Tool-use definitions

Classic working definition

A widely cited definition treats tool use as employing an external, detached object to change the form, position, or condition of another object, an organism, or the user itself, while holding or carrying that object. It is a working tool for comparison, not a law of nature.

Using vs making vs modifying

Tool use is acting with an object; tool manufacture is altering an object before use; modification is a degree of that altering. These are separate, layered questions — a behaviour can involve one without the others.

Detached vs attached object

Many definitions require the object to be detached from the substrate and under the animal's control. Behaviour using something still fixed to the ground, a plant, or a surface sits in a contested borderland often labelled 'borderline' or 'proto-tool' use.

Behaviour vs cognition

Whether an act counts as tool use is a definitional question about observable behaviour; what the animal understands about why the object works is a separate, harder question that the label alone does not answer.

Why the line is contested

There is no single agreed definition, so researchers state which one they apply. Boundary cases — bait, attached anvils, throwing, body parts — fall inside some definitions and outside others, which is why careful papers spell out their criteria.

It might seem obvious what a tool is, but science needs the term pinned down before behaviours can be compared, counted, or argued about. Without an agreed definition, two researchers watching the same animal could reasonably disagree about whether they had seen tool use — and a tally of "tool-using species" would mean little, because everyone would be counting different things. Definitions exist to make claims testable and comparable, not to capture some deep essence of toolhood.

A commonly cited working definition treats tool use as the use of an external object to alter the form, position, or condition of another object, another organism, or the user itself, with the object held or carried by the user. Each clause is doing work. "External object" tends to exclude an animal's own body. "Held or carried" tends to require control over a detached item. "Alter the condition of another object or organism" focuses on objects used to act on something else. A behaviour that misses one clause may fall outside that particular definition while fitting a broader one.

Crucially, this is a working definition, not a fact about the world. Other researchers adopt narrower versions (for instance, requiring that the object be manipulated in a goal-directed way) or broader ones (counting some uses of fixed objects). Because the definitions genuinely differ, the responsible practice — and the one careful papers follow — is to state explicitly which definition is being used, so readers can judge a claim against a known yardstick rather than an assumed one.

A frequent source of confusion is collapsing several distinct ideas into the single phrase "tool use." Tool use is the observable act of employing an object to act on something else. Tool manufacture is a further step: altering an object before use, such as stripping a stem or shaping plant material into a probe. Modification is the degree of that alteration — from a single removed leaf to substantial reshaping. These are layered, separable questions, and a behaviour can sit at one layer without reaching the next.

Picking up a suitable stick and using it as found is tool use without manufacture. Trimming that stick first adds manufacture. Reshaping it into a hooked form adds more extensive modification. Among the well-studied cases referenced elsewhere on FaunaHub, some chimpanzee (Pan troglodytes) populations are described as modifying probes before use, and some crows are reported crafting probing tools from plant material — but the point here is structural, not a claim about any one population: making and modifying are additional steps layered on top of using, and they should be reported as such rather than merged into a single undifferentiated label.

Keeping these layers distinct also keeps interpretation honest. Manufacture is often treated as more cognitively demanding than simple use, but "often" is not "always," and the inference is itself debated. Describing exactly what was observed — used, modified, or manufactured — lets readers see the behaviour without importing assumptions about what it proves. None of these layers, on its own, tells us what the animal understands about why the object works; that is a separate question this category was never designed to answer.

The definitional debate becomes sharpest at the edges, and a few recurring boundary cases show why. One is the object attached to a substrate. If an animal carries a rock and pounds prey against it, most definitions count that as tool use because the object is detached and controlled. But if the animal instead brings prey to a fixed stone or hard surface and strikes it there, the "object" is part of the environment, not held or carried — and several definitions exclude it, treating it as a related but separate category sometimes called substrate or anvil use. The sea otter (Enhydra lutris) case is often discussed precisely because shell-breaking can take both forms, sitting inside some definitions and outside others.

Other borderline cases stretch the term in different directions. Using bait to attract prey, throwing or dropping an object, building a structure, or using one's own body as an implement each satisfy some definitions and fail others — and authors disagree, in good faith, about where to draw the line. There is no neutral place to stand: every definition includes some cases and excludes others, and reasonable researchers weigh the trade-offs differently. This is why the boundary is described as contested rather than simply unsettled; the disagreement is about which distinctions are worth making, not about what the animals did.

Because of this, the most useful habit for a reader is not to ask "is this really tool use?" as if there were a yes-or-no fact, but to ask "under which definition, and what exactly was observed?" A behaviour can be a clear example under a broad definition and a borderline case under a strict one without anything about the animal changing. The honest summary is that the category has fuzzy edges by design, and the edges are where the science of definitions does most of its work.

Tool-use reports are easy to over-read, and the most common error is to slide from "used an object" to "used human-like technology." That framing imports a great deal the evidence does not carry. Comparing animal behaviour to human toolmaking as if along a single scale tends to obscure more than it reveals, because different species solve different problems in different ecological settings; a behaviour is best understood on its own biological terms rather than as a step toward a human benchmark. A definition that lets us count a behaviour says nothing, by itself, about planning, foresight, or understanding.

Two further cautions follow from how these cases are actually documented. First, tool use is typically reported for particular populations or individuals studied under particular conditions, often with small numbers and sometimes in captivity; it should not be generalised to an entire species or group, and a result on one task does not establish a broad, human-style competence. Second, tool use is scattered across distantly related animals rather than tracking a single ladder of cleverness, so it cannot be converted into an intelligence score or used to crown a "smartest" animal — the category is too narrow and too fuzzy at the edges to bear that weight.

Put together, this makes the definitional question practical rather than pedantic. When you meet a tool-use claim, the literate response is to notice which definition is in play, distinguish using from making from modifying, treat boundary cases as boundary cases, and resist the leap to technology or to ranked intelligence. That is what lets you take the genuine, well-documented cases seriously while reading the surrounding claims with appropriate care.

Headlines about animals "using tools" rise and fall on definitions readers never see. Knowing that the term has competing meanings lets you ask the right question — which definition is being applied? — instead of treating every claim as equivalent or assuming the word implies human-style technology.

Tool use is often used as a shorthand for intelligence, but the link is loose and the category itself is fuzzy at the edges. Understanding the definitional debate guards against over-reading a single behaviour and against dismissing real, documented cases just because they sit near a boundary.

• Treating "tool use" as a fixed, objective category — when in practice researchers choose among several definitions and often say which one they are using.

• Assuming using a tool and making a tool are the same thing; manufacture (modifying an object before use) is an additional step that not all tool use involves.

• Reading "tool use" as evidence of human-like technology or planning, rather than as a behavioural label that says nothing on its own about an animal's inner understanding.

• Counting borderline cases — bait, an anvil fixed to the ground, a body part used as an implement — as settled examples, when these are exactly where definitions disagree.

• Generalising a documented behaviour to a whole species or group, when tool use is typically reported for particular populations or individuals studied under particular conditions.

• Chimpanzees behavior
• Crows behavior
• Orangutans behavior

These methodology notes sit alongside FaunaHub's wider source practice. See animal research sources and how FaunaHub uses sources, and return to the animal intelligence & behavior hub.