MindFrame
Metacognition training — not a brain game

Train how you think.

MindFrame scores three things most tools ignore: not just whether you're right, but how you reasoned, and how well you knew your own confidence.

19
training modes
660
authored challenges
Beta
free while in public beta

Free plan · No credit card · All 19 modes included

Plus 5 free games, no sign-up: ASCENT, Evidence Showdown, Rule Shift, Signal Station and Memory Heist.

Choose a mode, see the estimated time, then launch straight into session setup.

Ten-second test

How sure are you — and are you right?

Did Cleopatra live closer in time to the Moon landing or to the building of the Great Pyramid of Giza?
50% · a coin flip99% · certain
Calibration chart: confidence on the horizontal axis, outcome on the vertical axis.confidence 50% → 100%rightwrong
Dashed line: perfect calibration.

5 games · free, no sign-up

Play one skill at a time.

Each game is built around one mental skill — seeking evidence, adapting when rules change, filtering noise, remembering a route. Signed in, every finished run counts toward your progress.

All games →

Measured, not marketed

What we can prove today, and what we can't yet

Evidence & limits →

660

Unique training challenges

as of 2026-09-29

16

Modes with a written challenge bank

as of 2026-09-29

19

Training modes

as of 2026-09-29

From the research, not from MindFrame

  • g = 0.63 academic performance after metacognitive strategy instruction · de Boer et al. 2018
  • g = 0.69 symptom change, metacognitive therapy compared with Cognitive Behavioral Therapy (CBT) · Normann & Morina 2021
  • 6–11% forecasting accuracy (Brier score) after under an hour of probability training · Good Judgment Project (Mellers et al. 2014)

Our own outcomes are not yet measured. MindFrame has not yet run a controlled study, so we do not claim it produces the effect sizes reported in the research. We will publish our own numbers, including weak ones, once we have them.

Product facts as of 2026-09-29. Precomputed from versioned sources; no per-visitor database query.

How a session works

One session, three readings.

A standard session takes about 4–5 minutes. You answer, rate your confidence before you see the result, and get three separate readings back.

✓

Were you right?

01

— the floor every tool measures.

⟁

Was your reasoning sound?

02

Explain your thinking and it is scored separately from the answer — by Claude for in-depth reasoning, by a published formula otherwise.

◎

Did you know how sure to be?

03

— the gap between how confident you felt and how often you were right.

Overconfident. Sure, and wrong more often than you expect.

Well-calibrated. Confidence matches accuracy. Learnable.

Underconfident. Right more often than you admit.

  1. 1

    Choose a mode and answer a challenge

    Metacognitive planning

  2. 2

    Rate your confidence before you see the result

    Self-monitoring

  3. 3

    Your reasoning is scored, not just your answer

    Reasoning evaluation

  4. 4

    See your calibration gap

    Calibration feedback

  5. 5

    Get one clear next move

    Metacognitive regulation

The full walkthrough →·What is calibration? →

Session Library

All 19 training modes, all free.

Each card opens a mode page with the skill it trains. The badge is a typical session length.

Bias Hunter

~4 min

Spot the cognitive error before it costs you.

View mode + start path →

Calibration Lab

~4 min

Make your confidence mean what it says.

View mode + start path →

Metacognitive Scan

~6 min

Name the meta-error, not the mistake.

View mode + start path →

Prediction Arena

~4 min

Stake a number. Watch it age.

View mode + start path →

Reframe Forge

~5 min

Replace distorted thoughts with grounded ones.

View mode + start path →

Decision Lab

~5 min

Make clearer calls under uncertainty.

View mode + start path →

Reflection Sprint

~5 min

Understand how you actually think.

View mode + start path →

Emotional Radar

~5 min

Notice when emotion is steering the wheel.

View mode + start path →

Contradiction Challenge

~5 min

Hold competing claims in view at once.

View mode + start path →

Signal vs Noise

~4 min

Separate what matters from what merely looks informative.

View mode + start path →

Memory of Thought

~5 min

Watch your reasoning change when new evidence lands.

View mode + start path →

Uncertainty Engine

~5 min

Reason clearly when the answer is not fully knowable.

View mode + start path →

Strategy Selector

~5 min

Choose the right cognitive tool for the problem.

View mode + start path →

Thinking Trap Escape

~5 min

Recognize the trap, then get out of it.

View mode + start path →

Assumption Audit

~5 min

Surface the hidden premise everything depends on.

View mode + start path →

Argument Gym

~9 min

Steelman the opposition, then update with precision.

View mode + start path →

Focus Guardian

~4 min

Defend your attention before the distractors feed.

View mode + start path →

Attention Ladder

~5 min

Climb from seconds of focus to sustained attention.

View mode + start path →

Distraction Gauntlet

~5 min

Reason while interruptions try to hijack the room.

View mode + start path →

Every mode and game in depth →

What sets MindFrame apart

Not a brain game. Not EQ training.
Metacognition training — the upstream skill.

Brain games

Trains speed on the trained task

Improvement tends to stay on the trained task (Simons et al. 2016).

EQ training

Trains emotional awareness

Adds little beyond personality in predicting performance (Joseph & Newman 2010).

MindFrame

Trains metacognition

Scores your reasoning and how well your confidence matches your accuracy, and shows your own trend over time.

Two ways in

Train a skill, or play one.

A training session takes about 4–5 minutes and scores your reasoning and calibration. A game takes a few minutes and trains one skill. Both are free.

✓ Free plan, no credit card✓ 19 training modes✓ 5 games✓ 3 AI coach messages a day free✓ Your data stays private