Wild dolphins were played copies of their own whistles with the voice stripped out, and called back. · published 8 October 2026
Synthetic copies of each animal’s own signature whistle, with voice characteristics removed so only the contour remained, drew the same whistle back from the group in 8 of 12 trials - significantly more than whistles belonging to familiar or unfamiliar animals.
These are the working notes, unedited. Everything the video
states as fact comes from here, and nothing else may be stated as fact.
The sections on limits and on what the finding does not show are
part of the brief, not an afterthought — see
the method.
The background. Bottlenose dolphins each develop a signature whistle: a distinctive melodic contour, invented early in life and kept for decades. Other dolphins occasionally copy a particular animal’s contour. The obvious guess is that the copy works like a name - a way of addressing that individual - but a guess is all it is until the animal being addressed does something about it.
The crucial control. A whistle carries two kinds of information: the tune, and the voice producing it. If a dolphin responds to a copy of its own whistle, it might be reacting to how it sounds rather than to the pattern. To separate those, the researchers used computer-generated whistles with the voice characteristics removed, leaving only the contour. Anything the dolphin did in response had to be a response to the shape of the tune.
The study. Stephanie King and Vincent Janik, “Bottlenose dolphins can use learned vocal labels to address each other”, Proceedings of the National Academy of Sciences, volume 110, 2013. The playbacks were done on free-ranging dolphins off the east coast of Scotland, in the Moray Firth and St Andrews Bay.
The design. Three conditions. A synthetic copy of the focal animal’s own signature whistle (12 playbacks). A synthetic whistle belonging to a familiar animal from the same population (12). And a synthetic whistle from an unfamiliar animal in a different population entirely, modelled on captive dolphins (10).
What they found. Hearing its own whistle, the group answered by producing that same whistle back in 8 of the 12 trials. That differed from the familiar control (Barnard’s exact test, P = 0.04) and from the unfamiliar control (P = 0.001). Responses were scored by five observers who did not know which condition they were rating, with moderate agreement between them.
Limits
These are small numbers: 12, 12 and 10 playbacks. The researchers could not tell which animal in the group replied, so the response is properly credited to the group rather than to the individual whose whistle was played - which is a real gap in a claim about addressing an individual. Group sizes averaged around 13 animals. A post hoc check failed to confirm four of the synthetic stimuli as signature whistles. And copying is rare in wild dolphins, so the behaviour being studied is uncommon to begin with.
What this does not show
“Name” is an analogy and the authors use the more careful phrase, learned vocal label. Nothing here shows a dolphin understands a whistle as standing for a creature, or uses it to talk about an absent one. What was demonstrated is narrower and still unusual: an animal responding specifically to a learned sound pattern associated with itself, after the voice cues that might have given the game away were deliberately removed.