Science

Research on mathematical talent runs from Julian Stanley / SMPY (1971–) through later critiques. Mechanisms are interacting systems.

Evidence percentages below are relative judgements for this atlas, not fabricated precise statistics. Named studies keep their published bands.

Findings

• 1971– · Julian Stanley / SMPY — The maths tail has a tail Above-level SAT-M still spreads at 12–13.
• 1990s– · Benbow & Lubinski — High SAT-M predicts some STEM outcomes Longitudinal, still a selected US sample.
• 2000s · Stanislas Dehaene — Approximate number sense An early starting point that training and language also shape.
• 1913 · Hardy on Ramanujan — Jagged, enormous production A case, not a national genetics lesson.
• 1985 · Françoys Gagné — Gift versus talent Aptitude still needs catalysts and practice to become olympiad or research skill.
• 2014 · Macnamara et al. — Education practice ~4% Do not treat maths homework like Berlin violin diaries.

Mechanisms

• Structure detection — Seeing the same form under new numbers.
• Working-memory for relations — Holding a transformation while another runs — a residual Hambrick-type variable.
• Proof stamina — Tolerance for being stuck, closer to Dweck’s difficulty story than to speed.
• Spatial–mathematical overlap — Some geometries lean on spatial talent; not all maths does.

Evidence

Debates

• Acceleration versus enrichment — Stanley favoured subject skip for some. Others fear social cost.
• Contest maths versus research maths — Olympiad craft is not Noether’s job. Identification often conflates them.
• Who gets the above-level test — CTY fees and who hears the invitation still shape the “gifted” file.

Mathematical talent: size the claim to the study. Interaction, not a 50/50 slogan, is the default model.

OriginsScienceSignsDevelopPracticeExemplarsMyths