Monkeys outscored preschoolers on shape tests — geometry may not be uniquely human

The ability to distinguish a square from a lopsided quadrilateral, to recognize symmetry and parallel lines, or to identify a right angle has long been considered uniquely human.

The prevailing view held that humans store shapes as mental rule lists — "four sides, four angles, bilateral symmetry" — while monkeys process only raw appearances. Because monkeys were thought incapable of the rule-based approach, the ability was cited as an evolutionary dividing line between humans and other primates.

New research directly challenges that assumption. Monkeys, it turns out, recognize shapes in much the same way humans do — and under certain conditions they handle geometric rules more consistently than young children.

A joint research team from Carnegie Mellon University and UC Berkeley published the findings in the Proceedings of the National Academy of Sciences, Vol. 123, No. 30, concluding that geometric intuition is not exclusive to humans but is shared with other primates.

The conventional view rested on a well-known experiment: six shapes are displayed in a row, one of them slightly different from the rest, and subjects must identify the odd one out.

Humans excelled at the task, quickly spotting, for instance, a slightly distorted quadrilateral among five perfect squares. Baboons, by contrast, performed poorly overall.

Scientists took those results as evidence that only humans understand shapes through rules. The new research team set out to reexamine that conclusion.

The team used a matching task instead. A single shape appeared at the top of a screen; subjects then chose which of two shapes shown below it was identical.

Participants included eight rhesus macaques and baboons, American preschoolers, and adult members of the Tsimane people of Bolivia — a farming and foraging society with little industrialization or formal schooling.

The researchers analyzed how each group processed shapes using three models: low-level visual processing, which handles raw appearances; higher-level visual processing, which captures more abstract form; and a "geometric" mode that stores shapes as rule lists — the capacity long assumed to be uniquely human.

The results defied conventional wisdom. Monkeys also used the geometric mode — less strongly than humans, but the signal was clear. They were not simply reading shapes by appearance; to a meaningful degree, they were applying rules.

Statistical analysis found no sharp boundary between monkeys and humans in this capacity.

The team ran an additional experiment in which shapes were rotated 20 degrees before subjects attempted the matching task. Under those conditions, the monkeys' geometric scores tripled compared with the unrotated version. In fact, the monkeys' scores on the rotated task exceeded the preschoolers' scores on the standard, unrotated version.

If geometric ability were truly a rule-processing capacity unique to humans, rotating the shapes should not have boosted the monkeys' scores at all — the animals would simply lack the capacity from the outset.

The research team noted that preschoolers recognized shapes in a manner similar to monkeys, concluding that whether a subject is human or monkey is not the decisive variable. They went on to say that the origins of human geometric ability may share a common root with those of other primates.

The team also said the findings challenge the hypothesis that human geometric reasoning is fundamentally different from that of other primates.

DOI: 10.1073/pnas.2532934123

J. Li, I. Boni, L.R. Sandwick, E.M. Sanford, C.M. DeLong, L. Wenjie, M.M. Henderson, S.T. Piantadosi, & J.F. Cantlon, "Continuity in geometric intuition between humans and monkeys," Proc. Natl. Acad. Sci. USA. 123 (30) e2532934123.