Crows are said to have an ability that was thought to be uniquely human

⇧ [VIDÉO] You might also like this partner content (after ad)

Crows are renowned for their extreme intelligence, which is among the highest in the animal kingdom. Experimental studies have shown that they are able to make and use tools to solve a given problem, and able to understand and exploit digital signals, among other skills. Researchers from the University of Tübingen, Germany, have found that they are also able to understand the concept of recursion – which until now was considered a uniquely human skill.

Recursion refers to the cognitive ability to integrate a structure of elements into other structures of the same type, with a certain hierarchy. The following sentence is an example: The mouse the cat was chasing ran » ; the phrase “the cat was chasing” is embedded here in another sentence. The grammatical rules of language use recursion to expand the variety and complexity of possible sentences. Thus, recursion is considered the key concept that distinguishes human language from all other forms of animal communication.

Its origins in evolution are however controversial and several researches have been carried out to verify whether or not animals were able to perceive and produce recursive sequences. In 2020, a team of researchers showed in an experimental study that some monkeys are able to understand the idea of ​​recursion, at least as well as three- to four-year-old children — suggesting that the concept was not ultimately not unique to humans.

An ability that seems to be mastered by non-human primates

In this experiment, two pairs of symbols — among ( ), [ ] and { } — were presented to human and animal subjects on a screen, in random order. Subjects were trained to reorder them as a recursive sequence of the “central integration” type — such as { ( ) } or ( { } ) for example. If successful, the humans received verbal approval, while the monkeys received a reward in the form of food.

The researchers then presented their subjects with an entirely new set of parentheses and observed how often they laid them out recursively. In this task, the monkeys proved to be as good as the children; however, they required an additional training session. Inspired by the results of this research, animal physiologists from the University of Tübingen set out to perform a similar experiment on crows.

These birds have repeatedly demonstrated complex cognition, through elaborate tool use, analogical reasoning, and numerical proficiency. ” As songbirds, their vocal communication abilities have interesting parallels to human speech, such as complex acoustic cues, sensitive learning periods, the need for auditory feedback, elaborate vocal production abilities, and social learning “, underline the researchers in Science Advances. All of these traits make the crow a promising candidate for researching an understanding of recursion.

(A) Training procedure. (B) Symbol pairs used for training. (C) Main types of response obtained. (D) Proportion of response types produced by crows (in saturated colors) compared to American adults, Tsimane adults, American children, and monkeys as reported by Ferrigno et al., Science Advances (2020). © D. Liao et al.

First, the birds were trained to order the pairs of symbols in the form of a centered recursive sequence (pecking with their beaks at each symbol in the correct order). If successful, they were rewarded; if unsuccessful, the screen flashed and a beep sounded. They were then tested on their ability to center nested structures against never-before-seen symbol pairs.

Performance equivalent to that of human children

Crows have been trained to master the structures { ( ) } and { [ ] }, until you get more than 70% success. Then, these are the pairs ( ) and [ ] — never shown together in training — submitted to them. Result: the crows were able to produce recursive sequences in about 40% of the trials, a performance “significantly superior to chance”, underlines the team. The performance of the crows was not significantly different from that of the children, but was superior to that of the monkeys — especially since the crows did not need additional training, unlike the primates.

The team then wanted to test whether the position of the symbols in training affected the birds’ responses. In another experiment, they therefore arranged for one of the pairs used for training to be either the inner pair or the outer pair. The birds were therefore trained with two sets of symbols: < [ ] > and [ ( ) ]. In testing, the pairs < > and ( ) were then presented together. Result: Not only did the crows provide fewer cross-type responses, but they almost always provided the structure: < ( ) >. ” This result suggests that crows may be sensitive to the bounded hierarchical structure of centrally embedded sequences. “, note the researchers.

Some experts remained skeptical of the results of the 2020 study, saying it was possible that the order of the symbols during training had a major influence on the monkeys’ responses. For example, if the training sequence was [ ( ) ], and were then offered the pairs ( ) and { }, they would tend to position the parentheses having memorized the previous scheme. The same problem arises here, with the crows. To dispel the doubt, the researchers therefore added a level of complexity, using training lists with three pairs of symbols, as described below.

Generation of recursive sequences from three pairs of symbols. © D. Liao et al.

With three pairs of symbols, the probability of producing the sequences without grasping the underlying concept of recursion becomes much lower said Diana Liao, first author of the study. Ravens performed as well as in the previous experiment. ” The majority of responses were by far centric sequences, with a proportion of 42.5 and 43.8% for crow 1 and crow 2, respectively “, reports the team.

This discovery raises the question of why crows (and possible other animals capable of recursion) might use this ability. ” They don’t seem to have elements similar to human language, so recursion might be useful for other cognitive functions. “, supposes Giorgio Vallortigara, professor of neurosciences at the University of Trento in Italy, who did not participate in this study. One hypothesis is that animals could use recursion to represent relationships within their social groups.

Source: DA Liao et al., Science Advances

Leave a Comment