In shows like Lassie, I was always impressed at the amount of information a dog was able to convey to a human: ‘What’s that, Lassie? A little girl trapped in a building that you tried to reach but then couldn’t owing to the fire that caught alight to the fence surrounding it?’

I never owned a dog myself, so I assumed this lack of understanding of dog-talk was down to some inadequacy on my part. However, a recent study has looked into exactly what it is that dogs do convey when communicating with humans with some surprising findings.

We know that dogs are able to understand a lot of what we say to them. A border collie, ‘Rico’, learned an impressive 200 human words. As well as spoken commands, dogs also understand a lot of gestures, such as pointing or even a glance from a human.

We also know that dogs are able to convey certain types of information to humans. When food is hidden in a place that a dog knows about, but the human doesn’t, dogs are very good at showing the human where this might be (through a lot of barking, running about, jumping around etc.)

We humans communicate to inform others of information we think they may not know. From an early age, children are able to assess whether other individuals know what they themselves know, and inform them only when necessary. Indeed, sharing knowledge with others that they already know is seen as somewhat of social faux pas (as someone should inform older family members who insist on repeating such knowledge again and again). We do this even when sharing such information has no direct benefit to ourselves, and only helps the person we’re informing.

But can dogs do this? A recent study using forty pet dogs looked to see whether they would not only request an object that they were interested in, but also communicate to a human information about an object she wanted.

Firstly the dog was put in a room with an experimenter and experienced one of four possible conditions. Some dogs got to play with their favourite toy, but were ignored by the experimenter. Other dogs played with another toy (not their favourite) with the experimenter. Still other dogs watched the human ‘playing’ with an object (a hole puncher they used to make holes in paper). The final group had an object in the room with them (a porcelain vase) which neither the experimenter nor the dog interacted with.

After 60 seconds in one these four conditions, the experimenter pretended to answer a phone call, and left the room. This isn’t because dogs are thought to understand phones, but rather to ensure that the dog continued to hear the experimenter’s voice when they left the room, so they were reminded of their presence.

Another experimenter then came into the room, picked up the object, and hid it in one of four possible cupboards. That experimenter then left the room, and the original one returned, with no idea where the object was hidden. They began to look for the object, in four distinct phases. Firstly, they sat on the spot where the object had been previously. They then ‘looked’ for the object whilst sitting in one place, raising their arms and saying ‘Hmm., that’s weird. It was there, and now it’s gone. I don’t understand it.’

Thirdly, the experimenter addressed the dog directly, asking where the item had gone. Finally, the experimenter stood up and looked around, but remained silent. After these four phases were complete (taking only 35 seconds in total) the experimenter had to guess the location of the object, using the behaviour of the dog (‘what’s that, Lassie? The vase is hidden in the cupboard on my left?’)

If the dog’s communication led the experimenter to the correct hiding place, she would exclaim, ‘wow! Here it is! Great!’ If, on the other hand, the item was not where the dog had signalled it might be, she would say, ‘oh, too bad! It’s not here.’ If she received no signals from the dog, she did not look for the object, but instead would just lift her arms and shoulders and say, ‘too bad, we can’t find it’.

Each dog received four sessions, each of these containing four trials (each of the trials in one of the four possible treatments with the four different objects).

So- did the dogs tell the humans where the objects were hidden? Well, sometimes. The interesting finding here is that what the dogs communicated depended on who the experimenter was. When the experimenter was a stranger, the dog would alert them to the location of the object only when it was something that they were interested in (the two different dog toys, but not for the hole punch or vase). However, when the experimenter was their owner, the dogs would alert them to the location of the object in all cases, even when it was of no interest to them.

Because it seemed that dogs might be trying to tell their owners things the owners didn’t know, the experimenters did a second follow-up experiment. This experiment was similar to the first, except this time the two objects were both ones that the dog had no interest in, but one of which was of great interest to the owner, and the other was not.

Both the objects were presented at the same time, and both were hidden, to see whether the dog would point out the relevant one to their owner. For example, the owner would pick up and use a pair of scissors, and then pick up a roll of sellotape but not use it. When she came back into the room she would be holding a piece of paper and ‘look’ for the object she needed following the same phases of ‘looking’ as in the previous experiment. They also had another condition where one of the two objects was the dog’s favourite toy, as a comparison to these two ‘human-interest’ objects.

It was found that the dogs did direct the humans to an object in most cases (e.g. scissors or sellotape) but not to the particular object they needed. However, when the object in question was the dog toy, they directed the person towards it overwhelmingly more often than to the other object.

So, was I right about Lassie? From the second experiment, it does seem that dogs do not differentiate between objects that an owner needs and ones they don’t need (which could lead Lassie leading little Tommy to a shovel rather than to a fire-extinguisher, although it seems most likely that he’d just lead him to a dog toy). On the other hand, the first experiment showed that dogs were more likely to lead humans to the object they wanted when the human in question was their owner rather than a stranger. However, this was even the case when the human had shown no interest in the object. So, it seems that dogs are not able to tell which object a human wants based on the person’s past behaviour, despite their ever endearing motivation to try.



Kaminski, J., et al., (2011) Dogs, Canis familiaris, communicate with humans to request but not to inform, Animal Behaviour, doi:10.1016/j.anbehav.2011.06.015

Image credits:

Chicken/ Lassie cartoon: Doug Savage,