Yes, it is true. Spatial ability (also called spatial intelligence, spatial skills, or spatial thinking) is one of the strongest and most consistently documented cognitive predictors of achievement, interest, and attainment in STEM fields, even after accounting for mathematical and verbal ability.

Key evidence

Large-scale longitudinal studies spanning decades and hundreds of thousands of participants show that spatial ability measured in adolescence (around age 13) predicts later STEM educational credentials and careers over and above math and verbal scores. Adolescents with higher spatial ability relative to verbal ability are especially likely to enter and succeed in STEM.

Childhood measures also predict outcomes: stronger spatial skills in 4th grade increase the likelihood of choosing STEM college majors years later, both directly and indirectly through later math achievement and motivation, after controlling for verbal skills, motivation, and family background.

Cross-sectional and correlational work links spatial skills to performance in specific STEM domains (engineering design and problem-solving, chemistry, physics/kinematics, mathematics, geology, etc.). The association is often moderate (e.g., around r ≈ .36 with math performance in meta-analyses) and remains after controlling for general intelligence or fluid reasoning.

Why the link exists

STEM disciplines rely heavily on visualizations—diagrams, graphs, 3D models, molecular structures, maps, number lines, engineering drawings, etc. Spatial thinking helps people mentally rotate, transform, and organize information in space, which is central to understanding and solving problems in these fields. Spatial skills appear especially important early in learning (as a “gateway”) and can become less rate-limiting once expertise develops specialized mental representations.

The association is stronger in some STEM areas (e.g., engineering, chemistry) than others (e.g., certain biology topics that rely less on mental visualization).

Causality and trainability

Most evidence starts correlational, but experimental and quasi-experimental studies show that targeted spatial skills training improves STEM course grades, STEM GPAs, and retention (with particularly notable benefits for women in engineering). Meta-analyses confirm spatial skills are moderately malleable (average training effect ~0.4–0.5 standard deviations), with transfer to other spatial tasks and, in many cases, to mathematics performance.

In short, spatial ability is not the only factor in STEM success (motivation, opportunity, prior knowledge, and other cognitive skills matter), but the link is robust, predictive beyond general ability, and actionable through training and early experiences (e.g., play with blocks, puzzles, and spatial toys).

Object-based spatial skills involving mental manipulation and visualization of shapes and objects are the kinds most consistently linked to later STEM success in children. Navigation or large-scale spatial orientation skills show weaker or less consistent unique prediction once other factors are controlled.

Main categories that predict STEM outcomes

Spatial ability is multi-faceted. Researchers often classify skills along dimensions such as intrinsic vs. extrinsic (object-focused vs. relating objects to the environment or self) and static vs. dynamic (fixed vs. involving transformation/movement). The strongest childhood predictors of STEM achievement and major choice fall in the intrinsic-dynamic and related visualization domains:

• Mental rotation and mental folding/transformation: The ability to mentally turn, fold, or reconfigure 2D or 3D shapes. These predict science achievement (including physics and biology) in 7- to 11-year-olds even after controlling for verbal ability and age. Mental folding has emerged as a particularly strong unique predictor.
• Spatial visualization and object assembly/manipulation: Skills measured by tasks such as Block Design (constructing patterns with colored blocks), paper folding, pattern assembly, 2D/3D drawing, or copying designs with LEGO/tangrams. These form part of an “object manipulation” domain that shows the strongest associations with STEM grades and degree choice in longitudinal data, remaining predictive after accounting for general intelligence and verbal ability.
• Spatial scaling: Judging or transforming relations across different scales (e.g., relating a small model or map to a larger space). This extrinsic-static skill uniquely contributes to science scores (especially biology and chemistry) in middle childhood.

Early measures of these skills matter. Spatial assembly ability at age 3 (e.g., accurately copying shapes or LEGO structures) predicts math readiness at age 5. Mental transformation skills around age 5 predict later calculation performance, partly through better number-line understanding (a spatially organized representation of magnitude).

Evidence linking childhood skills to STEM

• Middle childhood (around 4th grade): Broader spatial skills (often indexed by Block Design or similar tests) directly predict later choice of STEM college majors, above and beyond concurrent math/verbal achievement, motivation, and family background. They also operate indirectly by boosting intervening math achievement and motivation. High spatial performers in 4th grade show substantially higher odds of entering math-intensive STEM majors.
• Broader pattern: Object-manipulation skills (mental rotation, drawing, assembly) tend to outperform pure navigation or map-reading measures as predictors of STEM educational outcomes. All major spatial domains correlate with STEM, but object-focused ones show the most robust unique links.

What is less strongly linked

Large-scale navigation or wayfinding ability correlates with spatial cognition overall but contributes less unique variance to STEM success in the available data. Everyday spatial awareness (e.g., simply knowing where things are in a room) is related but not the primary driver identified in predictive studies.

Practical note

These skills are moderately malleable. Activities that build mental rotation, visualization, and assembly—such as construction toys, puzzles, block play, spatial language, or targeted training—show benefits that can transfer to math and science performance. The predictive power appears early and persists, making childhood a useful window for both assessment and support.