What animals can drop stones into a water-filled tube to bring floating food within reach or bend wire to form a hook? Obviously chimpanzees? No, New Caledonian crows have evolved sophisticated tool use too.

Humans are not unique in their ability to use tools, and many other animals employ tools. The precise occurrence and distribution of tool use depends on how it is defined, and such behaviours are not always easy to classify. In his seminal book ‘Animal Tool Behaviour: The Use and Manufacture of Tools by Animals’ (1980, Garland STPM Pub.), Benjamin Beck defined tool use as “the external employment of an unattached environmental object to alter more efficiently the form, position, or condition of another object, another organism, or the user itself, when the user holds or carries the tool during or just prior to use and is responsible for the proper and effective orientation of the tool”. According to this widely accepted definition, a bird that uses a twig to extract food from a crevice is using a tool. But what about a gull opening a mussel by dropping it onto a stone, a male bowerbird attracting a mate with its elaborately decorated bower or a heron using discarded crackers as a bait to catch fish? Such cases, where tools are not held or manipulated, are often referred to as borderline tool use or proto-tool use.

Of those species that are considered true tool users, most show relatively simple forms of tool use, while habitual and sophisticated tool use is rare. Significantly, however, it has arisen independently in birds and primates. This, of course, provides a splendid opportunity to study the conditions in which such tool use evolved as well as the adaptative context. An interesting (and controversial) question is whether tool use per se requires advanced cognitive abilities. Although it can be classified as purposeful, this does not automatically imply either consciousness or causal understanding. Several studies in primates and birds have found that tool use is correlated with an enlargement of the brain as a whole or of particular regions. For example, true tool-using birds have relatively larger brains than proto-tool users. As most of the large-brained, tool-using groups are only distantly related, it has been suggested “that the complex cognitive processes involved in tool use may have repeatedly co-evolved with large brains in several orders of birds” (Lefebvre et al. 2002, Behaviour, vol. 139, pp. 939-940). A more recent analysis has found that the cortex of the cerebellum, a brain region involved in motor control and motor learning (and hence probably linked to tool use), is more folded in tool-using birds than in non-tool users. Such correlations, however, do not rule out that these changes were due to different selective pressures in the first place. In addition, some observations have questioned the proposed relationship between tool use in the wild and physical intelligence. In problem-solving tasks in the laboratory, for example, tool-using animals do not consistently outperform non-tool using related species. It is also becoming increasingly clear that comparatively simple processes can generate seemingly sophisticated behaviour.

So it is all a bit of a minefield, but nevertheless, as in so many cases of convergence, the birds are highly instructive. Here, tool use is more widespread than previously thought and is found in at least thirty-three different families.

New Caledonian crow

Amongst those birds that have independently evolved tool use, the New Caledonian crow (Corvus moneduloides) stands out. This corvid is endemic to New Caledonia, a remote tropical island in the southwest Pacific that offers a wide spectrum of habitats. Throughout their range, the omnivorous crows manufacture and use tools, which allow them to obtain prey that would otherwise be inaccessible. Their tool-making skills exceed those of chimpanzees and are more similar to human tool manufacture than those of any other animal. Not only does their tool making involve considerable processing and require remarkable dexterity, but the crows have also evolved different tools and manufacturing techniques at different sites, probably to suit specific foraging needs.

Tool types and uses

Two types of tools predominate, stick tools and pandanus tools, the manufacture of which has probably evolved independently. Stick tools are made from twigs, grass stems or similar plant structures. They can either be non-hooked – being more or less straight and requiring only little modification – or hooked. Construction of the more complex hooked tools typically involves choosing a forked twig from which parts are removed and the remaining end is sculpted and sharpened. Pandanus tools are made from barbed leaf edges of screw pines (Pandanus spp.) through precise ripping and cutting. Intriguingly, crows show laterality in the manufacture of pandanus tools, making them preferentially on the left edges of the leaf. There are several distinct designs, which show a particular geographic distribution, with certain designs being more common in certain regions. In both stick and pandanus tools, there is a successive modification of the structure, leading to a high degree of standardisation, but their construction involves very different techniques. There seems to be no other word for this than a proto-technology.

Since New Caledonian crows are difficult to observe in the wild, less is known about the foraging situations in which they use the tools and how much they actually rely on them. The function of the pandanus tools is not really understood. Stick tools seem to be employed mainly for extracting large wood-boring beetle larvae from crevices in trees. The crows poke them until they bite the stick in defence and can be drawn out. This “larva fishing” is very similar to the “termite fishing” practised by chimpanzees and requires remarkable sensory-motor control, a task even difficult for a human. Such large beetle larvae are very rich in proteins and lipids and thus highly profitable prey. A recent quantitative assessment of the crows’ diet based on stable-isotope profiling has revealed that they are not only a substantial part of the adult diet, but also constitute most of the food that is delivered to nestlings. Proficient tool use therefore seems to offer considerable energetic rewards.

It has been argued that the tool use of New Caledonian crows is more context-specific than that of chimpanzees (which use tools made from different materials for very different purposes) – in the wild, crow tools are always stick-like, made from plant material and presumably employed for food extraction only. However, captive individuals have been observed to use a variety of materials, including feathers and garden wire. Laboratory experiments have furthermore shown that the crows are able to learn functional properties of novel tools such as stones. Having watched how stones falling into a water-filled tube raised the water level, bringing floating food within reach, they quickly learned to use stones in this way, even preferring large to small stones. Another study demonstrated the use of metatools, i.e. tools that act on another tool. Several individuals spontaneously employed a short tool to obtain an otherwise inaccessible longer tool that then allowed them to extract food from a hole. Even more remarkably, tool use has recently been demonstrated in a non-foraging context, providing the first report of multi-context tool use in birds. In one study, captive individuals used stick tools to make first contact with objects that were novel and hence potentially dangerous, while crows in another study used a tool when food was within reach but placed next to a model snake. These observations suggest that, similar to primates, these birds can distinguish between their body and the tool and value the tool less.

Development of tool use

Since tool use and manufacture are so complex, how do New Caledonian crows develop their skills? Experiments with juveniles have shown that there is a strong genetic predisposition for the use of basic stick tools and the making of simple pandanus tools. Hand-reared individuals deprived of opportunities to observe tool handling developed functional use of stick tools at the same point in ontogeny as others to whom a tutor was available. Despite this predisposition, learning plays a significant role, leading to individual differences in tool repertoires. Tool use is complicated, and juveniles need one to two years to refine their skills. During this period, they spend a lot of time using stick tools while having little or no success in food extraction. The significant learning costs are probably outweighed by the future benefits of competent tool use. Juveniles in the wild seem to learn directly (by observing tool-using adults) as well as indirectly (by examining discarded tools). It is possible that information on the complex tool designs is culturally transmitted, which would explain the differences in tool design between different sites. If so, this would be an example of cumulative technological evolution. New Caledonian crows seem to live in small family groups that stay together for an extended period of time, allowing the opportunity for tool-using skills to be transmitted vertically (from parents to offspring).

Tool use and cognition

So do these birds, which have large brains and are the only corvids to use tools in the wild, possess better cognitive abilities than their relatives? It has been suggested that certain features of their tool manufacture, such as the standardisation, laterality and successive modification, point to a cognitive sophistication that is reminiscent of early hominid stone cultures. There is also no doubt that the crows perform impressively well in tool-related tasks in the laboratory, demonstrating a remarkable degree of intelligence. When faced with a particular problem, they select the appropriate tools, combine different tools sequentially and even design new tools, if necessary (as famously demonstrated by a captive crow named Betty, which bent and unbent wire to construct a hook). Their ability to learn the properties of novel tools has led to the proposition that the crows are able to process causal information based on cognitive mechanisms other than simple associative learning. It has been suggested that they might use analogical reasoning (where previous experience is mapped onto a similar novel problem), which could explain their sophisticated tool-making abilities. It is therefore tempting to suggest that New Caledonian crows causally understand physical problems.

Uncovering aspects of physical understanding and reasoning is, however, extremely difficult in non-human animals. The exact cognitive mechanisms underlying tool use in New Caledonian crows remain unknown and the birds’ use of tools in a non-foraging context has led to the suggestion that these mechanisms could draw on more general principles than previously assumed. Corvids as a family are characterised by relatively large brains, remarkable behavioural plasticity (especially highly innovative foraging behaviour) and well-developed cognitive abilities. Other species, such as rooks (Corvus frugilegus), can also make and use tools in the laboratory, showing a degree of sophistication similar to that of New Caledonian crows. It is therefore considered unlikely that cognitive skills have been limiting in other corvids. So why has such tool use only evolved in the crows on New Caledonia?

Evolution of tool use

Rather than differing from other corvids in general cognitive capacity, New Caledonian crows might mainly differ in their motivation to use tools. An essential aspect of tool use is what can be referred to as plastic object combination. So too, there is evidence that this species playfully combines objects more often than other corvids. A study comparing captive-bred New Caledonian crows and common ravens (Corvus corax) showed that in the weeks after fledging, object combination behaviour increased in the crows, while it decreased in the ravens. There is, however, an interesting similarity between these two species. The ontogeny of tool-oriented behaviour in the crows is qualitatively very similar to that of food-caching behaviour in the ravens, which has led to the suggestion that the evolutionary origins of tool use in New Caledonian crows could lie in food caching.

So while certain behavioural and/or morphological features might have predisposed these birds to develop tool use, favourable environmental conditions could then have promoted the evolution of this behaviour. When the species’ ancestor colonised New Caledonia, the woodpecker niche was vacant, with profitable hidden prey and no direct competition for this resource. The absence of predators allowed the crows to spend considerable time and energy on interaction with and exploration of objects as well as to evolve a rather slow life history, with a prolonged juvenile phase, during which complex tool use can be learned. Hence, the evolution of tool use in New Caledonian crows probably required a combination of several factors.

Woodpecker finch

Habitual tool use also occurs (and shows a number of similarities to that of New Caledonian crows) in the woodpecker finch (Camarhynchus pallidus), one of the famous Darwin’s finches (Geospizinae) that are endemic to the Galápagos islands. This bird uses a tool, typically a cactus spine or a twig that may be shortened or otherwise modified and is then held in the beak, to extract insects from otherwise inaccessible tree crevices. It therefore fills the niche of woodpeckers (which are absent from the Galápagos) through a versatile behavioural adaptation, circumventing the need for less flexible morphological specialisations. Recently, some individuals have been observed to employ a different type of tool with novel functional features – barbed twigs from blackberry bushes, a plant that is not native to the islands. The twigs were first modified by removing side twigs and leaves and then used such that the barbs helped drag prey out of tree crevices. This observation highlights the finches’ ability to choose and modify tools flexibly and exploit novel environmental opportunities. Such behavioural flexibility is also shown by the other species of Darwin’s finch and was possibly a key factor in their success in the rather hostile environment of the Galápagos islands.

It has been suggested that the arid, unpredictable conditions in the coastal zone favoured the evolution of tool use in woodpecker finches. The importance of this behaviour differs between vegetation zones. In the arid zone, where food is not only limited but also hard to access, with insects retreating to hiding places, tool use is essential, especially during the dry season. Up to half of the finches’ prey is acquired with the help of tools, rendering them even more routine tool users than chimpanzees. The tools allow them to extract large, nutritious insect larvae from tree holes, making tool use more profitable than other foraging techniques. In contrast, in the humid zone, woodpecker finches rarely use tools, since food availability is high and prey is more easily obtainable. Here, the time and energy costs of tool use would be too high.

Interestingly, laboratory experiments have demonstrated that young woodpecker finches from both arid and humid habitats are capable of using tools. Juveniles showed this behaviour in absence of a tutor and at the same stage in development as individuals that had the opportunity to learn from a model. Adults from humid habitats that did not use tools, however, did not acquire tool use in the presence of a tool-using conspecific. These findings indicate that social learning is not necessary for tool use to develop (although opportunities for social transmission do arise in the wild). There seems to be a genetic predisposition for this important behaviour, which is then refined by individual trial-and-error learning during a sensitive phase early in development. This means that, rather than following a stereotypical behavioural pattern, tool use can be modified and adapted by learning. The differences in adult tool use in arid and humid habitats are likely due to different opportunities for learning in these different environments. The notion that a genetic predisposition for a particular learning process might be beneficial for behavioural adaptations in unpredictable environments is also supported by research in the fields of song learning, memory and navigation.

Although tool use in woodpecker finches is flexible, selective and occurs at high frequency in certain conditions, there is no convincing evidence to date that it requires special cognitive abilities. The finches can learn new tasks in the laboratory quickly and easily, but this might simply reflect enhanced general learning abilities. In certain physical tasks resembling their natural tool use, tool-using woodpecker finches neither outperformed non-tool-using conspecifics nor closely related non-tool-using small tree finches (Camarhynchus parvulus). Hence, this species might lack sophisticated physical cognition in conjunction with tool use. Importantly, the finches seem to be unable to assess problems in advance. Like in New Caledonian crows, the evolution of tool use might have been favoured by an unexploited resource, combined with characteristics such as high flexibility or explorative tendency that could have been present in the common ancestor of Darwin’s finches.

Other bird species

Simpler forms of habitual tool use are found in several other bird species, including Egyptian vultures (Neophron percnopterus), brown-headed nuthatches (Sitta pusilla) and hyacinth macaws (Anodorhynchus hyacinthinus).

Egyptian vulture

It is well known that this small Old World vulture employs stones as tools to crack open hard-shelled ostrich eggs (a similar behaviour is observed in chimpanzees and sea otters). The bird takes a stone into its beak and forcefully throws it at the egg until the shell is broken, usually taking a few minutes. This behaviour seems to be largely innate and is also displayed by naïve individuals. Its origin could be related to the throwing of eggs and, interestingly, rounded (egg-like) stones are preferred to jagged ones. In an Israeli population, Egyptian vultures are exploited by brown-necked ravens (Corvus ruficollis), which are unable to open the hard-shelled eggs themselves. Pairs of ravens have been observed to repeatedly attack a vulture that had just opened an egg, finally driving it away. This is another example of the well-developed cognitive abilities of corvids.

In a small population in Bulgaria, Egyptian vultures even use twigs to collect sheep wool for padding their nests. Although both twigs and wool can serve as nesting material, this appears to be deliberate tool use. The birds approached bits of discarded wool with a twig in their beak, which was then either used as a rake, to gather the wool into heaps, or to roll up the wool. Interestingly, wool was collected only after shearing or simulated shearing of sheep had taken place, but not after wool had simply been deposited in sheep enclosures.

Brown-headed nuthatch

In longleaf pine forests in Louisiana, foraging brown-headed nuthatches have been observed to methodically use bark pieces to remove other flakes of bark from a tree. The birds inserted the bark piece underneath an attached bark scale, using it like a wedge and lever, so as to expose hiding insects. Occasionally, they reused the same piece of bark several times before dropping it and sometimes even flew short distances with the bark flake in their beak. Such tool use was not seen on other tree species, where the bark does not detach so easily, or when nutritious pine seeds were abundant. It had been assumed that this was a local behaviour limited to this particular study area, but similar behaviour is now reported from other regions. The evolutionary origin of this tool use might be related to these birds frequently wedging seeds into cracks in the bark to hammer them open with their beak, which can lead to bark coming off. In a few cases, brown-headed nuthatches have also used a bark flake for concealing a seed cache.

Hyacinth macaw

These large South American parrots mainly feed on hard-shelled nuts and have been repeatedly observed to use tools when breaking open the nuts. Several birds wrapped a piece of leaf around the nut so as to hold it in place. Such behaviour is also shown by palm cockatoos (Probosciger aterrimus) in Polynesia, where it is likely to be convergent. Observations of captive hyacinth macaws have demonstrated that several other objects can aid the opening of nuts – for example, pieces of wood were employed as wedges (and possibly for other functions). It seems that this species has an innate tendency to use tools during manipulation of nuts, as naïve juveniles tried out a variety objects in combination with nuts. There should, however, be ample opportunities for social learning in this highly social parrot, and adults opened nuts more quickly and more successfully than juveniles (although this could simply be due to their greater strength).