Can Dogs Eavesdrop?
Do dogs prefer nice people? Researchers look at whether dogs notice who is 'nice' and 'not nice', and how they respond.
By Zazie Todd, PhD
![]() |
Photo: Sophie Louise Davis / Shutterstock |
By Zazie Todd, PhD
Several studies suggest that dogs pay attention when
humans are nice to someone, and preferentially approach the ‘nice’
rather than ‘not nice’ person. A new study by Esteban Freidin et al
investigates dogs’ eavesdropping abilities in the search for further evidence
on this ability.
Studies of canine eavesdropping typically involve a scenario
in which two people have food and another person, playing the role of ‘beggar’,
approaches to ask for some. One person gives food to the beggar, while the
other refuses. After observing this, dogs are released to see which person
they will approach first.
In a study by Sarah Marshall-Pescini et al (2011), dogs that observed generous versus selfish donors later chose to approach the generous person. This preference was strongest when both donors spoke and gestured towards the beggar. It was still found when the information available to the dog was verbal only, but dogs did not seem to be able to interpret gesture on its own. This study manipulated the behaviour of the donor, but the beggar’s actions remained the same.
In a study by Sarah Marshall-Pescini et al (2011), dogs that observed generous versus selfish donors later chose to approach the generous person. This preference was strongest when both donors spoke and gestured towards the beggar. It was still found when the information available to the dog was verbal only, but dogs did not seem to be able to interpret gesture on its own. This study manipulated the behaviour of the donor, but the beggar’s actions remained the same.
The current experiment uses a version of this in which the
dog observes how the ‘beggar’ responds. At the start, the two experimental assistants
went up to the dog and showed it their plates, which contained cornflakes (for
humans) and sausages (for dogs). Then they took up positions at opposite sides
of the room, and began to eat the cornflakes at a regular pace.
In the main condition, a person playing the role of beggar
came in, approached each person in turn and asked for food. They always gave
the beggar a cornflake, but he responded differently to each person. With the
“positive” person, he ate the cornflake and said “So tasty!” With the
“negative” person he rejected the cornflake, put it back on the plate, said “So
ugly!” and turned his back to them. This condition therefore included both
Gestural and Verbal information.
There were two other conditions. The Gestural condition was
the same, except the beggar did not speak. In the Verbal condition, the beggar
spoke the same words as in the main condition, but he did not gesture (and
hence did not receive cornflakes).
In each condition, the beggar had six interactions (i.e.
three with each person). After the beggar had left the room, the dog was
released and was free to approach the people.
The experiment was conducted at a location familiar to the dog, whether that was its home or a day-care. Seventy-two dogs took part, although some had to be excluded because they did not approach anyone (some of these dogs did not pay attention to the interaction and some seemed fearful). After this, there were fifteen dogs per group.
![]() |
The experimental set-up. Source: PLoS One |
The experiment was conducted at a location familiar to the dog, whether that was its home or a day-care. Seventy-two dogs took part, although some had to be excluded because they did not approach anyone (some of these dogs did not pay attention to the interaction and some seemed fearful). After this, there were fifteen dogs per group.
In the main condition, 13 out of 15 dogs approached the
person who had received a “positive” response from the beggar. This suggests
the dogs had successfully eavesdropped on the interactions.
However, in the Gestural and Verbal conditions, the dogs’
choices were not significantly different from chance. This suggests that
information from both gestures and speech was needed for the dogs to be able to
choose the person who was most likely to give them something nice to eat.
Another version of the experiment was conducted, in which
the two people changed places three times during the interaction sequence.
After this, the dogs did not choose at a rate different from chance. It may be
that they used the location to help them remember, that they were not able to
discriminate easily between people they had only just met, or that they were
confused by the multiple changes of place.
In a further version, in which the beggar acted the same but
the donors were replaced by bowls of food, the dogs’ choices were no different
from chance.
The researchers say, “we found that dogs could choose which
donor to beg food from based, not on the behaviour of the target individuals
(the donors in our protocol), but on the reaction that an interacting person
(the beggar, who was absent at the time of choice) showed towards them. This
finding may indicate a level of subtlety in dogs’ eavesdropping not found
before.”
The dogs could use gestural and verbal information combined
to decide who to approach for food, but without both types of information they
did not know who to approach. And unlike previous studies, this was based solely on the reactions of the beggar.
When do your dogs beg for food?
Reference
Freidin, E., Putrino, N., D’Orazio,
M., & Bentosela, M. (2013). Dogs’ eavesdropping from people’s
reactions in third party interactions. PLoS One, 8(11), e79198.
Marshall-Pescini, S., Passalacqua, C.,
Ferrario, A., Valsecchi, P., & Prato-Previde, E. (2011). Social
eavesdropping in the domestic dog. Animal Behaviour, 81(6), 1177-1183.