Spatial Training Improves Math Performance¶

Claim · All claims
Evidence · 1 study · quant-synthesis r2 · q3 · i1 small · n=3,765 participants across 29 studies, k=89 effect sizes
Currency · newest study 2022 · not yet checked against its sources · not reviewed by a person

Training learners' spatial skills — through mental rotation, spatial visualization, and similar tasks — transfers to improved performance in mathematics, particularly in areas with strong spatial components such as geometry, measurement, and graph interpretation.

Subclaims¶

q3 i1 A meta-analysis of 29 controlled pre-post training studies finds that spatial training produces a small-to-moderate transfer improvement in mathematics performance, with larger transfer when training uses concrete manipulatives and when outcomes are "near" (closely aligned to the spatial training) rather than "far". → Hawes et al. 2022

Evidence¶

Hawes et al. 2022¶

Hawes, Z. C. K., Gilligan-Lee, K. A., & Mix, K. S. (2022). Effects of spatial training on mathematics performance: A meta-analysis. Developmental Psychology, 58(1), 112–137. doi:10.1037/dev0001281

q3 · systematic meta-analysis of controlled pre-post training studies · i1 · small-to-moderate effect, Hedges's g=0.28 (SE=0.07) · n=3,765 participants across 29 studies, k=89 effect sizes · quant-synthesis · r2

This meta-analysis pooled 29 controlled pre-post studies (N = 3,765, k = 89 effect sizes) testing whether training spatial skills (e.g., mental rotation, spatial visualization tasks) transfers to mathematics performance. Relative to control conditions, spatial training produced an average transfer effect on math of Hedges's g = .28 (SE = .07), alongside a larger direct effect on spatial thinking itself (g = .49, SE = .09). Effects were moderated by participant age (larger in older participants, ages 3–20), by use of concrete manipulatives (larger than computerized training), and by transfer distance ("near" transfer to spatially-aligned math content produced larger effects than "far" transfer); training dosage, spatial gains, posttest timing, control-group type, experimental design, and publication status did not moderate effects, and publication-bias/selective-reporting checks were nonsignificant.

Discussion¶

Mechanism. The most common explanation is that spatial and mathematical thinking share working-memory and representational resources: many mathematical concepts (number lines, place value, geometry, data displays) are inherently spatial, so strengthening spatial processing frees capacity or improves the quality of mental models used during math work. This places the claim in the same family as Cognitive Load Theory and Chunking reduces working memory load — the claim is about improving the efficiency of processing, not adding content.

Transfer is selective. Transfer from spatial training to math is strongest when the target math content has an obvious spatial structure (geometry, graphing, estimation of magnitude) and weaker for purely symbolic or procedural content. Practitioners should match the type of spatial training (e.g., mental rotation vs. spatial visualization) to the spatial demands of the target math topic rather than expecting uniform gains. This selectivity is consistent with the general pattern that analogical reasoning improves transfer only when the trained and target tasks share deep structure — surface-level similarity between a rotation task and, say, fraction arithmetic is not enough.

Dosage and durability. Open questions remain about how much training is needed, how long gains persist, and whether embedded approaches — using spatially supportive tools such as block play, manipulatives, or gesture — outperform decontextualized spatial drills. Until evidence entries are added, effect sizes and boundary conditions on this page should be treated as provisional.

Who benefits most. Learners with weak baseline spatial skills have the most room to grow, which makes spatial training a candidate for accommodations and targeted support rather than universal replacement of math instruction. It is best positioned as a supplement within cognitive load reduction strategies, not as a substitute for explicit math teaching.

Design implications. Where spatial training is used, it should be brief, frequent, and paired with explicit connections to the spatial structure of the math content being taught — for example, using mental rotation practice before a geometry unit, or number-line estimation activities alongside magnitude comparison. Embedding spatial language ("above," "rotate," "parallel") into math instruction is a low-cost complement that requires no separate training block.

Related Claims¶

• Block play predicts math achievement — spatial construction play is an ecologically valid form of spatial experience linked to math outcomes.
• Chunking reduces working memory load — a candidate mechanism: spatial training may improve how mathematical information is chunked and represented.
• Cognitive load reduction improves learning — spatial skills may reduce intrinsic load in spatially demanding math tasks.
• Analogical reasoning improves transfer — transfer from spatial training to math depends on structural alignment between the trained skill and the target content.
• A dynamic geometry model of the quilt problem helped an informal educator understand its spatial relations — a narrower finding that bears on this claim
• Hands-on learning improves achievement — related
• Discrepant training conditions (+1/+2) provided limited evidence of greater transfer to related spatial reasoning tasks than own-level training — a narrower finding that bears on this claim
• A systematic review argues that concrete materials aid mathematics and science learning when they are explicitly and gradually connected to the abstract symbols they represent — related
• Spatial Skills Improve Through Block Building — related
• Training effects were larger on outcome measures aligned with the trained skill (blending, segmenting) than on less aligned measures — a narrower finding that bears on this claim
• A spatial STEM curriculum yields far transfer to improved verbal deductive reasoning in high school students, including on nonspatial relation trials — a narrower finding that bears on this claim
• Some aspects of spatial skills are more predictive of mathematical development than others, so the construct must be differentiated — related
• Effects of two-dimensional mental rotation training on children's mathematics performance are mixed, with significant correlations between rotation and mathematics measures — a narrower finding that bears on this claim
• Children develop locative literacy through informal collaborative digital mapping with family members using Google Earth — related
• Five hours of Tetris play improved some mental rotation measures but not most spatial skill measures — a narrower finding that bears on this claim
• The spatial curriculum produced near transfer to improved spatial scanning and increased self-reported spatial habits of mind, but not to mental rotation — a narrower finding that bears on this claim
• Evidence for spatial training improving STEM achievement is promising but insufficient; rigorous research is still needed — related
• Working memory training shows short-term near-transfer gains but no compelling evidence of consistent long-term academic achievement gains — related
• Transfer-outcome format can favor symbolic controls over manipulative conditions (transfer-appropriate processing), potentially explaining inconsistent transfer findings — related
• Trained students outperformed controls at selecting the more reliable of two contradictory texts, at both post-tests — related
• The article argues that chemistry offers opportunities for mental training in manipulation, directed observation, correlation of ideas, and logical reasoning beyond its factual content. — related