What can change, what tends to remain stable, how education affects cognitive ability, and why improving on a test is not automatically the same as increasing general intelligence.
Yes, cognitive abilities and intelligence-test performance can change. But intelligence also shows substantial stability, especially in people’s relative standing compared with others, and not every apparent improvement represents a broad change in general ability.
The question becomes clearer once we separate four different kinds of change.
Four things people mean by “my intelligence changed”
Imagine Leo takes cognitive tests at age 12 and again at age 17.
1. His raw performance changed
At 17 he solves more items, remembers more information and answers harder questions than he could at 12.
That is genuine cognitive development.
2. His performance relative to peers changed
Suppose his age peers improved by roughly the same amount. Leo’s raw performance rose, but his standardized standing might remain close to the same percentile.
This is why children can become dramatically more capable while obtaining broadly stable standardized IQ scores: the norms change with age.
3. His obtained IQ score changed
Perhaps Leo scores 103 at 12 and 111 at 17. That change may reflect real development, measurement error, differences between instruments, practice, health, education, testing conditions or some combination.
A score change is evidence, not a complete causal explanation.
4. His underlying broad cognitive ability changed
This is the strongest claim. To support it convincingly, researchers want evidence that change generalizes across multiple sufficiently different measures, exceeds expected measurement error and is not merely practice on the same task.
These four statements are related, but they are not interchangeable.
How can intelligence be stable if it changes?
Because stability usually means rank-order stability.
Imagine a class of children all become much better at reasoning between ages 8 and 14. If the children who were relatively strong at 8 also tend to be relatively strong at 14, the group can show large developmental gains and high rank-order stability at the same time.
Stability therefore does not mean a frozen level of cognitive performance.
Long-term studies find substantial continuity in individual differences from later childhood into older age. The Lothian Birth Cohort, for example, found a sizeable association between cognitive test performance at age 11 and performance on the same test decades later. That is remarkable stability—but far from perfect identity.
Does education actually increase intelligence?
The strongest available evidence is consistent with a causal effect of additional education on measured cognitive abilities.
Ritchie and Tucker-Drob reviewed 142 effect sizes from 42 datasets involving more than 600,000 participants. Importantly, they did not rely only on the ordinary correlation between education and IQ, because more cognitively able students may also stay in education longer.
The review included quasi-experimental designs such as changes in compulsory-schooling laws, school-entry cutoffs and analyses controlling earlier cognitive ability. Across approaches, an additional year of education was associated with approximately 1 to 5 IQ points of improvement on cognitive measures.
The range is broad because estimates differ by design and outcome. It should not be read as a guarantee that one extra school year gives every individual a fixed number of IQ points.
Taken together, these quasi-experimental results provide strong evidence that measured cognitive ability is not environmentally inert, while still leaving uncertainty about the exact causal effect in every population and setting.
If schooling can raise scores, does that mean g itself changed?
Possibly—but the inference requires care.
If an intervention improves a wide range of cognitive tests that load on g, researchers may interpret the pattern as evidence for broader cognitive change. But simply observing a higher full-scale score does not reveal exactly which latent processes changed.
An increase could be concentrated in knowledge, reasoning, processing efficiency or several domains. The term “intelligence increased” is most defensible when the improvement is broad, reliable and not limited to narrow test familiarity.
What about brain-training apps?
Here the evidence is much less encouraging.
People usually improve on tasks they practice. If you repeatedly train a working-memory exercise, you may get better at that exercise and sometimes at very similar tasks. This is called near transfer.
The more ambitious claim is far transfer: training one narrow process produces durable improvement in substantially different abilities such as general reasoning or everyday cognitive performance.
Meta-analytic work on working-memory training has found little convincing evidence that such training produces reliable far transfer to intelligence when appropriate active control groups are used.
This teaches a general lesson: practice effects are not the same as intelligence change.
What else can affect cognitive functioning?
Development, education, ageing, neurological disease, injury, severe deprivation and some health conditions can affect cognitive functioning. Population-level changes in test performance across generations also show that cognitive-test scores are responsive to historical and environmental conditions, although the mechanisms behind such trends remain debated.
Genes also contribute substantially to individual differences in intelligence, but heritability does not mean immutability. Heritability describes variation within a population under particular environmental conditions; it does not tell us that a trait cannot change when environments change.
A worked example: training versus broad change
Imagine Priya practices a computerized memory task for six weeks.
Before training she correctly completes 6 levels. After training she completes 10.
Conclusion 1: Priya improved on the trained task. Strongly supported.
She then improves on another task that is almost identical except for different pictures.
Conclusion 2: Some near transfer occurred. Plausible.
She then takes independent tests of abstract reasoning, vocabulary, processing speed and spatial ability. They are essentially unchanged.
Conclusion 3: The evidence does not support a broad increase in general intelligence.
This is why the word “change” has to specify what changed and how we know.
Intelligence is neither fixed like a serial number nor freely adjustable at will.
Cognitive abilities develop. Education can causally improve cognitive-test performance. Illness and ageing can alter functioning. At the same time, individual differences show substantial long-term stability, and narrow training gains often fail to generalize broadly.
A scientifically useful claim about intelligence change therefore needs to specify the measure, timescale, comparison group and breadth of transfer.
References
- Ritchie, S. J., & Tucker-Drob, E. M. (2018). How much does education improve intelligence? A meta-analysis. Psychological Science, 29, 1358–1369. DOI
- Deary, I. J., Pattie, A., & Starr, J. M. (2013). The stability of intelligence from age 11 to age 90 years. Psychological Science, 24, 2361–2368. DOI
- Deary, I. J. (2014). The stability of intelligence from childhood to old age. Current Directions in Psychological Science, 23, 239–245. DOI
- Melby-Lervåg, M., Redick, T. S., & Hulme, C. (2016). Working memory training does not improve performance on measures of intelligence or other measures of far transfer. Perspectives on Psychological Science, 11, 512–534. DOI
- Deary, I. J., Cox, S. R., & Hill, W. D. (2022). Genetic variation, brain, and intelligence differences. Molecular Psychiatry, 27, 335–353. —. DOI
