Stating the Obvious: We Are All Unique!

We can tell people apart because we know them. We can tell the difference between family, friends, and strangers. We can distinguish people by looking at their faces: the shape of the nose and the mouth, the color of the eyes and hair, freckles, moles, and scars. These physical characteristics help us to recognize each other.

It is the same for animals: individuals need to recognize each other to tell their partner or family apart from their rivals. From a human point of view, you might think that African penguins are all identical, but this is far from the case. Individuals can identify each other and recognize those they know thanks to the black spots on their chests, which form a different pattern for each individual [1]. The same might be true for other species. For instance, from a distance, all greylag geese look very similar—the same gray feathers, the same pink-orange beak, and the same size. Graylag geese live in big groups and form lifelong bonds with their partners. Geese need to be able to distinguish their partners from dozens of other geese.

Konrad Lorenz, a Nobel laureate who worked with geese for most of his life, claimed to be able to recognize each individual in the flock he was working on [2]. It is also possible for a human who works permanently with animals, like zookeepers for example, to tell the difference between individual animals. Can we tell the geese apart? Are there any clues? Are these clues the same for the geese as for humans? That is what we tried to understand in this study: we used artificial intelligence software to find out if there are distinct features on the faces of the greylag geese. Then, we observed how geese reacted to photos of other geese, including their mates, familiar flock members, and strangers. We did this to figure out if the geese can differentiate among individuals. We predicted that geese would respond more warmly to photos of their partners than to unfamiliar faces.

Where did we Conduct the Study?

To conduct such a study, it is necessary to individually mark animals, otherwise it would not be possible to distinguish one family, or one goose, from another. Our study was conducted at a research center of the University of Vienna in the Alm valley in the northern part of the Austrian Alps, located in central Europe. A flock of graylag geese was introduced there in 1973 by Lorenz. The birds are fully free and can move around as they wish—but they receive food from us so they can stay over the winter, while their conspecifics in other areas are migratory. All geese are tagged with colored leg rings and are used to the close presence of humans. Data about every individual goose has been collected since 1973, and therefore we know their relationships, friends, and relatives within the flock (Figure 1) [3]. We also know that over the period of the study, the total number of geese in the flock was on average 110, and we collected more than 500 photos of flock members (89 geese within the same year across days or weeks, and 84 geese across 2–4 years).

In general, graylag geese are highly social. They live in big flocks for most of the year and pair-partners (male and female) usually stay together for several years, similar to the way humans form couples with strong family ties. There are earlier contributions about these flocks in Frontiers for Young Minds (see here and here).

What did we Investigate?

We photographed the faces of each individual several times over the course of a year. We focused on beak shape because, unlike feathers, beaks are a part of the body that hardly changes at all over an individual’s lifetime. We took multiple photos for two reasons: first, we could vary the light and the quality of the photo; and second, we have photos of the same individuals at different ages. Then the photos were carefully analyzed to highlight the details (Figure 2). To determine whether the goose faces were visually unique, we developed a software to identify similarities between the beaks of the geese.

Finding differences among the beaks of individual geese does not imply that the geese identify their friends according to their beak shapes. They could use other cues or sound-related signals, for instance. Therefore, the next step in our investigation was to check if and how the geese reacted to photos of themselves and others. For this purpose, wooden boards were set up at five spots near the feeding area. We took care that the birds got used to the wooden boards by placing them there long before the experiment started.

During the experiment, we pinned life-size goose pictures to the boards. Each board had a food bowl in front of it, filled with the same food the geese usually eat. We also placed cameras nearby to record how the geese behaved (Figure 3). We used a random process to decide which type of picture to show: a blank board or a picture of a goose, which might be the goose’s partner, a general flock mate, or even the goose itself. To distinguish between the birds involved in the observation, we will use the term “focal goose” for the bird standing in front of the wooden board and being observed by us; and we will talk about the individual in the photo as the “goose in the photo”. Once a focal goose entered the test area, we recorded its behavior for 5 min. We looked for friendly or aggressive behaviors that showed how much the geese liked or disliked the photo.

We combined all the friendly behaviors into one score and called it the affiliative response. A high score meant the goose approached the photo quickly, stayed longer, and vocalized more. We then used a computer program to see if the individual in the photo made a difference in how the focal goose responded.

What we Found

The results confirmed our expectations. The software we developed could identify individual geese by their beaks with a very high accuracy: 97%. So, we have proof that there are physical beak traits that are specific for each individual goose, and these can be used to tell the geese apart. However, our software focused solely on beaks, and it is highly likely that geese also use other characteristics to recognize their fellow creatures, for instance sound-related or other visual cues (e.g., voice, body shape and size, etc.).

Furthermore, the focal geese showed the strongest friendly behavior toward photos of their partners. This means that they moved faster toward and got closer to the photo, spent more time eating near the photo, and vocalized often and longer than they did next to photos of non-partner geese. We also observed that the geese rarely showed aggressive behavior toward the photos, possibly because they somehow realized they were not real geese.

What our Results Mean for Science

We found that geese recognize each other in photographs. They acted differently when they saw pictures of their partners compared to other geese, showing that they could tell individual geese apart. Recognizing who is who could help geese in many ways. For example, being able to identify their goslings or remembering who supported them in the past could improve their survival. Some animals use these kinds of skills for teamwork and helping others [4]. For geese, individuality may also play a role in how they respond to threats, or how they act in stressful situations. More research is needed to understand how geese use their ability to recognize each other in their everyday lives.

We confirmed that each goose’s face is distinct enough to be detected by an algorithm. Although our program only focused on the beak morphology in geese, other studies show that it is possible to differentiate individuals according to other body features, such as whole-body spot patterns in Serengeti cheetahs [5]. Software like ours could also provide a better understanding of preference mechanisms in adult geese. Perhaps there are physical traits that are more interesting than others in terms of fitness, such as traits that improve the survival of the individual, that would therefore be interesting for a goose to pass on to its offspring. A similar program could be used to identify these traits, and we could thus gain a better understanding of certain characteristics of the species.

Our results also show that graylag geese are smarter than previously thought, as facial recognition requires well-developed intelligence skills. Because graylag geese are social animals, groups have a hierarchy. So, for the group to function well, it is important for geese to recognize each other and distinguish between high- and low-ranking individuals. This not only reduces conflict but also allows close individuals to bond and support each other.

In conclusion, we succeeded in proving evidence for visible differences in the beaks of graylag geese using artificial intelligence software. Our results have taught us that it is necessary for a social species like the graylag goose to differentiate between its conspecifics. Furthermore, the development of digital tools capable of distinguishing between individuals could prove very useful for population monitoring, for example. Such tools would make the identification of individuals faster and more efficient than it is now, especially as monitoring endangered species is now essential for nature conservation. With so many species on the brink of extinction, there is a lot of monitoring work to be done, and using software could be of great help to scientists.

Glossary

Conspecifics: ↑ Members of the same species.

Affiliative Response: ↑ A set of behaviors hinting at a positive social relationship.

Gosling: ↑ A young goose less than three months old and not able to fly.

Algorithm: ↑ A set steps a computer follows to perform a task, like identifying patterns in images.

Morphology: ↑ The study of shapes and structures of living things, such as the shape of a goose’s beak.

Hierarchy: ↑ A form of organization often found in social species, where dominant individuals have better access to resources compared to low-ranking individuals.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

Many colleagues helped contribute to the success of our study. We gratefully acknowledge the Friends of the Konrad Lorenz Research Center (Verein der Förderer der Konrad Lorenz Forschungsstelle) and the Cumberland Foundation (Stiftung) for ongoing and long-term support: Josef Hemetsberger, Julia Rittenscho ber, Stefanie Filz, Larissa Schweiger, Helene Vesely to mention a few.

AI Tool Statement

The author(s) declare that no Gen AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.