Stella and Bunny really talk? Recently, the internet has been flooded with videos where they are seen pressing buttons on a keyboard that plays pre-recorded words like “outdoor” or “play”.
And what’s better, in some videos, seem to form phrases by pressing several buttons in a row (play with Dad). These videos were immediately interpreted as proof that any dog, with a little training, can have a conversation with their master.
But what does science say?
Several researchers had previously tried to dialogue with animals, such as the great apes Washoe, Koko and Kanzi, or the parrot Alex, with quite convincing results. So it’s not impossible for Stella and Bunny to actually use keyboards to communicate.
However, it should be noted that to understand the words, Alex and the others were trained intensively by teams of professionals (psychologists, ethologists), up to 8 hours a day, for several years. This is not the case with Stella and Bunny: their owners are not scientists, they are not trained in training, and the have been trained at a very restful and irregularly shaped pace for about a year, limiting the chances of having such impressive results.
A study released by FluentPet, a company that manufactures animal keyboards, is being conducted in collaboration with researchers at the University of California to understand how animals can use this media. Bunny is one of many participants. To enroll your dog in the studio, you need at least one camera, but also, and above all, a keyboard that is strongly recommended to buy from FluentPet.
This is quite surprising, because it is very rare for study participants to put money out of their own pocket to participate in a study, it is usually the opposite! In addition, FluentPet’s involvement suggests a problematic conflict of interest: the company could choose to keep only the positive results, to better sell its keyboards.
Since Bunny is part of the study, his videos posted on social media are often interpreted as part of the study and are therefore understood to be reliable. However, the study is still ongoing and the results have not yet been published. Bunny and Stella’s videos, which are not scientifically validated, come from the teachers’ personal stories. The recordings are not raw, but cut and edited: does this serve to eliminate the leftovers, or rather serve to tell a nice story? We also don’t know how many of the essays are available: the impressive published essays may only be rare chances. In fact, this is recognized by Bunny’s own owner in this video:
Video, What About Bunny.
These videos that create expectation and make animal lovers dream can potentially be marketing tools. Stella and Bunny’s official websites, gift shops and books on demand suggest that this phenomenon can be a lucrative business, and urge to be careful with the content of the videos. Similarly, FluentPet, the company that manufactures some of these keyboards, pays some influencers (including Bunny’s mistress) for the number of clicks on shared links.
Potential biases
In this video, Bunny’s owner speaks to him in English, and Bunny responds with the keyboard:
Bunny, the ‘talking’ dog / What About Bunny.
However, the language of the “keyboard” is different from spoken English: it is derived from simplified English (there are few words on the keyboard, there are no prepositions like “a” or “from”, etc.) and is based on the use of buttons, not sounds.
Therefore, you have to start from three premises to claim that talk to their owners. The first is that the understand English (they know that the sound /pa.ase.o/ refers to walking). The second is that they dominate the keyboard language (a given button refers to walking). Third, they understand the correspondences between the two languages (they know that the spoken word “walk” refers to the same as, for example, the pink button on the left). Thus, a dog is taught a “keyboard” language using a third language (a human language), which represents an undoubted difficulty for the animal, which does not dominate either of these two systems.
For the first hypothesis, it’s hard to know if understand English, it’s just not proven. For the second hypothesis, it is impossible know if actually master keyboard language, for several reasons.
These videos do not control some well-known and feared biases of researchers. The meaning of the phrases produced by is often reconstructed by the amas themselves: this bias of anthropomorphism is to attribute human ideas or behaviors to animals. For example, when Bunny combines “sound” and “adjust” in the video below, its owner explains that Bunny recently decided to use these words to say “shut up”. How can we be sure bunny doesn’t mean “lower the sound of the keyboard!”, for example? The original “phrase” is too vague to accurately express what the bitch means, and what we interpret then is usually… what’s good for us.
Bunny ‘talking’ about cats / What about bunny.
Second, the keyboard, which is supposed to materialize simple English, is also problematic because the layout of the buttons is anthropocentric. For example, words are grouped by grammatical class (a group of buttons for names such as cat, another for verbs such as playing, etc.), which does not necessarily make sense from the canine perspective. Although are likely to have their own language, this keyboard is unlikely to be a good materialization of it, for intuitive use.
A classic bias in psychology is the “clever Hans” effect, named after a horse famous for its ability to answer complex questions by hitting the ground with its hoof, but which actually used the behavioral signals of its audience (e.g. the inclination of its heads) to know when to stop hitting the ground. In Bunny or Stella videos, we cannot dismiss the hypothesis that their owners are intentionally giving clues about the “right button” that they must press (by the look, the orientation of the body…). Especially since we can’t see what’s going on behind the camera.
Morgan’s “canon” states that we should not attribute high-level cognitive abilities to animals if their actions can be explained by lower-level capabilities. Therefore, in the absence of repeated controlled psychological experiments, we should avoid claiming that these speak with keyboards, and formulate simpler and safer hypotheses.
The simplest explanations
It is possible to account for these observations without concluding that really understand the meaning of the words used. In particular, it is not possible to rule out the idea of associations being established only between a behavior (pressing the “toy” button) and a consequence (playing with the toy).
In other words, it is possible to learn how to claim the toy by pressing the right button, without really understanding that it is a word or understanding the word, just as B. F. Skinner’s rats knew which lever to press to get food or drink without knowing how to speak. This is especially true for buttons with abstract or complex meanings, such as “foreign” or “I love you”.
Human language is a productive and generative system: once the basic rules are mastered, an infinite number of phrases can be generated with a limited number of words. In this case, there’s no evidence that have that ability. It is quite possible that these combinations are fixed sequences, learned by the to trigger a specific response. For example, they may have learned that when they press “play” and then “ball”, the ball is played. Similarly, when you press a button and insert a coin into a vending machine, you get a can of soda, but you can’t conclude that you’re talking to the machine.
Some of the word combinations also seem to suggest that arbitrarily pick them, until they hit. For example, in this video, the owner interprets the combination “come play” followed by “I do want to eat” as “Stella wants food to be put into her toy”. Stella, like any dog, is happy to receive food, but this doesn’t mean she’s deliberately ordered it.
A more parsimonious explanation is that, in the face of her mistress’s lack of reaction to “watching”, Stella merely pressed other buttons, without consistency, to trigger in her owner a response that benefits her. In addition, in this video, Bunny’s owner explains that the less she responds to her dog’s requests, the more messy the combinations produced.
How to check if dogs talk?
We have seen that must master two languages, English and keyboard, to communicate properly with their owner. To facilitate learning and verify that master keyboard language, they must be fully immersed in it: all communication should be done only through the keyboard, without ever using spoken words.
It must then be shown that understand the meaning of the words used. In the same way that has been tested with the Chaser dog, you can give orders through the keyboard: if you understand combinations you have never heard before, you can conclude that they understand the meaning of each word. For example, if from “touching ball” and “collecting stuffed animals”, a dog can understand “touching stuffed animals”, it is able to extract the meaning of each word. Once this check is done, we can do other experiments to analyze your production capacity.
It is not impossible for Stella and Bunny to use these keyboards as language, but in the absence of further scientific evidence, we must be cautious. However, we can recognize that these keyboards stimulate the intelligence of dogs, keep them busy and strengthen the bond with their owners.
One last crucial point: we don’t need marketing tricks to communicate with our pets: many studies show that we already understand each other very well. So, trust yourself!
Mélissa Berthet, Docteur en biologie spécialisée en comportement animal, Ecole normale supérieure (ENS) – PSL and Léo Migotti, Doctorant at Sciences Cognitives, Ecole normale supérieure (ENS) – PSL
This article was originally published in The Conversation. Read the original.
