With Callie in the scanner, the shimming and localizer sequence took less than a minute. She knew the routine. For the functional scans, we modified the hot dog experiment. Instead of holding up a hand signal for ten or fifteen seconds, Andrew would hand me a swab, and I would hold it in front of Callie’s nose for a few seconds. She would continue holding still to allow enough time for the hemodynamic response to peak, and then I would reward her with a hot dog treat. It was the same as before except we would insert a smell during the middle of the repetition.
To get her used to a cotton swab being shoved in front of her face, Callie and I had practiced this at home. The first few times she backed away, but she soon realized nothing bad was going to happen and just sniffed.
With the functional scans running, she performed flawlessly. Each swab was presented eight times in random order. It took two functional runs lasting six minutes each, and then we were done. Another four hundred scans in the can.
McKenzie, on the other hand, was not having a good day. She did not like the smells or the swabs coming at her. From the control room, I could see that her brain images were moving. Even though we acquired nearly five hundred images, most were unusable. She would have to come back another day after more training with the swabs.
Like we did in the hot dog experiment, Andrew and I analyzed the smell data individually for the two dogs as well as combining their brains. Both analyses yielded surprising results. When we combined their brains, we were able to identify the parts of the brain that reacted to the different smells in both dogs. This showed the common regions of activation. In contrast, the individual analyses told us how the dogs reacted differently. We focused on two comparisons.
First, we compared the brain activity of dog scents to human scents. This was done by ignoring whether the scent was familiar or strange. We simply averaged all the dog scents together and all the human scents together and compared the two brain patterns. The first thing that popped out was that the canine smells strongly activated the olfactory bulb and the frontal cortex above it. I suspected that this was because dog urine is a more potent stimulus than human sweat.
When we compared the familiar scents to the strange scents, ignoring whether it was from a dog or a human, once again, we found more activation of the olfactory regions to the strange smells. This demonstrated that the olfactory activation is controlled not just by the potency of the smell but also by its familiarity. Familiar smells don’t require much brain processing. Strange ones do. Consistent with this interpretation, the dogs’ own urine didn’t evoke any detectable brain activity. Just like humans aren’t aware of the smell of their own breath, dogs seem to tune out the smell of their own pee.
Strangely, we also observed strong activation to the unfamiliar smells in the cerebellum, a part of the brain usually associated with movement. When I was presenting the cotton swabs to Callie, sometimes she sniffed more intensely. The cerebellum activation was most likely the neural origin of the sniffing, which would be more intense for smells the dogs hadn’t encountered.
The most interesting finding appeared when we subdivided the dog and human scents into their subcategories of familiar and unfamiliar. One, and only one, type of smell activated the caudate: familiar human. This was especially true for Callie. In her case, the familiar human was Kat.
Kat’s sweat activated Callie’s caudate—same as the signal for hot dogs. But Kat wasn’t even at the scanner. This meant that Callie had identified the scent as Kat even though she wasn’t physically present. And if Callie had a mental category for Kat that didn’t require her physical presence, then this suggested that Callie had a sense of permanence for the people in her household. She knew who her family was, and she remembered them. We found further evidence for this interpretation in an area called the inferior temporal lobe. This part of the brain is closely associated with memory function, and like the caudate, the inferior temporal lobe was strongly activated by the smell of a familiar human.
The inferior temporal activation told us that the dogs remembered their human family, and the caudate activation, more prominent in Callie, told us that her remembrance of Kat was a positive one. Could it be longing? Or love? It seemed entirely possible. These patterns of brain activation looked strikingly similar to those observed when humans are shown pictures of people they love.
The results of the smell experiment expanded our understanding of the dogs’ mental world. All through the Dog Project, we had been focused on the nature of the dog-human relationship. We love dogs, but what do they think of us? Even with just two dogs, a picture was beginning to emerge. The pattern of activations in the cortex suggested that they concocted mental models of our behavior, which might be due to mirror neuron activity. But regardless of the mechanism, the smell data showed that their mental models included the identity of important people in their lives that persists even when the people aren’t physically present.
I was willing to accept that as an acceptable demonstration of love from Callie. But even if I was being too generous, the fact that the dogs knew who we were, and that they had categories for us, indicated that we humans make a lasting impression on our dogs. We are appreciated.
22
First Friend
WHEN WE BEGAN THE DOG PROJECT, we had no idea what we would find. What started as a half-baked idea to scan dogs’ brains mushroomed into a full-fledged research program faster than I ever expected. Even with just hot dogs, and then smells, we had found evidence that dogs mentalized about the humans in their lives. I suppose this should not have been surprising. Many dog owners are convinced that their dogs know who they are and return their love for them. But, for the first time, we saw direct evidence of reciprocation in the dog-human relationship and social cognition in the canine brain.
This was truly exciting, but, in the interest of scientific objectivity, we had to be careful in generalizing from our experiments. The World Health Organization estimates that the population of dogs is 10 percent that of humans. That translates into roughly 700 million dogs worldwide. And we had studied the brains of precisely two of them. Although we had expanded the enrollment in the Dog Project since our initial experiments, we were still studying a very selective group of dogs. These were dogs that were loved by their humans. But even that is not enough. Most dogs aren’t willing to go into an MRI, and most people aren’t willing to train them to do so. That still leaves the 700 million dogs of the world. What did our experiments tell us about those dogs and their relationships to humans?
From an evolutionary perspective, dogs are incredibly successful. Their numbers speak to that. Given that dogs share their environmental niche with humans, their success must be a result of learning how to read us. Not just reading human behavior but, I believe, learning to read our intentions, which means that they have a theory of mind for humans. And that is exactly what we found in the Dog Project. So even though Callie and McKenzie were rarified representatives of the world dog population, what we found in their brains showed the defining characteristic of dogs: social learning. Their brains showed that they cared about human intentions.
Proof of social cognition means that dogs aren’t just Pavlovian learning machines. It means that dogs are sentient beings, and this has startling consequences for the dog-human relationship.
Most of the dogs in the world are village dogs. They are not anyone’s pets, although they might look that way at first glance because they often gather near humans. People know who the dogs are, but they might not have names. Village dogs insinuate themselves into the fabric of human societies. They feed on scraps, sometimes garbage, and sometimes food that is deliberately left by humans for them to eat. In some parts of the world, people let them hang around just so that they can eat them later.
If Callie had lived anywhere else in the world, she would have been a village dog. She had that rangy appearance—not too big, not too small—and the eyes of an opportunist. For the first year that she lived in our house, I was convince
d that she would run away at the first opportunity of something better. But after the Dog Project, I no longer thought that. The project had changed both her brain and mine.
Indeed, if there is one thing dog ethologists can agree on (and it might indeed be only one thing), dogs are masters of change. If nothing else, dogs’ defining characteristic is their adaptability. Apart from vermin, dogs are the only mammalian species that is found everywhere humans are, and humans have occupied every habitable niche on the planet. As the ethologists Raymond and Lorna Coppinger observed, “The rapidity with which the dog has changed form, and the seemingly endless varieties of its form, challenge the theory of Darwinian evolution, that adaptation must be a slow process.” The Coppingers were referring primarily to dogs’ changing physical form, but the same can be said for dogs’ behavior.
When scientists speak of behavioral change, they are really talking about learning. And, as far as we know, there are only two mechanisms of learning that animals employ: associative learning and social learning. For a century, Pavlovian behaviorists have argued for the predominance of associative learning. Animals, dogs included, are great at learning associations between neutral events and things that they like, such as food, or things that they don’t like, such as pain. But associative learning cannot explain all of animal behavior. For one, it is inefficient. For an animal to learn associations, it has to actually experience the events. This is a trial-and-error process. By this learning mechanism, a dog would actually have to touch its paw to a hot stove to learn that that is something to be avoided.
Social learning is far more efficient. Many animal species employ social learning. Songbirds, for example, learn their species-specific calls from each other. But besides humans, dogs may be the best of all. By watching other dogs, Fido can learn a great deal. He doesn’t have to burn his paw to learn that the stove is dangerous if he sees another dog (or human) do the same. And, of course, puppies learn from each other and their mother, copying behaviors like pulling toys.
I have often wondered how dogs got so good at social learning. While many animals learn from members of their own species, dogs are one of the few that can learn from other species. Herding dogs, for example, learn by observing sheep and cattle. And all dogs learn by observing humans and other members of their households, just like Callie learned how to open doors. Village dogs, even though they are not attached to specific humans, exemplify this ability for social learning. There is no other way they could keep up with the ever-changing form of human society.
In the hot dog experiment, we found that the meaning of the hand signals had transferred to the caudate—a brain region known to be associated with positive expectations. While a cool scientific finding, it was not really unexpected given what we had known about Pavlovian learning. What was more revealing, and what we never commented on in our academic papers, was all the other stuff going on in Callie’s and McKenzie’s brains. The motor cortex activity. The inferior temporal lobe. Those were the regions that pointed toward a theory of mind, and they were the same regions that popped up in the smell experiment associated with familiar humans.
These cortical regions showed that the dogs might be constructing mental models of our actions. The inferior temporal lobe suggested that they were recalling memories, perhaps what one hand pointing up meant, or the identity of the person associated with a sweat sample. These are the types of mental processes that any sentient being would use on a daily basis. Humans use memories and ascribe meaning to people and actions all the time. Apparently, so do dogs.
Even though we found evidence for canine theory of mind in our experiments, Callie and McKenzie were not exactly the same in this regard. They showed differences in how their brains reacted to the hand signals and to the smells. With only two subjects, it is difficult to draw sweeping conclusions, but I will take scientific license to speculate.
In the hot dog experiment, McKenzie had stronger caudate activation to the “hot dog” hand signal. Strangely, Callie was the food lover, while McKenzie much preferred toys as rewards during training. Because of her great love of hot dogs, I had expected Callie to show the stronger caudate activation. But she didn’t. One possibility is that because Melissa and McKenzie had competed in agility competitions, McKenzie was more attuned to hand signals. I had not taught Callie a hand signal before the Dog Project, which might have put her at a relative disadvantage. Another possibility, which I think is very likely, is a genetic basis.
Even though we thought she was a feist, Callie was more like an adopted village dog. A mutt. McKenzie, on the other hand, was meticulously bred to be a herding dog. Border collies are known for their stares, what the Coppingers refer to as the eye-stalk. Border collies don’t just see with their eyes; they use them to control other species. There may have been much more going on in McKenzie’s brain as she not only interpreted Melissa’s signals but also returned them with her eyes. While I had noticed a flicker of that in Callie when her eyes dilated in anticipation, it was nothing like being stared down by a border collie.
In the smell experiment, though, the pattern reversed. Callie had the stronger response to the familiar human smell. Maybe that was because Callie slept with Kat and me in our bed, while McKenzie slept in her crate. Or maybe the bond between Callie and Kat was stronger than that between McKenzie and Melissa’s husband. Could it be that our dogs tell us more about our human relationships than we tell ourselves? The term therapy dog would take on a new meaning.
The evidence for social cognition in dogs’ brains has important implications for the dog-human relationship. Dogs watch us constantly, even though we may not be aware of it. With furtive glances, they take in their surroundings and form mental models of what we humans are intending to do. It is the humans who are unaware of the dogs. And that is where misunderstandings can arise.
Humans are sloppy creatures. Like the proverbial bull in a china shop, we are oblivious to our body language. We bump into objects. We accidentally step on our dogs’ tails. We emit a constant stream of sounds with frequently inconsistent meanings. It is a wonder that dogs can pull anything consistent out of this barrage of signals. And yet they do.
The whole purpose of the Dog Project was to understand the dog-human relationship from the dogs’ perspective, and the most important thing that we learned was that dogs’ brains show evidence of a theory of mind for humans. This means that they not only pay attention to what we do but to what we think, and they change their behavior based on what they think we’re thinking. They are the Zeligs of the animal kingdom.
Zelig was a fictional character created and played by Woody Allen in his 1983 movie of the same name. Zelig had no personality of his own. Instead, he took on the personality and physical form of people around him. Because doctors thought he was crazy, Zelig was institutionalized in a mental hospital, where he took on the form of a psychiatrist. (His real psychiatrist, a woman played by Mia Farrow, falls in love with Zelig and they eventually run off together at the end of the movie.) Apart from being a terrific film, Zelig is a case study in theory of mind. Zelig’s problem was that he had no sense of self. He had a sense of only others. The sense was so strong that he knew what was in other people’s minds, and he became them.
If dogs are like Zelig, then the form they take depends on the people they live with. If they live with calm, consistent humans, they will pick up on those qualities. If they live with people who talk constantly, without saying anything, dogs will quickly learn that there is no useful information in their chatter. With their social cognition skills, dogs do not need an excess of jabbering. Patricia McConnell, the well-known animal behaviorist, has written extensively about the effectiveness of the less-is-more approach to dog communication. The takeaway is that humans should pay more attention to what their body language communicates than what their mouths say.
Dogs’ sensitivity to social signals also puts a new twist on the old notion of human as “pack leader.” While it is easy to confuse being a pack
leader with being dominant, that is a mistake that has harmed more dogs than any other piece of advice.
The better analogy for being a pack leader comes from management literature. While there are different styles of leadership, the most important characteristics of a great leader are clarity and consistency. Without those two qualities, people (and dogs) cannot know your intentions. Great leaders are also respected, not because of their position, but because of their inner strength and integrity. Leaders do what they say. Leaders listen to people, and although they may not always agree, they have respect for others. Great leaders help people.
It wasn’t until I started working with Callie in the Dog Project that I realized just how much she could be dialed in to my signals. Like a catcher and pitcher, we became a team. She had always had that ability. I just hadn’t given her any clear direction before.
Eventually, I came to the conclusion that the key to improving dog-human relationships is through social cognition, not behaviorism. Positive reinforcement is a shortcut to train dogs, but it is not necessarily the best way to form a relationship with them. To truly live with dogs, humans need to become “great leaders.” Not dictators who rule by doling out treats and by threatening punishment, but leaders who respect and value their dogs as sentient beings.
Even though I couldn’t have known about the depth of dogs’ social cognition when we started, respect for dogs had been built into the Dog Project. Early on, we had made the presumptive decision to give the dogs the right of self-determination. If they didn’t want to be in the MRI, they could walk out. Same as a human. We created a consent form. Although the dogs did not have the capacity to understand its contents, their human guardians did. The guardians were able to weigh the risks, however minimal, against the benefits and decide whether it was in the dogs’ best interest to participate. The legal model we used for this process was lifted from the manual on human experimentation. We treated the dogs as if they were human children. But nobody had ever done this before. In the eyes of the law, dogs are still considered property.
How Dogs Love Us: A Neuroscientist and His Adopted Dog Decode the Canine Brain Page 19