Strengthening Memories During Sleep

Even though we spend about a third of our lives asleep, the function of sleep is still not fully understood. One of the leading hypotheses concerns the beneficial role of sleep for the consolidation of memories. Thousands of years ago, philosophers and early psychologists realized that memories are strengthened over a period of sleep. However, the mechanism by which this benefit is achieved still remains somewhat of a mystery.

In the mid-1990s, a curious scientific discovery in rodents ushered in a new age of sleep and memory research. Researchers noticed that brain cells in the hippocampus, a brain region involved in memory formation, were activated when the animals were running in a maze. Curiously, these same cells were activated in the same order when the animals were later asleep. This phenomenon, called hippocampal replay, has since been predominantly linked with memory consolidation during sleep.

But what about humans? Recording the activity of these cells in human brains requires access to deep brain structures, which is a non-trivial experimental task. However, in 2007 Björn Rasch and his colleagues developed a new scientific paradigm that provided some crucial evidence for memory “reactivation” in humans. Rasch first asked participants to learn the locations of pairs of images while smelling a rose-like odor. After learning, the participants went to sleep in the laboratory. Unbeknownst to their participants, the researchers presented the odor during a specific stage of deep sleep called slow-wave sleep. Upon waking up, the participants were tested on their memory and – despite having no memory of the odor presentation during sleep – had significantly better memory relative to a group that did not receive the odor during sleep.

The researchers hypothesized that presenting the odor reactivated the related memories during sleep and primed them for consolidation. This paradigm was later named targeted memory reactivation (TMR). Two years later, John Rudoy and his colleagues expanded on this design by replacing the odors with learning-related sounds, unobtrusively presented during sleep. Their findings revealed the memory-specific nature of TMR: cueing during sleep selectively enhanced memory for a single item, whereas a different item whose sound was not presented during sleep did not enjoy the same benefit.

Ever since these findings were first published, they have sparked the imagination of sleep and memory researchers. Follow-up studies have found that cueing during sleep can improve memory not only for spatial locations but also for motor-related and vocabulary learning. Dozens of studies have considered the application of TMR for various types of memories and explored related questions, such as the ability to learn new information during sleep. Simultaneously, we and other research groups aim to reveal the boundary conditions of TMR and trace back its connections to the better-understood phenomena of hippocampal replay.

Recently, we sought to explore whether sets of items can be simultaneously reactivated using auditory cues. Whereas Rudoy’s study had a single congruent sound for each item (e.g., a cat picture was coupled with a “meow” sound), we paired up arbitrary items (for example, a telephone and a burger) and taught separate spatial locations for both, while presenting the sound of the first item (a dial tone). Later during sleep, we presented the sounds associated with half of the learned pairs. We were curious to see whether the cueing benefit would carry over to the item that was not directly related to the presented sound (the burger, in our example). Our results proved that was, in fact, the case. Both items benefited from the presentation of the sounds. Moreover, a larger change in memory in one of the items was associated with a larger change in its paired companion.

In many ways, this result is reminiscent of the original study by Rasch and colleagues, in which they used odor – not sounds – to improve memory for multiple unrelated items. Items may be reactivated together regardless of cue modality, even when only a subset of sets are cued during sleep. The notion of multiple parallel reactivations of memory may imply that the benefit of cueing is divided between the different items, whereby each receives a “slice of the cake.” An alternative explanation is that all items benefit independently of one another in a parallel manner. Preliminary, unpublished results suggest the latter and raise the possibility that the forest – rather than the trees – is what is being activated. In other words, the benefit of TMR to a specific item may not depend on the number of items reactivated.

Theoretically, these findings may also suggest TMR is a wider, more generalized phenomenon than hippocampal replay, which involves a specific activation of cells associated with a specific memory. Most importantly, these new insights regarding TMR may vastly extend its possible benefit to memory in life-like scenarios. Imagine if a student could selectively manipulate sleep to enhance her memory of material covered over multiple classes throughout the day. This exciting idea remains to be empirically tested, but the prospects are mind-blowing. After thousands of years of exploring the benefit of sleep to memory, we may just be on the verge of finally harnessing it for our benefit.