One of the tragedies of aging is the slow but steady decline in memory. Phone numbers slipping your mind? Forgetting crucial items on your grocery list? Opening the door but can't remember why? Over half of adults aged 64 years or older report memory complaints. For many of us, senile moments are the result of normal changes in brain structure and function, instead of a sign of dementia, and will inevitably haunt us all.

Fortunately, scientists are devising interventions to help keep the elderly mind sharp. One popular approach—borrowed from the training of memory experts—is to teach the elderly mnemonics, or little tricks that help to encode and recall new information using rhythm, imagery or spatial navigation.

By far the most widely used mnemonic device is the method of loci (MoL), a technique devised in ancient Greece. It involves picturing highly familiar routes through a building (your childhood home) or a town (your way to work). Walk down the route and imagine placing to-be-remembered items at attention-grabbing spots along the way; the more surreal or bizarre you make these images, the better they can help you remember. To recall these stored items, simply retrace your steps. Like fishing lines, the loci are hooked to the memory and help you pull them to the surface. Though generally used to remember objects, numbers or names, the MoL has also been used in people with depression to successfully store bits and pieces of happy autobiographical memories that they can easily retrieve in times of stress.

Yet with age, the brain gradually loses the flexibility to change in response to training. Nevertheless, many studies show that MoL successfully slows memory decline in the normal aging population, but why this happens is a complete mystery. That is, until recently.

Thickening of the brain
In 2010 a Norwegian team set out to look for the most obvious sign of MoL-induced neuroplasticity: macrostructural changes in the brain. Expert instructors led 23 volunteers with an average age of 61 through an intensive eight-week-long program. By the end of training, these volunteers were able to use MoL to remember three lists of 30 words in sequential order in no more than 10 minutes. The control group—matched in age, sex and education—went about their daily lives as usual for the same duration. During the final test, researchers challenged both groups with a tough task: they first flashed a list of 15 unrelated words, each for only a second, and asked the volunteers to remember the words in order. Researchers then showed them a list of 30 words—15 previously presented and 15 new—and asked both groups to not only pick out the matching words but also to identify their position in the first list. Because the second task involves spatial recognition, researchers reasoned that it should particularly benefit from MoL training.

As expected, MoL-trained volunteers correctly named the position of six more words, on average, than did the volunteers who were not trained, although both groups recognized a similar number of matching words. Structural MRI images taken before and after training identified a surprisingly large morphological change in the MoL-trained volunteers. These individuals showed significant thickening of the right fusiform cortex and the right orbitofrontal cortex, areas particularly involved in encoding abstract visual memory. The degree of thickening correlated with memory improvement: the larger the change, the better the performance.

The control group showed a trend toward thinning of their cortices, perhaps as a result of cortical atrophy in normal aging. A later study from the same group showed that learning MoL also preserves the structural integrity of white matter—the myelinated axons of nerve cells that form communication highways between brain regions.

This initial sleuthing shows that the elderly brain is still remarkably adaptable as it learns to process verbal information using spatial imagery. While intriguing, these studies only just scratched the surface: the volunteers were self-selected through answering a newspaper ad, and the small sample size cautions against generalization. Furthermore, these changes may reflect a general increase in cognitive load rather than the use of MoL per se. Still, compared with other mnemonics, the benefits of MoL often persist after initial training even without subsequent practice, strongly suggesting that structural changes underlie the method's long-term benefits.

Rewiring the brain
structural changes do not necessarily lead to changes in function. To directly assess whether the brain operates differently following MoL training, two groups of researchers decided to look at brain-activity patterns instead. In a 2003 study, Swedish scientists recruited volunteers in their twenties and sixties and used positron-emission tomography (PET) scans to follow changes in their brain activity as they adopted MoL to remember a list of random words. Several mentally grueling sessions later, all of the young volunteers—but only half of the elderly—remembered roughly four more words than they had in their initial test, with the younger crowd performing much better on average.

Those who improved showed increased activity in the occipitoparietal cortex and left retroplenial cortex during encoding. These brain regions are associated with spatial mental visualization and navigation. A similar pattern of activation was previously found in world-class memory champions as they spontaneously used MoL for memory tasks. This suggests that the observed changes were in fact a result of adopting the mnemonic, and that despite old age, the brain is still capable of a qualitative shift in the way it encodes new memories. A later study using functional MRI confirmed these results and further implicated the fusiform and lingual gyri in memory encoding.

What about the half of elderly participants who did not improve? One clue was their complete lack of activation of MoL-associated brain regions during testing, prompting researchers to wonder whether these volunteers actually used MoL during the experiment. As it turned out, they did not! A subsequent informal chat revealed that many older participants found it difficult to associate loci with words under the experiment's tight time constraints, felt frustrated and gave up. This seems to be a general response: a recent study involving more than 700 mnemonic-trained older adults found that only 25 percent made an effort to use MoL after initial learning in a series of follow-up sessions spanning five years.

Though a promising technique for many, MoL is particularly difficult for the elderly, who are less able to generate and rely on a mental map of distinctive landmarks. Is there any way to lower the barrier of entry for learning MoL?

Modernizing the mnemonic
In early 2012 a team of Canadian researchers gave the ancient MoL mnemonic a 21st-century facelift. Inspired by the first-person shooter game Half-life 2, the team constructed several detailed virtual-reality environments to serve as loci, rather than letting MoL learners generate their own. This technique deviated from the long-standing rule that the loci need to be of personal significance for the method to work.

But it paid off. Researchers allowed 142 undergraduate volunteers only five minutes to explore the virtual environment before asking them to memorize 110 unrelated words. Before the task began, the researchers taught two thirds of the volunteers how to use MoL and asked some of them to pick a familiar environment for their loci; the rest of this group was told to use the virtual environment they had just navigated. The other third of the volunteers did not receive any specific instructions on memory techniques.

Both MoL groups outperformed the controls. They were 10 to 16 percent more accurate in their recall, and students who used the virtual environment performed just as well as those told to generate their own landmarks. The interesting bit is that in both groups, participants reported that they sometimes used memory methods that they were more familiar with during the task. Compared with participants in the standard MoL group, however, those using the virtual environment took to the mnemonic more readily, suggesting that this approach was easier to adopt.

Extrapolating data from young adults to the aging population warrants caution, but researchers are optimistic. By using software that creates many diverse environments, the elderly may be able to tailor the richness and theme of each space to both their learning ability and to what they wish to remember. A young scientist from Belgium is dreaming big: in a 2013 TEDx talk, Kasper Bormans described using virtual-reality replicas of the homes of Alzheimer's disease patients to help them “store” the memory of their loved one's faces using MoL. By combining the mnemonic with training in other cognitive domains—logic and reasoning, organization and attention—as well as antidementia medication where needed, researchers hope to slow (if not halt) the memory loss that haunts our twilight years.