According to some estimates, there are about 1 billion dogs sharing Planet Earth with us. Most live in largely unrestricted conditions as village or pariah dogs, while only a small percentage of the world’s dog population live independently from humans. Most dogs have to maintain regular connections with humans to share our resources, which is obvious for companion and working dogs, but also holds true for strays and village dogs. The amazing evolutionary and cultural success of dogs would not be possible without their intricate and complex socio-cognitive skills, enabling them to flexibly adapt to the exceptionally variable anthropogenic niche.

But how do dogs manage to maneuver in a world full of strangers? Do they possess ‘ready-to-use’ preferences (and dislikes) to sort unknown people, just as humans do, into friend or foe? Or do dogs judge unfamiliar people based on a decisive ‘first impression’? These were the core questions of two of articles in the special issue ‘Canine Cognition’ in the journal Learning & Behavior.

There is ample evidence that in all likelihood, each dog that belongs to a particular human or small group of humans (traditionally these dogs are considered as having an ‘owner’) develops a unique relationship dubbed as ‘attachment’ to its owner. This is a strong and highly discriminative bond, because only the so-called attachment figure can serve as a secure base and safe haven for the dog when it faces mildly stressful new environments and strangers.

Although most dogs are basically friendly with unknown people, they get anxious when their owner leaves them alone with a stranger and they typically show elevated excitement upon reunion with their returning owner – unlike when they meet with a stranger. The capacity for the quick and training-free formation of attachment behavior is considered one of the basic features of the dog as a domesticated species – especially since it has been found that socialized tame wolves lack this relationship capacity with their caregiver.

However, we also know that attachment does not last for a lifetime, if the original bond is somehow interrupted. It has been found that adult dogs living at a shelter are ready for quick development of a new relationship with an unknown person – which requires no more than a few 10-minute long, dedicated sessions of friendly encounters between dog and human. These results suggest that although dogs have their exclusive bonds and preferences towards their owner, they retain considerable openness for positive interactions with strangers as well.

The study by Tan and colleagues and reported in the special issue gives further support to the theory of ‘first impression’. According to this theory, in those social species that initially treat strangers in a friendly manner, the benefits of doing so may outweigh the risks of occasional encounters with unfriendly strangers. Tan and colleagues found that although dogs prefer to rely on the signals from their owner, compared to a completely unknown person, in a food choice situation, in every other measure they turned to be very trusting with strangers – the dogs followed strangers’ pointing signals and trusted them enough to eat in their vicinity. In other words, dogs approached previously unknown human beings with a rather positive expectancy, which is definitely an adaptive feature for a species that has coexisted with humans for many millennia in quite variable environments.

Dog trainers, working dog owners, veterinarians and ethologists alike, rely on the basic behavioral feature of companion dogs that enables these animals to accept previously unknown people as temporary group members with whom they will more or less readily cooperate. This phenomenon makes it almost incomparably easier to conduct behavioral and cognitive experiments with dogs than with basically any other animal, mammalian or avian species. This is also true when we want to examine the human-related behavior of the other very popular companion animal, the cat – in most cases these animals can be reliably tested only at the home of their owners, and still, researchers have to expect high drop-out rates because of the incompatibility of cats with strangers. Meanwhile dogs basically can be tested anywhere and with anyone (while their owner is present), while the removal of individual primates from their usual environment and/or social group would be unimaginable in the case of behavioral or cognitive research. The same is true for the research of social interactions (such as social learning) in birds – keas, ravens, or budgerigars are routinely tested with other individuals whom they are familiar with to avoid initial conflicts with strangers.

The ‘first impression’ hypothesis suggests that while the initial approach to strangers can be open and positive, dogs will perform a quick assessment of the unknown individual and adjust their behavior based on the positive or negative experiences they collected. It was found earlier that dogs are capable of so-called third-party observations or ‘eavesdropping’, and consequently form a positive or negative expectation of the outcome of future interactions with the observed humans. Tan and colleagues set up an experiment that probed whether a dog would discriminate between a ‘totally unknown person’ and a stranger with whom they gained positive or negative experience just before the main test.

Contrary to the hypothesis, dogs remained basically equally trusting with each of these persons – they did not prefer the positive-stranger over the totally unknown one, nor did they avoid the negative-stranger in favor of the totally unknown one. Underlying this result may be the fact that the ‘negative exposure’ was (understandably) rather mild – here the assistant simply avoided any contact-making attempts from the dog in the pre-training phase, or in another scenario, provided false information about the location of hidden food. It is quite possible that if the dogs had experienced a harsher negative behavior from the stranger, they would have developed a more characteristic attitude against that person (as was suggested by Bhattacharjee and colleagues in case of the pariah dogs in India).

Of course there are other possible mechanisms that dogs might use in their assessment of strangers. In another article in the special issue, Kiss and colleagues tested whether dogs would use their owners as a template when they faced strangers –favoring those who are more similar to their owner against the dissimilar ones. Of course ‘similarity’ can stem from many sources. The recognition of kin and conspecifics can involve a variety of mechanisms such as early learning, phenotypical matching, or inherited cues. In human infants it has been shown that they prefer people who talk the same language, belong to the same race (i.e., have the same skin color) and wear similar items of clothing as their caregivers or friends do.

Many of these attributes could also be used by a dog to assess the similarity between their owner and a stranger. Kiss and colleagues concentrated on two types of features: inherent (way of moving, language spoken) and occasional (apparel) cues. They set up a scenario where the dog initially learned that its favorite toy can be freed from an inaccessible location with the help of a human assistant (the trainer). Then the dog was introduced to two equally friendly strangers, who showed similarities and differences to the dog’s owner. To instantiate occasional similarity, the stranger and the owner wore the same type of conspicuous apparel. To instantiate inherent dissimilarity, the stranger talked a foreign language and moved in a strange way (visibly limping). After the familiarization, dogs faced the ‘inaccessible toy’ scenario, where the two strangers were available as possible (although passive) helpers. The question was whether dogs would prefer to gaze at the stranger who showed occasional and/or inherent similarity to the owner.

According to the results, the gazing behavior of the dogs was influenced by the inherent features of the people – they preferred to look at the stranger who walked in a normal way and talked the same language as the owner. Occasional features (the same conspicuous apparel of the owner and the stranger) did not have an effect. These results are logical if we consider that clothing is highly variable even during the course of a day, therefore similarities or differences between the owner’s and a stranger’s apparel may be beyond the level of importance for a dog. However, integral features such as the usual locomotion and the spoken language can be useful for a dog when it assesses the safety of an unknown person. Notably, Kiss and colleagues, did not find that dogs avoid the oddly moving/talking person – they rather showed a more positive expectancy towards the stranger who had the same integral features as the owner had.

The two papers showcased in this post provide interesting information about how dogs possibly direct their initial steps towards unknown people. The scientific community usually readily accepts and exploits the trustfulness of dogs with strangers, without asking how this works. These studies took a deeper look at this phenomenon. Another major part of the studies is concerned with the different aspects (good and bad) of the dog-owner relationship, which is a popular research topic. However, there is still a need for more research such as these from Kiss and colleagues and Tan and colleagues –which reveal important facets of the dogs’ social inventory that helps them to navigate within a world full of strangers.