When an Animal Picks Up a Stone

AThe list of animals said to use tools has lengthened quickly. A survey published in 1981 recognised 21 species in the wild; a comparable audit in 2019 named 96. The growth is usually read as a correction of an older condescension towards animal minds, and in part it is. But the two lists were not compiled in the same way. The later one draws on camera traps and on observers who stayed at a single site for a decade, and it admits as a tool any object detached from the substrate and used to alter another object — a definition wide enough to include the fish that carries a clam to an anvil rock. Counting species is in any case a poor measure of anything much. What separates a New Caledonian crow, which trims a barbed twig into a hook before probing for grubs, from a sea otter that keeps a favoured stone in a fold of skin is not captured by adding both to a total, and the total tells us nothing about why either of them began.

BAnneke Vorst, a behavioural ecologist at the Weller Institute, has spent her career asking not which animals use tools but when they begin. Her longest dataset follows 19 groups of bearded capuchins across two habitats in north-eastern Brazil. In the 11 groups at sites where fruit availability drops below a quarter of its annual peak for at least four months a year, stone hammering on palm nuts was routine in nine. Among the eight groups in humid forest, where fruit never falls that far, a single group hammered, and it did so only during the drought of 2015, abandoning the habit within a year of the rains returning. Tools, Vorst concludes, are what an animal reaches for when its preferred food fails. The capacity may well be widespread; the behaviour surfaces where the ordinary diet does not.

CRafael Duarte, an anthropologist at Coimbra, thinks the correlation real but wrongly explained. Open dry ground, he points out, is also where loose quartzite cobbles of the right mass — his measurements put the effective range at 0.9 to 1.6 kilograms — lie uncovered, whereas a humid forest floor buries them beneath leaf litter. Scarcity of food and abundance of hammers arrive together, and Vorst's design cannot prise them apart. Between 2017 and 2021 his team therefore carried suitable cobbles into two humid-forest sites where fruit remained plentiful, and left them beside palm nuts. At one site a juvenile female began striking nuts after 19 months, and four other animals had copied her within a further year. The second site produced nothing at all in four years of watching. Materials, on this account, are what limits tool use; hunger merely gets the credit.

DNeither camp disputes that what one individual invents can spread through the animals around it. Across the pooled Brazilian sites, 2,100 hours of camera footage show juveniles placing themselves within a metre of a hammering adult far more often than chance allows, and handling the discarded shells once the adult has moved away. The same footage carries a condition that is easy to read past: a new technique persisted in a group beyond a single season only where at least three adults had taken it up before the inventor's own attempts ceased. Vorst grants that transmission explains how a behaviour is kept, then insists that it says nothing about why the first animal tried. Duarte replies that the distinction flatters the question of origins, since an invention that nobody copies leaves no trace for either of them to study.

EEfficiency data unsettle both accounts. Adult capuchins at Vorst's dry sites extract on average 2.4 kernels a minute; animals under four years old manage 0.3, and full adult proficiency is not reached until roughly the ninth year. Measured against the alternative of foraging for insects and soft fruit, hammering returned 38 per cent more energy per hour in the dry months when fruit is scarce, and 4 per cent more in the wet. The margin is therefore real but narrow, and it is negative for any animal still learning the technique. Duarte reads this as support for his own position: a skill so costly to acquire is unlikely to be summoned by hunger alone, and requires years of unhurried practice with the materials to hand. Vorst reads the same figures as evidence that only a genuine shortage could justify so long an apprenticeship.

FTwo features of the dispute deserve more attention than they receive. The first is that the two researchers are not answering the same question: Vorst's data describe the conditions under which hammering appears in a population, Duarte's the conditions under which an individual can practise the thing at all, and a finding about one need not embarrass the other. The second is the traffic between field and laboratory. Captive birds and primates, given clean workshops and unlimited time, produce feats that are then reported as species-typical, though the animals concerned were reared where nothing else was on offer. Duarte concedes that his seeded sites numbered two, both inside one reserve, and that the cobbles he supplied were cleaner and more uniform than any river would deliver. That admission is more useful than another species added to the list.