What GAI leaves out · Reviewed 2026-09-23
Working memory, processing speed, and GAI
GAI is higher or lower than IQ mostly because working memory and processing speed are in one composite and not the other. This page is that mechanism. It is not a diagnosis of a memory disorder, and it is not the GAI vs IQ calculator.
Why GAI drops memory and speed
Full Scale IQ is designed to summarize the core battery, including how much information you can hold and how quickly you can do simple visual work. Those skills matter. They also move around more with fatigue, anxiety, motor demands, and some clinical conditions than vocabulary or matrix reasoning do. The General Ability Index was published so a clinician could describe reasoning when those two indexes make the full-scale average a bad portrait. That rationale is the one Saklofske and colleagues set out for the WISC-IV GAI, and later manuals kept the idea while changing the subtests.
Leaving them out is not a claim that memory and speed are unimportant. It is a claim that they should not be forced into every sentence about “ability.” CPI, the Cognitive Proficiency Index, is the companion composite that keeps them in. GAI without a glance at CPI is half a paragraph.
What those two indexes measure
Working memory (WMI) is holding information and working on it for a few seconds: digits, letters and numbers, picture sequences, mental arithmetic, depending on the edition. On this site’s nine tasks the analogs are computation span and visual reversal. Processing speed (PSI) is simple, timed visual decision: symbol search, coding, cancellation. The analog here is symbol search. Both are broad abilities in CHC terms (Gwm and Gs). The map is on the method page.
A low index is not a trait name. A PSI of 88 can be a slow pencil, a timer the examinee feared, poor scanning, or a stable weakness. The report’s behavioral notes are part of the data. A website that sees only 88 cannot tell those apart.
How they move GAI away from IQ
Picture five index scores. Three reasoning indexes sit around 118. WMI and PSI sit around 90. Full Scale IQ, which has to honor all five, lands nearer the middle. GAI, which uses only the reasoning side, stays near 118. The gap is positive and it has a visible cause. Flip the pattern — reasoning near 95, memory and speed near 120 — and FSIQ rises above GAI. Efficiency is carrying the average. Both patterns are readable. Neither pattern says which number to throw away.
The size worth mentioning is discussed on GAI vs FSIQ: a few points can be statistically real on a Wechsler, and about 10 points is commonly worth a conversation. If you have the two composites, paste them. If you have CPI too, the CPI vs GAI calculator is the direct comparison of reasoning against efficiency. Gap there is GAI minus CPI, a wider lens than GAI minus FSIQ, because FSIQ already contains some of the reasoning that GAI contains.
Which page to use for which pair
| You have | Question | Open |
|---|---|---|
| FSIQ and GAI | How far apart are the two summaries? | GAI vs IQ calculator |
| GAI and CPI | Is efficiency far below reasoning? | CPI vs GAI calculator |
| VCI VSI FRI WMI PSI | Which index peaked or dragged? | Index calculator |
| No scores | A personal estimate of the same shape | Nine-task test |
Do not run all four and average the gaps. Pick the pair the report actually printed. Index means are not a back door to official GAI.
How to talk about a low memory or speed index
In plain language: reasoning (GAI) and the skills that keep information moving (WMI, PSI, CPI) are telling different stories. Classrooms feel that as “knows the material, loses the steps” or “understands, runs out of time.” Those sentences are hypotheses. They are not accommodations. Extra time, a quieter room, or a different task format is a decision for the school and the clinician who saw the protocol, not a coupon printed by a calculator.
If the only evidence is this website’s nine-task estimate, hold the hypothesis even more lightly. Symbol search in a browser is not Coding on a WAIS. A noisy phone and a thumb will dent processing speed. Retake nothing the same hour. If the official report is the evidence, go back to that examiner with the specific indexes, not with a rewritten IQ.
The ingredients change; the idea does not
Every current edition discussed on the WAIS / WISC page keeps working memory and processing speed out of GAI. The reasoning list changes. WAIS-IV folds visual and fluid work into PRI. WISC-V splits them and then keeps Visual Puzzles out of GAI even though it is spatial. Do not explain a child’s WISC gap with an adult WAIS subtest list you memorized. Match the edition printed on the report.
Common questions
Does GAI include processing speed?
No. Processing speed and working memory are the indexes GAI is built to leave out. They remain inside Full Scale IQ.
Is CPI the opposite of GAI?
CPI is the efficiency composite (memory and speed). It is the other description, not a rival IQ. Compare them on the CPI page when both are printed.
Can low processing speed mean ADHD?
It can be part of a clinical picture. It is not a diagnosis. A score, especially an online estimate, cannot make that call.
Should I want GAI to be higher?
No. A higher GAI means the full-scale average was pulled down. It is a split to understand, not a score to maximize.
Where to go next
The GAIvsIQ laboratory keeps one URL for each job. This note does not replace the instrument it points to.
Ready when you are
Have CPI and GAI?
That pair is reasoning versus memory and speed. It is a different page from GAI vs IQ.