Based on the findings of our previously published positron emission to
mography study, we proposed that recorded eye movements during REM sle
ep are visually targeted saccades. In the present study, we examined t
he correlation between the number of eye movements in REM sleep (EM) a
nd visual imagery in dreaming (V) and provided further support for our
proposal. All the observations (N = 11) were made with one individual
to eliminate interindividual variation and were made during the secon
d REM sleep period to control for a time-of-night effect. V, with or w
ithout dream report length partialled out, was strongly associated wit
h EM only in the 1-min interval immediately preceding awakening. The t
ime course of the association suggests that the strong EM-V associatio
n reflects a phasic, localized activation of the eye-movement-control
system in association with REM sleep eye movements.