01 Digit Span
Hear a sequence of digits and recite it back — forward at first, then in reverse once the difficulty climbs. The textbook benchmark for verbal short-term memory.
MindForge is a seven-exercise gym for working memory — the small, slippery buffer behind reading, conversation, problem solving, and everything else that asks you to keep a thought alive while another one arrives.
Each mode targets a specific component of working memory as described by the Baddeley–Hitch model. Modes unlock progressively as you level up — so the seventh tool only arrives once you can actually use it.
Hear a sequence of digits and recite it back — forward at first, then in reverse once the difficulty climbs. The textbook benchmark for verbal short-term memory.
Cells light up across a grid in sequence. Tap them back in order. The Corsi Block Tapping paradigm, adapted for a touchscreen and a sleepy Tuesday morning.
Hold a string of letters while you adjudicate live math prompts. Trains the executive — the part that juggles.
Letters arrive one at a time. Press when the current letter matches the one N steps ago. N rises and falls within a session, always at the edge of your span.
Press on green, freeze on red. The pacing tightens as you improve. Builds the inhibitory muscle — the one that stops you from typing the email you just thought of.
Sort cards by a hidden rule — color, shape, or count. The rule changes after you've gotten comfortable with it. Inspired by the Wisconsin Card Sorting Test; tracks perseveration so you can see when your brain wants to keep playing yesterday's game.
Study a long raw sequence, then reconstruct it from a chunked view. Teaches the single most useful memory technique we have — squeezing twelve items into four.
Working memory isn't a single thing. The most enduring model splits it into a few cooperating systems — and MindForge has an exercise for each of them.
Most adults can hold three to five chunks in active memory at once. Training won't move that ceiling much — but it sharpens the noisy buffer beneath it.
Adding a new item evicts an old one. Modes like N-Back drill that swap — the moment-to-moment refresh that decides whether you'll remember what someone is still saying.
After every session, accuracy above 80% nudges the difficulty up a notch. Below it, the mode holds. You spend your minutes at the edge of what you can do, where the gains are.
Training improves the tasks you train. Transfer to unrelated tasks is modest and contested. MindForge doesn't promise to make you smarter — it promises a precise gym for a precise muscle.
Working memory is the bottleneck behind focus, comprehension, and clear thinking under pressure. These are the people who tend to feel the difference most.
Holding a lecture, a formula, and the next question in your head at once. Daily reps make that juggling act feel less crowded.
Developers, writers, analysts, clinicians — anyone who loses the plot when interrupted mid-task. Train the buffer that picks the thread back up.
Working memory naturally dips with age. Short, deliberate practice keeps it engaged — the same logic as keeping any skill warm.
Frayed by notifications, recovering from burnout, or just out of practice at sitting with one task? Begin at an easy level and let it grow with you.
XP grows from real performance, not from showing up. Streaks compound, difficulty climbs, and unlocked modes stay unlocked — but only when your accuracy actually earned them.
Every correct answer is worth ten XP base. Consecutive correct answers stack a multiplier — a long run of right answers is worth far more than the same answers spread across mistakes.
Levels follow a ×1.35 scaling factor up to 50. The next level is always visibly closer than the one after — so the progress bar never lies, and there are no hidden cliffs.
Hit 80% accuracy in a session and the mode bumps up — longer sequences for Digit Span, faster pacing for Go/No-Go, shorter windows before a Rule Switch flip. Below 80% it holds steady. You never grind the same wall twice.
No accounts. No sign-in. No analytics. No third-party SDKs. No network calls at all. Save data is a plain JSON file in your app sandbox, written atomically so a crash mid-game won't corrupt it.
No setup, no account, no plan to choose. Open it and play — the app handles the rest.
It's the one mode open at level 1. Digits flash one at a time; type them back in order.
Short, consistent sessions beat the occasional marathon. Frequency is what moves working memory.
Score 80% in a session and it nudges up a notch; below that, it holds. You always train at the edge of your span.
Earn XP, level up, and the other six modes open one by one — pacing the new tools to when you can use them.