After five minutes of walking up the steps and jumping off, Callie tentatively placed her paw on the patient table. With great enthusiasm I encouraged her to keep going.

“That’s it, Callie! Good girl! Want more hot dogs?”

She got it. Once up on the table, she saw it wasn’t scary at all and that there was a ready supply of hot dogs. Now she had to go into the bore of the MRI.

I had already secured the foam chin bar in the head coil, located dead center in the magnet tube. Now I placed, Hansel and Gretel style, a trail of hot dogs leading from the entrance of the MRI to the head coil. Without a thought, Callie continued walking down the patient table into the bore, lapping up hot dogs as she went.

My colleague Lisa, who was filming the event, gasped in excitement at the sight of a dog walking into the MRI.

I quickly circled the scanner so I could face Callie from the other end of the bore. She was crouched down in a sphinx position just short of the head coil. Her tail was swishing back and forth. I reached in with a hot dog in my hand and immediately felt the room spin.

Callie saw the hot dog and scooted forward into the head coil.

“Good girl!” I said with my highest, most excited voice.

She took the hot dog and backed up a little bit, but she didn’t leave the bore. With a steady stream of hot dogs, Callie quickly adapted to the new environment and was soon happily eating treats while nestled in the head coil. There was no indication that the magnetic field bothered her.

With Callie comfortable in the magnet, we had accomplished the first goal of the session. Since that had been relatively easy, it was time to see how she would react to an actual scan.

The scanner software was created for human subjects, so it had no way of knowing that Callie was a dog. Inputting an accurate weight of the subject was the most important piece of information because that determined how much radio power the scanner would emit.

Too much power would cook her like so much meat in a microwave.

With hot dogs, I once again coaxed Callie into the MRI. When she was comfortably settled in the head coil, I gave a thumbs-up. The scanner made a series of clicks and hums as it revved up.

Callie’s eyes narrowed.

Then, like the onslaught of a thousand bees, the scanner started buzzing. This was the initial preparation phase, called shimming. The scanner automatically adjusts the magnetic field to compensate for the distortion caused by whatever is placed inside. Normally, shimming takes a few seconds, but with Callie inside, the buzzing continued. Even though she was wearing earmuffs, she wanted no part of it and headed for the exit.

I waved my arms back and forth, signaling the MR tech to abort the scan.

“What was that noise?” I asked.

“Shimming,” he said.

“Why was it going on for so long?”

“The scanner was having trouble compensating,” he explained. “Probably because it expects a human.”

We hadn’t thought of this. We hadn’t even recorded the shimming noises for our training sessions. We had assumed they would take only a few seconds, a minor blip compared to the lengthy functional scans that would follow. Callie reacted to these novel sounds as any dog would: she got scared.

We tried a dozen times, but Callie scooted out as soon as the scanner started buzzing. We even tried starting the scan before she went in, figuring that if she got used to the ambient noise, I could coax her into the head coil. Eventually, with enough repetitions, the scanner was able to cobble together a crude compensation for her canine form.

Next up were the functional scans. These are a series of scans that individually take about two seconds to capture the brain. By continuously acquiring these functional scans while Callie was in the MRI, we could measure changes in her brain activity and figure out what she was thinking. At least, that was the plan. Finally, we would do a structural scan, which is a high-resolution picture of the brain used to identify brain anatomy.

It was tough for her. The earmuffs kept sliding back, exposing her ears to the full onslaught of the noise. Even so, Callie managed to hold her head in position for a few seconds at a time. We stopped the scanner after three minutes’ worth of scanning. That, we felt, would be enough to evaluate the quality of the data.

Before she got too tired, we decided to make one attempt at a structural image. The structural scan takes thirty seconds, and Callie would have to hold still the entire time. After the scan, she bounded out of the magnet and pawed off her earmuffs. She jumped up and licked my face and then ran over to Lisa, who gave Callie a big hug.

“What a good girl!” she exclaimed.

We all went into the control room to see what the images looked like.

The structural image looked remarkably good. There were ghost images throughout, which occur when the subject moves, but it was clearly recognizable as a dog’s brain. The functional images were a different story. Out of 120 images, only one contained anything that looked like a brain. Mostly they were jumbles of digital snow with an occasional eyeball peeking into the field of view.

I hugged Callie and said, “I’m so proud of you.” But in reality, I didn’t know if this was going to work.

The next scan—with Callie, the other dog, McKenzie, and the whole entourage—was in three weeks. I hoped we could figure it out before then. If we didn’t, I would have to pull the plug on the Dog Project and acknowledge that the naysayers had been right: you can’t scan the brain of an awake dog.

1

Dia de los Muertos

TWO YEARS EARLIER

EVERY NOVEMBER 1, I push aside the remains of the Halloween candy and erect a shrine on the dining room table.

I begin with a vase that Kat and I bought in Mexico on our honeymoon. It’s a cheap thing, with a stylized owl painted on one side, but the vase has somehow survived multiple moves across the country, and I have come to value it for its resiliency rather than its beauty. It also provides the necessary ethnic authenticity for the ritual and functions as an ideal centerpiece to prop up photographs.

We keep the photos in a drawer all year long, only to be brought out on this day. Kat and I surround the vase with them: pictures of family members who have passed away over the years. Then, to complete the offering for their spirits, we scatter a cornucopia of the sweetest, most delicious baked goods.

Our two daughters, Helen and Maddy, had never questioned why we did this. They had, after all, lived with the ritual all their lives. But when they achieved the age of enlightenment, preteen-hood, they realized that celebrating Dia de los Muertos—the Day of the Dead—was not a normal thing to do. At least not the way we did it.

We included the dogs.

Although I had grown up with dogs, it wasn’t until I finished medical school that I had the opportunity to acquire a dog that I could truly call my own.

Kat and I had been married for five years, and we were putting off children until I completed my training. So, in celebration of completing my first year of medical internship—a grueling year of hundred-hour weeks—we answered an ad for puppies. Pug puppies, actually. I note this with some qualification, because to many, pugs are a grotesque distortion of the canine form. Of course, Kat and I didn’t see them that way. Their large heads, with pushed-in noses and bulbous eyes, were almost human—a sort of baby substitute.

We named our new puppy Newton.

Like all pugs, Newton’s face was brachycephalic, meaning short-nosed, but his was foreshortened in the extreme, with his nostrils forming mere slits. He was what breeders call an apple head because of the taper of his skull. His panoply of malformations only made him more endearing to us, and his constant snuffling and snoring became a welcome background noise to our lives. At night, he slept with his unusual dome nestled in my armpit.