Let me tell you a tale of two grandfathers, Irv and Murray. For decades, Irv, an introverted, quiet, retired bartender and former military engineer, had the same morning routine: coffee and cream; a roll; and the puzzle page of the Press of Atlantic City. He methodically and religiously worked his way through each one, from the crossword to the jumble to the cryptoquip, a substitution cipher that asks solvers to decode clues and figure out the pun.

Extroverted and spontaneous Murray, a successful businessman and local politician, also had his morning routine: coffee with lots of sugar; oatmeal; and tinkering on one of his many writing projects, such as a loosely autobiographical musical about a traveling salesman. Murray swam a few times a week, devoured books and loved to travel. But he never did crosswords.

Irv died at age 94, and he barely experienced any cognitive loss before the last six months of his life, when he exhibited rapid mental decline. Murray lived to be 91, but the last several years of his life were marked with severe dementia.

When I was researching my book Thinking Inside the Box: Adventures with Crosswords and the Puzzling People Who Can’t Live Without Them, I was fascinated by my family’s case study. The evidence, it seemed, couldn’t be clearer: doing crosswords late in life prevents dementia. And at first, all the studies I found seemed to bear this hypothesis out. “Regular crosswords and number puzzles linked to sharper brain in later life,” a May 2019 Science Daily headline proclaims. According to a University of Exeter study, older adults who regularly did word and number puzzles had increased mental acuity. A 2011 experiment with members of the Bronx Aging Study found that a regular regimen of crosswords might delay the onset of cognitive decline. Belief in puzzle power has fueled multimillion-dollar industry of brain-training games like Lumosity or Dakim.

But as I dug deeper, I found that the narrative swung just as persuasively in the other direction. “Crosswords and puzzles do not prevent mental decline, study says,” a December 2018 CNN headline declares. A longitudinal Scottish study showed that a regular puzzle habit didn’t seem to affect test subjects’ mental sharpness one way or the other. Other crossword-bashing studies point out that people would be better served flexing their brains in a more creative activity—say, writing a musical.

Indeed, the more I researched, the more I realized that the story about crosswords and dementia could be told in either direction, and in equally compelling ways. Media coverage about the crossword and dementia is completely black and white: crosswords are either the brain’s secret weapon or they’re a gigantic waste of time.


The crossword does seem to have a special relationship to memory: not as medicine, but as a sort of lens for understanding our mental storage units. (Maybe it’s no accident that the layout of every self-storage warehouse resembles nothing so much as a crossword grid.) How the mind does crosswords provides a snapshot of the hard-to-define limbo between short- and long-term memory, where you form integrated units somewhat more complicated than just rapid-fire perceptions but that don’t get filed into long-term storage.

Consider H.M., the most famous memory patient of the twentieth century. A 1953 operation to remove his hippocampus left H.M. with nearly perfect anterograde amnesia, meaning he could form short- but not long-term recollections, rendering him the ideal subject for memory studies. H.M. also loved crossword puzzles. A 2004 study led by Duke scientist Brian Skotko proved that even though H.M. couldn’t consciously lay down new tracks in long-term memory, when he did the same crossword over and over, he improved at clues he shouldn’t have been able to know how to solve (because they referred to post-1953 events). H.M. didn’t know why he knew these answers, but he knew that he knew them. Crosswords reveal something going on in this space between the short-term working memory and long-term permanent recollections.

In 1974, psychologists Alan Baddeley and Graham Hitch divided short-term memory into three parts: the central executive, which controls the flow of information; and two “slave” systems, the phonological loop and the visuo-spatial sketchpad, that feed perceptions to the executive. But that model didn’t account for everything that working memory could do. Long-term amnesiacs, for example, can recount events that had just happened to them in far more sophisticated narrative detail than the phonological loop can process. So in 2000, Baddeley revised the model to add a third slave system: the episodic buffer, a liaison between short- and long-term memory that basically integrates both verbal and visuospatial information and lets you hold a narrative in your head.

But the episodic buffer’s been hard to study: how can you tell what kinds of processes are combined or separate? In 2014, researchers in Claremont, Calif., examined how Scrabble and crossword experts process and store information in short-term memory. They found that these puzzle whizzes had extremely strong working memories as compared to their control group, college students who had scored 700 or higher on the verbal section of the SAT.

It’s hardly surprising that people very good at word games would do well when tested on their verbal skills, but things got interesting when the researchers looked at how their short-term memories were firing. Both groups, but especially the crossword experts, appeared to use both verbal and visuospatial components of their short-term memory—that is, that instead of separating out visual cues from verbally processed ones, the crossworders were integrating both types of perceptions in their short-term memories. Here was the episodic buffer, loud and clear.

Even though the episodic buffer makes intuitive sense—I feel like I process short-term memories in integrated units—this Scrabble and crossword study is one of the first to provide support to the model. In other words, crosswords give us an MRI scan into this odd void between short-term and long-term memory: the gray space in our gray matter.


Whatever the effect on their mental agility, Irv’s puzzle-loving brain and Murray’s puzzle-eschewing brain did bond over one memory-testing routine: Jeopardy! They’d watch together a few times a week and shout at the television in friendly competition. Irv knew sports and history; Murray, literature and famous people. Even when Murray was having trouble with short-term memories, and even when Irv couldn’t drive at night anymore, they could both tap into facts from the deep past, often amazingly accurately. Combined, they were formidable.