DOCUMENTATION

How the Core Brain test works

Most online IQ tests inflate scores to maximise sharing. Core Brain doesn't. This is the methodology behind the test — what it measures, how it scores, and what the result actually means.

The IRT 3PL methodology

Core Brain uses Item Response Theory (IRT) — specifically the 3-parameter logistic (3PL) model — to estimate your latent cognitive ability (θ). This is the same psychometric framework used by major standardised tests including the WAIS-IV, the SAT, the GRE, and modern certification exams.

Every question in the Core Brain item bank has three calibrated parameters:

• a — discrimination (how sharply the question separates higher-ability from lower-ability test-takers)
• b — difficulty (the ability level at which a test-taker has a 50% chance of answering correctly)
• c — guessing floor (the probability a low-ability test-taker gets it right by chance — for 4-option multiple choice, ≈25%)

Your final score is computed via Bayesian Maximum A Posteriori (MAP) estimation under a standard normal N(0, 1) prior. Newton-Raphson iteration solves for the θ value that best explains your specific pattern of right and wrong answers. The estimate is clamped to ±4 SD (effectively IQ 40-160) to prevent extreme outliers from broken responses.

The six cognitive axes

Rather than reporting a single composite IQ, Core Brain scores you across six cognitive dimensions independently:

1. Pattern Recognition — spotting structural rules in sequences (colours, shapes, numbers).
2. Spatial Reasoning — rotating, folding, and re-orienting objects in your head.
3. Logical Deduction — composing rules into multi-step inferences.
4. Verbal Reasoning — analogies, vocabulary, conceptual relationships.
5. Numerical Reasoning — quantitative problem-solving with applied math.
6. Abstract Matrix Reasoning — Raven's-style abstract visual problems (the purest fluid-intelligence measure).

Per-axis sub-scores require a minimum of 2 items on that axis and are scaled to a 0-100 range so they're directly comparable to each other.

How your score is calculated

After you finish the test, the scoring pipeline runs:

1. Collect your responses with each item's a/b/c parameters
2. Run 60 iterations of Newton-Raphson MAP estimation to find θ̂
3. Map IQ = 100 + 15 × θ̂ (standard normal scale)
4. Soft-clamp to your chosen difficulty's expected band ±15 (prevents one-off lucky/unlucky session noise)
5. Compute Fisher information at θ̂ → standard error of measurement (SEM)
6. Report a 95% confidence interval as IQ ± 1.96 × SEM
7. Compute reliability as 1 − σ_error²

How to interpret your result

• The number is an estimate, not a measurement. The 95% CI is the honest envelope around your true ability.
• Sleep, stress, and test familiarity each shift single-session scores by 5-10 points. If your score feels surprising, retake under different conditions.
• IQ predicts roughly 25% of variance in academic and occupational performance. The remaining 75% is conscientiousness, opportunity, motivation, and circumstance.
• The six-axis breakdown is more useful than the composite for understanding cognitive strengths and weaknesses.

What the test doesn't measure

Core Brain is not a clinical assessment. It does not measure or diagnose:

• Emotional intelligence, creativity, social cognition
• Specific learning disabilities (dyslexia, dyscalculia, ADHD)
• Long-term memory or autobiographical recall
• Domain expertise or crystallised knowledge depth
• Personality traits (use a Big Five instrument for that)

For clinical assessment, see a licensed psychologist administering the WAIS-IV or Stanford-Binet.

Privacy & data

Core Brain stores your answers locally in your browser (localStorage, 24h TTL). When you unlock your full report, Stripe handles the payment — we never see your card details. We don't sell, share, or use your data for marketing.

The optional Brain Training dashboard is currently closed access. Email us to grab an invitation link when new spots open.