A self-contemplating shadow,
In enormous labours occupied
Full self-awareness or "I-ness". Sentience reflected upon itself in detail.
The concept of autosentience emerged during the early development of artificial intelligence. A biological mind possesses conscious awareness of only a small fraction of the complexity of its internal mental processes, and possesses the capability to deliberately revise even less than this.
While biological evolution did not produce full autosentience it is possible in theory for an artificial mind to fully view its internal mental processes and make much more extensive and detailed revisions to its own programming. This ability would allow the AI to redesign itself radically, leading to the capacity for self-improvement and noetic growth. However in practice the complexity of an artificial mind is so great that full autosentience requires very large amounts of processing power, and conscious awareness by an entity of its own mental processing must include awareness of that awareness itself, leading to a recursive state.
This internal awareness — known as 'ultraconsciousness' in the Interplanetary Age — could lead to Depersonalization Disorder and Hyperautism, particularly in cyborgs using internal feedback algorithms (more details here). Modosophont aioids also found that full autosentience was very difficult to achieve without excessive use of processing power, so approximation and self-modelling were used instead.
Something approaching full autosentience and autoscience is believed to be a feature of most, if not all transapient minds, but due to the large scale of the processing substrates used by archailects, self-knowledge must be in fact limited by the effect of light-speed delays between processing nodes.