Long-term memory of words has a crucial role in the developing abilities of young children to acquire language. In dogs, the ability to learn object labels is present in only a small group of uniquely Gifted Word Learner (GWL) dogs. The ability of these dogs to acquire large vocabularies consisting of hundreds of names of dog toys through naturally occurring interactions in human families presents them as a valid model for studying language-related cognitive mechanisms. As they are very rare, little is known about the mechanisms through which they acquire such large vocabularies. In the current study, we tested the ability of five GWL dogs to retrieve 12 labelled objects two years after the object-label mapping acquisition. The dogs proved to remember the labels of between 3-9 objects. The results shed light on the process by which GWL dogs acquire an exceptionally large vocabulary of object names. As memory plays a crucial role in language development, these dogs supply a unique opportunity to study label retention in a non-linguistic species.

Subjects: 5 dogs participated in this study (2 females, all Border collies). These dogs all proved to possess a vocabulary of object labels (Fugazza et al., 2021). The owners reported that their dogs learned the labels through naturally occurring play interactions. During these play interactions, the owners would introduce a toy to the dog by saying the toy's name, and then allow the dog to fetch the toy and pull on it several times, while they repeatedly say the name. An example of such a teaching process is available on this link. All of these dogs participated in a study conducted in 2020 (Dror et al., 2021).

The toy names learning procedure conducted as part of Dror et al., (2021): In December 2020, each dog owner received a box containing 12 toys. On a predetermined date, the owners were instructed to open the box and from that day on, they had one week to teach their dogs the name of the 12 toys. The owners were given the freedom to teach the dogs the names of the toys in any way they saw fit and spend as much time as they had available on the task. For a more detailed description see Dror et al., (2021).

Testing procedure for the current study and those conducted by Dror et al., (2021): At the end of the week, the dog's knowledge of these toy's names was tested. Parts of the tests conducted in December 2020 were broadcast and are available on this link. Those tests, as well as those conducted in the current study, were done online. The owner placed the toys in one room and sat in a different room out of the toy's view. In each room, the owner placed a tablet/smartphone/laptop and connected this device to the online meeting platform. In this manner, the experimenter could monitor the dog's reaction and instruct the owners on what to do, in real time.

The test was carried out as follows:

1. The owner was sitting with the dog in a room.
2. The experimenter told the owner which toy should be retrieved (the order of the toys was randomly determined).
3. The owner asked the dog to retrieve the toy by pronouncing its name (typically: ‘Bring < object name>!’).
4. The dog left the room and entered the room with the toys on the floor, where it selected a toy by picking it up and bringing it to the owner.
5. If the dog retrieved the correct toy, the owner praised the dog and briefly played with the retrieved toy.
6. If the dog did not retrieve the correct toy:
   • In the initial test examining the dog's label knowledge conducted by Dror et al., (2021); the trial was repeated, and the repetition was not included in the data analysis. If the dog made a second mistake, the owner retrieved the toy without showing it to the dog or stating its name.
   • In the memory tests conducted by Dror et al., (2021) and in the two-year memory test of the current study; the experiment was interrupted for a few minutes while the owner told the dog to wait in the room and went to remove the toy without showing it to the dog.
7. After the dog (or the owner, in case of repeated mistakes by the dog) retrieved the correct toy, the experimenter instructed the owner to ask for the next toy.
8. Whenever there were only 3 toys left on the floor, the experimenter instructed the owner to place all the toys back. Therefore, in the initial test examining the dog's label knowledge conducted by Dror et al., (2021), the number of toys from which the dog could choose varied between 12-4. However, in the memory tests conducted after one and two months by Dror et al., 2021, and in the current two-year memory test, the 12 toys were divided into two sets of 6 toys, and each set was tested on a different occasion. Therefore in the memory tests, the number of toys from which the dog could choose always varied between 6 and 4.

References:

• Fugazza, C., Dror, S., Sommese, A., Temesi, A., & Miklósi, Á. (2021). Word learning dogs (Canis familiaris) provide an animal model for studying exceptional performance. Scientific reports, 11(1), 14070. DOI: 10.1038/s41598-021-93581-2
• Dror, S., Miklósi, Á., Sommese, A., Temesi, A., & Fugazza, C. (2021). Acquisition and long-term memory of object names in a sample of Gifted Word Learner dogs. Royal Society Open Science, 8(10), 210976. DOI: 10.1098/rsos.210976