Amygdala Neurons Predict Which Images People Remember

The amygdala decides what you remember, and it isn't only deciding about frightening things. In a study published October 27, 2026, in Nature Neuroscience, researchers at Cedars-Sinai Health Sciences University recorded individual neurons in the amygdala while patients looked at ordinary images, and found that the strength of those neurons' first response predicted whether the image would still be remembered minutes later.

The recordings came from 112 patients who'd had electrodes implanted in their brains to find the source of drug-resistant epileptic seizures. The electrodes let the team listen to single neurons in the amygdala, a walnut-sized structure deep in the temporal lobe that's best known for processing fear and aggression. The work ran through a multi-institution consortium led by Cedars-Sinai and backed by the National Institutes of Health's BRAIN Initiative.

Patients viewed images of everyday objects in five categories, including food and animals. About 10 to 30 minutes later, they were asked which images they'd already seen. Certain amygdala neurons preferred one category over the others, and when a neuron fired strongly for an image in its preferred category, that image was the one the patient later recalled. The pattern held in two separate datasets inside the study.

The team's reading of the process has two steps. First the amygdala categorizes what the eyes are seeing. Then the strength of the category neuron's response sets the tag: strong response, and the experience gets earmarked for longer-term storage; weak response, and it fades. The finding turns the amygdala into a sorting gate that runs at the moment of seeing, not a storage vault that fills up afterward.

The surprise is the address. Memory research usually starts at the hippocampus, and senior author Ueli Rutishauser, a professor of neurosurgery at Cedars-Sinai, said it surprised him that the relationship between memory formation and amygdala activity showed up so clearly. Earlier work had tied the amygdala to memory mainly for emotionally charged events. This study extends the job to neutral, everyday visual experience.

The limits are plain. The study shows a correlation between a neuron's response and later recall; it doesn't show that changing amygdala activity would change what a person remembers, and the exact mechanism that turns a strong first response into a stored memory isn't nailed down. The recall window was also short, half an hour at most, and the experiments covered visual information only.

First author Jiayang Xiao and the team frame the result as a map coordinate, not a treatment. A structure that tags experiences for keeping is a plausible place to look when memory fails, including in Alzheimer's disease and temporal lobe epilepsy, but that step is research the paper points toward rather than work it does.

The data exists because epilepsy care created the opening. Patients already carried implanted electrodes so clinicians could find where their seizures start, and the memory experiments rode on that clinical setup with the patients' participation. That route is why the study could record single neurons in the human amygdala at all, a measurement animal work can approximate but can't replace, and why assembling recordings from 112 patients took a consortium rather than a single lab.