Scientists find a walnut-sized area of the brain has a giant role in memory

Whether a new image becomes a memory may depend partly on how strongly a particular group of brain cells responds when it first appears. Recordings from people undergoing epilepsy monitoring reveal that certain amygdala neurons fire more strongly for images that participants later recognize. The finding broadens the role of a region usually associated with fear and emotion.

The Cedars-Sinai Health Sciences University–led study, published in Nature Neuroscience, examined memory formation at the level of individual neurons. Its results suggest that the amygdala helps prioritize specific content for memory, rather than simply responding to emotionally intense experiences. The strongest signals came from cells tuned to particular categories of images.

The work points toward possible treatment targets for memory disorders, including Alzheimer’s disease and temporal lobe epilepsy. However, the researchers recorded activity rather than testing a memory-restoring intervention. Their experiments measured image recognition after delays within the same day, not whether experiences remained accessible for months or years.

The project drew on recordings from 112 patients who had electrodes implanted to monitor epileptic seizures. These clinical circumstances allowed researchers to observe signals from individual neurons deep inside the brain. The electrodes served a medical purpose, while the memory tasks provided an opportunity to study how the recorded cells behaved.

First author Jiayang Xiao led the research with corresponding author Ueli Rutishauser, director of Cedars-Sinai’s Center for Neural Science and Medicine. A multi-institutional consortium supported the large collection of recordings. Funding included the National Institutes of Health’s BRAIN Initiative, which supports tools and research for understanding brain activity.

The team tested its findings in two datasets using different tasks. One combined a short-term working memory exercise with a later recognition test. The other used a more conventional learning-and-recognition procedure, providing a separate test of the main result.

In the first task, participants viewed unfamiliar pictures from five categories: people, food, animals, cars and landscapes. After a 10–30-minute delay following the working memory task, they saw 400 images. Half had appeared earlier, and participants judged whether each picture was old or new while reporting their confidence. Participants’ overall accuracy in that recognition test averaged about 70%, compared with a 50% chance level.

The second task began with 100 unfamiliar images. Following another 10–30-minute interval, participants viewed 100 pictures, half previously presented and half new. Again, they identified old and new images and rated how certain they were.

The investigators identified neurons that responded more strongly to one image category than to others. A cell might show its strongest response to animals, for example, while another preferred people or landscapes. The researchers called these category-selective neurons because their activity reflected a preference for a type of visual content.

Those preferences alone were not the central discovery. Within a neuron’s preferred category, its response was stronger when an image was later remembered than when it was forgotten. The same memory-related difference did not appear reliably when images belonged to categories that the cell did not prefer.

“Our results demonstrate that the activity of specific kinds of neurons in the amygdala predicts, while the person sees an image for the first time, whether that image will later be remembered,” Rutishauser said. That relationship links the content being viewed with later memory performance. It does not mean that one neuron independently decides which experiences the brain keeps.

The first task yielded recordings from 1,871 neurons across six brain regions, including 575 in the amygdala. About 40% of the recorded amygdala neurons met the study’s criteria for category selectivity. The researchers then compared responses to subsequently remembered and forgotten images within those preferred categories.

The amygdala has long attracted attention for its involvement in emotional memory. Powerful experiences can receive priority, making this region a natural focus for studies of memorable events. The new work examined whether its contribution also involves the specific information being encoded.

To address emotion, the team included ratings of the images’ emotional valence and arousal in its analyses. Valence describes whether content is positive or negative, while arousal describes its stimulating quality. The relationship between category-cell activity and later memory remained after accounting for those ratings.

The effect also appeared when positive and negative images were analyzed separately, and in both higher- and lower-arousal groups. Those results argue against emotional content alone explaining the pattern. They support a broader role for the amygdala without implying that emotion is irrelevant to memory.

Stronger responses also predicted whether participants would later recognize an image with high rather than low confidence. Additional modeling linked the neural signal to the recollective component of memory, beyond a general sense of familiarity. The result connects activity during the first encounter with the strength of the memory expressed later.

The memory-related difference appeared in amygdala activity about 350 milliseconds after an image appeared in the first task. A related signal emerged later in the hippocampus, around 1,300 milliseconds after presentation. This timing suggests that the amygdala participates early in the process, although timing alone does not establish that it causes subsequent hippocampal activity.

Successful encoding also involved greater coordination between the two regions. The timing of individual neuronal signals aligned more closely with slower electrical fluctuations recorded across the amygdala–hippocampal circuit. That coordination offers a possible mechanism for connecting representations into an episodic memory.

Both the left and right amygdala showed memory-related signals, with a stronger effect on the right. The researchers suggest that this difference could reflect the visual nature of the tasks. Further studies must establish whether the asymmetry holds for other kinds of information.

The findings identify a more specific candidate for investigation than the amygdala as a whole. Future approaches might try to influence cells representing the information a person is trying to remember. Whether such targeting can safely improve memory remains an open question.

“Our findings have important clinical implications, as the memory impairments seen in disorders such as Alzheimer’s disease and temporal lobe epilepsy might be partly due to disrupted processes within the amygdala,” Rutishauser said. He identified neuromodulation and cellular therapies as possible research directions. Neither approach was established as a treatment by this study.

The patient population also defines the evidence’s limits. Recordings from people undergoing epilepsy monitoring offer unusually detailed access to human neurons, but do not represent every population affected by memory problems. Laboratory picture tests likewise capture only part of the complexity of autobiographical memory.

Rutishauser said future work could investigate whether the amygdala plays a similar role for sounds and smells. That would test how broadly these category-specific signals extend beyond vision. Another key challenge is determining whether changing the signals improves memory, rather than merely predicting which images someone will remember.

These studies connect recent human findings with foundational research on memory circuits and experimental stimulation.

Direct electrical stimulation of the human amygdala enhances recognition memory for objects but not scenes
This experiment examines how stimulation can affect different kinds of visual memory differently, highlighting the specificity of potential interventions. (Communications Biology, May 2026)

Exploring individual differences in amygdala-mediated memory modulation
Researchers found substantial variation in responses to stimulation, including impairment in some participants, underscoring why treatment development requires caution. (Cognitive, Affective, & Behavioral Neuroscience, December 2024)

Concept and location neurons in the human brain provide the ‘what’ and ‘where’ in memory formation
Single-neuron recordings explore how representations of items and locations contribute to forming memories of events. (Nature Communications, September 2024)

Research findings are available online in the journal Nature Neuroscience.

The original story "Scientists find a walnut-sized area of the brain has a giant role in memory" is published in The Brighter Side of News.

Like these kind of feel good stories? Get The Brighter Side of News' newsletter.