What Is Tool Use? A Definition That Keeps Having to Be Revised

The standard definition requires using an unattached object to alter the form, position or condition of another object or of the user, with the user holding and directing the object during use. Each clause exists to exclude something, with unattached ruling out using a fixed branch as a lever and holding ruling out simply dropping something. Those restrictions are contested, since several behaviours that are plainly comparable fall outside, including a bird dropping a shell onto rocks to break it and an animal using an anvil it does not hold. Broader definitions include such cases and then admit behaviours most people would not call tool use. The difficulty is that the category was defined to capture something distinctive rather than to describe a natural class, which is why every version leaks.

The documented range is wide and includes behaviours that satisfy any reasonable definition:

• •Chimpanzees stripping sticks to fish termites from mounds, which was the observation that first broke the supposed human monopoly
• •New Caledonian crows manufacturing hooked tools from twigs and leaves, shaping them to a consistent design
• •Sea otters using stones to break shellfish, held on the chest while floating
• •Octopuses carrying coconut shell halves to assemble into shelters, which involves anticipating a future use
• •Elephants using branches to swat flies and modifying them for the purpose
• •Capuchin monkeys using stone hammers and anvils to crack nuts, with archaeological evidence of the practice going back centuries

Making a tool is a stronger criterion than using one and fewer species satisfy it. Manufacture requires modifying an object before use, which implies representing what is needed before it exists, and the crows and the chimpanzees are the clearest cases, with both shaping material to a form and with the crows producing designs that vary between populations in ways transmitted socially. Using a tool to make another tool is a further step and has been demonstrated in a small number of experiments. Anticipating a future need by transporting a tool to where it will be used, which requires planning beyond the present situation, has been shown in several species. Each of these was proposed at some point as the real boundary and each has been met, which is the pattern the whole field displays.

Distinguishing genuine problem solving from trial and error requires careful experimental design and the standard tasks are worth knowing. The trap tube presents food in a tube with a hole the animal must avoid pushing it into, which tests whether the animal understands the situation or has learned a rule. The floating object task presents food out of reach in a narrow tube of water, solved by adding stones to raise the level, and performance reveals what the animal grasps about the mechanism. Multi-step problems require selecting a tool to obtain another tool. Presenting a novel problem rules out prior learning. Controls for social learning establish whether the animal worked it out or copied. And field observation supplies what laboratories cannot, since a captive animal with nothing else to do may solve problems that tell you little about its life.

The repeated failure of proposed boundaries is informative about the enterprise rather than about the animals. Each definition was constructed to capture something human and each turned out to describe a capacity distributed more widely, which suggests that the search for a single qualitative difference is misconceived and that the differences are matters of degree, of combination and of the cumulative transmission that lets human tool use build on itself. That last point may be the real distinction, since human technology accumulates across generations in a way that animal tool traditions largely do not, with each generation starting from where the last left off rather than reinventing. That is a claim about culture rather than about cognition, and it is better supported than any of the boundaries that preceded it.

The standard definition requires an unattached held object used to alter something else, and each clause exists to exclude a case that looks comparable. Chimpanzees, crows, otters, octopuses and capuchins all satisfy it. Manufacture, tool use to make tools and transporting for future use were each proposed as the real boundary and each has been met, which suggests the distinction is cumulative culture rather than any single capacity.