The default network (DN) is associated with a variety of introspective cognitive processes. Recent developments support that the ‘canonical’ DN comprises at least two parallel distributed networks, DN-A and DN-B, which are closely interdigitated in the individual brain. However, the two networks can be clearly distinguished by their connections with different medial temporal lobe (MTL) structures, which help explain each network’s involvement in different aspects of introspection: DN-A plays a role in mental scene construction and is prominently connected to the anterior hippocampus, parahippocampal cortex, and medial entorhinal cortex. Meanwhile, DN-B plays a role in social cognition and is prominently connected to the amygdala and more lateral entorhinal cortex. These different MTL connections help explain the heterogeneity of functions within the canonical DN, and putatively may shape the fractionation of these distinct networks during brain development, such that different cortical networks end up handling interactions with different MTL regions.