Infants respond to alterations in their own voice months before they recognize themselves in the mirror

Recognizing our body as our own grounds our perceptions, actions, and social interactions. Even newborns detect cross-modal links between self-related perceptions,¹ considered necessary for body ownership.² Yet, clear evidence for a sense of ownership only appears once children not simply detect but also correct mismatches between predicted and perceived self-related sensorimotor inputs. So far, this has mostly been measured using mirror self-recognition, which revealed that children typically do not start recognizing themselves before 18 months of age.^3,4,5 Infants rarely see their own face, however, and mirror self-recognition has been related to sensorimotor⁶ and social experiences.⁷ We therefore hypothesized that a sense of ownership over the voice might develop earlier, because most infants routinely perceive themselves vocalizing from birth.^8,9,10 This might allow them to build sensorimotor models supporting the identification of self-generated signals earlier in this modality. Related research with adults has used the altered auditory feedback paradigm to show that they respond to transpositions of the fundamental frequency (f₀) of their voice by restoring the alignment between predicted and perceived feedback.^11,12,13,14 Crucially, these compensatory changes depend on the voice being perceived as self-generated.^{15,16,17,18,19} Using an adaptation of this paradigm with 5-month-old infants, we found that they can already perceive and correct upward transpositions of their f₀, at an age when they remain unable to recognize themselves in the mirror. This suggests that hearing ourselves routinely allows us to learn what we sound like months before we recognize what we look like, which might be especially crucial for speech acquisition.