Gregory Berns is a neuroscientist who started to study his best friend's brain. After years of looking at human rewards systems, Berns realized he could teach dogs to sit – or, more precisely, lay down – in an MRI machine to study their brain in real time and see what they really think about their human owners. In his new book, 'How Dogs Love Us,' Berns writes about his personal experience with canines and how it influenced his research, leading him to train dozens of dogs to be studied in a lab.

Why did you start studying dogs?
Most of my career was spent using an FMRI to study the human rewards system. We were trying to map out the reward system of the brain, looking at fruit juice, money, and other rewards. We came to the conclusion that there's a fairly restrictive reward system in the brain and there's a common neural currency. I don't know why I didn't think of it sooner, but then we had this idea: Why can't we do this with dogs? I've always lived with dogs, so it made sense. The main difficulty with it is that the subject has to hold absolutely still. Before, you either had to anesthetize the dog or find some invasive way to strap them down. Instead, we trained them.

How did you do that?
First, I built a simulator at home and practiced on my terrier, Callie. Eventually, we created an MRI certification for the dogs. Now we have 13 certified MRI dogs and another 12 are now starting the program. We have rescue mixes, a few border collies, a boston terrier, and probably an overrepresentation of golden and golden lab mixes. It reflects the overall popularity of the dogs and the use this research could have for service dogs.

How will your research help service dogs? 
There's a high washout rate for service dogs and eventually, you could use a brain scan to see what dogs are most trainable for hand signals or voice commands. The MRI could be used as a screening tool or a measure of the effectiveness of the communication. Are you getting through to the dog? When you don't get through, you don't know why. Is your training being received and understood but the dog just isn't interested?

So your research is focused on dog training?
Our experiments to date have focused on the dog and human interaction. Everything is focused on communications with hand signals and smells. In a study being reviewed right now, we presented smells from other humans and dogs from the dog's household next to strange human and dog smells. It's clear from the scans that they're identifying the familiar human. And we're seeing activation in the same reward pathways as the hand signal and food. There's a positive association, even when the humans aren't there.

But is that association emotional or connected with where they're fed?
A lot of dog people say, "duh, dogs love us – you don't have to put them in the MRI to know that." But that is anthropomorphizing. There's a contingent of hardcore behaviorists who think that the things dogs do they do just for food and shelter. To me, there's no way from looking at a dog to tell the difference. You can infer it, but we're working to measure it. We're about to launch a new experiment to see how much is social reward and how much is Pavlovian conditioning. In it, we'll have dogs watch computer screens. There will be hands on the screen giving signals and a machine delivering treats when the signals are followed, without humans being involved. It's a baseline measurement of learning pathways. Then we'll do the same with actual humans and compare the results.

What can we learn about dogs that helps us train them?
We can work to find what aspects of the communication process are most effective at eliciting response. Ultimately, it's all going to map on the reward or motivational system and certain modalities may be more direct pathways. Predominately we use visual or vocal signals, and we don't even know the degree to which they understand language. Border collies are said to have a huge vocabulary. But do they have cognitive understanding or are these words just sounds to them?

Is there an innate difference between breeds and temperament?
We've run the statistics on breed, but we only have a dozen dogs so far, so we can't answer that question. Like people, dogs have a lot of variability in their temperaments, or personalities, if you want to call it that. Some dogs are just never going to be able to do this kind of study. We're specifically looking for a temperament to do this. Anxious dogs in particular – it's sort of inhumane to get them in the study.

Why do we find dogs so . . . lovable?
I think dogs are special in terms of other animals in the world. We don't know how it's happened, but they've been living with humans the longest – 12,000 years by some estimates, or maybe closer to 30,000 years. That's a long time, considering that modern humans have been around nearly this long. Evolutionarily speaking, they have a big head start on cats or any other animal. And it's quite possible they've had an influence on human evolution. It's easy to imagine that a nomadic group of humans that took in a dog had some advantages – maybe it helped offer protection or social benefits. If there were a prehistoric dog whisperer that could train dogs, it's theoretically possible that they were preferred and that we evolved to have this trait.