Cross-Section Inference Across Grades 4 to 6: Developmental Patterns, Individual Differences, and Cognitive Demands

Cross-section inference, the ability to visualize the two-dimensional shape resulting from slicing a three-dimensional object, is a foundational yet cognitively demanding spatial skill. It has been studied mainly in adults and young children, leaving middle childhood unexamined at scale. This study examined problem characteristics contributing to difficulty, developmental differences across Grades 4 to 6, and gender-related variation under varying cognitive demands. A sample of 1,339 Israeli children from Jewish and Arab educational sectors completed 29 multiple-choice problems varying in object complexity and cutting-plane orientation. Guessing was modeled directly through a guessing-adjusted link function, with testing session as a random effect. Beta regression explained 71% of the variance in problem difficulty, identifying oblique cutting planes, discontinuous cross-sections, shape mismatch, limited visibility of cutting lines, and competing visual distractors as significant predictors. Approximately half of the children performed at or below chance, yet errors were systematic rather than random, with 86% of errors reconstructing the object's external appearance rather than its internal structure. Mean accuracy rose from Grade 4 to Grade 6, but this rise was not separable from differences among the classes tested; individual and classroom differences exceeded grade-level variation. A significant gender × demand interaction showed that boys outperformed girls only on the most demanding problems. Across two school years, cross-section inference showed no reliable spontaneous development, while a small set of design features accounted for most variance in difficulty. The ability appears to wait on instruction rather than maturation, and these features identify concrete targets.