Why Your Old Search History Reads Like a Stranger's

In 2000 a 34-year-old woman emailed James McGaugh, a memory researcher at the University of California, Irvine, asking for help. Since she was eleven, she wrote, she had been able to remember what she was doing on almost any day of her life. Whenever she saw a date flash up on television, or anywhere else, she was taken straight back to it. “Most have called it a gift but I call it a burden. I run my entire life through my head every day and it drives me crazy!!!”

McGaugh’s group studied her for years, and in 2006 they published her case in Neurocase under the initials AJ, describing a memory that was “nonstop, uncontrollable, and automatic”. She later went public under her own name, Jill Price.

Her complaint was not that she remembered things wrongly. It was that she could not stop remembering them accurately, and could not stop being taken back. A growing line of neuroscience argues that the ordinary capacity to forget, which she lacked, is not a shortfall. In 2017 Blake Richards and Paul Frankland argued in Neuron that memory research had concentrated on how memories persist, and that forgetting is part of what memory is for. On their account the goal of memory is not to carry information through time. It is to make good decisions now. Forgetting helps in two ways: it weakens the pull of information that has gone out of date, and it stops the mind fitting itself too closely to single past events at the expense of the general pattern.

Very few people have Price’s memory. Almost everyone now carries something that behaves like it.

Open an old search history and it reads like someone else’s. The queries are recognisably yours, down to the misspellings and the habit of typing a whole question into a box built for keywords, but the person who needed those answers has gone. A search history is not a record of what someone knew. It is a record of what they lacked: the symptom typed in at three in the morning, or the same question asked several ways because none of the answers settled it. Each entry is a want caught at the moment it reached for something, kept long after the want moved on, and still available to the systems that model you from records like it.

The drafts folder works the same way. It holds the versions of messages that were never sent, and an earlier version can be closer to what was meant than the one that went out. The camera roll holds the moments someone decided were worth keeping, which is a different list from the moments that mattered.

Unless an auto-delete setting is on, none of it fades. That is the design.

Does taking photos make you remember less?

The popular version of this story says the phone is replacing memory: we photograph things so we no longer have to remember them. The research supports something stranger.

In 2014 the psychologist Linda Henkel led people on a guided tour of an art museum and told them to observe some objects and photograph others. Afterwards they remembered fewer of the objects they had photographed, and fewer details about them and about where in the museum they had been. Henkel called it the photo-taking-impairment effect. There was an exception. When people zoomed in to photograph one part of an object, the impairment disappeared, and their memory for the parts they had not zoomed in on held up as well.

Three years later Alixandra Barasch and colleagues complicated the picture. Across a field study and three lab studies, people who were free to take photos during an experience recognised more of what they saw and less of what they heard than people without a camera, even though they never looked at their photos again. Taking “mental photos”, with no camera at all, did the same thing. The camera did not wipe the experience. It moved attention toward the eyes and away from the ears while the experience was still happening.

The obvious explanation for Henkel’s result was offloading: the camera remembers, so the mind lets go. In 2018 Julia Soares and Benjamin Storm tested that by having people photograph objects on Snapchat, where the image disappears, or delete each photo the moment it was taken. Memory was impaired anyway, by as much as or more than when people expected to keep the photos. They concluded that offloading may not be the main mechanism. The act of taking the photo seems to pull attention away at the moment of capture, whether or not anything is saved. Diana Tamir, Jamil Zaki and colleagues found a similar pattern with recording and sharing: across three studies, people who went through an experience without media remembered it more accurately, while the evidence that recording changed how much they enjoyed it was inconclusive.

None of this makes external memory simply a loss. In 2015 Storm and Sean Stone showed that saving one file before studying another improved memory for the second, as long as people trusted the save. Putting something somewhere reliable frees the mind to take in what comes next. That is what external memory is for, and it works.

What the studies support is not replacement. Documenting changes the experience before there is anything to remember. The phone is already reshaping what gets noticed while the moment is still going on, and the version of the moment that gets kept is the one the camera was pointed at.

What an archive cannot do

In 2000 Karim Nader, Glenn Schafe and Joseph LeDoux showed in rats that a consolidated fear memory, once recalled, returns to an unstable state and has to be stored again, and that blocking the restorage left the animals unable to remember on later tests. How far and how reliably this applies to human memory is still argued over. But it changed the field’s picture of recall. Remembering is not playing back a recording. A memory is rebuilt when it is used, by the person using it, and the person using it keeps changing.

Human memory still works that way. What is new is that it now runs alongside a record that does not. A memory of a bad year is rebuilt each time it comes back, by someone who knows how the year ended. The camera roll from that year was filled by someone who did not. The search history from the month someone was frightened about their health can still hold every query, in order, exactly as typed. When the two meet, the rebuilt memory and the unrevised record, the record wins on detail and the memory wins on meaning. That is where the stranger comes from. The archive is accurate about a person you have spent years revising.

Phones now arrange these meetings on purpose. Photo apps assemble the past into slideshows and offer them unasked: this day, some years ago. A date appears on the screen, and you are taken back to it. It is a mild, occasional version of what Price described, set off the same way hers was.

Richards and Frankland’s point was that a mind which forgot nothing would decide worse, held to every past event as if it were still current. That part is well grounded: forgetting is part of how memory does its job, not a fault in it, and photographing something changes how it is attended to whether or not the photo survives. What is much less studied is the long run, what it does to a person to carry an outside memory with almost no transience, one that keeps every three a.m. search exactly as it was. The research so far mostly measures the moment of capture, not the years after.

Price is the nearest thing to a test case, and she had no doubt about her answer. Whether the rest of us would give the same one, or have simply stopped opening the folder, is not something the studies so far were built to find out.