A widely used experimental paradigm in psychological research and clinical assessments involves identifying emotions from facial expressions, typically using posed expressions as stimuli. Perceptions of such stimuli are assumed to mirror those of naturally occurring emotional expressions. However, this assumption has been questioned because the perceptual equivalence of posed and spontaneous expressions has not been empirically established. To address this, we directly compared perceptual judgments of posed and spontaneous facial expressions produced by the same expressers in three preregistered studies. A total of 2,408 perceivers judged the emotions displayed in 1,244 dynamic facial expressions of eight emotions (anger, disgust, fear, sadness, joy, pride, compassion, and love). Consistent with our main hypothesis, emotions were much better recognized from posed compared to spontaneous expressions, by both Western (Study 1, N = 470) and non-Western perceivers (Study 2, N = 438). This pattern was replicated in a cross-cultural context in Study 3 (N = 1,500). Furthermore, in all three studies, we observed a "negativity bias" with only posed expressions. Specifically, negative emotions were better recognized than positive emotions from posed expressions, while the opposite was true for spontaneous expressions, such that positive emotions were better recognized than negative emotions. Our findings present clear evidence that perceptions of posed and spontaneous facial expressions meaningfully differ, and raise questions about the generalizability of findings from existing research that uses posed emotional expressions.