Context shapes emotion perception of real-life laughter and crying vocalizations regardless of their diverse perceptual properties

Affective vocalizations are central to human social experience yet researchers debate their nature. Across two preregistered studies (N=842), we tested the diagnosticity and contextual malleability of real-life laughter and crying vocalizations. Using human data and machine learning methods we demonstrate that real-life laughter and crying vocalizations are perceptually diverse, with ~40% of the vocalizations perceived as conveying atypical affective messages. This vocal diversity was also reflected by the distinct acoustic features of typical versus atypical vocalizations. Finally, we demonstrate that co-occurring visual context has profound impact on the perception of real-life laughter and crying vocalizations, especially when vocal cues are ambiguous. Together, we show that real-life vocalizations convey both typical and atypical affective cues and highlight the striking role of contextual information and multimodal integration in facilitating emotion understanding in everyday life.