Crows and their relatives have lived alongside us for millenia, becoming an integral part of our mythology and culture, from Gilgamesh and Odin to Chinese myths about the sun.  Many of the birds in the corvid family (which includes magpies, rooks, jackdaws, ravens and, of course, crows) show an impressive level of intelligence.  In addition to using tools, these remarkable birds have shown the ability to recognize individual humans and warn each other about “bad” humans. Now a study by Simone Pika and Thomas Bugnyar has shown that ravens make referential gestures — pointing out items and offering them to other ravens.  While this may sound like relatively mundane behaviour, it’s actually surprisingly rare in non-human animals; our closest relatives, the great apes, don’t seem to do anything similar.  Humans begin to produce and understand referential gestures like pointing around the same time that they start learning language; it’s been suggested that this link isn’t accidental, but represents that it is around this age that human children begin to acquire a theory of mind.  By understanding that other people are also intentional agents — that is, that they also have a mind and act in order to accomplish goals — children are able to overcome many of the difficulties involved in acquiring a language.  Observing referential gestures in ravens not only enhances our understanding of (and respect for) these majestic birds; it also casts our view of the evolution of linguistic pre-requisites in a different light.

Human children begin to learn language when they are about one year old; some researchers have suggested that this is because it’s at that age that they begin to attend to the communicative intent of the people around them.  By conceiving of their social partners as beings with a mind like their own, children are able to interpret their communicative actions within a framework that allows them to extract a great deal more information from an interaction than they might otherwise.  The distinction between this kind of learning and straightforward associative learning (for example, a dog associating the sound of a bell with dinner) can be hard to appreciate because of how deeply ingrained theory of mind is in adult humans.  We reflexively think of other humans as being like ourselves and appreciate that their actions derive from the mental states which motivate them; we automatically make inferences based on these assumptions.  If someone points at something while saying a word in a foreign language, you automatically assume that they are trying to draw your attention to the object and convey some information about it.  This is because we conceive of other people as having minds similar to our own and so we frame the interaction accordingly.  If someone points to a fruit and says a word, we assume there is an intended connection between  the two, rather than just a statistical co-occurence.

Referential gestures such as pointing are also a good example of another aspect of theory of mind.  When we point at something (either with a gesture or by naming it) we often aren’t trying to affect someone’s behaviour, but rather to direct their attention.  This is a remarkably important distinction.  We can alter the behaviour of all sorts of things (such as a car or a computer) by manipulating them, but we can only direct the attention of something with a mind, since attentiveness is a mental state, not a behavioural one.  The relatively mundance act of pointing is one way we see that one year old children are developing a theory of mind.  When a young chimpanzee (or a young human) grabs at an adult, she is trying to get a specific behavioural response (such as being fed); on the other hand, when a young human points towards an object, she is trying to direct the attention of an adult without necessarily trying to elicit any particular behavioural response.

Perhaps surprisingly, several studies have shown that non-human apes perform quite poorly in comprehending such gestures.  In general, they don’t produce referential gestures in the wild and, although they can be taught to recognize pointing or the use of a marker, they seem to learn these as ritualized behaviour patterns rather than as a general behaviour with a communicative intent.  In one experiment, chimpanzees and orang-utans failed to use communicative clues provided by a human (such as pointing or the use of a small wooden marker) to help determine where an object was hidden; by contrast, human children aged 2.5 and 3 years old performed quite well at the same task.  The only chimpanzees and orang-utans who did well were those who had been previously taught a gesture; however, even these individuals failed to generalize their knowledge to another gesture.  In these experiments, the non-human apes seem to be percieving the gestures as behavioural cues rather than communicative gestures.

Although our closest relatives seem to have difficulty with referential signals,  Pika and Bugnyar found that ravens do show and offer each other items, which suggests that they do conceive of one another as intentional agents.  They observed interactions between members of seven pairs of ravens; each pair usually consisted one of member of each sex.  They saw 25 instances of “showing” (picking up and holding a non-food item in its beak) and 10 instances of “offering” (picking up and holding an item in its beak while moving its head up and down).  The ravens predominantly picked items up only in the presence of an attentive partner, which demonstrates that this beahviour isn’t simply because the ravens enjoy picking up items; likewise, these activities clearly weren’t an attempt to feed their partners.  In fact, the ravens often waited for a response from their partner, which often consisted of playing with the object together or  a friendly behaviour such as pretend feeding.

These behaviours do seem to show that the ravens conceive of each other as intentional beings with a mind of their own.  That is, they don’t interact simply to modify one another’s behaviour, but also to change each other’s mental state.  This is a remarkable finding that challenges our traditional, straightforward conception of the mental states of other animals.  We often tend to rank the consciousness or sentience of other animals based on their apparent similarity to us; research like this challenges the naïve assumptions behind this attitude.  Not only do corvids show clear signs of intelligence; this research demonstrates that they behave as though they are aware of one another as intelligent entities and so try to direct and share one another’s attention, even going to the extent of playing with an object together.

As discussed above, the awareness of one another as intentional agents is an important precursor to the development of human langauge.  After all, one of the most remarkable things about human language is how we use it to modify and direct other people’s attentional and emotional states and not just their behaviour.  We don’t just tell people to do things; we also share memories, tell jokes and stories and point out things for attention.  In addition, the conception of others as intentional agents seems to play an important role in language acquisition.  In understanding the evolution of human language, it’s important to try to understand the development of these pre-requisites.  We have often considered our closest relatives when we try to understand such traits, including theory of mind.  Results like those found by Pika and Bugnyar suggest that we may have to cast our gaze farther afield.

The researchers make the interesting point that new kinds of inferential processes may arise when communication becomes governed by cooperative motives.  It’s worth considering our somewhat narrow viewpoint when we think about such questions.  Although we spend a great deal of time and energy studying communication, we tend to limit our enquiries to species that resemble us. Ants and bees are some of the most socially co-operative animals we know, and yet we tend not to think of them as having a theory of mind or intentional behaviour.  While some may argue that this is because of their comparatively small brains, I suspect that the crudeness with which we interpret their cues reflects more about our anthropocentric bias than it does about their mental states (or lack thereof).  Ants and bees use chemical signals extensively in their communication; it may well be that our failure to observe any evidence of “shared attentional states” is because we have such a poor understanding of their communicative interactions.  However difficult it may be for us to interpret the actions of birds or other mammals, it is vastly more difficult to imagine the world through the eyes of an ant — indeed, my choice metaphor underscores that difficulty, since ants tend to rely on their antennae far more than their eyes.

Pika and Bugnyar do a nice job of showing that ravens use referential gestures to direct each others attention to objects.  This is a wonderful discovery which forces us to broaden our conception of the mental states of other animals.  Clearly, some of the pre-requisites for human language can be found in animals which are only distantly related to us, raising the question of what set of circumstances brought all of these requirements together to flower in our little corner of life.  Furthermore, recognizing that ravens view each other as intentional beings with minds of their own raises important questions about our interactions with them and the kind of respect we should afford them as conscious beings in their own right.

Ref: Pika, Simone & Bugnyar, Thomas. (2011) Pika, S., & Bugnyar, T. (2011). The use of referential gestures in ravens (Corvus corax) in the wild Nature Communications, 2 DOI: 10.1038/ncomms1567

[Edited Feb 7 to remove a repeat of the text which somehow got pasted under the post.  Sorry about that!]