Hypnagogia – memory loop or live sensory input?

My acquaintance (MA) who gets MWR experiences (microsleep with REM - see here) was recently sitting holding a cup. MA noticed the cup begin to sag, though still firmly held, as if MA had fallen asleep. MA lifted it up again. Despite the ‘sag’ MA felt fully conscious and could see the cup perfectly throughout this incident.

However, an observer said MA looked asleep with eyes apparently closed during this incident. It looks like an example of hypnagogia that MA gets in addition to MWRs. It is interesting because it may answers one of the questions I’ve had about such hypnogogic episodes for a long time.

During such incidents the witness sees the real scene around them throughout. Sometimes a fictitious element, like a ghostly figure or alien, might be added to what the witness sees. This appears to be a dream element added by the witness’s own brain. But is the real scene coming from continuous sensory input or from a sort of memory loop in the witness’s brain? If MA’s eyes were indeed closed it suggests the latter.

If this is the case it means that anything that changes in the real physical scene during the incident will not be noticed by the witness. Should the witness then wake up any such changes might occur suddenly, like an object apparently appearing or disappearing instantly. That would, obviously, be viewed as paranormal by the witness. It would be nice to repeat this accidental experiment to confirm this idea. It’s even possible that hypnagogia uses live sensory input sometimes and a memory loop on other occasions.