A couple of years ago, in the middle of the night, I crept downstairs to find my
father sitting at the kitchen table, sobbing like a child.

My mother was beside him, trying to comfort him, an activity that took up more and more of her time. He was 87 and had dementia. It wasn’t unusual to find him upset or confused. But on this night, something seemed to be happening to him in real time – in 1941.

He was 6 years old, and was leaving Pittsburgh, the only home he had ever known, for an Air Force base in San Antonio, where his father had been ordered for duty. He and his parents were travelling there by train, transferring in Chicago.

It was the beginning of a lonely, difficult time for my father’s family, moving between Air Force bases in the South, where landlords sometimes turned them away because they were Catholic. An only child, he had been allowed to take one pet with him, a canary he was carrying in a birdcage.

As they were changing trains in Chicago, the bottom fell out of the cage. The canary flew out, up into the vaulted atrium of the station’s Great Hall. There was no way to get the bird – there was no time, they had to board a train to Texas. So my 6-year-old father shuffled after his parents, holding an empty cage.

In the years that had elapsed, he had negotiated arms treaties with the Soviets, had advised presidents, had served as a US ambassador, all with the same watchful, wisecracking reserve. I thought I knew who he was. I could count on one hand the times I had seen him cry. Now here he was, sobbing over the canary as if it were yesterday.

This was all, it seemed, because of his brain. He had fallen hard in their house in Washington DC, smacking his head on the hardwood floor. Blood rushed into spaces in his brain, and cells starved of oxygen began to die. Eventually, he was diagnosed with vascular dementia, which is most often caused by strokes.

For five years after that, my parents lived with my family outside Boston, and we learned first-hand how brain injury affects behaviour. My father recovered in some ways, but he became chaotic, his thoughts broken into mirror shards.

The biggest problem was that he had no idea where he was. Specifically, he did not know why he was living with us in Massachusetts, and no matter how many times we tried to remind him, over and over, he tried to leave. We would catch him packing the car, and gently – or not so gently – guide him back into the house.

This child-father was full of surprises. He bought surprising things: five laptops! A cruise on the Norwegian fjords! Recurring $2 donations to every Democrat running for any office, anywhere! Once, in a week-long cascade of Amazon deliveries, we received seven identical birdbaths from China.

He remembered things about his own life, but in a scrambled way. “I am physically in the Soviet Union and subject to their rules and regulations,” he told me, tartly, when I asked him to do something he considered unreasonable, like put on pants.

During a stay in a rehab facility, he reported that his roommate had diverted a smuggled shipment of black caviar and that the nurses were all KGB. “How did you know they were KGB?” I asked. He looked at me slyly. “Because they spoke perfect English.”

But the canary, that was something different – a fully formed, untouched memory from the remote past. My brother and I had never heard this story; it hadn’t appeared in my father’s written accounts of his childhood.

Somehow, the brain injury had broken it loose and allowed it to rise to the surface. From a man whose thoughts had become so fragmented, this sequence was gloriously complete, as if we had reached into the smoking debris of a car wreck and withdrawn an Easter egg.

Last month I climbed the hill to a laboratory at the Rockefeller University in New York City, overlooking the East River, carrying images of my father’s brain to show to a scientist I had never met.

Even to me this seemed like a crazy idea. My father died two years ago, in his bedroom in the house we shared. We did the things you do – a cremation, a funeral, a memorial service, flowers, casseroles. My mother withdrew, and for six months it seemed she might slip away after him, but then something, maybe the garden, recaptured her interest, and she was back.

What I am saying is that the death had come and gone. But I still felt a little stunned by what had preceded it, nearly five years of trying to manage my father’s impulsiveness and confusion.

Our coping mechanisms were elaborate: patrolling the neighbourhood, collecting him, distracting him by planning trips we would never take. Sometimes arguing with him was so exhausting that we drove him to the airport and let him walk up to the terminal, then waited for him to return sheepishly to the car.

On one particularly hopeless day, my husband taped a note on the inside of the front door.

YOU LIVE HERE, it said.

DON’T MOVE.

WE LOVE YOU.

We regaled our friends with funny stories, but let’s be honest – it was a battle. We were desperate for sleep; my mother was so worn out trying to keep him from leaving that she occasionally ended up in the hospital herself. Finally we called in a locksmith to put up padlocks on the insides of the doors to keep him from leaving in the night.

His death brought an end to this circus. The house quieted. We were grieving, but there was something else. What in the world had just happened to us?

Then, last Thanksgiving, I received a press release with the headline “How the Brain Decides What to Remember.” It described research by Dr Priya Rajasethupathy, a neuroscientist at The Rockefeller University who studies how memories form and are maintained in the brain. By stimulating the thalamus in mice, her team had found a way to transform a fleeting memory into a lasting one.

Her research could eventually help people with dementia. In the early stages of the disease, neurons in the hippocampus, where memories are formed, are dying by the millions; on average, at the point of diagnosis, almost a third of them are already dead. What if it were possible to tag important information and offload it to the parts of your brain that are still healthy?

At the end of the interview, I told her about my father, and she asked if I had images. And so here we were, four months later, gazing at ghostly cross-sections of his brain.

The scans had been taken hours after he hit his head. At first, we were looking at chronic damage from the outer layers of the brain, swollen ventricles and a shrivelling cortex. That was age, stress and alcoholism, which he once told me was “the family disease”. Then, in deeper cross-sections, something new appeared, the bright white of a bleed. That was the fall.

As Rajasethupathy scrolled, the white spot expanded into the meat of the brain. We knew there had been a haemorrhage, but we didn’t know where.

“Boom,” she said. “There it is.”

It was at the edge of the right hippocampus.

Much of what we know about the physiology of memory can be traced back to a 27-year-old man from Connecticut, known in the literature as HM, who approached a neurosurgeon for help controlling his epileptic seizures. In 1953, the surgeon removed parts of his temporal lobe, including the hippocampi, a pair of small, curved structures nestled just above each ear.

The surgery had a dreadful, unexpected consequence: after his hippocampi were removed, HM was unable to form new memories. He retained new information for about 30 seconds before it vanished. He got lost going to the bathroom. His doctors had to reintroduce themselves every time they entered the room.

And yet HM could recall his early life, his parents, his friends in high school. He could recount in cinematic detail – down to the colour of the upholstery – a ride he took in a single-engine airplane when he was 13 years old.

A similar paradox appears in patients with dementia, which typically begins with massive cell death in the hippocampus. First to go are memories from the last minutes or hours — Where did I put my keys? As the disease progresses, major life changes may be forgotten — Why am I in this strange place? And who are you? But even then, remote memories may be preserved.

My father, like HM, had held on to material from his early life. He could still hold conversations in Russian and Bulgarian, and gloated about outliving Henry Kissinger. He regaled the hospice nurses with the ridiculous story of how he and my mother got together, on an ill-advised mission to assist in the Hungarian revolution.

And like HM, he had suffered a sudden injury to the hippocampus. “This is where we see the blood kind of leaking out into the brain,” Rajasethupathy told me. Volume loss would have followed, as cells in the right hippocampus, deprived of oxygen, died. She had to look up what that meant, since the right and left hippocampi control different functions.

It is thought that the left one controls “verbal and autobiographical and storytelling”, she told me, and that the right one controls the memory of space and location. In both cases, it is the hippocampus that stabilises signals into memories. I asked her: what could you understand based on this scan, without knowing anything about my father?

She paused momentarily. An “inability to form new memories”, she said.

Walking away from her office, I thought of all the strange days we had with him, as we tried to guess what internal signals he was following. He had to go to Philadelphia; he had to go to Kyiv. Once, my husband found him at 4 in the morning sitting quietly in the dark outside a plant nursery, because I had happened to mention that I wanted a dwarf Japanese maple tree.

Once, he told us he had to get to the Votivkirche, in Vienna, to accept a medal for preventing the assassination of Archduke Franz Ferdinand. But toward the end, all he knew for sure was that he was in the wrong place, and that he had to leave. Sometimes he told us he had to get to another house – his real house – in the same town.

“Do you have another wife there?” I once asked him.

“Yes,” he said starchily. “And she’s nicer to me.”

What Rajasethupathy told me was strangely thrilling, a biological explanation for that whole exhausting, topsy-turvy time in our family life. But when I told my husband about it, he was sad. He was the one who used to drive around the neighbourhood trying to find my father when he disappeared.

He was sad, he said, because when we brought my father home to live with us after his fall, no one warned us that he would spend the rest of his life trying to leave, and that we would spend the rest of his life trying to convince him to stay.

It could have been predicted. It was right there on the scan.

In graduate school Rajasethupathy studied the case of HM and its lesson, the central role of the hippocampus. But she was certain that there were other important memory centres in the brain, and set out to find them.

Around 10 years ago, she began training waves of mice on virtual reality mazes, selecting out the mice with the best memories. These mice, it turned out, all shared a gene found to be functioning in the thalamus, a pair of walnut-shaped structures that have long been viewed as the brain’s sensory switchboard.

A second discovery, using a different approach, came three years later, when her team tracked the brain activities of mice as they retrieved different kinds of memories. The team watched seven different pathways out of the hippocampus as the mice tapped into remote memories. Only one lighted up: the thalamus.

Rajasethupathy dreams of opening up a gate in the thalamus to help offload memories into protected parts of the brain. Her team has already found that if a light-sensitive gene in a mouse’s thalami-cortical loop is activated while it learns a task, the mouse will recall the task a month later. A weak memory is transformed into a strong one.

This raises a tantalising possibility. What if we could smuggle the important stuff – say, where you live, or whom you married, or that you completed all the work you had set out to do – into a safe room, before the flood comes and washes everything away?

Rajasethupathy has her own reasons for trying to solve this problem. When she was 3 years old, her mother died in a car crash. Her father was left to raise Priya and her brothers.

For years afterwards, her father found it too painful to speak about her mother. So when they were alone, the children exchanged their memories of her, stories and images, as if doing so could bring her back into the room.

Priya was so tiny that she had only two or three memories. They were just flashes – her mother sitting beside her at the dining table, drinking coffee. Her smile, her laugh, her nose ring. These were the things she had replayed in her mind. They were permanent, she knew that. But she was never quite sure if they were real.

This struck her as a question she could devote her whole career to. How is it that the brain, this biological machine, can convert a single, transient stimulus – a trace pulled out of a coursing torrent of information – into something that lasts forever?

We say, of certain memories, that we will take them to the grave. But we do not yet have the tools that Rajasethupathy hopes to build, and I saw, in my father, that we do not decide what to remember. Some primitive musculature kicks in, and the people around you must simply adjust.

I left her office on the East River carrying an image of the rupture in his brain, a burst of brilliant white inside dark folds of tissue. The image had been taken just as his old self was slipping away.

After that he was a different person, stuck in the moment before one sets off on a journey. He didn’t brood, he no longer wanted to drink, but it was hard for us to hold on to him. After dinner, he would push his chair away from the table and thank us for a lovely evening, compliment us on our handsome family, and announce that he was leaving.

I can’t say we ever convinced him otherwise. By the end he was too weak to attempt escape, and he only wanted to go to one place. It was a place none of us knew anything about, because he had last been there in the 1930s, before the war, before the canary.

He wanted to go to Tonawanda, a city on the Niagara River, north of Buffalo, New York. He had gone fishing there with his father and his uncle.

Everyone else who was on that trip was dead, so it was hard to say what was so wonderful about Tonawanda; all we knew was that Tonawanda was the last strong signal, the one that drowned out everything else. There was nothing we could do but let him go.

This article originally appeared in The New York Times.

Written by: Ellen Barry

Images by: Graham Dickie

©2026 THE NEW YORK TIMES