Scientists make shocking discovery: dogs cannot read our minds

It's like The Village of the Damned. Image Credit: Tanakawho

“Canine telepathy?” asked one press release, referring to the similarly provocatively titled “Can your dog read your mind? Understanding the causes of canine perspective taking”, published online this week in the journal Learning and Behavior. The answer, of course, was no, although that was hardly the question being posed. The paper was not a cross-species ESP study, with dogs sitting in one room and being asked to guess what geometric shapes humans sitting in the adjacent room were being shown, but something a bit more practical. The authors examined how well different types of canids*(domestic dogs as well as wolves) were able to spot which of two humans was the attentive one – the one that would be more likely to reward them with treats.

While many animal lovers claim that dogs can tell when their owners are angry or sad – or merely plotting to take them to the vet – there is little more than anecdote to back this up. Yet it does often seem like dogs know what we’re up to. The dogs in my father’s household, who wouldn't exactly be described as well behaved, transform into circus-level performers when “chew-chews” (dog treat jargon) are about to be doled out. Treats are distributed after the correct execution of a command. Having spent far more time on “sit” than alternatives such as “roll over” or “play dead”, the dogs immediately place their butts on the ground as soon as a human picks up the biscuit jar. But do they believe that sitting somehow pleases us (not entirely false, though technically the desired behavior was obeying of commands) or have they just learned through trial and error that sitting = chew-chew? Forget whether dogs can read our thoughts, can they even comprehend that we have thoughts, that we pay attention (or don’t pay attention) and, if not, how do they manage to adjust their behavior to suit our wishes?

To answer some of these questions, scientist tested four types of canids – pet dogs (tested indoors), pet dogs (tested outdoors), animal shelter dogs, and tame wolves.† After preliminary “pre-training”, dogs were given the choice of begging for treats from either an attentive human – one that could see the dog – or a human whose attention was somehow diminished. The obscuring of attention was achieved by one of four means – the human 1) having their back turned to the dog, 2) reading a book, 3) holding a camera over their eyes or 4) wearing a bucket over their head. If that last one sounds a bit ridiculous, that was sort of the idea – to create a situation the animals were unlikely to have encountered previously.

Results showed that not only could domestic dogs read some signs of inattentiveness, so could wolves. All animals fared well in the task that involved choosing the attentive human over the one with their back turned. This provides possible evidence against the argument that dogs’ ability to read human mannerisms is a product of their domestication. In fact, in general, animal shelter dogs (whose home environments lacked the enrichments afforded to pet dogs) performed closer to wolves than to pet dogs. Response to the obstacles of camera and bucket were not extraordinary in most subjects, but in tasks where the inattentive human held a book, pet dogs (in both indoor and outdoor settings) did notably better at begging from the attentive human rather than the distracted reader. The book scenario was chosen specifically to simulate the kind of hurdle to getting attention that the average pet dog experiences in their daily routine. As expected, wolves and shelter dogs largely missed this subtle cue.

Additionally, when the challenging bucket task was repeated for long enough that animals might have a chance to learn from reinforcement (20 trials each), both dogs’ and wolves’ performance improved in tasks where they were rewarded for begging food from the attentive human. However, most failed to get the hint when treats were given out by the inattentive bucket-headed human.

Overall the researchers concluded that conditioning is important for canids to learn which human behaviors and mannerism will improve their chances of getting rewards, but that “a willingness to accept humans as social companions” was also key. Domestication, it seems, was not a requirement. What does all this say about whether dogs can conceive of thoughts in humans? Nothing. The authors readily admit that their tests were not designed to determine the mental process behind the animals’ decisions. That’s probably a wise move given that it’s pretty challenging (impossible?) to study cognition in non-verbal animals. Nevertheless they’ve managed to provide some insight as to how dogs “know” when we’re more likely to be receptive to their needs. Meanwhile, domesticated cats have long known that they can turn an inattentive reader into a receptive provider of affection and dinner simply by sitting on the distracting book. I draw no conclusions from that last (non-experimentally obtained) observation. Any inferences about the relative intelligence of cats and dogs are entirely of your own making.

* Canidae is a family of mammals that includes domestic dogs (Canis lupus familiaris) and wolves (various subspecies of Canis lupus) and well as foxes, coyotes and jackals.

† The whole indoor/outdoor business with the pet dogs was done to match the necessary testing conditions for the other animals. Wolves were tested outside, whereas shelter dogs were tested inside.