The human voice is one of our most powerful social signals: in a split second we can recognize who is speaking, sense their emotion, and infer intent. This ability is not unique to humans—macaques, marmosets, and other primates also rely on voices and calls to guide their complex social lives. Such parallels point to a neural machinery for decoding vocal signals that is deeply rooted in primate evolution. Yet we still lack a detailed account of how voices are represented at the level of single neurons, and how these circuits differ—or remain conserved—across species.
EVOICE will bridge this gap by moving from cortical maps to neuronal mechanisms. In WP1 we will use high-density, fMRI-guided recordings in macaques and marmosets (and, where feasible, in humans) to identify “voice cells” and map their anatomical distribution with cellular precision. In WP2 we will model the coding principles of these neurons using acoustic descriptors and deep neural networks, allowing us to disentangle universal strategies from lineage-specific adaptations and to identify neural precursors of key human perceptual abilities. WP3 will modulate the voice-cell systems, combining developmental exposure, adult training, and reversible perturbations to test how flexible, generalizable, and necessary these systems are.
EVOICE constitues the first comparative neuron-level investigation of cerebral voice perception. It will clarify how cortical circuits encode, adapt, and sustain vocal communication across primates. The results will shed crucial light on the evolutionary foundations of human voice perception, provide a principled basis for translational research on disorders of voice communication, and inform future brain–machine interface technologies, including cortical implants designed to restore or enhance speech and voice perception.
