RESEARCH GUIDE / 05
Toy words: exceptional dogs, specific evidence
In a 2025 study, selected Gifted Word Learner dogs extended familiar labels to new objects used in the same way during play. The result is about unusual word-learning dogs in a defined task. It is not a vocabulary target, an IQ scale or proof that a dog thinks like a child of a particular age.
Editorial team / by PawLine. Source-checked November 13, 2026. Not independently expert-reviewed.
Who reached the final task?
The primary paper reports eleven dogs recruited, ten participating after one exclusion, eight meeting the initial learning criterion and seven completing all phases. Those stages matter. The final result does not describe an unselected group of ordinary household dogs, and the authors explicitly warn against extending it automatically to the general dog population.
Fugazza et al. (2025): functional object-label extension ↗Source for this explanation, not a review of Taskwise.
What was new about the objects?
The researchers used objects assigned to different ways of playing. Later, dogs experienced new objects being used in those ways without hearing their labels during that exposure. The subsequent choices provided evidence for extension based on use rather than a shared appearance. This is a summary of the research question, not a home training protocol.
Fugazza et al. (2025): functional object-label extension ↗Source for this explanation, not a review of Taskwise.
Two categories do not mean two possible choices
Although the study concerned two functional categories, the classification tests included familiar objects alongside the new ones. The dogs had at least eight objects available. A casual calculation based on two answer options would describe the wrong task. Our comparison avoids converting the result into a percentage that supposedly ranks a pet.
Fugazza et al. (2025): functional object-label extension ↗Source for this explanation, not a review of Taskwise.
Similarity is not equivalence to a child
The authors discuss functional similarities to aspects of early human word learning while leaving underlying mechanisms and the breadth of the ability open. They also note that dog categories may be more limited. A parallel in one process is not a human mental-age conversion. A dog that does not learn toy names has not failed a general intelligence requirement.
Fugazza et al. (2025): functional object-label extension ↗Source for this explanation, not a review of Taskwise.