Joint attention (JA) is a cornerstone of adaptive human social functioning. Little functional magnetic resonance imaging (fMRI) research has examined, in interactive paradigms, neural activation underlying bids for JA, met with a congruent or an incongruent social response. We developed a highly naturalistic fMRI paradigm utilizing eye-tracking to create real-time, contingent social responses to participant-initiated JA. During congruent responses to JA bids, we observed increased activation in the right amygdala, the right fusiform gyrus, anterior and dorsal anterior cingulate cortices, striatum, ventral tegmental area, and posterior parietal cortices. Incongruent responses to JA bids elicited increased activity localized to the right temporoparietal junction (TPJ) and bilateral cerebellum. No differences in eye-gaze patterns were observed during congruent or incongruent trials. Our results highlight the importance of utilizing interactive fMRI paradigms in social neuroscience and the impact of congruency in recruiting integrated social, reward, and attention circuits for processing JA.