鈥淐hicken. Chicken. Chicken.鈥� I鈥檓 woken by the sound of my own voice looping madly. My sleep-addled mind can鈥檛 work out what鈥檚 going on. What time is it? Am I still dreaming? Then Laika, my dog, bounds into my bedroom.
Laika is a good dog, or at least I thought she was. I recently bought a touchpad called FluentPet, which allows you to record your own voice onto an array of buttons so that pets can push them and 鈥渟peak鈥� words. Laika can ask me to fill her water bowl, go to the park or scratch her belly. The trouble is, the whole thing has backfired and she has started asking for treats at all hours. That has made me wonder just how far her mental abilities go. Is her doggy mind really clever enough to guilt-trip me into giving snacks? Come to think of it, does she know what a snack is?
Advertisement
Most dog owners, like me, have wondered about these deep questions, and we will probably never get fully satisfying answers. But thoughtful experiments and emerging technologies are starting to provide a clearer picture of the canine mind. So, with Laika at my side (and a pocketful of treats), I set out to discover what dogs really have going on between their fluffy ears.
On one level, it is surprising that dogs and humans can communicate at all. Our two species separated on the evolutionary tree roughly 95 million years ago, and our last common ancestor had a brain smaller than that of a mouse. This means humans and dogs found their own distinct paths to their large brains, and consequently there are myriad differences between them, says evolutionary neuroscientist at Harvard University.
Clearly, though, dogs and humans can understand each other to some extent. This can probably be partly explained by two things. First, dogs evolved from wolves, which traditionally live in large family packs. 鈥淲olves have all these behavioural traits to organise a family, so that gives dogs the ability to join a family, which in this case is human,鈥� says at E枚tv枚s Lor谩nd University in Hungary, who directs the Family Dog Project, a long-running research endeavour. Second, dogs have lived alongside humans for millennia, which has created selection pressures for dogs that are better able to communicate with humans. 鈥淭hey can let us know what they want and they can understand, to some extent, what we鈥檙e saying to them 鈥� so that鈥檚 probably produced changes in their brain,鈥� says Hecht. We know, for example, that that respond strongly to human vocalisations like laughing and crying.
But it is all too easy to project meaning onto these interactions that isn鈥檛 there. Social media is awash with videos in which dogs appear to combine words into sentences on touchpads. In one, a dog called Bunny at which point his owner finds a splinter-like object between his toes. Other dogs use the pads to express emotions like disgust, love and joy. It is fair to say all this is far beyond Laika鈥檚 abilities, and I can鈥檛 help doubting how much these other dogs really know what they are saying. 鈥淵ou鈥檙e getting into very dangerous waters here,鈥� says Mikl贸si, who says he is 鈥減essimistic鈥� about such claims.
That said, recent research suggests dogs do have some understanding of what words mean. In August, Federico Rossano at the University of California, San Diego, and his colleagues found that dogs trained to use soundboards , regardless of whether these were said by their owner or a stranger or spoken live or from a recording. This suggests that the dogs weren鈥檛 reading their owner鈥檚 body language or randomly pressing buttons in the hope of something nice happening, but that they do associate some specific context with the button, says canine psychologist at the University of Portsmouth, UK.
Often, behaviours like these can be explained by association, with dogs learning to read our actions and make predictions about what will happen. Based on subtle gestures, your dog can know that you are about to go on a walk, perhaps before you are even consciously aware of it, says Mikl贸si. But that is distinct from dogs having a clear mental representation of what a 鈥渨alk鈥� is. When dogs press buttons, 鈥渋t doesn鈥檛 demonstrate that the dogs have an expectation that it changes the owner鈥檚 mind鈥�, says Kaminiski.
Do dogs have a theory of mind?
All of which makes me less concerned that Laika is manipulating me 鈥� but I have begun to worry whether she even knows I exist. In an effort to find out, I put Laika on her lead and take the train to visit Kaminski at her Dog Cognition Centre in Portsmouth. There, along with her PhD student , she has been testing the extent to which dogs understand others鈥� thoughts, an ability referred to as 鈥渢heory of mind鈥�.
For her first test, Laika sits in front of two open boxes with treats inside. The Dog Cognition Centre has sourced the 鈥渟melliest sausages鈥�, says West, to really motivate Laika. The box on the left is opaque, except for the front panel, which allows her to see inside. The second box is transparent. With Laika waiting expectantly, I enter and stand on the opposite side of the room, so that I can鈥檛 see the treat in the opaque box, but can see the one in the transparent box. If Laika is aware of what I can see, then she is more likely to stare at the treat in the transparent box, anticipating that I will pick that one up and give it to her. 鈥淲e are asking if she makes the inference that your behaviour is based on what you can and cannot see,鈥� says Kaminski. Laika looks at neither treat but simply gazes at me, which doesn鈥檛 exactly instil me with confidence about the richness of her inner world. For context, human infants reach the level of theory of mind required for this test at around 12 months.
Still, we press on to the next test, named 鈥渃himpanzee chess鈥� because it is more often used to test the cognitive abilities of primates. It involves me competing with Laika for pieces of sausage hidden under flowerpots, some of which I sneakily remove. If Laika guesses which piece of sausage remains by choosing the right pot, it indicates she has some awareness of what I have seen in the past. Human children reach this level at around 24 months. Against my expectations, Laika beats me in two consecutive games. Of course, we can鈥檛 draw conclusions based on these two trials alone. But I鈥檒l take it: there may be mental representations in the old dog yet.
More rigorous experiments that applied these kinds of tests to many dogs have found they are , and there is some evidence that dogs have the level of theory of mind required to play chimpanzee chess. Yet even chimpanzee chess falls short of testing a higher-level theory of mind, which involves the ability to understand that other beings might hold false beliefs about the world. Human children attain this level at roughly the age of 4. 鈥淚nterestingly, it鈥檚 also the age when children start lying because they understand there are certain things in the world that you do not have knowledge of, which they can manipulate,鈥� says Kaminski. All of this probably means that Laika doesn鈥檛 have a hugely well-developed mental grasp of who I am and what I know about the world.
Having learned what I can from behaviour studies, I turn to the booming field of canine brain imaging. For years, researchers have been scanning people鈥檚 brains while they perform mental tasks, and these studies have revealed the patterns of neural activity that underlie everything from recalling memories to conjuring up mental imagery. People are asked to picture an orange, for example, and we then find out the neurons that do that job. More recently, researchers have tried the same techniques on dogs, examining their , faces and emotions as well as identifying brain regions responsible for language representation, attachment and relationship dynamics with humans.
Canine brain scans
The difficulty with doggy brain imaging, however, is ground truth. There is no way of getting dogs to picture an orange or even asking them what they are experiencing. A dog might be wagging its tail inside an MRI machine, but we can鈥檛 know if that actually means it is happy 鈥� or even if happiness is a valid emotion to project onto a dog.
However, a start-up in India called Dognosis is making progress by building EEG headsets for dogs that aim to 鈥渄ecipher the canine mind in real time鈥�. The idea is to connect dogs鈥� internal neural signatures with outward behaviours in their natural environments. If the dog is wagging its tail while running through a field, we can be more confident of what鈥檚 going on in its inner world.
Founder Akash Kulgod is currently refining a prototype device using dogs鈥� impressive ability to sniff out cancer samples. He shows me a video in which Dognosis鈥檚 鈥渃anine experimenter in chief鈥� 鈥� a beagle called Snow 鈥� can roam freely among eight different boxes in a room, one of which contains a mask that someone with cancer has breathed into. When Snow smells this box, it sparks a fairly consistent electrical signature in the EEG headset. In the first instance, this could enable Kulgod to identify when dogs smelled a disease much more reliably. (Dogs are legitimately great at identifying disease, but can start to conveniently sniff it out everywhere after a while, in the hope of getting more treats.) Dognosis is now partnering with hospitals in low-income countries, where biopsies and blood tests for cancer are harder to access than dogs.
A dog might wag its tail, but we can't truly know if that means it is happy
This research highlights one of the hardest problems when it comes to knowing what鈥檚 going on in the minds of dogs. Because of their incredible olfactory abilities, so much of dogs鈥� perception is based on smell. That means there is a huge gulf between the human and dog 鈥渦mwelt鈥�, or sensory experience of the world, says Kulgod. 鈥淭hey鈥檙e seeing with their nose, quite literally.鈥�
So even if there is some kind of mental representation of 鈥渃hicken鈥� when Laika presses the relevant button on FluentPet, it is probably vastly different from my own. This difference surely extends to how dogs perceive other beings in the world, say Hecht. 鈥淲hat are humans for dogs? Are we another dog? Are we a strange creature? It鈥檚 very complicated to find out how they [imagine] us,鈥� says Mikl贸si. He welcomes the research possibilities that wearable brain imaging would open, but argues that such technology could get in the way of everyday human-dog relationships. 鈥淚f you are a dog owner, you don鈥檛 really need that,鈥� he says.
Still, Kulgod says he is confident that technology can 鈥渁mplify our ability to connect with dogs鈥�. He hopes that Dognosis鈥檚 headset will, at some point, be able to find consistent neural signatures of mental states, so the device can tell you when a dog is experiencing anxiety, say. And now Laika and I are helping with the effort. Kulgod sent me one of his headsets and Laika (pictured above) looks fantastic in it. I鈥檝e been recording her brainwaves as I say her name to see what鈥檚 going on in her mind. But let鈥檚 face it: she鈥檚 probably just thinking about chicken.
