The published literature suggests that systematic self-observation may
be a suitable method for clarifying the nature and correlates of hypn
agogic imagery and thus a useful adjunct to psychophysiological and co
gnitive studies of sleep onset. The potential applicability of one rec
ently proposed self-observation procedure (Nielsen, 1992) to such stud
ies is demonstrated in the present work. The procedure permits numerou
s hypnagogic images to be collected during spontaneous drowsy periods
occurring during the day. The observer sits in an upright, head-unsupp
orted position, fixes an observational intent, and pays attention to i
nternal events; images are observed, transcribed and then assessed for
their likely memory sources. The procedure has been pilot-tested by t
he author in four exploratory studies comprising over 250 hypnagogic i
mages. Neuromuscular events accompanying these images (e.g., head nods
, leg jerks) and EEG correlates of the images are described. Certain d
istinctions among imagery types are suggested, e.g., fleeting vs. full
y-formed, images with self movement vs. images with non-self movement.
Silberer's conclusions regarding the 'autosymbolic' function of hypna
gogic images are supported and extended by the results. Four types of
memory element (immediate, short-, medium-, and long-term) appear to h
ave contributed causally to the formation of these hypnagogic images a
nd are illustrated. To demonstrate how the self-observational method m
ay be used to model the formation of hypnagogic imagery from such memo
ry sources, a single sample image and its multiple memory sources are
described and analyzed in detail.