Joint attention is a keystone in social cognitive development and a skill acquired early in life. It is the triadic coordination of attention between two people and an object or event in which they are commonly interested. Language development follows in its tracks and is dependent on this early acquired skill. Its deviation from typical development is considered one of the earliest signs of autism. Consequently, its remediation has gained intensive focus in therapy. In this review, the development of joint attention skill in initiating (IJA) and responding (RJA), and its atypical development in autism and related spectrum disorders would be discussed. This would include existing problems in pointing, sharing attention with the participant, and facial recognition; and the rationale behind these deviations as covert attention. Related fMRI findings would also be reviewed, outlining the integration between the posterior involuntary parietal and superior temporal cortices (RJA) and the anterior volitional prefrontal and orbital frontal areas (IJA) in typical development, and the long-distance underconnectivity and local overconnectivity in autism. Several cortical regions are implicated in autism, revealing the heterogeneity of the findings, but general conclusions could be drawn.