Skip to main content

memory retrieval

How Our Brains Replay Memories

Posted on by

Retrieving a Memory
Caption: Encoding and replaying learned memory. Left panel shows the timed sequence of neurons firing in a part of a person’s brain involved in memory as it encodes the random pair of words, “crow” and “jeep.” Colors are assigned to different neurons to differentiate their firing within the sequence. Right panel shows a highly similar timed sequence of those same neurons firing just before a person given the word “jeep,” recalled and said the correct answer “crow.” Credit: Vaz AP, Science, 2020.

Note to my blog readers: the whole world is now facing a major threat from the COVID-19 pandemic. We at NIH are doing everything we can to apply the best and most powerful science to the development of diagnostics, therapeutics, and vaccines, while also implementing public health measures to protect our staff and the patients in our hospital. This crisis is expected to span many weeks, and I will occasionally report on COVID-19 in this blog format. Meanwhile, science continues to progress on many other fronts—and so I will continue to try to bring you stories across a wide range of topics. Perhaps everyone can use a little break now and then from the coronavirus news? Today’s blog takes you into the intricacies of memory.

When recalling the name of an acquaintance, you might replay an earlier introduction, trying to remember the correct combination of first and last names. (Was it Scott James? Or James Scott?) Now, neuroscientists have found that in the split second before you come up with the right answer, your brain’s neurons fire in the same order as when you first learned the information [1].

This new insight into memory retrieval comes from recording the electrical activity of thousands of neurons in the brains of six people during memory tests of random word pairs, such as “jeep” and “crow.” While similar firing patterns had been described before in mice, the new study is the first to confirm that the human brain stores memories in specific sequences of neural activity that can be replayed again and again.

The new study, published in the journal Science, is the latest insight from neurosurgeon and researcher Kareem Zaghloul at NIH’s National Institute of Neurological Disorders and Stroke (NINDS). Zaghloul’s team has for years been involved in an NIH Clinical Center study for patients with drug-resistant epilepsy whose seizures cannot be controlled with drugs.

As part of this work, his surgical team often temporarily places a 4 millimeter-by-4 millimeter array of tiny electrodes on the surface of the brains of the study’s participants. They do this in an effort to pinpoint brain tissues that may be the source of their seizures before performing surgery to remove them. With a patient’s informed consent to take part in additional research, the procedure also has led to a series of insights into what happens in the human brain when we make and later retrieve new memories.

Here’s how it works: The researchers record electrical currents as participants are asked to learn random word pairs presented to them on a computer screen, such as “cake” and “fox,” or “lime” and “camel.” After a period of rest, their brain activity is again recorded as they are given a word and asked to recall the matching word.

Last year, the researchers reported that the split second before a person got the right answer, tiny ripples of electrical activity appeared in two specific areas of the brain [2]. The team also had shown that, when a person correctly recalled a word pair, the brain showed patterns of activity that corresponded to those formed when he or she first learned to make a word association.

The new work takes this a step further. As study participants learned a word pair, the researchers noticed not only the initial rippling wave of electricity, but also that particular neurons in the brain’s cerebral cortex fired repeatedly in a sequential order. In fact, with each new word pair, the researchers observed unique firing patterns among the active neurons.

If the order of neuronal firing was essential for storing new memories, the researchers reasoned that the same would be true for correctly retrieving the information. And, indeed, that’s what they were able to show. For example, when individuals were shown “cake” for a second time, they replayed a very similar firing pattern to the one recorded initially for this word just milliseconds before correctly recalling the paired word “fox.”

The researchers then calculated the average sequence similarity between the firing patterns of learning and retrieval. They found that as a person recalled a word, those patterns gradually became more similar. Just before a correct answer was given, the recorded neurons locked onto the right firing sequence. That didn’t happen when a person gave an incorrect answer.

Further analysis confirmed that the exact order of neural firing was specific to each word pair. The findings show that our memories are encoded as unique sequences that must be replayed for accurate retrieval, though we still don’t understand the molecular mechanisms that undergird this.

Zaghloul reports that there’s still more to learn about how these processes are influenced by other factors such as our attention. It’s not yet known whether the brain replays sequences similarly when retrieving longer-term memories. Along with these intriguing insights into normal learning and memory, the researchers think this line of research will yield important clues as to what changes in people who suffer from memory disorders, with potentially important implications for developing the next generation of treatments.


[1] Replay of cortical spiking sequences during human memory retrieval. Vaz AP, Wittig JH Jr, Inati SK, Zaghloul KA. Science. 2020 Mar 6;367(6482):1131-1134.

[2] Coupled ripple oscillations between the medial temporal lobe and neocortex retrieve human memory. Vaz AP, Inati SK, Brunel N, Zaghloul KA. Science. 2019 Mar 1;363(6430):975-978.


Epilepsy Information Page (National Institute of Neurological Disorders and Stroke/NIH)

Brain Basics (NINDS)

Zaghloul Lab (NINDS)

NIH Support: National Institute of Neurological Disorders and Stroke; National Institute of General Medical Sciences

Discovering the Brain’s Nightly “Rinse Cycle”

Posted on by

Getting plenty of deep, restful sleep is essential for our physical and mental health. Now comes word of yet another way that sleep is good for us: it triggers rhythmic waves of blood and cerebrospinal fluid (CSF) that appear to function much like a washing machine’s rinse cycle, which may help to clear the brain of toxic waste on a regular basis.

The video above uses functional magnetic resonance imaging (fMRI) to take you inside a person’s brain to see this newly discovered rinse cycle in action. First, you see a wave of blood flow (red, yellow) that’s closely tied to an underlying slow-wave of electrical activity (not visible). As the blood recedes, CSF (blue) increases and then drops back again. Then, the cycle—lasting about 20 seconds—starts over again.

The findings, published recently in the journal Science, are the first to suggest that the brain’s well-known ebb and flow of blood and electrical activity during sleep may also trigger cleansing waves of blood and CSF. While the experiments were conducted in healthy adults, further study of this phenomenon may help explain why poor sleep or loss of sleep has previously been associated with the spread of toxic proteins and worsening memory loss in people with Alzheimer’s disease.

In the new study, Laura Lewis, Boston University, MA, and her colleagues at the Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston. recorded the electrical activity and took fMRI images of the brains of 13 young, healthy adults as they slept. The NIH-funded team also built a computer model to learn more about the fluid dynamics of what goes on in the brain during sleep. And, as it turns out, their sophisticated model predicted exactly what they observed in the brains of living humans: slow waves of electrical activity followed by alternating waves of blood and CSF.

Lewis says her team is now working to come up with even better ways to capture CSF flow in the brain during sleep. Currently, people who volunteer for such experiments have to be able to fall asleep while wearing an electroencephalogram (EEG) cap inside of a noisy MRI machine—no easy feat. The researchers are also recruiting older adults to begin exploring how age-related changes in brain activity during sleep may affect the associated fluid dynamics.


[1] Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep. Fultz NE, Bonmassar G, Setsompop K, Stickgold RA, Rosen BR, Polimeni JR, Lewis LD. Science. 2019 Nov 1;366(6465):628-631.


Sleep and Memory (National Institute of Mental Health/NIH)

Sleep Deprivation and Deficiency (National Heart, Lung, and Blood Institute/NIH)

Alzheimer’s Disease and Related Dementias (National Institute on Aging/NIH)

NIH Support: National Institute of Mental Health; National Institute of Biomedical Imaging and Bioengineering; National Institute of Neurological Disorders and Stroke

Unlocking the Brain’s Memory Retrieval System

Posted on by

Memory Trace in Mouse Hippocampus

Credit:Sahay Lab, Massachusetts General Hospital, Boston

Play the first few bars of any widely known piece of music, be it The Star-Spangled Banner, Beethoven’s Fifth, or The Rolling Stones’ (I Can’t Get No) Satisfaction, and you’ll find that many folks can’t resist filling in the rest of the melody. That’s because the human brain thrives on completing familiar patterns. But, as we grow older, our pattern completion skills often become more error prone.

This image shows some of the neural wiring that controls pattern completion in the mammalian brain. Specifically, you’re looking at a cross-section of a mouse hippocampus that’s packed with dentate granule neurons and their signal-transmitting arms, called axons, (light green). Note how the axons’ short, finger-like projections, called filopodia (bright green), are interacting with a neuron (red) to form a “memory trace” network. Functioning much like an online search engine, memory traces use bits of incoming information, like the first few notes of a song, to locate and pull up more detailed information, like the complete song, from the brain’s repository of memories in the cerebral cortex.