(1) Substituting (as Solms does) forebrain for brainstem in the search for a dream controller'' is counterproductive, since a distributed system need have no single controller. (2) Evidence against episodic memory consolidation does not show that REM sleep has no role in other types of memory, contra Vertes & Eastman. (3) A generalization of Revonsuo's threat simulation" model in reverse is more plausible and is empirically testable. [If you have access to it, it is easiest to follow the
full special issue or the
book version in its entirety.]