Furthermore, no significant relationship (correlation coefficient: r < 0.3) was observed between beta 1 receptor occupancies of the drugs and the number of dreams. On the other hand, good relationships (r > 0.95) were observed between central and peripheral beta 2 or central 5-HT receptor occupancies and the number of dreams. These findings suggest that beta 2 and/or 5-HT receptor occupancy is superior to beta 1 receptor occupancy as an index for the sleep disorders.
This suggests that a beta 2 agonist may be appropriate for sleep.
Note: they appear to be talking about the number of dreams recalled (due to awakenings) rather than the actual number of dreams.