Bizarreness effect in dream recall.

This study aimed to ascertain a) whether morning reports of dream experience more frequently reproduce bizarre contents of night reports than nonbizarre ones and b) whether this effect depends on the rarity of bizarre contents in the dream or on their richer encoding in memory. Ten subjects were awakened in rapid eye movement (REM) sleep three times per night for 4 nonconsecutive nights and asked to report their previous dream experiences. In the morning they were asked to re-report those dreams. Two separate pairs of judges scored the reports: the former identified the parts in each report with bizarre events, characters or feelings and the latter parsed each report into content units using transformational grammar criteria. By combining the data of the two analyses, content units were classified as bizarre or nonbizarre and, according to whether present in both the night and corresponding morning reports, as semantically equivalent or nonequivalent. The proportion of bizarre contents common to night and morning reports was about twice that of nonbizarre contents and was positively correlated to the quantity of bizarre contents present in the night report. These findings support the view that bizarreness enhances recall of dream contents and that this memory advantage is determined by a richer encoding at the moment of dream generation. Such a view would seem to explain why dreams in everyday life, which are typically remembered after a rather long interval, appear more markedly bizarre than those recalled in the sleep laboratory.