In several previous studies, 18‐month‐old infants who were directly addressed demonstrated more robust imitative behaviors than infants who simply observed another's actions, leading theorists to suggest that child‐directed interactions carried unique informational value. However, these data came exclusively from cultural communities where direct teaching is commonplace, raising the possibility that the findings reflect regularities in infants' social experiences rather than responses to innate or a priori learning mechanisms. The current studies consider infants' imitative learning from child‐directed teaching and observed interaction in two cultural communities, a Yucatec Mayan village where infants have been described as experiencing relatively limited direct instruction (Study 1) and a US city where infants are regularly directly engaged (Study 2). Eighteen‐month‐old infants from each community participated in a within‐subjects study design where they were directly taught to use novel objects on one day and observed actors using different objects on another day. Mayan infants showed relative increases in imitative behaviors on their second visit to the lab as compared to their first visit, but there was no effect of condition. US infants showed no difference in imitative behavior in the child‐directed vs. observed conditions; however, infants who were directly addressed on their first visit showed significantly higher overall imitation rates than infants who observed on their first visit. Together, these findings call into question the idea that child‐directed teaching holds automatic or universal informational value. We considered 18 month old infants' imitative learning from child‐directed and observed interaction in two cultural communities: a Yucatec Mayan village where infants have been described as experiencing relatively limited direct instruction (Study 1) and a US city where infants are regularly directly engaged (Study 2). Infants participated in a within‐subjects study design where they were directly taught to use novel objects on one day and observed actors using different objects on another day. Mayan infants' imitation did not relate to condition, whereas US infants who were directly addressed on their first visit showed significantly higher overall imitation rates than infants who observed on their first visit.