Relatoria

Cognitive training, built from the research

Train whatthinking ismade of.

Reasoning, working memory, reading and processing speed: the processes beneath nearly every hard thing you do. Relatoria trains each one with methods taken from published studies, and tells you plainly what those studies found.

Start with Frames, all 55 stages of the SMART protocol, free forever.

Start training freeSee the evidence

Everything else free for 14 days. No card.

Frames/Stage 06 of 55

  1. Question 1 of 3
  2. Question 2 of 3
  3. Question 3 of 3

Premises

  1. ZIDis the opposite ofVEK
  2. VEKis the opposite ofLOF

Question

Is ZID the same as LOF?

In the app, that line is a 30-second clock. Here, take your time.

Three real questions from Frames. The words mean nothing. That is the point: the only way to the answer is to reason.

01The premise

Most brain training makes you better at brain training.

It is the field’s most consistent finding. Across 50 studies of Cogmed, the best-known working-memory program, gains were largest on tasks like the training, smaller on other memory tasks, and indistinguishable from zero on everything else.

So what you train decides almost everything. Relatoria doesn’t train games. It trains the few processes nearly every hard task leans on, with methods that were measured on tests the training never touched.

0.57
0.25
0.05

Tasks like the training

memory tasks resembling the trained ones

Other memory tasks

memory tasks unlike the trained ones

Everything else

intelligence, attention, language, math

Fig. 1Working-memory training across 50 studies of Cogmed: average gain right after training, with 95% confidence intervals. Aksayli, Sala & Gobet (2019), Educational Research Review.
Average effect of Cogmed working-memory training by outcome type (Hedges’ g, 95% CI)
Outcomeg95% CI
Tasks like the training (memory tasks resembling the trained ones)0.5660.472 to 0.66
Other memory tasks (memory tasks unlike the trained ones)0.2460.102 to 0.391
Everything else (intelligence, attention, language, math)0.048-0.017 to 0.113

Why processes? Because tasks share them.

Scores on very different tests rise and fall together. Process Overlap Theory explains why: every test draws on several cognitive processes, and the same processes turn up across many tests.

That overlap is the route a trained gain can travel. A trick learned for one game has one place to show up. A stronger process has every task that uses it.

Kovacs & Conway, 2016, Psychological Inquiry

Process Overlap Theory, explained
TaskReas.Mem.Verb.SpeedTrick
A matrix-reasoning test
Reading a dense report
Mental arithmetic
Keeping up in a fast conversation
Driving in heavy traffic
Learning a new subject
A brain-training game
Fig. 2Schematic. Each task draws on several processes, and each process serves many tasks. A trick learned for one game serves one.
02The programs

This is what the training looks like.

Real tasks built from the research, adaptive to your level, each with its evidence beside it. One subscription holds every one of them.

01Reasoning

Derive what was never said, from words that mean nothing.

Frames rebuilds the SMART protocol: 55 stages of same, opposite, more and less, from one premise to three. You move on only after 16 right in a row, then 16 of 16 on a test. Relatia picks up where the studies stop: nine relation families, a placement test to find your level, and no top level.

Frames
The SMART protocol, stages 1–55Free
Relatia
Nine relation families, adaptive, open-endedBeyond the studies
Manual
Choose the relations and up to ten premises

In a randomized trial where more than half the students finished all 55 stages, the trained group gained 20 verbal IQ points while an untrained group stayed flat. Across six controlled trials of SMART and its variants, trained groups gained about 8 verbal IQ points more than comparison groups.

Colbert et al., 2018 · Relatoria review of 6 trials, 2026

Relatia/2D Space

Level 11

Premises

  1. VOKis one step north-east ofTAL
  2. TALis one step straight north ofRIM
  3. RIMis one step straight west ofPEZ

Conclusion

VOK is north of PEZ

TrueFalse
NPEZRIMTALVOK

Nine relation families

In SMART

  • Opposition
  • Linear
  • Disposition
  • Stepped
  • Nodal
  • Tree
  • 2D Nodal
  • 2D Stepped
  • 2D Space
Fig. 3Relatia, 2D Space. Each premise places a term; the conclusion is checked against the picture.

02Working memory

The room you hold a thought in while you use it.

Remember is a Cogmed-style library of 39 modes: spatial patterns and tracking, spoken sequences in three recorded voices, bound associations, and fine color memory. Picture N-Back adds a harder kind: track a photograph and the window it filled, several steps back, across 2,522 scenes.

Remember
39 modes in four families, every setting open
Picture N‑Back
Scene and place at once, up to 8 back

Across 50 studies of Cogmed, working-memory training reliably improved memory, including memory tasks unlike the trained ones. It did not improve intelligence or grades.

Aksayli, Sala & Gobet, 2019

Remember/Four families

PresentHoldRecall

Visual

13 modes

Matrix

Auditory

8 modes

Sequence · 3 voices

Associative

8 modes

Triple Associates

Hue

10 modes

Color Flash

Fig. 4One mode from each of Remember’s four families: present, hold, recall.

03Reading

Read faster without losing the meaning.

Accelerate erases each sentence just behind your eyes, then asks what it said. Six right in a row and the eraser speeds up. Twenty-four sessions, in English, Spanish, French and Hebrew. Four shorter programs train the fast retrieval underneath: of meanings, words, facts and ideas.

Accelerate
24 sessions, 1,200 sentences, four languages
Retrieve
Same number or not, across 21 languages, at a glance
Recall
Find the missing word, across 16,040 sentences
Flash Fluency
Math facts until they are automatic
Diverge
Unusual uses for 120 everyday objects

Reading against a shrinking time limit beat the same practice untimed, for speed and for comprehension, and the gains held six months. It has since been replicated in English and Italian.

Breznitz et al., 2013 · Horowitz-Kraus et al., 2014

Accelerate/Session 07/24

46 ms/char
Englishעברית

Although her delivery arrived late,

the herb seller set up

before the gates opened.

What did the herb seller manage to do?

  • Get her stall ready only after the gates opened
  • Set up her stall in time
  • Sell yesterday’s herbs until the van came

כשהחל הגשם, גררו הסבלים

את שקי תפוחי האדמה

אל מתחת לסככה

לאן הועברו השקים?

  • למחסן קירור בקצה השוק
  • אל מקום מקורה
  • מהסככה אל השטח הפתוח שבמרכז השוק
Fig. 5Accelerate. The mask erases at a set rate; read ahead of it, then answer.

04Processing speed

Take in more of a single glance.

Dual flashes an animal at the center and a star at the edge, then wipes both with noise. Name one, place the other. Three right and the next flash is shorter; a miss and it lengthens. Forty levels add distance, distractors and lookalikes.

Dual
40 levels across five stages, flash from 500 ms down

Dual adapts the speed-of-processing task from the ACTIVE trial of older adults. We quote that trial’s outcomes only from primary papers we hold, and Dual has not been tested on its own.

ACTIVE · see the evidence ledger

Dual/Level 01

Flash 300 ms

Try it

That is the display, held still. Press Flash and it appears for 300 ms, then noise wipes it. Keep your eyes on the box.

Fig. 6Dual, playable. One real trial, with the app’s own staircase setting the flash.
03The evidence

What the studies found. And where they stop.

Brain training earned its reputation by overclaiming. Here is the study behind each program: who was tested, what moved, and how solid it is.

  • Meta-analysis
  • Randomized trial
  • Controlled study
  • Adapted from a trial
  • Not yet tested

  1. Frames

    Colbert et al., 2018 · Relatoria review of 6 trials, 2026

    6 controlled trials · 132 people, mean ages 9–22 · lead trial: 26 students

    Found

    In a randomized trial where more than half the students finished all 55 stages, the trained group gained 20 verbal IQ points while an untrained group stayed flat. Across six controlled trials of SMART and its variants, trained groups gained about 8 verbal IQ points more than comparison groups (95% CI −1 to 17).

    How solid it is

    All five gains came from trials run by SMART’s developers; the one independent trial, in which few finished, found none.

  2. Remember

    Aksayli, Sala & Gobet, 2019

    50 studies of Cogmed · children to older adults

    Found

    Better memory, on tasks like the training (g = 0.57) and on memory tasks unlike it (g = 0.25). Most of the gain was still there months later.

    How solid it is

    50 studies, 637 effect sizes. The gains stayed in memory: none on IQ or grades.

  3. Picture N-Back

    Rudebeck et al., 2012

    27 adults trained · 28 untrained

    Found

    After 20 days, trainers beat controls on a matrix-reasoning test of fluid intelligence by d ≈ 0.7–1.0. The test is among the most g-loaded there are and looks nothing like the training: gains this far from the training are rare (Fig. 1).

    How solid it is

    One study of 55 adults; the comparison group did no training.

  4. Accelerate

    Breznitz et al., 2013

    110 adults · 55 with dyslexia

    Found

    Faster reading and better comprehension than the same practice untimed; the gains held six months later.

    How solid it is

    Replicated since in English-speaking children (Horowitz-Kraus et al., 2014) and Italian adults (Zamfira et al., 2024).

  5. Retrieve

    Zhang et al., 2024

    60 students · active control

    Found

    About five hours improved working-memory updating, task switching and reading speed, none of which were trained.

    How solid it is

    Randomized and double-blind, against an active control.

  6. Diverge

    Sun et al., 2016

    40 participants · untrained control

    Found

    Twenty sessions made ideas more original than an untrained group’s.

    How solid it is

    Randomized, but small: 28 people completed it.

  7. Dual

    The ACTIVE trial

    Speed-of-processing training · older adults

    Found

    We quote the trial’s outcomes once its primary papers are in our library.

    How solid it is

    Dual adapts the trial’s task; it has not been tested on its own.

  8. Relatia, Frames 56+, Recall, Flash Fluency

    Our design

    Built on the methods above

    Found

    No trial yet.

    How solid it is

    Relatia and Frames 56+ extend SMART in the direction its researchers proposed.

The long version

04The details

The details are the training.

What separates a protocol from a game is mostly small rules, kept strictly.

  • Mastery, not minutes

    Frames moves you on after 16 right in a row, then a 16-question test you must pass without a miss. Time spent counts for nothing.

  • Difficulty that follows you

    The training adjusts as you answer: harder the moment you are ready, easier when you are not. Every session is spent at the edge of what you can do.

  • Answers that are proven

    Several of Relatia’s engines will not issue a question until a proof kernel has checked its answer: a derivation if true, a counterexample if false.

  • Material worth training on

    2,522 photographed scenes. Three recorded voices. Thousands of sentences in each of four languages.

  • Progress you can read

    Every session is recorded: level, speed, accuracy. Leaderboards rank confirmed levels, never a lucky run.

  • No ads. Ever.

    Not banners, not rewarded videos. Your attention is the thing being trained; nobody else gets to bid on it.

05Pricing

Try everything. Keep the protocol.

Every account starts with 14 days of the complete program, with no card. When the trial ends, Frames stays free forever. The rest is one subscription.

Free

Forever

$0

  • Frames: the SMART protocol, all 55 stages
  • Leaderboards, streaks and your full training history

The complete program

First 14 days free

/ year

That is a month.
Or billed monthly.

  • Remember: all 39 working-memory modes, every setting
  • Accelerate in four languages, plus Retrieve, Recall, Flash Fluency and Diverge
  • Dual processing-speed training, all 40 levels
  • Relatia, manual reasoning, and Frames stages 56+
Start the 14 daysFull pricing

Relatoria is free for anyone who can’t afford it. Request a scholarship.

For comparison

Working memory
Cogmed: often $1,500+ per program
Processing speed
BrainHQ: $14 a month
Relational reasoning
RaiseYourIQ: $140 a year
06Questions

Fair questions.

01

Does brain training work?

Mostly, it improves what you practice and little else, which is why most of it disappoints. A few methods have moved scores on tests they never trained: IQ after relational training, matrix reasoning after n-back, reading after acceleration. Relatoria is built on those, and the ledger above shows each study. Read the full case.

02

Is Frames the real SMART protocol?

Frames rebuilds the published SMART design: the same relations, the same 55-stage order from one premise to three, and a strict mastery rule. The trials used the original commercial software, so Frames is a faithful rebuild, not the tested program itself. Stages 56 and up are our extension.

03

How much time does it take?

Weeks, not days. The studies ran short, regular sessions: 24 sessions of 15–20 minutes for reading acceleration, two 45-minute sessions a week for twelve weeks for SMART. Relatoria is built for the same rhythm.

04

What is free, and what costs money?

Frames stages 1–55 are free forever, with leaderboards, streaks and your history. Everything else is one complete program, and every new account gets 14 days of all of it, with no card.

05

Who were the studies done with?

The SMART trials were mostly with school students, with a few in adults and older adults; reading acceleration, adults; the retrieval trial, university students; ACTIVE, older adults. The ledger names each population, so you can judge how close it is to you.

06

What if I can’t afford it?

Then it’s free. Relatoria offers scholarships with no proof of income and no explanation: pick the price that works for you, down to $0. Request a scholarship.

Start where the studies started.

Stage 1 of Frames is free, and so are the 54 after it. Everything else is yours for 14 days, with no card.

Start training free