Dream reports collected from 14 participants after planned awakenings during REM sleep show that story-like organization depends on both time of night and REM duration. Reports from the 4th sleep cycle were longer than those from the 2nd, and reports after 10 minutes of REM were nearly significantly longer than after 5 minutes. The number of dream-stories per report did not change, but the sequential and hierarchical complexity of individual dream-stories increased only after 10 minutes of REM in the 4th cycle. These results indicate that building a long, coherent dream narrative requires sufficient time and cognitive resources.
Dream reports occur after awakenings from all sleep stages, not just REM sleep. In a nap study with 10 healthy young adults, dream recall length, measured by total word count, was greater after REM sleep than after brief NREM sleep periods (10 seconds and 1 minute), but recall length did not differ among NREM awakenings of varying durations. Delta EEG activity, an indicator of sleep depth, was lower after REM than after longer NREM periods, while Beta activity did not vary. The consistent recall length across NREM conditions suggests that arousal level alone does not determine dream report length; memory processing or limitations of EEG as an arousal measure may explain the longer REM reports.