[Editor's update: A February 9 episode of Nova will feature Chaser.]
In a recent study in the journal Behavioural Processes, John W. Pilley and Alliston K. Reid have demonstrated an impressive level of language ability in a Border Collie named Chaser. It’s certainly easy to be most impressed by the fact that she knows the names of 1,022 objects, which she was taught over a nearly three-year period. Yet, from a scientific point of view and especially for scientists who study language acquisition and cognitive abilities, that is not as interesting as the other conclusions from the study “Border collie comprehends object names as verbal referents.”
These scientists who studied Chaser also conclud that she can distinguish between the names of objects and commands. In other words, she understands that names refer to objects, regardless of the action she is told to perform to those objects. She was asked to either nose, paw or take one of three toys in an experiment, and could successfully do so. Years ago, the study of a Border Collie named Rico amazed the world with reports that the dog had a vocabulary of over 200 words, did not demonstrate this ability. Though Rico may have been able to do so, the experimental design did not allow a definitive conclusion.
Pilley and Reid also concluded that Chaser understands categories of objects such as “ball,” “Frisbee” and “toy.” When asked to retrieve an object of one of these types, she was successful at choosing an item from the correct category. She is familiar with many items in each group. Interestingly, “balls” and “Frisbees” are categorized based on overall shape, but “toys” are those objects she is allowed to play with as opposed to those with which play is forbidden. The function, but not the form, of toys and non-toys is distinct.
The final conclusion in the study was that Chaser can learn the name of a new object by inferential reasoning by exclusion. That is, she can learn the name of a new object based on the fact that it is the only novel object in a group of objects whose names are all already known by her. This kind of learning cannot be based on associative learning mechanisms because the novel name and the novel object are not presented together.
What do you think about this study and what does it make you wonder about your own dog?