like 80%, not 60% or 95%. Measured via Brier Score and calibration error across sessions.

⟁

Reasoning Quality

The logic behind your conclusions

The structural validity of your arguments. Good reasoners identify the right evidence, weigh it correctly, and avoid unsupported jumps. AI-scored on every session.

⊘

Bias Recognition

Catching systematic errors before they compound

The ability to spot cognitive biases — confirmation bias, availability heuristic, anchoring, framing effects — in your own thinking and in presented arguments. Trained through Bias Hunter and related modes.

↻

Belief Updating

Changing your mind in proportion to evidence

The ability to revise your beliefs when evidence warrants it — neither too much nor too little. Trained through Contradiction Challenge and Memory of Thought modes.

⊡

Working Memory Utilization

Using your attention efficiently

How effectively you use limited attentional resources. High utilization means you hold the right information at the right time, avoid cognitive overload, and structure complex problems well.

✦

All five together

Your Cognitive Fingerprint

After 20 sessions, these five skills feed an eight-axis profile — your Cognitive Fingerprint. Shareable, measurable, specific to you.

Start your first session

See your calibration gap — free

Your first session takes about 4–5 minutes. You will see your Brier Score, your Calibration Error, and an AI debrief of your reasoning. No credit card required.

Try it now

One challenge. One confidence rating. No sign-up.

This is step 2 of the session loop — the attempt + confidence capture. Experience the thing every explanation above is talking about.