Humans can recollect past events in details (recollection) and/or know that an object, person, or place has been encountered before (familiarity). During the last two decades, there has been intense debate about how recollection and familiarity are organized in the brain. Here, we propose an integrative memory model which describes the distributed and interactive neurocognitive architecture of representations and operations underlying recollection and familiarity. In this architecture, the subjective experience of recollection and familiarity arises from the interaction between core systems (storing particular kinds of representations shaped by specific computational mechanisms) and an attribution system. By integrating principles from current theoretical views about memory functioning, we provide a testable framework to refine the prediction of deficient versus preserved mechanisms in memory-impaired populations. The case of Alzheimer's disease (AD) is considered as an example because it entails progressive lesions starting with limited damage to core systems before invading step-by-step most parts of the model-related network. We suggest a chronological scheme of cognitive impairments along the course of AD, where the inaugurating deficit would relate early neurodegeneration of the perirhinal/anterolateral entorhinal cortex to impaired familiarity for items that need to be discriminated as viewpoint-invariant conjunctive entities. The integrative memory model can guide future neuropsychological and neuroimaging studies aiming to understand how such a network allows humans to remember past events, to project into the future, and possibly also to share experiences.