How does your brain know who someone is?

Imagine recognizing someone you know across a room. You can recognize them by their face. You can recognize them by their voice when you can’t see them. But your brain doesn’t normally treat those as two unrelated pieces of information. They are both the same person.

How does the brain transform completely different sensory signals—a face and a voice—into an abstract representation of who someone is?

We recorded individual neurons from the hippocampus while marmosets saw photographs and heard calls from familiar animals in their social environment. Sometimes they saw only a face, sometimes heard only a voice, and sometimes saw a face while simultaneously hearing either the matching or a different individual’s voice.

Some hippocampal neurons responded selectively to a particular individual whether that animal was identified by face or voice. Other neurons encoded whether the face and voice belonged to the same individual.

Across the broader neural population, identity could be decoded across sensory modalities, and the representation also contained information about social relationships such as family membership.

The ability to connect a face, a voice, and past experience to the same person is fundamental to memory and social life. These functions depend on medial temporal-lobe systems that are among the earliest and most severely affected in disorders such as Alzheimer’s disease and other forms of dementia, where recognizing people and maintaining knowledge about relationships can progressively deteriorate.

By identifying how the primate brain links different sensory cues to a common representation of identity, this work provides a mechanistic framework for asking how disease disrupts the representations that allow us to know who someone is. That could ultimately help connect changes measured in neural circuits to clinically meaningful losses of memory, recognition, and social knowledge.