I am a lifelong dog owner, and like many dog owners am often impressed with how smart my dogs have been. They pick up on subtle body language and non-verbal cues, they seem to understand specific words, and they are capable of successfully communicating their wants and desires. My latest dog is an Australian shepherd, who is both smart and willful. Any attempt at training him to do what we want results in him equally training us to do what he wants. An of course we love them and the emotional connection is real and bidirectional. Dogs and humans have evolved a symbiotic relationship.

Still, I was very skeptical when I heard about a recent social media phenomenon – posting videos of dogs using a soundboard to communicate. After watching the videos I am completely unimpressed, and my skepticism has been supported. It turns out that this is mostly just the old “Clever Hans” effect, falling into the same trap that all attempts to teach animals to communicate have risked.

In the early 20th century, Wilhelm von Osten, who was a mystic and phrenologist (among other things) showcased his horse, Hans, who he claimed could not only do arithmetic, but could read, solve problems, track a calendar, and other tasks. Hans would communicate by tapping his hoof the correct number of times. Osten probably really believed in Hans’s abilities, and he showcased them far and wide. However, when psychologist Oskar Pfungst investigated Hans he found that the horse was simply responding to non-verbal cues from his owner, essentially noted when to stop on the correct answer. He initially removed the trainer from the area, but Hans was still able to perform. However, he then made sure that no one present knew the answer, and then Hans could not perform. Hans needed cues from people to know when to stop.

Perhaps a more complex example is Koko the gorilla, who his trainer claimed could use 1,000 signs and understand 2,000 spoken words. The evidence presented to backup these claims is mainly in the form of videos. But if you watch these videos you notice that Koko’s communication is very hit or miss, and requires a lot of clever interpretation on the part of the trainer. So much so that it is possible to conclude that most if not all of the communication is happening in the mind of the trainer.

Since Koko is not making full coherent sentences, just stringing together 1-3 words, a lot is left up to interpretation. Is Koko really combining words logically to convey new meaning, or just signing until they get what they want. It does seem that Koko has associated some signs with physical objects or with actions. But is this association really language? Sometimes Koko makes mistakes, and has to keep going until the trainer gives them positive feedback. Sometimes they are “kidding” when they get things wrong, and sometimes you have to infer what they might mean.

There is no questions gorillas are very smart animals, and have an ability to learn a lot of information. But the evidence simply did not convince the scientific world that they possess that level of human-like language. Koko turned into a cautionary tale that has hung over animal language research ever since. Again – I am not saying that animals have no language. They clearly do. Dogs and primates in particular have been shown to understand words spoken by humans, not just intonation and body language. The real question is – can they use words to communicate, and can they string words together logically? So far, the scientific community is not convinced.

Let’s get back to the dog videos – what I see in them is a clear Clever Hans / Koko phenomenon. Copper in particular is unimpressive. When asked a question he puts his paw out and hits whatever button is right in front of him, without even looking at the buttons (shades of facilitated communication). Also, there is a very limited set of buttons, and everything relates to doggy interests. So the dog could hit literally any button and the owner could make sense of it. In the “Bunny” video the dog hits the button for “home” several times and the owner says – “Yes, we are home.” What’s that button even there for? When at home, the owner concludes the dog is telling them they are home, and if not at home I presume they would conclude the dog wants to go home. It doesn’t seem like there is any possible wrong answer.

This is a pretty clear example of wishful thinking, and the videos are certainly not compelling evidence that actual communication is going on. At most these dogs are learning to associate certain actions with getting a reward of some kind. There is also likely a selection bias going on in terms of which videos segments are uploaded to social media – it’s reasonable to conclude we are seeing the best evidence there is.

This all may seem harmless and fun, but the underlying phenomenon can get very serious and have profound consequences. Similar methods are used to communicate with non-verbal people, and suffer all of the problems of animal language research. It is very easy for the communicator or facilitator to impose their own mental processes onto their client. Far from giving a voice to a non-verbal person, they are stealing their voice (even if they mean well). Sometimes this can also lead to very dark places, such as using such methods to make serious accusations against others.

I suspect this is a lesson we will have to learn over and over again. More than a century after Clever Hans and the true phenomenon underlying his performance was revealed, it’s happening all over again.