Bees and numbers

Quantity discrimination: solid, and narrower than it sounds

Trained to fly towards a card carrying a particular number of shapes, bees choose correctly above chance for small quantities, typically up to about four, and the performance degrades as the quantities get larger and closer together. That pattern — good at small numbers, worse as they grow — is familiar from many animals and is consistent with an approximate magnitude sense rather than with counting.

The persistent difficulty is that stimuli differing in number also differ in total area, total edge length, density and spatial frequency. Careful designs vary those independently, and the results survive some of that control. Whether they survive all of it is the live argument, and it is a real argument rather than a quibble: a bee tracking total edge length would produce much the same choice data.

Zero: one study, one very interesting result, one open question

A 2018 study trained bees on a rule — choose the card with fewer elements — and then presented an empty card against a card with shapes on it. Bees chose the empty card above chance, and did so more reliably the larger the comparison quantity, which is the signature you would expect if the empty set were being placed at the low end of a magnitude scale.

That is a genuinely striking design, and it has been argued over in print by two groups. In 2021 MaBouDi and colleagues published a critique in Proceedings of the Royal Society B: they re-examined the stimuli, argued that the choices track continuous properties such as spatial frequency, edge length and convex hull rather than number, and built a nine-neuron model with no numerical representation in it that reproduced the reported behaviour.

In 2026 the same journal published a reply. Zanon and colleagues argued that the critique's stimulus analysis excluded the low spatial frequencies that dominate bee vision, which introduced spurious correlations between visual features and number, and that recalculating across the range a bee actually resolves leaves a sensitivity to numerosity the low-level account does not explain.

Arithmetic: the claim this site does not make

A 2019 study reported that bees trained to associate colours with adding one or subtracting one performed above chance on a matching task. If it holds, it would mean an insect holding a quantity in working memory and operating on it according to a symbolic instruction, which would be a remarkable result.

What bees plainly do use quantity for

Away from the number-card literature, magnitude estimation is doing obvious work in a bee's life. Foragers assess how much nectar a patch is returning and adjust dance vigour accordingly. Scouts assess nest cavity volume — measured by walking around inside it and integrating the distance travelled. Colonies assess quorum size at a candidate nest site by how often a scout encounters another scout there.

None of that is counting either. All of it is more consequential to a bee than a card with four dots on it, and it is what a page on this subject should leave a reader with.