Dr. Gregory Berns – TRANSCRIPT
How many of you are dog people? A show of hands. Excellent! How about cat people? OK, you guys can go to the break early. So, of the dog people and the cat people who want to be dog people, how many of you have thought, “Wouldn’t it be great to know what my dog is thinking?” I think everyone else already knows what their dog is thinking, right? I got into this project, and I’m going to tell you a little bit about how – This is basically a stupid dog trick story.
It really started with this dog named Newton, who was really my favorite dog. I’ve had many dogs through my life, but Newton was my favorite, and he lived to be about 15 years old. After he passed away, I thought, I have these tools, this MRI machine, that I have been using for decades to study human decision making and what motivates people, why haven’t we used this on other animals?
Certainly, other animals have many of the same feelings and motivations that people do. But this is kind of an area of science that people don’t like to talk about. So I embarked on this project about four years ago to try to figure out what dogs think, and specifically what dogs think of us. If we’re talking about humans, we have kind of two ways we can think about what other people are thinking: we can either ask them, and sometimes they will tell us if they know, and they want us to know what they are thinking; or we can observe actions, we can observe behaviors, we can try to infer things about what people are thinking from their actions.
With animals, and dogs, of course, we can’t really ask them. We can ask them, and we may think that they tell us, but we really don’t know what they’re thinking. So we’re kind of left with their behaviors: we can observe their actions, and we can try to infer what they are thinking. This is the foundation of behaviorism, and it’s been around since Pavlov. But there are, of course, very tricky issues here, and humans being humans, we tend to anthropomorphize everything. It’s kind of in this area that I became very interested in intrigued with the possibility of trying to figure out what dogs are thinking by using MRI. The technique is straightforward. It’s been around for decades.
The idea is: if we were studying a human, we would put a human in an MRI, have them do some type of task, and we’d measure blood flow or brain activity and then try to figure out what parts of the brain do what. Very straightforward, if you’ve had an MRI, it’s not terribly pleasant, but people will do it. How do we do this with other animals? How do we do it with a dog? I’m going to show you what we found.
Here’s a short video. It’s a what we call our training video, and it demonstrates how we did this. Before I start it, you’re going to see two dogs in this video. The first dog, Callie, is my dog. She was actually the replacement for Newton. She was adopted here in Atlanta from the Humane Society. We loved Newton so much; we could never get another pug, so Callie is the anti-pug. The other dog is McKenzie, a border collie. We just kind of get right into it. I’ll narrate as we go along. [Callie – Introduction to head coil] This’s Mark Spivak. He’s my partner in this endeavor, he’s a dog trainer.
The first thing that we had to do is figure out how do we get dogs to go into a tube, to put a head coil around their head to pick up the brain waves, and hold absolutely still. What you are seeing here, is that Callie is not a particularly obedient dog; she has no particularly special skills. But she does have one very good trait, and that is: she likes hot dogs. Mark is doing what we call clicker training. Every time she approximates what we want her to do, he clicks, and then she gets a hot dog. This’s the very first time she’s been introduced to the thing we call the head coil, and we didn’t know at this point whether this was even going to be possible.
[McKenzie – Introduction to head coil] This dog, McKenzie, a border collie, is highly trained. She’s very skilled in agility, and her owner, as you’ll see, gets her to sit in this coil very quickly. (Video) Dog owner: Good girl! Yes! Is she too far out now? (Video) Gregory Berns: Yeah, basically, we are looking for the brain case to be in the center, right there. That’s good.
(On stage) If you’ve had an MRI, you know that you’re told not to move, right? This is the big challenge of doing this. [Mckenzie – Holding without any chin rest] Up until this point, I didn’t know if this was going to be possible until I saw this. This was literally after about five minutes of training. When I saw that, I knew we could do this. [Callie – Training with chin rest] What you saw McKenzie doing was close but not quite good enough. What we are going after if we’re to achieve data that compares to humans – (Video) GB: You are perfect! Excellent! Perfect job! (On stage) GB: Mark told me I had to be more demonstrative than I am normally. (Video) GB: Perfect! Yes! (On stage) GB: What you notice we did was we introduced a little chin rest because we have to give the dogs a target to put their head on. McKenzie adapts this very quickly. She’s actually in a simulator for an MRI that we built. She’s doing quite well, but this is actually still too much movement. The really difficult part of this is the noise that the scanner makes, playing in the background. These are recordings that we made to acclimate the dogs to the training. It’s very loud. This’s being played at low volume just to get her used to it. But it’s really about 95 decibels, and it’s like jackhammer loud.
(Video) GB: That’s it, that what we are doing! (On stage) GB: This is after about a month or two of training. [Callie – Scammer training] We’re at the real MRI now. This’s probably the most expensive training session ever performed. (Laughter) We get charged about 500 dollars an hour to use the MRI. But we had to use the real thing at a certain point. At this point, we didn’t even know how they would react to the magnetic field. The key thing I want you to notice is these dogs are doing it willingly, and they enjoy it. That is the whole point of this project. We treat these animals as family members. We don’t sedate them, and we don’t restrain them. [Callie – Final training] This’s actually after about two months of training. We made some modifications to the chin rest, and even a shelter dog like Callie can do this. [Full chin rest, ear muffs, tube, hand signals] If you look carefully, you also notice that she’s wearing ear muffs. It’s very important because the scanner is so loud, and the dogs hearing is quite sensitive. [This means “hot dog”]
The other thing that we did – This’s a scientific experiment, really. [This means “no hot dog”] That’s the training video. The “hot dog, no hot dog” hand signals, we started with this because we didn’t know if this was going to work, so we decided we needed to do something really simple. This’s just straight up Pavlovian conditioning where we taught the dogs two hand signals: this means “hot dog,” and this means “no hot dog.” If this technique works, what we should see is activity in the reward system of their brain to this hand signal but not this signal. I also put up a slide here.